Mounting Panels And Laboratory Modules For West Point
Document: 102737385 Page 208 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
The following records are a very incomplete listing selected from the available correspondence, letters, memoranda and reports published by the Computer History Museum. Refer to their Collections and contents for the complete records. The references below may only contain the title or a brief summary of a Memorandum's contents. A subset will contain an extract of a memorandum, and rarely, some will have nearly most of a record. Do not rely on any of the information below, always refer to the source material published by the Computer History Museum. The listings below are only a fraction of the material published by the Computer History Museum. Any copyright of the Computer History Museum remains with the Computer Museum.
DEC records are not organised as a single stream organised into a simple timeline but consists of multiple batches of records in multiple, ovelapping, forward or reversed timelines. There can be multiple copies of a record in different batches. A single batch can appear in different subsets of the DEC Collection. The following was selected to trace the history and produce a timeline for the PDP-1. Other records are included purely out of curiosity, or a whim. You have been warned.
| 1959 |
| 1960 |
| 1961 |
| 1962 |
| 1964 |
| 1965 |
| 1966 |
| 1967 |
| 1968 |
| 1969 |
| 1970 |
Document: 102737385 Page 208 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 209 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 186 of 190 Published: By: Al Blumenthal - Digital Equipment Corporation
Document: 102737385 Page 189 of 190 Published: By: Henry Crouse - Digital Equipment Corporation
Document: 102737385 Page 190 of 190 Published: By: S. Mikulski - Digital Equipment Corporation
Document: 102750017-05-0006-src.pdf, Page 196 of 217
Source: DEC BiWeekly Report, Part 6 of 8
Published:
Published By: Digital Equipment Corporation
Curve drawing display equipment consists of several projects at present. The first is an incremental scope - that is, a Type 30 with counting buffers. In order to get the speed up to l MC with the LO speed line and avoid a jam buffer "DC" counting logie has keen developed that doesn't use a ripple carry. The diode boards are complete, the block schematics are complete, but the detailed layout awaits more help, and the diode drivers are still in the works. The second item is a pulse train suitable for generating lines and circles to drive the incremental scope. (The computers need instructions to deal with the scope directly as well as with the generator). The generator is in the block diagram stage.
Most of my time the last two weeks has been spent on development of a new high A-D Converter for N.S.A. Their original order for a 32 channel Analog-Digital Linkage (PDP-1 - PACE) calls for using the Raytheon AD 10-A Analog to Digital Converter, a Raytheon Multiplexer and 32 DEC D to A converters. The Raytheon AD 10-A was originally speced as a 10 bit, a 2 µs unit, with an overall accuracy of .097% ±½ LSB. It eppears now, however, that this unit is really a 4 µs device with an accuracy (on the lab model) of about .1% + 5 counts and -1 count plus a random error which they can't tie down.
In view of the situation, we plan to try to develop a 5 µs, 10-bit unit with appropriate accuracy. Two of these units operated in parallel would provide the necessary speed. I have begun work on a prototype and so far no major difficulties have come up (many minor ones, of course). The present schedule calls for completion and evaluation of this unit before the end of this month. At that time we will present the latest data on both our unit and Raytheon's unit to N.S.A.
At the present time we are also checking out the Raytheon Multiplexer which arrived this week.
The Adams machine which was supposed to go out on Thursday, January 3rd, was held up at the last minute because the building it was to have been housed in was not ready in time. We expect another call from Adams around the 7th or 8th of this month to confirm a new installation date for this machine
The JPL machine has come back from its tempcraxy home at Raytheon and is being prepared to be shipped to JPL sometime around the 15th of this month. The things remaining to be done on this machine are the installation and checkout of the Data Channel and the final checkout of the Mag Tape System which includes a Mag Tape 52 and three Tape Transports.
Gorden Bell's teletype modules have been patched into a DEC 1906 mounting panel. The incoming line unit has bean completely debuged and turned over to Dick Best for module packaging. The outgoing line unit is near completion and. will be sent to engineering shortly.
A first run-through on cost estimates fcr the new memory system indicates that adding 4K fielde to the basic 4K system in PDP-1 and PDP-4 will cost a third of the present cost of a Type 112 module. The stacks themselves make up the bulk of this figure. The cost of the memory extension control goes down in about the same proportion.
Document: 102737385 Page 206 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 207 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 181 of 190 Published: By: Digital Equipment Corporation
Document: 102737385 Page 174 of 190 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 176 of 190 Published: By: Dave Denniston - Digital Equipment Corporation
Document: 102737385 Page 177 of 190 Published: By: Richard F. Mills - Digital Equipment Corporation
Document: 102737385 Page 184 of 190 Published: By: J. Smith - Digital Equipment Corporation
Document: 102737385 Page 164 of 190 Published: By: Ted Johnson - Digital Equipment Corporation
Document: 102737385 Page 165 of 190 Published: By: Richard F. Mills - Digital Equipment Corporation
Document: 102737385 Page 205 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 203 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 204 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
When we visited ITT last week we promised thet on our machines which we deliver to them from now on would include outiet panels on the computer for oscilloscopes for use during servicing. We hove always been reluctant to do this because the circult breakers in the computer cannot take the additional load of an oscilloscope. They however asked that they be included in thelr wire of separate power to them. If they do this, it is relatively easy for us to add the outlets.
We way never sell more computers te ITT but I think we should lookk into this problem so that we can do it later on when desirable. One 5 1/4 inch panel fitting in a presently blank slot would probably be sufficient. Ail scopes have cables long enough to run the full length of the computer if they can be plugged in somewhwere near the middle. I also believe that there is a blank panel here and there thet we can use. This pene! could have two er three outlets plus a filter and a Jones strip to which they can tie power.
We offered te do this for Foxboro but at additional cost. If we work out a nice solution, I think think we should offer it to Foxboro also at no additional price. Maybe we should include this when we tell them we are thinking of changing to three bays.
Document: 102737385 Page 151 of 190 Published: By: Harlan Anderson - Digital Equipment Corporation
Document: 102737385 Page 153 of 190 Published: By: Harlan Anderson - Winston Hindle - Digital Equipment Corporation
Document: 102737385 Page 169 of 190 Published: By: Digital Equipment Corporation
Document: 102737385 Page 171 of 190 Published: By: R. J. Beckman - Digital Equipment Corporation
Document: 102737385 Page 185 of 190 Published: By: Nick Mazzarese - Digital Equipment Corporation
Document: 102737385 Page 150 of 190 Published: By: ?.?.?. - Digital Equipment Corporation
Document: 102737385 Page 199 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 200 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 201 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
I think it would be a good Idea if you would perlodically check with ITT and Ben Gurley to see how likely the two large jobs are which they have asked us to bid on. Ed Fredkin visited California last week and should have a good feeling for that project now. The duplex system which ITT asked us to bid on might just have been an exercise, or they may have something definitely in mind. In either case, I think they should tell us now because we will be planning our production accordingly.
Document: 102737385 Page 202 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 145 of 190 Published: By: George Rice - Digital Equipment Corporation
Document: 102737385 Page 143 of 190 Published: By: Stan Olsen - Digital Equipment Corporation
Document: 102750017-05-0006-src.pdf, Page 179 of 217
Source: DEC BiWeekly Report, Part 6 of 8
Published:
Published By: Digital Equipment Corporation
In his PDP-4 automatic module tester, Ken Wakeen is planning to use an A to D converter made from our newest modules. The converter was finished last week and we had an opportunity to test it with our new Kin-Tel power supply. The results were very encouraging. In the past we have avoided pretty much the question of the absolute accuracy of our converters and stuck more to terms that were easier to measure-like the number of bits which can be obtained with monotonicity. I think that we will always be able to manufacture a finished converter with greater accuracy than that which we want to specify to a module customer who is building his cwn converter. The reason for this is that the errors produced by specific modules are somewhat hard to define, and even more difficult to test thoroughly.
If we were to make an educated guess as to how accurate a converter would be made with our modules, we might add up figures something Like the following
Old Modules New Modules 4677A .1 % 4678 : .04 % 1547 common mode .05 1572 common mode .03 " sensitivity .1 " sensitivity .02 " current .02 " current .02 1562 .15 1565 .005 1564 misadjustment for 20 bits ≈ .01 1564 misadjustment for 20 bits ≈ .01 misc. .01 misc. .44% total error .06 to .16 total error .135% safety factor .06 to .16 safety factor .035 - .085 Estimated max. error .5 to .6% Estimated max. error .15 to .2% These figures are quite conservative, principally because they are simply educated guesses.
The tests that we made on Ken Wakeen's converter were for DC absolute accuracy. We set the converter at its normal operating speed but initiated conversion from external clock as a somewhat lower rate. The input of the converter was from the new Kin-Tel power supply which is accurate to within &pludmn;.01%. Results were read on indicator lights driven by the converter register. We measured the input voltage which caused the converter to oscillate between two successive states. This was done by eyeballing that point where the lights seem to be flickering between the two states with equal intensity. When the speed of the converter itself was fairly slow, these points were extremely well defined, i.e. 0.1 or 0.2 millivolts. As the speed of the converter was increased up to the published speed, these points became more spread out and would sometimes depend on the direction from which you approached them. For each test, we measured 19 points corresponding to the converter switching between such numbers as 00011 and 00100 or the points 11011 and 11100. The results of the tests were as follows
Accuracy (all figures ±.01% due to Kin-Tel) Speed 3.7 µs/bit* 7.5 µs/bit 25 µs/bit* temperature +29°C ±.09% ±.038% ±.026% +61°C ±.1% +18°C ±.06% *theoretical speed These results, of course, are not the whole picture but they are encouraging, particularly the results of high and low temperatures. The biggest factorr in the accuracy of the converter is the comparator. The comparator, which we used, had quite a low temperature coefficient. Therefore, we should not expect this type of temperature coefficient in all finished units. However, the decreasing aceuracy at the high temperature can be explained almost completely in terms of the compparator which means that the other components do not exhibit a significant temperature coefficient (particularly the low level amplifiers).
Even though we now quoted a temperature coefficicnt on our ladders, we really shouldn't expect them to be anything like this because the temperature coefficients are in themselves a function of temperature; in this relatively limited range, the resistors should change extremely little. However, resistor manufacturers do not like to specify this type of thing. The only way that we can guarantee it is with a direct measurement.
The improved accuracy at low temperature is probably due to faster switching and hence, a similar increase in accuracy would not be expected at lower operating frequencies.
The low speed accuracy figures of .038% and .026% are prohably better than can be expected in the general case. This is due to the fact that the gain bandwidth product cf the input stage on the comparator ia pretty nearly a constant. The particular comparator used in this test had high-gain in the input stage, and was relatively slow. A comparator which had low-gain speeds but high bandwidth would probably operate the same at the lcwer speeds as at the higher speeds. However, the very fact that we had a converter that tested out this good is in itself pretty impressive when you consider that the best units sold on the market right now have absolute accuracy of .03% plus a temperature cefficient. (Of course, these units are expected to keep this accuracy over an extensive length of time, something that we have not tested for yet. They also have a considerably large number of bits, which is a separate question.)
For the time being, I think the number that is the most meaningful is the .09% accuracy number. This is the area in which we must be thinking for the time being.
One thing that doas not really show up in the above tables is the distribution of the error. The error was almost linear with respect to the voltage. At very small inputs, error was just slightly negative (i.e. switching occurred when the voltage was slightly less than it should be}. The error increased to a maximum positive value somewhere between full scale, and half of full scale. Once it reached the maximum, it remained fairly constant until full scale. This is considerably better than a converter where the error on alternate points would be + and - full value. The error figures could be made to look somewhat better by adjusting in the converters so that the error was centered equally positive and negative, but in the figures above the converter is adjusted so that the error is minimized at small voltages. I believe this is what is generally desired.
It was particularly uscful to be able to test this particular converter since it will be used here in the plant and we can check repeatedly to get more information about the stability of the unit. I also want to measure the repeatability on the computer by putting in a specific voltage and having the computer print the high and low values which occur during 100 samples or so. Another test which would be quite important is how the converter reacts to a square wave input, and to a variation in a conversion repetition frequency. At first blush, it appears that the accuracy should not be affected by these factors; however, past experience shows that they sometimes are and certainly must be checked for before we start quoting any fixed figures on absolute accuracy.
New Business We are working on a bid for Atomic Energy of Canada which is a control system for Atomic Energy plant. It involves the use of the PDP-12 or PDP-4 Computer, our standard Type 30 Display, our serial drum and Adage analog-to-digital converter and a special alarm signal monitor system. This is due on February 15.
Harvard University has finally placed a firm purchase order for the Spark Chamber Reader. Work is progressing fine on the necessary schematics and mechanicsl layout. It looks as If we now have a marker for the SW1250's that could not be matched for more critical use.
Tuesday, January 15, the JPL machine was shipped by air freight to California and Bob Beckman and myself went to the airport with the machine in order to supervise the loading and gain information and experience so that we might properly evaluate this method of transportation. After observing the rough handling this unit received while we were standing there, we both wondered what kind of treatment it would have received if we were not around. Also in order to ship anything from this factory to a customer by air, the goods must be handled (loaded - unloaded) a minimum of six times. It is quite possible that this figure could go to eight or ten depending upon the number of times the airlines would have to shufle them around in their warehouse while making up loads for various aircraft. It is my opinion that shipping by air is not the best possible method of shipment
Document: 102737385 Page 139 of 190 Published: By: J. L. Atwood - Digital Equipment Corporation
Document: 102737385 Page 197 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 198 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 140 of 190 Published: By: Stan Olsen - Digital Equipment Corporation
Document: 102737385 Page 137 of 190 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 137 of 190 Published: By: Gordon Bell - Digital Equipment Corporation
Date January 22, 1963
Subject Progress Report PDP-1 and PDP-4 Construction (Nov. and Dec.)
From J. Smith
PDP-1
| System No. | Schedule Date | Date Complete | |
|---|---|---|---|
| PDP-1-34 | 9000-5129 | 11/2/62 | 11/2/62 |
| PDP-1-35 | 9000-5863 | 11/21/62 | 11/21/62 |
| PDP-1-36 | 9000-5864 | 12/7/62 | 12/10/62 |
| PDP-1-37 | 9000-5865 | 12/21/62 | 12/21/62 |
| PDP-1-38 | 9000-5150 | 1/4/63 | 1/4/63 |
PDP-4
| System No. | Schedule Date | Date Complete | |
|---|---|---|---|
| PDP-4-7 | 8000-7437 | 12/14/62 | 12/17/62 |
| PDP-4-8 | 8000-7718 | 12/18/62 | 12/21/62 |
| PDP-4-9 | 8000-7719 | 12/31/62 | 12/31/62 |
| (Module Test) | |||
| PDP-1-38 | 9000-5150 | 1/11/63 | 1/15/63 |
Document: 102737385 Page 146 of 190 Published: By: Dit Morse - Digital Equipment Corporation
Document: 102737385 Page 131 of 190 Published: By: Harlan Anderson - Digital Equipment Corporation
Document: 102737385 Page 128 of 190 Published: By: Gordon Bell - Digital Equipment Corporation
Date January 28, 1963
Subject Tape Controls 57 and 54
From Gordon Bell
57 System Description
According to my schedule of November 30 from Steve Lambert for the Tape Control 57, it is four weeks overdue. Fortunately the purchase order from JPL has nort arrived, and as such their machine probably won't be late. There may be some correlation between JPL not receiving specifications on the 57 and our not receiving their purchase order.
The most significant error made so far with the 57 is that a write-up has not been presented for outside comment resulting in a tape conbtrol design that may be weak. It may not be too late for such a review.
A description of the control (programming and system viewpoint) must be sent to JPL by Friday, February 1, 1963. The preliminary draft should be submitted for my approval on Wednesday, January 30, 1963.
A description of the control in a form suitable for publication should be available within the next month for marketing use.
57 Maintenance Description
Steps should be initiated now to provide a maintenance manual with the first delivery. If the project leader can not make these arrangements, we should meet immediately to remedy the situation.
57 Control
This control was originated by me in August, 1962 and turned over to Steve Lambert. The description of the control is not available in a semi-professional looking document
The programs which operate the control though partially described are not ready for distribution. These routines will be distributed through the Program Library, and as such can have a second or third writeup. Here, the routines and their listings are the most important thing.
A control is to be delivered to Foxboro on Wednesday, January 30. Both the control and the program description should be ready by then.
Future Development
The enclosed figure outlines the procedure to be carried out by project engineers. While there are many new and interesting projects upon which to embark, this must not be allowed to confuse the development of any single project.
Following an outline for each project accomplishes:
The project engineer must be aware of the considerations shown in the following figure. There are too many new and interesting developments and projects upon which we are to embark to allow sloppy planning to shackle us. This not only shakles the project engineer, but his conteporaries who use his sub-system design in their system. An engineer should be prepared to assist customer service, but in most cases, continued service calls by a project engineer merely indicates incompetent or understaffed field service, poor design or documentation, or improper training of technicians or incompetent technicians. All those situations can be avoided by some planning. Hardwork in the design area is necessary, and is certainly appreciated, but project planning is necessary too.
Document: 102737385 Page 194 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 195 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 196 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 196 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 133 of 190 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 134 of 190 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 136 of 190 Published: By: Win Hindle - Digital Equipment Corporation
Document: 102737385 Page 135 of 190 Published: By: Frank Kalwell - Digital Equipment Corporation
Document: 102737385 Page 120 of 190 Published: By: George O'Dea - Digital Equipment Corporation
Document: 102737385 Page 125 of 190 Published: By: Harlan Anderson - Digital Equipment Corporation
Document: 102737385 Page 127 of 190 Published: By: Personnel Committee - Digital Equipment Corporation
Document: 102737385 Page 191 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 192 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 193 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 190 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 119 of 190 Published: By: A Blumenthal - Digital Equipment Corporation
Document: 102737385 Page 189 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 111 of 190 Published: By: Gordon Bell - Digital Equipment Corporation
Date February 4, 1963
Subject Additional Personnel to Handle PDP-1, 4 Changes
From Gordon Bell
I would like to borrow Don Smith from the Sales Department for a six month period. The Projects I presently contemplate are:
| 1. | Design of Data Channel for Type 59 Magnetic Tape Control. |
| 2. | Modify PDP-1 Central Processor for BBN job and design Memory Switching. |
| 3. | Design Teletype Line Scanning Processor for BBN System. |
| 4. | Assist on BBN System. |
| 5. | Assist on PDP-1A design changes: |
| a) Reversal of polarity of AC outputs. b) Modify shift/rotate instruction. c) Install exchange AC/IO instruction. d) Install normalize instruction. e) Install exchange memory instruction. f) Install character processing instruction g) Install index instruction. |
Document: 102737385 Page 118 of 190 Published: By: Robert Savell - Digital Equipment Corporation
Document: 102737385 Page 104 of 190 Published: By: Robert Maxcy - Digital Equipment Corporation
Document: 102737385 Page 110 of 190 Published: By: Roland Boisvert - Digital Equipment Corporation
Document: 102737385 Page 103 of 190 Published: By: E. T. Johnson - Digital Equipment Corporation
Document: 102737385 Page 102 of 190 Published: By: Harlan Anderson - Digital Equipment Corporation
Document: 102737385 Page 112 of 190 Published: By: Robert Savell - Digital Equipment Corporation
The color T.V. in the Precision Oscilloscope which are still on the floor, were always considered Engineering Projects by most of us but apparently we have been keeping an inventory. It now looks like we will sell that Precision Scope, but the color T.V. should be taken out of inventory and put into an Engineering Development Project.
Document: 102737385 Page 188 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 101 of 190 Published: By: R. Mills - Digital Equipment Corporation
Document: 102737385 Page 186 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
I am still thinking seriously of developing our own tape unit. One thing I would like to have you do, or have one of your people do, is to collect very approximate statistics as to where the maintenance has been on the Potter Tape Unit.
We would like to know how often their brushes and commutators on the DC motors have to be changed, and how often the brakes have to be adjusted on the reel motor and what other experience has been on the mechanical parts. Some of these same mechanisms we would use on our next tape unit if they were free of failure.
Document: 102737385 Page 187 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 99 of 190 Published: By: Jack Atwood - Digital Equipment Corporation
Document: 102737385 Page 96 of 190 Published: By: Bob Savell>/span> - Digital Equipment Corporation
Document: 102737385 Page 97 of 190 Published: By: Jack Atwood - Digital Equipment Corporation
Document: 102737385 Page 184 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 185 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 93 of 190 Published: By: H. O. Painter - Digital Equipment Corporation
Document: 102737385 Page 94 of 190 Published: By: Jack Atwood - Digital Equipment Corporation
Document: 102737385 Page 84 of 190 Published: By: Bob Hughes - Digital Equipment Corporation
Document: 102737385 Page 86 of 190 Published: By: Bob Lassen - Digital Equipment Corporation
Document: 102737385 Page 88 of 190 Published: By: A. Blumenthal - Digital Equipment Corporation
Document: 102737385 Page 89 of 190 Published: By: George O'Dea - Dick Mill - Digital Equipment Corporation
Document: 102737385 Page 75 of 190 Published: By: George O'Dea - Digital Equipment Corporation
Document: 102737385 Page 78 of 190 Published: By: Jack Atwood - Digital Equipment Corporation
Document: 102737385 Page 81 of 190 Published: By: N. Mazzarese - Digital Equipment Corporation
Date 3/11/63
Subject Computer Sales Forecast
From N. Mazzarese
| PDP-1 | Quantity | Value | Probability | When |
|---|---|---|---|---|
| CRC | 1 | 97K | 100% | 0-3 |
| Raytheon | 1 | 200K | 100% | 0-3 |
| BBN | 1 | 600,000 | 85% | 0-3 |
| Lincoln Labs. | 1 | 200K | 50% | 0-3 |
| Princeton | 1 | 250K | 100% | 0-3 |
| LRL | 1 | 350K | 50% | 0-3 |
| Stanford | 1 | 280K | 75% | 0-3 |
| ADX | Quantity | Value | Probability | When | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| S.T.C. | 1 | 300K | 50% | 0-3 | |||||||
| PDP-4 | Quantity | Value | Probability | When |
|---|---|---|---|---|
| Foxboro (Fitchburg) | 1 | 160K | 85% | 0-3 |
| Foxboro (Westinghouse) | 1 | 120K | 50% | 0-3 |
| Foxboro | 2 | 120K | 50% | 3-6 |
| JPL | 3 | 250K | 90% | 0-3 |
| AECL | 1 | 200K | 75% | 0-3 |
| Worcester Foundation | 1 | 80K | 50% | 3-6 |
| Leeds & Northrup | 1 | 80K | 50% | 0-3 |
Less than 50% 0-6 Months
| Type of Machine | ||
|---|---|---|
| Univ. of Rochester | 200K | PDP-1 |
| Systems Development Corp. (Info. International) | 300K | ADX |
| Sylvania | Rental | PDP-1 |
| JPL | 120K | PDP-1 |
| Beckman Instruments | 120K | PDP-1 |
| Litton Systems | 80K | PDP-4 |
| Michigan Uni. | 120K | PDP-1 |
| Maryland Univ. | 120K | PDP-1 |
| Raytdeon Co, | 120K | PDP-1 |
Document: 102737385 Page 83 of 190 Published: By: Harlan Anderson - Digital Equipment Corporation
Document: 102737385 Page 180 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Harlan Anderson hed a visit from Minnesota Mining and Manufecturing Company several months ago. They tried to sell magnetic tape to us but also had one of their research people describe some new tape they working on. This is exceedingly thin metallic film on a plastic base, and they sald they were getting 6,000 lines to the inch. We should visit them soon because this type tape on a LINC type transport would give exceedingly high data rate and a lot of information on our small reels. We may have to order serial recording but this would be very reasonable.
Document: 102737385 Page 181 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Andy was out visiting CRC and saw a demonstration of our color display there. He thought this was very impressive and it seems obvious that we ought to exploit this. Many magazines are very hard up for interesting colored pictures for their cover. I suggest that we encourage Bob Savell to finish this display up and then get some interesting colored photographs and then try to give these away to magazines like Electronics te use on their cover.
Document: 102737385 Page 182 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 183 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 178 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 179 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Several months ago I asked Al Blumenthal to prepare a short article on the techniques used for testing computers. He started this but I never followed through on it. We do testing techniques which I think are quite unique and very useful. What also helps to sell our computers is to let people know how thoroughly we test the units. I am thinking particularly of the techniques of putting things in a tent to raise the temperature.
I think it would be a good idea to outline our techniques and to see what kind of an article we could generate out of it, or we might concentrate on the tent idea and get a short section in the production techniques of Electronics magazine.
Document: 102737385 Page 72 of 190 Published: By: George O'Dea - Digital Equipment Corporation
Document: 102737385 Page 175 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 176 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 177 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 70 of 190 Published: By: Dit Morse - Digital Equipment Corporation
Document: 102737385 Page 174 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Devid Caldwell called on Monday, February 18 to say that he was having visitors from University of Michigan on Friday and that he would like to bring them out here to show our Precision Display and PDP-1 Computer. They are interested in reading spark chamber photographs and Professor Caldwell has told them that curs is obviously the only machine to get. The man's name will be Professor Harry Jones and he will have two others with him. I am not sure whether David Caldwell will be out here or not.
You might call the day befere to see if we can help in any way in bringing them out.
Document: 102737385 Page 170 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
We recelved a quote from Amphenol on a 36 pin connector for use In large system plug-in units but this will not work out well because it has to be thicker and therefore will not fit In our standard construction. Loren Prentice is now making a model of a double width plug-in unit which will have two 22 pin connectors on it which will make a total of 44, This looks like a reasonable approach to a large plug-in unit.
Gordon Bell suggests that we do all our register transfers through one common register. This is the way the MTC Computer worked originally. This would cut down the number of gates and they might end up using the very high speed transistor gates.
I asked Bob Savell to consider repackaging the reader, punch and typewriter control panels to make them less expensive. We might put much of it on a very small number of large plug-in units. We might also Include the micro-tape logic in the same place.
I told Bob Savell to start working on the new punch timing control for PDP-1 but to plan to have it in the new computer.
Dit Morse feels that the teletype typewriter is a satisfactory typewriter for computer use. He of course would like a more extensive character set but a typewriter that works has a very definite advantage. I can't see that we'll have time to evaluate any other typewriter in time.
Loren Prentice has been working on a new design for the PDP-1 and PDP-4 console fronts. I suggested that they drop all work on that and work on the console front for the new computer. This one should include space for punch, reader, LINC and control panel.
Some people like the idea of having an extra register to store the contents of the accumulator when it is not being used. This would allow the accumulator to be used for index adding and other things. The extra register could then be used as a carry register which would allow very fast multiply. If this carry register {s used as an accumulator buffer, the accumulator might then be used as the register which transfers information between registers. Several people have told me they would like to have a pointer register.
We have to decide whether we want indicators on all flip-flops or not. I have asked Jack Smith to estimate what it would cost to add an indicator.
It is a real chore to change cabinet design. Our present mounting panels hold 25 plug-in units and if we move the marginal checking panel, it will hold 26. It would therefore be convenient to keep the digit length of the machine 26 or less bits long.
Document: 102737385 Page 171 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
A new computer Is long overdue at DEC but we have not been in a position to build one because we have been so long in winding up the details from our present computers. However, now we do have the techniques and the time and the money for a new computer, I think we should go ahead and make one In a reasonably fast time schedule.
The proposal is to do all aspects of the computer design In parallel. This means that at the end of the time schedule whether it is four or six months, the job should be done. Then after a rest of a month or two we could if we wanted to go off and make another computer. Here is a list of the items which should be carried on in parallel:
We have never looked at competition before but I think as a result we have lost out because we don't know the points in which our machines are significantly better than others. I think that we should consider doing this parallel effort sub-contracting a survey out to someone like I.I.I. to compare our machine in detail with others.
Document: 102737385 Page 172 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
We understand that there is someone making equipment identical to ours which we will have to compete with in a bid to Applied Physics Laboratory. One thing we should consider is to have them bid on equipment which might be difficult for others to bulld. We are now laying out one plug-in unit which has eight flip-flops with read-in gates. This will be priced relatively inexpensively and might be difficult for someone else to build. In addition, we are making a six bit counter and a six bit shift register. All of these will only have one side of the flip-flops come out. If this will help, please let me know. Document: 102737385 Page 173 of 209 Published: By: Ken Olsen - Digital Equipment Corporation New Computer Design
We are considering the design of a new computer. There is a question as to whether we want indicators in all flip-flops or not. Will you estimate for me what it costs per indicator to be added in production to a computer
Document: 102737385 Page 69 of 190 Published: By: R. Mills - Digital Equipment Corporation
Document: 102737385 Page 60 of 190 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 67 of 190 Published: By: J. Smith - Digital Equipment Corporation
Document: 102737385 Page 157 of 209 Published: By: Tom Stockland - Digital Equipment Corporation
DATE February 22, 1963
TO Ken Olsen
FROM Tom Stockland
Subject New Computer Design
My apologies for form and content of this mnemo, it is a rushed job. In particular it does not include enough evaluation of the competition nor enough filtering of the ideas presented. While I am on vacation, I will try to sketch out more of the machiune design.
Comitments on delivery dates, prices and so on should be to Ken Olsen and the company and not to customers.
The machine should be specifically designed to do the job as listed below superlatively well rather than in any way "look like the competition" or be an answer for them.<
This machine is to fill a vacuum we believe to exist at the present time in the computer market.
We must make no comprimises in carrying out the ides which are involved in its design. The implication of the above is that, as usual with DEC effort, the ideas shall be limited to those which are eminently easy to do, general, straightforward extensions of the art..... In fact, "today's technology today."-----God.
The source of the ideas presented in this note are indicated in an effort to provide "source data" while I'm gone. If the general ideas are agreed upon, future administration of the project will be vastly improved.
If we turn out machines regularly, we need some more official advanced development - that is answers to specific how-can-we-do-this-job questions. (Coax delays, micro-logic, serial, majority logic circuits, etc.)
It is time the Programmer was giuven real power in sub-routine writing ability so that no modifications of instructions are ordinarily necessary during program relocation.
Multi-programming, time sharing, fast break-in or what-you-will is necessary in the eyes of most users of our equipment and in fact necessary (though they don't know it) to many users who are comtemplating using our equipment.
Data words need to match today's data requirements in accuracy. The analog people are almost entirely concerned with 14 bit accuracy for what they call four significant digit precision.
Large memories are here. Index registers are here.
Some fair expansion of the machine should be planned for at the beginning though we understand that wholesale revisions of the machine are out of order.
The rest of this memo is a list of specifics pertaining to the generalities listed above.
Routine Relocation Power - The ability to operate routines wherever they may be located in memory after a dump from, say, the drum can be provided by the ability to (1) modify each memory reference by a constant while (2) checking the result against specified bounds and trapping to a particular memory location or executuive program if the required location is outside the bounded area. This feature can be achieved reasonably easily during the initial design of a machine by allowing the index adder, or its equivalent, to do the work. Dit says this feature would make programming "ten to a thousand times easir." Ed says that if if you can use the arithmetic element more and memory less, you're way ahead and this feature would leap in that direction. (Dit, Shelly, Kotok, Ed and Ben) This feature is considered by advanced type people to be crucial to the machine design.
Trapping - Trapping means to execute and {sic) instruction located at, for instance, the address indicated on the op code. This trapping would be done on non-used instructions or memory addresses outside of the bounds set by the executive routine in the relocation of power indicated above (Dit, Ben.)
Character Handling Power - The ability, in one form or another, to address characters stored in memory hopefully to deal with character strings in I/O transfers such as is done in the Lisp and Comet Programs. Dit, Ben and Ed are in favor of this, Ivan goes even further and says that bit addressing features are of great power. However, Len disagrees.
On Oolescence - Trapping also allows optional expansion by do-it-now-with-program, after with wires. Also de-bugging and checking power is automatically incorporated. The machine should be built of modular parts of course like different memories and AE's and an extra bit or two should be assigned in the instruction word for future variations not thought of now when you absolutely have to have that bit!
Multi-Processing - Multi-Programming - First and foremost, a fast break - this means primarily no need for many accesses of a clean-up variety to store away stuff in preparation for operations in response to a break request. The most potent feature here seems to be an extra register in the AE to allow either exchanges with the AE for saving purposes, or as an address calculator (Dit) or as a multiply index by, or as an addend register, or as a carry register depending on your exact orientation. The second thing which would help this process out is probably a separate index adder though I believe a machine try should be made to use one adder for everything. Since it is reasonably certain that two groups of wide modules will be used, however, it is probably not unreasonable to suggest the index adder. In the future, that means perhaps with the development of another machine, separate program counters may be in order. For now, core program counters should certainly be enough if they are necessary. To hell with data gather. The idea here is to eliminate control problems from the channel and put them in the program where they belong. Channels should be only high-speed data gathering devices. (Dit) System capability is an okay phrase. (Dit)
List Processing - This is a program technique which has general power which goes well together with our ideas of a processor with general power. It requires index registers and increment and decrement by more than one and, ideally, registers which can be packed with several addresses each (that is, word length equal to two times the address length.) However, | think a cleveruse of the relocation feature or of Dit's multiple indexing (1+ 2+ 4 scheme) will allow the shorter pack base address that this too short word machine will have. (Len) In general, this processing seems to be for the next machine though a small look into the future is probably in order. Similarly, floating point AE's will probably have to wait until the next machine or at the very best, be planned as a different kind of AE attachable to this memory.
Index Registers - These are clearly necessary. Dit feels that three register which could be added together in a micro-program fashion that is, any combination of the three according to MACRO programmed bits in the word, would be of more use than seven registers addressed directly by the same three bits though Kotok disagrees. 1 have no feelings. Whether the three could be added together and in fact the complete design of the index adder might depend crucially on the ability to build a simple circuit which would detect four out of seven to provide carry for carries. If this circuit were easily available | believe that five registers could be added together simultaneously and stored in a fifth and the sketch accompanying this memo shows the powerful use that could be made of this feature.
Addressable Registers - These would be very useful according to Len for much easy processing without complicated instruction and could perhaps be implemented to do the character addressing without using extra bits in the word by allowing certain kinds of character type transfers between registers. The most important addressable registers would perhaps be the in/out registers such as, for example, the scope buffer for use with the light pen -- especially if it were an incremental scope plus generator type. In this case too, the feature would allow sine, cosine and hyperbolic and parabolic function generation with no extra hardware. It would save on the IOT read-in bits but cost some address decoding.
Data Channel - Fast break SI, Data Channel SI, I/O Channel, no, - do it with program. (Dit)
Cute Instructions - Ben feels that load and deposit AC in push down list would be a useful instruction at least to the prospects of a clever turn of mind if not to real users. Instruction (Y+ )AC)) ==> AC is reasonably necessary for multi-dimensional matrices when indexing is not readily available and would implement easier list processing. Ben likes an instruction called execute effective address however, Len doesn't go along with him. Dit makes the comment that we should avoid doing things in little pieces.
Word Length - There are two criterion for word length, one is the data word that will usually be of necessity, and the other one is the number of bits that you need in your instruction. For floating point work, 48 bits seems to be a minimum and for graceful manipulation of the text this also seems like an appropriate word length. 1 do not believe that it is necessary to have precisely a multiple of six though this may be, in some cases, graceful for character processing. Many people would just love to have an extra bit or two to indicate whether this set of characters is to be considered in the list and for other marking purposes, ask Dit for example. I, myself, have run into this problem many times when programming character strings. Len will also agree I think. As far as the packaging limitations go, | agree that it is essential to keep the packaging the same which means no more than 25 units in a rack panel wide; notice that if the address portion is 16 or 17 bits, even, there are 8 bits left over in the mounting panel supporting the "short-word™ AE in which to provide extensions of the full register portion of the AE. Since the floating point people need 48 bits and we can't possibly take this much of a jump in the present machinery, we should either leave them out of consideration or consider two-word data accesses floating point words. To this end, Dit suggests a single bit in the data words to tell whether the word is to be interpreted as floating point or not. This might be an example of the use of a spare bit location in the word for use when a floating point processor might become available. How about word lengths for ordinary users of fixed point type calculations? The competition seems to feel that 24 bits is a reasonable length however, | submit that in many practical cases 14 or so bits is a reasonable length based on my discussions with various analog and hybrid types. This is because 14 bits represents four decimal digits which is the current okay number in that industry, though there as here okay numbers do not necessarily represent the best in engineering philosophy or power. Analog people further state that they need higher data rates than we can get and if we are to capitalize on our parallel computing and data handling power in order to try to overcome some of the taint of the current serial flap, we should consider, I think, 28 bits minimum so as to be able to pack two 14 bit words per register and thus, double our data output rate to digital to analog converters and the like - also to scopes.
Now on to word length as determined by the instructions. Certainly 16 bits represents a reasonable address length to address 65 kilowords of memory. Everyone agrees that this would be a desirable number. 3 bits for index register seems about right and one bit for deferring. 6 bits seems like a minimum for op code, 1 bit for a programmed operator - primarily to catch up to the competition of SDS. I insist on one spare bit and many people who feel character addressing is important would want to use my spare bit plus two others to do the character addressing in those instructions where it matters, and leave it for instruction modifications where it does not matter. This would give a total of 28 or 30 bits depending how you look at it. If you really believe that there should be a multiple of six, then | would recommend a 30 bit machine. However, 28 bits I think is my current recommendation. Incidentally, if you allow 7 bit characters for 128 character set, which is quite a reasonable number. and a "step forward", then this even meets the criterion that 2 bits of character addressing is enough and comes out even. In any case we have room for 33 bits and 17 address bits in the two mounting panels which have double trays so this gives us three extra slots for odds and ends.
I STRONGLY RECOMMEND A 28 OR A 30 BIT WORD.
Concurrent Programming - In this area | am not an expert but Dit seems to feel that the FORTRAN four Tanguage, which looks like the ALGOL language, is the language to use for all programming. I am not aware of the details of the character set required or like that. He wants to do it all in ALGOL. I would have a good discussion with Dit on the subject. All agree that a full-time programmer should be working from the start of the project.
More Work - Very soon, more work should be done in the following areas before the design is completely hard.
Conclusions -
I think a tentative example of the breakdown of parallel tasks in the developments of this machine would be somewhat as follows:
pictures 3...
Document: 102737385 Page 166 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
We are now considering building a new computer. You are one of the few people who have thought about color coding and other standards for the power-line end of the computer. Will you please write a memo outlining what you think the standards should be and anything else that should be Included when we build the computers.
LRL has had some trouble with noise between cabinets In their very large system. You might also consider the possibility of how we tie them all together and to ground. Just what we should do as far as grounding units and where they should be grounded, and to how many grounds, is not an easy problem.
Document: 102737385 Page 167 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
We are now considering building a new computer. You are one of the few people who have thought about color coding and other standards for the power-line end of the computer. Will you please write a memo outlining what you think the standards should be and anything else that should be Included when we build the computers.
LRL has had some trouble with noise between cabinets In their very large system. You might also consider the possibility of how we tie them all together and to ground. Just what we should do as far as grounding units and where they should be grounded, and to how many grounds, is not an easy problem.
Document: 102737385 Page 168 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
I understend thot we fixed the price on the transmitter delay line at $200 and the receiver at $170. I am very surprised at this number because I cannot conceive of us ever coming out ahead if we estimate all of the time that Dick Best apent In laying out the unit, end George Geralds making the layouts and the models, and all the testing that has gone inte lt. When compared to any other way of making a line unit, thie is just too much ef a bargain.
I am afraid thet we get this price by looking af the manufacturing cost and simply multiplying it by some small number. I would suggest in this case that we shouid add another factor. Document: 102737385 Page 169 of 209 Published: By: Ken Olsen - Digital Equipment Corporation New Computer Reliability
Now that we are using computers for Control Application, our need for reliability is much more severe than anything we have faced up to before. I am now particularly concerned about our drums that we are using in Control. U. S. Steel uses a drum for some reason or another and it takes two months to replace it. They are likely to be very unhappy. I suggest that we carefully consider the possibility of buying and keeping on hand a spare drum. This might be a rotating inventory item so that one drum does not stay on the shelf too long.
Document: 102737385 Page 68 of 190 Published: By: H. Crouse - Digital Equipment Corporation
Date 2/22/63
From B. Maxcy
The following prices are now in effect:
| Ultra Precision Display | Type 31 | $41,200 |
| Remote Computeriter and Control | 7,730 | |
| Precision Display | Type 30C | 19,460 |
| Soroban Punch Including: | GP2-300 | 26,600 |
| Punch Control Cabinet Echo Checking |
Document: 102737385 Page 59 of 190 Published: By: H. Crouse - Digital Equipment Corporation
Document: 102737385 Page 58 of 190 Published: By: Bob Oakley - Digital Equipment Corporation
Document: 102737385 Page 153 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 154 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
DATE February 28, 1963
TO Jim Burley (DCO)
FROM Kenneth H. Olsen
I was pleased to receive your note on our new modules. We are going ahead with an 8-bit Buffer which has read-in gates, a 7-bit Counbter which can be preset and a 6-bit Shift Register which can be preset. In addition, I think we should announce at the IRE Show our new modules which perform all the transfer to receiver functions on one module. I also suggested to Stan that he offer our PDP Indicator Lamp Panels because these would be very useful.
I am interested in your suggestion for a more general decoding module. I would like to have you make out some sketches of what you think they should be and then we will consider what we should do. It is possible for us to mount a socket in the back end of our modules and with a separate plug, of our standard 22 pin variety, we could then make up the jump ring to make this decoding unit into the configurations which are desired. This way when somebody replaces a module unassisted they don't have to solder in jumpers but they only have to pull out this jumper plug and put it in the new module. I would like to see in detail your ideas on what this module should look like and whether or not it should use this plug.
I am sure that you are right in pointing out that the proliferation of our module line is causing confusion to potential customers. I think it would be a good idea if you talk to Stu Grover about the problem and see if you could work out some new system. We might end up with a simplified catalog and a simplified how-to-do dookj describing only diode capacitor gates and the 500 kc line. However, I think that if you could spend a little timne with these people, it might be very helpful.
Document: 102737385 Page 155 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 156 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 21 of 190 Published: By: D. Smith - Digital Equipment Corporation
Document: 102737385 Page 32 of 190 Published: By: Digital Equipment Corporation
Document: 102737385 Page 21 of 190 Published: By: Digital Equipment Corporation
Document: 102737385 Page 151 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 152 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102750017-05-0006-src.pdf, Page 119 of 217
Source: DEC BiWeekly Report, Part 6 of 8
Published:
Published By: Digital Equipment Corporation
Block schematics of the D to A Converter are being drawn up by Drafting and construction of this portion should not be difficult.
PDP-1 and PDP-4 have been tested at Actan Labs for radiation output, both conducted and radiated from 14KC to 1 KMC and for susceptibility to radiation over the same range. The test followed the general outline of MIL-I-26600 (USAF). Both PDP-4 and 1 passed this spec. as far as susceptibility to radiation was concerned, PDP-4 flunked on conducted radiation output and both computers flunked on radiated output. The results will be published in memo form shortly and the exact test parameters and results can be obtained from Arthur Hall.
Document: 102737385 Page 56 of 190 Published: By: Arthur Hall - Digital Equipment Corporation
Document: 102737385 Page 44 of 190 Published: By: Henry Crouse - Digital Equipment Corporation
Document: 102737385 Page 49 of 190 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 42 of 190 Published: By: A. N. Blumenthal - Digital Equipment Corporation
Document: 102737385 Page 19 of 190 Published: By: Jack ?????? - Digital Equipment Corporation
Document: 102737385 Page 54 of 190 Published: By: N. Mazzarese - Digital Equipment Corporation
Date 3/11/63
Subject Computer Sales Forecast
From N. Mazzarese
| PDP-1 | Quantity | Value | Probability | When |
|---|---|---|---|---|
| CRC | 1 | 97K | 100% | 0-3 |
| BBN | 1 | 500,000 | 50% | 0-3 |
| Lincoln Labs. | 1 | 200K | 75% | 0-3 |
| Princeton | 1 | 250K 100 more | 100% 95% | 0-3 3-6 |
| Stanford | 1 | 280K | 75% | 0-3 |
| PDP-4 | Quantity | Value | Probability | When |
|---|---|---|---|---|
| Foxboro (Fitchburg) | 1 | 160K | 98% | 0-3 |
| Foxboro (Westinghouse) | 1 | 120K | 50% | 0-3 |
| JPL | 1 | 250K | 90% | 0-3 |
| AECL | 1 | 200K | 75% | 0-3 |
| Worcester Foundation | 1 | 80K | 50% | 3-6 |
| Bell Labs. | 1 | 80K | 50% | 3-6 |
| Type of Machine | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Univ. of Rochester | 200K | PDP-1 | |||||||||
| Systems Development Corp. (Info. International) | 300K | ADX | |||||||||
| Sylvania | Rental | PDP-1 | |||||||||
| Litton Systems | 80K | PDP-4 | |||||||||
| Michigan Uni. | 120K | PDP-1 | |||||||||
| Maryland Univ. | 120K | PDP-1 | |||||||||
| Raytheon Co, | 120K | PDP-1 | |||||||||
| Rutgers Univ. | 250K | PDP-1 | |||||||||
| V.P.I. | 50K | PDP-4 | |||||||||
| Foxboro | 90K | PDP-4 | |||||||||
| Data Trends | 250K | ADX | |||||||||
| Electrical Intpectionrs | ||
|---|---|---|
| PDP-4 | Stock | Intermediate |
| Display 30A | Stock | Intermediate |
| Mag intermediate Tape 50 | Stock | Intermediate |
| Precision Display 31B | 31B Lincoln Labs (MIT) | Intermediate and final |
| PDP-4 | Foxboro | Intermediate |
| Display 30H {Logic only) | BB&N | Intermediate |
| PDP-4 | Stock | Intermediate |
Document: 102737385 Page 14 of 190 Published: By: George Rice - Digital Equipment Corporation
Document: 102737384 Page 71 of 247 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 6 of 190 Published: By: Ken Olsen - Digital Equipment Corporation
The 4702 Telegraph Receiver is basically a serial to parallel converter. Data consisting of seven elements (7.0 unit code, 7.5 unit code, 8 unit code) is received in serial form. A five bit character is assembled in the shift register. Two and one-half units (start 1, stop 1.5; 7.5 unit code) accompanying each character are removed.
The receiver flag is turned on when the fifth element of the data character is in the shift register. The flag indicates to an external device that a data character has been assembled in the shift register and that the shift register output can be sampled. The time allowed to the shift register (7.5 unit code) is 25.0 milliseconds for 75 baud and 37.5 milliseconds for 50 baud. Bit 5 is the least significant bit.
The shift register is reset when the start element of the next character is received. If an external jumper is connected, the flag will turn off at the start element of the next character. The flag may also be turned off (DC or pulse input on Pin P) by the external device (computer, etc.).
The enable input (Pin U) must be at ground if the receiver is to be activated bythe start element (data on the telegraph line). This connection permits the receiver to be de-activated when the receiver and transmitter are on the same telegraph line (simplex operation). A negative level applied to Pin U while the transmitter is sending prevents the receiver from accepting the transmitted Telegraph signals.
The receiver examines the telegraph line eight times per unit. When the start element is recognized the receiver locks onto the signal. At the first one-half unit, the start element is gated into the shift register. From this time on, the data elements are gated into the shift register at full unit intervals. When the last element of data, element6, (1 start, 5 data) is gated into the shift register, the flag is turned on, and the receiver waits one more unit before attempting recognition of the next start element.
The clock input (Pin S) is such that eight pulses occur each element time.
Clock pulses for telegraph speeds are as follows (7.5 unit code):
| a. | 100 words per minute | (75 baud) | - - - - | 600 pulses per sec. |
| b. | 66.6 words per minute | (50 baud) | - - - - | 400 pulses per sec. |
It should be noted that the receiver will accept data at rates (450 baud) up to 642 words per minute using 7.0 unit code or 600 words per minute using 7.5 unit code. Only the clock pulses change.
The receiver may be connected to devices other than telegraph, i.e. interface between two computer systems, input from measuring devices, etc.
The data input (Pin T) requires that ground be present during the mark (idle, one, closed circuit, stop) and negative 3 volts be present during space (run, zero, open circuit, start).
An indicator light mounted on the handle provides a means of observing the state of the flag. A switch mounted on the handle opens the flag output circuit.
The 4703 Telegraph Transmitter is basically a parallel to serial converter. Data consisting of seven elements (7.0 unit code, 7.5 unit code, 8 unit code) is transmitted in serial form.
A five element data character is gated into the shift register (bit 5 is the least significant bit) by an external device (Pin F, H, J, K, L; gate pulse on Pin N). The start input must be ground at the time of the gate pulse in order to start the transmitter sending (Pin E). The transmitter will send out the start element (one element) followed by the five data bits (five elements). At the completion of sending the sixth element (1 start, 5 data) the flag is turned on. The flag indicates to an external device that the character has been sent, and that a new character may be set into the shift register for transmission. The flag must be turned off (DC or pulse input on Pin S) by the external device (computer, etc.).
The length of the stop element is determined by jumpers on the module (one unit in length for 7.0 unit code, 1.5 units in length for 7.5 unit code , or two units in length for 8.0 unit code). As previously indicated, the flag is turned on at the completion of sending the sixth element, indicating that the shift regisster can be reloaded. The start element of the next character will not be transmitted until the stop element time of the previous character has elapsed. A pulse on Pin S will delay the sending of a character .5 units (7.0, 7.5 unit codes). A negative DC level will inhibit sending while present on Pin S.
The maximum transmission rates using 7.5 unit code, the time allowed to reload the buffer after the flag is turned on is 18.0 milliseconds for 75 baud, 27.5 milliseconds for 50 baud.
The output of the transmitter (Pin W) is ground during mark (idle, one, closed circuit, stop) and a negative 3 volts during space (run, zero, open circuit, start). The Pin V is opposite. The clear input (Pin M) is used to ensure the state of the transmitter output when power is turned on (idle).
The active ouput (Pin U) is negative from the start of transmission until transmission of the sixth element (1 start, 5 data) has been completed. This level may be connected to the receiver (enable input, Pin U) to inhibit the receiver while transmitting on simplex lines.
The clock input (Pin T) is such that two pulses occur each element time,
Clock pulses for telegraph speeds are as follows (7.5 unit code):
| a. | 100 words per minute | (75 baud) | - - - - | 150 pulses per sec. |
| b. | 66.6 words per minute | (50 baud) | - - - - | 100 pulses per sec. |
It should be noted that the transmitter will send data at rates (450 baud) up to 642 words per minute using 7,0 unit code or 600 words per minute using 7.5 unit code. Only the clock pulses change.
The transmitter may be connected to devices other than telegraph, i.e., interface between two computer systems, output to recording device, etc.
An indicator light mounted on the handle provides a means of observing the status of the flag. A switch mounted on the handle provides a means of opening the flag output circuit.
Drawinds x2
Document: 102737385 Page 2 of 190 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102750017-05-0006-src.pdf, Page 75 of 217
Source: DEC BiWeekly Report, Part 6 of 8
Published:
Published By: Digital Equipment Corporation
The usage of trichlorethelyne has increased to ten (10) barrels (450 lbs. per barrel) every six (6) weeks. Price per barrel has decreased five (5) dollars, an ovesall saving of fifty (50) dollars per lot.
Proposals have been made in regard to the use of trichlorethelyne in bulk form. This understandably would mean a cost savings, use of one vendor, and stationary equipment. The elimination of moving barrels would also save time. However, our greatest disadvantage is the weight capacity of the floors, Estimated weight would be between 6,660 and 10,000 lbs.
We currently have five PDP-1's and five PDP-4's under checkout. We are also building a special system for Honeywell which is on the fourth floor of Building 5. Four of the PDP-1's and three of the PDP-4's have been committed to customers. One of the PDP-1's under checkout is due to go to CRC this week. We are not shipping the display and lite pen at this time and the A/D converter will follow sometime in May.
Another of these PDP-1's, the NSA System, is a very large system which has, beside the computer, three bays of equipment which Jon Fadiman and Barbara Stephanson are building; and also the new Data Control and Mag. Tape Control which Roland Boisvert is designing. We have just looked over the schedule on this system and the over-all schedule has slipped by about two weeks. All the people involved seem to feel that we still will be able to meet the delivery date of June 15th.
Micro Tape 555 A meeting was held with Tom Stockebrand and Henry Crouse to discuss the more pertinent and most pressing problems in regard to purchases for quantity production of micro tape 555. The most important thing is generating specifications for the motor and the heads. The head specifications have been gone over at least twice and they shouid be in fairly good shape. The motor specifications are really almost nonexistant and have to be most carefuily specified. These are the most important items in cost accounting for slightly more than half the cost of the unit. Design changes under way are the cast aluminum door which has been completed and released to the pattern maker. The meeting was held this week with the people of Meade Foundry in Bedford to discuss the changes and advisability of the use of a metal pattern and orders for the pattern have been issued fo Maynard Pattern. He agrees to deliver this pattern April 29th. Nearly ready for production release are the two types of control panels, one for the desk model and one for the cabinet model.
Changing of the motor necessitates some change in the machine drawing for the tape hub, I believe we can make 40 of these from a hand pattern now in the hands of Marvin Foundry. A most economical method would be to make up a match plate with 12 of these pattems on a single plate. The quality of the casting would be improved and the cost would be reduced. It now seems possible that we can complete all design changes by the end of next week. Two desk type enclosures have been completed by the shops and are ready for assembly to house the two tape units thet were in the cabinet that went to the IEEE Show.
Light Pen Mechanic parts for the light pen have been completed and delivered to inspection and another unit has been furnished to Derick Chin for assembly. Possible changes to the plastic parts were discussed with the Norman Jones Company today. They will submit proposals tomorrow on two plastic pieces that are part of the ligh pen assembly.
Light Pen The Meade Foundry people who were here yesterday, were contacted by Ron Cajolet and patterns and castings were discussed for most of the mechanical parts for the Display 31. Our estimates show that we can save the pattern cost on stock and machine time on the firt set of castings for the Display 31. Aluminum stock for this unit now costs us approximately $200.00.
During the past two weeks the mechanical inspection depariment has had the following final and intermediate inspections: Intermediate Final DEC - PDP-4 CRC PDP-1 Foxboro - PDP-4 J.P.L. - PDP-4 Harvard University - Data Reader 2318
There has been considercble computer activity recently with both Princeton and Yale about to obtain PDP-1's for their nuclear physics interest. There is a gcod chance of selling a PDP-1 to the University of Michigan for both the Department of Psychology and Physics use. Harvard also has expressed its desire to obtain a PDP-4 for its Dept. of Psychology and is showing definite interest for another computer to be used by its nuclear physics people.
With all the interest shown, especially by Physics people, in the PDP-1 it would be a shame for this company to concentrate all its efforts on new machines. Especially when we are not sure what kind of a non-rental non-renegotiable market there is for the proposed equipment. There are at least a dozen Universities which have expressed an interest in the PDP-1 either for next year purchase or even later. That is an awful good potential for the PDP-1, but if some newer machine should ever break into the physics field the PDP-1 will have a tough time. With the use of our new circuits and our memory knowledge we should be able to make both the PDP-1 and the PDP-4 more attractive from a hardware point of view and the software of course is already there. It has become my feeling that ihe users group and the CRT displays are our two most powerful sales tools at the present time.
Document: 102737385 Page 109 of 113 Published: By: Steve Lambert - Digital Equipment Corporation
Document: 102737385 Page 108 of 113 Published: By: Ed Harwood - Digital Equipment Corporation
Document: 102737384 Page 37 of 247 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737384 Page 37 of 247 Published: By: Ken Olsen - Digital Equipment Corporation
I have become quite concerned about our proposed plan to place our prototype PDP in Ed Fredkin's plant. Some of the potential customers that we heve talked with have shown some concern of being to close to Ed because they consider him potential competition. If this is a common attitude it could be quite serious to us. I am also afraid that if Ed carrles out all the services he promised, it might also be limited in the to make the installation lock god.propose that consider putting the POP tn present auditoriumwath room in the present storagethe fourth floor of Building 4. This is very nicely prepared with « tile@ weil iliuminated erea. This is so large thet we could set the computervery nicely end heve for people special experlmants and forlerge claswoom. This could work out very well fer Ed Fredkin because we couldtake the lock off present robbie door tn Building 4 end put it the door inwe could give Ed Fredkin direct to this We couldpartition along the fer wall that the Machine Shop would have access to thelrwash We could then allow Ed Fredkia, customers, to in freelythet end this fecility. If necessary, might make « email co-educationalKonneth 4. Olean
Document: 102737385 Page 106 of 113 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 147 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 104 of 113 Published: By: Harlan E. Anderson - Digital Equipment Corporation
Document: 102737385 Page 146 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 145 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 101 of 113 Published: By: Pete Bonner - Digital Equipment Corporation
Date May 10.1963
Subject Attached Arithmetic Execution Times Table
To "PDP Distribution List" From Pete Bonner
Due to the encounted increased demand for arithmetic operating times for both of our computers, the attached tablehas been compiled. The figures comprising this table were gathered from presently existing tables and "educated guessses" as to function execution times.
The table is being circulated only within DEC with the intent of having missing times filled in, correct times certified and incorrect times corrected. Where it is applicable, it would seem highly desirable to have minimum, maximum and average times.
It is hoped that serious consideration will be given to correcting and updating this table. Toward this end I would appreciate having all responses directed to me.
| Arithmetic Execution Times for PDP-1 (In microseconds, Including All Access) | Arithmetic Execution Times for PDP-4 (In microseconds, Including All Access) | ||||||
|---|---|---|---|---|---|---|---|
| Function | Fixed 18 | Float 18-18 | Fixed 36 | Float 36-18 | Fixed 18 | Fixed 36 | Float 36-18 |
| Addition | 10 | 720 | 180 | 790 | 16 | 96 | min. 360 max. 336 ave. |
| Addition* | 1,000 | ||||||
| Subtraction | 10 | 750 | 180 | 710 | 24 | 128 | min. 360 max. 336 ave. |
| Subtraction* | |||||||
| Multiplication | 14 min. 25 max. 20 ave. | 470 | 795 | 1,500 | 40 min. 256 max. 148 ave | 700 | min. 700 max. 500 ave |
| Multiplication* | 2,500 | 8,500 | 10,000 | ||||
| Division | 30 min. 40 max. 35 ave | 470 | 950 | 1,625 | 200 | min. 720 max. 650 ave | min. 820 max. 750 ave |
| Division* | 3,100 | 10,000 | 10,000 | ||||
| Sine-Cosine | 390 | 12,000 | 2,550 | 6,840 | |||
| Sine-Cosine* | |||||||
| Tangent | |||||||
| Tangent* | |||||||
| arcsine | 1,035 | ||||||
| arcsine* | |||||||
| arccosine | 965 | ||||||
| arccosine | |||||||
| Arctangent | 2,035 | ||||||
| Arctangent | |||||||
| Square Root | 715 | 3,025 | 1,420 | 1.1,630 | |||
| Square Root | |||||||
| Exponential | 660 | 825 | |||||
| Exponential | |||||||
| Logarithm | |||||||
| Logarithm | |||||||
Document: 102737385 Page 95 of 113 Published: By: Robert F. Maxcy - Digital Equipment Corporation
Document: 102737385 Page 98 of 113 Published: By: Dit More - Digital Equipment Corporation
Document: 102737385 Page 99 of 113 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 64 of 113 Published: By: T. H. Griffin - Midwestern Instruments
Document: 102737385 Page 85 of 113 Published: By: S Lambert - Digital Equipment Corporation
Document: 102737384 Page 21 of 247 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 144 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 81 of 113 Published: By: Jack Atwood - Digital Equipment Corporation
Document: 102737385 Page 83 of 113 Published: By: Dave Denniston - Digital Equipment Corporation
Document: 102737385 Page 83 of 113 Published: By: Dave Denniston - Digital Equipment Corporation
Document: 102737385 Page 91 of 113 Published: By: Digital Equipment Corporation
Document: 102737385 Page 68 of 113 Published: By: R. Beckman & N. Mazzarese - Digital Equipment Corporation
Document: 102737384 Page 12 of 247 Published: By: Harlan Anderson - Digital Equipment Corporation
Document: 102737384-05-001-src.pdf, Page 12 of 247
Source: https://www.computerhistory.org/collections/catalog/600000624/record/102750017/
Published:
Published By: Digital Equipment Corporation
9. Jet Propulsion Labs., Pasadena, Calif. (Computer Section) This section at JPL is by far our best customer for both computers and modules. PDP-1's and the four PDP-4 Computers are at the Lab. Due to budget cuts at JPL and the building of systems using the PDP-4's we are not expecting as much in the coming fiscal year as we did in the last for computers. Module orders should remain the same.
(Instrumentation Section) This section operates the windtunnel and central recording systems. The PDP-1 system being built by Astrodata in Anaheim, Calif. is to be delivered to JPL this month. The CRS-PDP-4 job is going well and more options will probably be purchased in the next few months, Some module orders are also anticipated.
11. System Development Corporation, Santa Monica, Calif. A large sized blanket order for modules has been issued covering the next six months. Much applications work is necessary for modules with extra tie-in to the PDP-1 which is to be installed this month.
Stanford University's Computation Center was dedicated recently at a large luncheon-reception attended by California business and professional people, This is called the "outstanding research computer setup West of the Rockies". The PDP-1 was demonstrated with the display scope. Stanford has also a complete Burroughs B5000 and an IBM 7090 in the Center, but the PDP-1 seemed to draw the most attention. A heat equation describing the temperature in a rod, the mathematical model of a plasma and the projection of 3-dimensional surfaces were demonstrated using the PDP-1 with the 7090. The 7090 under the control of the PDP-1 did the computation using John McCarthy's "time stealing scheme". Closed circuit television with several monitors was used enabling all the viewers to observe the scope display. The PDP-1 was used alone to demonstrate Spacewar, a complex Function Plotter, and Snowflake. Cleve Moller of Stanford gave an excellent presentation of the PDP-1 and explained its function in each part of the demonstration.
Quality Control
The ADA-1, which was shipped at the end of July, has been operating extremely well. We are still cleaning up loose ends however. Next week Jim Reardon is putting in a field modification to inhibit overflow when the input signal exceeds the allowable range. The manual is also being corrected and updated.
Electrical Mechanical PDP-1 PDP-1 PDP-1 Tape Control 510 Tape Control 510 PDP-1 PDP-1 46 PDP-1 47 Type 142 High speed A to D Converter PDP-5 Memory Tester 1516 PDP-5 Finals Electrical Mechanical PDP-1-46 PDP-1-46 Kie Corp. Mag tape 50-51 Mag tape 50-51 142 A-D Converter PDP-1-47 2010 Memory Buffer 142 A-D Converter 2010 Memory Bufferer Joe Nangle has made a nice product bulleting on the ADA-1 to give customers an idea of our capabilities in this area. Bulletins are also underway on the 138 General Purpose A to D, the 139 General Purpose Multiplexer, the 142 High-speed A to D, the Models Type 4679, 1704 and 1572. For the meanwhile I have written a memo summarizing the characteristics of these new items.
Roger Gagne is working on the 138 and 139 for Raytheon. It is operating quite well and will be delivered this week, We are preparing test procedures etc., so that future units can go directly through production.
The Model 142 for sales is nearly assembled, but we have not started checkout yet.
The two major limitations on our ADA equipment right now are the lack of a dual ended input and the capacitance seen at the input when the multiplexer is switched. The dual ended input will be handled with an operational amplifier.
Engineering The ADA-1, which was shipped at the end of July, has been operating extremely well. We are still cleaning up loose ends however. Next week Jim Reardon is putting in a field modification to inhibit overflow when the input signal exceeds the allowable range. The manual is also being corrected and updated.
Joe Nangle has made a nice product bulleting on the ADA-1 to give customers an idea of our capabilities in this area. Bulletins are also underway on the 138 General Purpose A to D, the 139 General Purpose Multiplexer, the 142 High-speed A to D, the Models Type 4679, 1704 and 1572. For the meanwhile I have written a memo summarizing the characteristics of these new items.
Roger Gagne is working on the 138 and 139 for Raytheon. It is operating quite well and will be delivered this week, We are preparing test procedures etc., so that future units can go directly through production.
The Model 142 for sales is nearly assembled, but we have not started checkout yet.
The two major limitations on our ADA equipment right now are the lack of a dual ended input and the capacitance seen at the input when the multiplexer is switched. The dual ended input will be handled with an operational amplifier.
Document: 102737385 Page 55 of 113 Published: By: John L. Atwood - Digital Equipment Corporation
Document: 102737385 Page 56 of 113 Published: By: Roland Boisvert - Digital Equipment Corporation
Document: 102750017-05-0006-src.pdf, Page 24 of 217
Source: DEC BiWeekly Report, Part 6 of 8
Published:
Published By: Digital Equipment Corporation
NSA System The interconnecting equipment for NSA is about ready to start checkout. We have completed all the logic diagrams and the wiring has been done on all of the panels except one. A large part of the interpanel wiring has been done and the drafts of the manuals have been turned over to the technical writers. The new modules have been our biggest problem and we now have most of these. Many of these were started a long time ago but hit a stop at one stage or another along ihe way. These included the Type 6102 Inverter, 4219 Quintuple Flip-Flop, Types 4678 and 4679 Level Amplifiers, Types 1574 and 1576 Ladder Networks, the 1578 Multiplexer Switch and the Type 1572A which is a variation cn our 1572 for use in systems where the capacitive load presented by the input is of a large degree of importance.
Computer A-D Options The Computer Guidance Committee approved four general A to D options which we want to have for the computers. The first is the Model 138 General Purpose A to D converter. This will convert up to 11 bits at varying speeds depending upon the number of bits and desired accuracy. The price of this unit is $5,000 and Bob Maxcy has some general specs on the system. (The specs on the system are extremely conservative and I have a prototype on the bench now which I hope to test out so that perhaps we can give some better specs at a later date. Certainly there is no problem in meeting the specs that we have put on the unit now.) The next option is the Model 139 General Purpose Multiplexer Control. This Multiplexer Control will handle up to 64 channels of multiplexer inputs, with either individual addressing or cycling. If the cyclic mode is chosen, the multiplexer will automatically return to Channel 0 after the last channel has been sampled. This option sells for $3,600. The multiplexer switches themseleves are available in groups of four and sell for $425 per group.
The next option is the Model 140 High Speed ADC. This is the 10 bit 6 microsecond unit which we developed for NSA. It has not yet been priced out. With this wiil be a high speed Multiplexer Control, Type 141, which also has not been priced out.
Prospective Customers We have received quite a few inquiries recently for ADC's and hybrid computation. Most of these have been people who have already surveyed the market and determined what they wanted. However, a few of these are prospects for us. These include four - Raytheon wants and A to D to A converter system to tie various devices onto a PDP-1. I gave them a price for our standard A to D option and multiplexer and am now in the process of figuring a price for a D to A system for them.
The University of Washington wants to buy a PDP-1 and they want four channels of D to A on this. Bob Maxcy is handling this and I gave him the price information on this D to A. Also I talked with Mr. Schelderoff out at ASCRC. They are presently buying an Adage converter to hook onto the PDP-1 as we had suggested to them earlier. They are also interested in purchasing a high speed A to D in the Fall so I talked to them about our high speed unit.
Teletype had their new Modal 33 and 35 there. The 33 does have some drawbacks over the 35 ($1500) such as no tab or backspace and apparently a few other differences that should be looked into. The man in the booth was not too sure about the differences. He said that althought the machine has been described as a 2 hour a day machine that really there is no limit to the number of hours per day. As a sidelight we found that the coding keyboard is available separately at a price that Teletype seemed somewhat apoligetic about -- $35:
Electrical Inspection Unit Customer Inspections PDP-1 Stanford Univ. Final PDP-1 M.I.T. Final PDP-1 D.E.C. Intermediate PDP-4 J.P.L. Final 30G Display D.E.C. intermediate 30G Display Stanford Final 30A Display DEC Intermediate & Final Electrical Inspection since last report Unit Customer Inspections PDP-1 Stanford Univ. Final PDP-1 M.I.T. Final PDP-4 DEC Final PDP-4 Adams Final PDP-4 Foxboro Final PDP-4 J.P.L. intermediate Display 30H OAL 2 panel special Display 30G DEC Intermediate 3371.1 Tel. Type Rec. & Trans. Adams Intermediate & Final
Document: 102737385 Page 141 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 142 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 143 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 140 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 133 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 135 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 136 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 138 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
DATE May 27, 1963
TO Gordon Bell
cc: Bob Savell
FROM Kenneth H. Olsen
At the SJCC Bob Savell and I talked to Mr. George G. Keyes who was supervisor, Sales and Applications, Teletype Corporation, 5555 Touhy Skokie, Skokie, Illinois. He is very much interested in our applications that are used and I think he would be a good one to direct our questions to.
One of the most significant developments at the Trade Show was the fact that Teletype is developing a competitive spirit. This is truly revolutionary. They are not only feeling competitive with Western Union but with the new IBM system. They know the prices on all these things; they know how their equipment compares.
After careful questioning it turns out that the new Light Duty Page Printers are not really light duty but because of the lack of experience with them they don't want to promise too much. They do, however say that after 500 hours of use they should be lubricated and after three or four lubrications the thing should be checked over mechanically.
Now it is not the number of hours of use, of course, but it is really the number of characters typed and so we might install a mechanical counter which after so many characters are typed would indicate that a lubrication is necessay.
They gave us inconsistent stories as to the availability of these typewriters and I think that depends partly on the pressure we give them and how interested they are in our applications. One man said six months, another one said four weeks. Keyes was the man that said four weeks and I think if we challenge him on this, he could deliver in four weeks. I think we should immediately get an automatic send-receive set with Punch and Reader and use this to standardize our paper tape preparation.
They can make a new character set for very little money. If they have the characters engraved already it only costs $15 to setup for moulding the cylinder and it probably costs very little to have them engrave new characters. However, I suggest that we use the most standard character set available.
This unit was designed by Mr. Dreyfus who is one of the famous industrial designers. If I remember correctly he is the one who designed the present telephone. They also have their Model 35 unit which is equivalent to the present 23, redesigned by Dreyfus. It is the same base and the same general lines as the Type 32; however, it is much heavier and bigger. And also, cost $1,500 comnpared with $500 for the Model 32. However, it has more freedom because of the stunt box. It has both Vertical and horizontal tab. The vertical tab is programmed in approximately the same way as a line printer and I believe it is used mainly to skip lines when you come to the end and beginning of pages. They have a disc with pins in it to do the vertical tabing. It also has aa few other features but I don't think these are very critical.
Because of the low price of the Type 33, people can readily afford to have spares so the relatively short life of the unit isn't as important. However, for continuous long-life applications we might suggest that people use Type 35. I think they might be able to plug it in the same socket.
At the Trade Show the Type 35 had a spocket feed paper and they said it will be available on the Type 32.
Document: 102737385 Page 135 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 35 of 113 Published: By: J. L. Atwood - Digital Equipment Corporation
Document: 102737385 Page 103 of 113 Published: By: K. Wakeen - Digital Equipment Corporation
Document: 102737385 Page 47 of 113 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 49 of 113 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 50 of 113 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 44 of 113 Published: By: Ed Harwood - Digital Equipment Corporation
Document: 102737385 Page 127 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 131 of 209 Published: By: J M Graetz - Digital Equipment Corporation
Document: 102737385 Page 42 of 113 Published: By: D. A. White - Digital Equipment Corporation
Document: 102737385 Page 38 of 113 Published: By: Allan Kotok - Digital Equipment Corporation
Document: 102737385 Page 39 of 113 Published: By: Gordon Bell - Digital Equipment Corporation
Document: 102737385 Page 40 of 113 Published: By: Allan Kotok - Digital Equipment Corporation
Document: 102737385 Page 41 of 113 Published: By: Harlan Anderson - Digital Equipment Corporation
Document: 102737385 Page 31 of 113 Published: By: J.Smith - Digital Equipment Corporation
Document: 102737385 Page 34 of 113 Published: By: George O'Dea - Digital Equipment Corporation
Document: 102737385 Page 36 of 113 Published: By: Dit Morse - Digital Equipment Corporation
Document: 102737385 Page 130 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102750017-05-0006-src.pdf, Page 12 of 217
Source: DEC BiWeekly Report, Part 6 of 8
Published:
Published By: Digital Equipment Corporation
Purchasing In the past few weeks we have experienced delivery problems with Teletype equipment, particularly the BRPE-11 Punches. This was due principly to an engineering hold Teletype placed on the production of their punches. It seems there has been an excessive amount of tape breakage in the field and redesign in the tape reel area was necessary.
Since DEC does not use the tape reels that are causing all the trouble, we have had no probiems with tape breakage. A telephone canversation between Bob Savall and one of ihe engineers responsible for the hold resulted in a waiver for six of DEC's units to be shipped immediately.
We purchased three of Teletype's new model 33 Printers equipped with dial phones for PDP-6; another with tape perforator and reader is expected in mid June.
At this time delivery is relatively palatable, but in the near future it is expected to jump to three months, perhaps more.
Unfortunately it will be several months before a 50 cycle unit will be made available. At present, Teletype's European customers are forced to convert the equipment themselves.
The Adage equipment, both the Bidirectional Data Link for Minneapolis Honeywell and the A-D Convertor for CRC, have had several delays. The Adage people were here this week to assure delivery of June 12 for the Data Link and June 15 for the A-D system.
The Holley printer, in similar checkout straits, will be delivered the week of June 10, 1963.
A new committment will be made this week to Soroban Engineering Inc. for twenty-five Computerwriters. Deliveries are scheduled during the period of September 1963 through September 1964.
The Magnetic Tape project for the 555 Magnetic Tape Units is now underway. One thousand reels, inserts, and boxes were shipped to 3M Co. for the mounting of the #489 tape on reels, and insertion into the boxes. 3M can apparently supply us with tape in this manner less expensively than selling us the tape on their own reels. The cost per 3600 feet of Magnetic Tape on their reels is $69.38 per roll. The cost for mounting this tape on our reels and inserting the reels into boxes is $60.03 (per 3600'} a unit price of $4.35 per reel (260' per feel). Delivery should be made in 4 to 5 weeks.
On May 31, Jack Smith and I visited Pastoriza Electronics to check progress on our 728 Power Supplies and Tape Control 510 Wiring. The work thus far was very acceptable and it appears as if Pastoriza will meet our delivery requests
Two new teletype modules have been designed. These modules are for use with the 8 bit teletype code and are the 4706 Incoming Line Unit and 4707 Teletype Transmitter. Connector wiring for these modules is similar to the wiring for the 4702 and 4703 5 bit teletype modules thus allowing for system expansion with a minimum of rewiring. A production release has been started on the 4706 module and the 4707 will be released very shortly.
Several improvements have been made in the 4702 and 4703 5 bit teletype modules and will be included in the 4706 and 4707 8 bit units. The 4702 incoming Line Unit will have jumpers on the etched board that will allow the user to select either assertion or negation output signal polarity. Dummy outputs have been provided on three pins so that the 5 bit and 8 bit Incoming Line Units can be directly interchanged. The 4703 Teletype Transmitter will contain a line driver that is capable of driving the line relay directly. Either output polarity can be obtained by moving a jumper on the etched board.
Construction of the PDP-5 prototype was completed Wednesday and checkout has been started.
Final design of the Type 135 (PDP-4) Memory has been completed except for the pin assignments on the new sense amplifiers. The wiring layout is done except for those connectors and the order placed with production. We hope to equip every PDP-4 with this memory beginning with #21 which is scheduled to come out of production on July 22. The selection scheme for this memory has been changed, at the expense of a couple of 1987 modules, to be the same as that for the PDP-1 memory under design. This makes it possible to use the type of stock for both computers as compared to the original scheme which required one type for PDP-1 and two others for the PDP-4 (one for each of the two 4K segments).
We hope to have the layout for the PDP-1 Memory complete by July 15 so that production can start on the first unit and a debugged system operating by the end of August.
Before the first of July, we plan to ship five PDP-4's and two PDP-1's. This means around the end of June and the beginning of July, we will have practically our whole group out doing field installations.
At the present time, it looks like the NSA computer will be at least one month late and a new projected delivery date is July 15th.
During the past two weeks, the Mechanical Insp. Dept. has had the following Final and Intermediate Electrical and Mech., Inspections.
Electrical Mechanical Final Intermediate Final Intermediate PDP-4-13 PDP-4 JPL PDP-4 Mag Tape Control 67 Mag Tape 50 Mag Tape 50 Mag Tape 51-50 PDP-4-20 PDP-4-18 PDP-4B-11 PDP-1C-30 Mag Tape 50 51 Control PDP-4B-16 PDP-4B-13
Document: 102737385 Page 22 of 113 Published: By: G. Moore - Digital Equipment Corporation
Document: 102737385 Page 25 of 113 Published: By: George O'Dea - Digital Equipment Corporation
Document: 102737385 Page 16 of 113 Published: By: Sandy Moore - Digital Equipment Corporation
Document: 102737385 Page 129 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 12 of 113 Published: By: George O'Dea - Digital Equipment Corporation
Document: 102737385 Page 14 of 113 Published: By: Arthur Hall - Digital Equipment Corporation
Document: 102750017-05-0006-src.pdf, Page 2 of 217
Source: DEC BiWeekly Report, Part 6 of 8
Published:
Published By: Digital Equipment Corporation
Two more Model 33 Printers with tape reader and perforator have been ordered from Teletype and are scheduled to arrive August 7, 1963.
The Monroe 16 column line printer purchased for Ken Wakeen will be shipped the first week in July.
Delivery of Standard Equipment The model 2500 Readers from Digitronics are being delivered according to schedule, Two were received this past week and Digitronics has confirmed that the balance will be delivered according to our required schedule of two per month - July through October. Although Digitronics missed delivery of the two Model 3500 Readers in June, they will be able to deliver three each in July and August to bring us up to schedule. The balance of the Model 3500's will be shipped two per month continuing through October.
Soroban Engineering, Inc. is meeting delivery of the computeriters with no problem - two each July and August, three September, two per month October through January.
Teletype will be shipping three BRPE-11 Punches by June 30, 1963 and will complete the present order for punches with three in July and three in August. The remainingtwo Modes 28 Printers on open order will be shipped in July.
The 555 Magnetic Tape, reels, inserts and boxes presently being worked on by Minnesota Mining and Manufacturing Company should be ready for shipment Monday, June 24th.
Micro Tape Two weeks were spent working very closely with the people in the machine shop and mechanical inspection in order to mechanically fabricate and complete ten micro tape decks. Quite a bit of difficulty was experienced in making and inspection of jigs and fixtures necessary for this work. However, the job was completed and the units are ready for final assembly. The only hold up has been in receiving motors and other hardware from our vendors. Due to the problems with these jigs and fixtures, the prints have been changed which should help to make this unit a little bit easier to build without sacrificing any of the necessary accuracy.
DEC Paper Tape Reader The past two weeks have been spent assembling the mechanical parts for a paper tape reader utilizing a stepping motor for drive and eventually we hope fo be able to use our own design for light source and sensitive reading surfaces. If this unit works out as visualized, it could very well mean that the paper tape reader could be mounted on the same panel where the present paper tape punch is located and take 1/3 less space then the present reader. I have been working with Bob Savell on this and we are just about ready to start making mechanical tests of the tape moving mechanism.
Document: 102737385 Page 7 of 113 Published: By: Dick Mills - Digital Equipment Corporation
Document: 102737385 Page 128 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 2 of 113 Published: By: Roger Melanson - Digital Equipment Corporation
Document: 102737385 Page 5 of 113 Published: By: R. L. Lane - Digital Equipment Corporation
Document: 102737385 Page 18 of 113 Published: By: Klaus Doering - Digital Equipment Corporation
Document: 102737385 Page 125 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102737385 Page 124 of 209 Published: By: Ken Olsen - Digital Equipment Corporation
Document: 102750017-05-0007-acc.pdf, Page 139 of 174
Source: DEC BiWeekly Report, Part 7 of 8
Published:
Published By: Digital Equipment Corporation