PDP -8/S  - Desktop

(the highlighted text is corrections I have made to the original FAQ as new information from the original designer has surfaced).

Basic information:

Date of introduction:  1966 (Unveiled, Aug 23, WESCON, Los Angeles).
Technology: All transistors (Qty: 1001 in CPU)- no ICs.
Date of withdrawal:    1970.
Total production run:  ????? (desktop model - very few).
Price: $9,995 (did not include core memory)
Claim to fame: First CPU to break $10,000 barrier, smallest transistor desktop computer!,
shared data path techniques seeded later architecture in 16 bit machines,
actually spawned the concept of bus architectures used in all the current
microcomputer chips. (Saul Dinman was granted Patents in later years).
The PDP-8/S desktop mini is the rarest PDP-8 minicomputer there is.

Designed solely by: Saul B. Dinman, Product Line Manager, Module Product Line. 1966

Technology:  DEC Flip Chip modules and core memory, as in the PDP-8.
        Unlike the PDP-8, the PDP-8/S memory module was mounted between
        a pair of quad-height single-width boards that plugged into the
        standard flip-chip sockets (this was sold separately as the H201
        core memory unit and supporting tech., at $3890 for 4K by 13 bits).

Reason for introduction:  "It was at one of those meetings that Ken went off on a
        rip about not being able to get the cost of the PDP-8 down any further. He issued
        one of his famous challenges, stating that he foresaw the day when a PDP-8, software
        compatible computer should be able to fit on a desktop and cost less than
        $10,000. Having been a CPU architect prior to my joining DEC, I went back to
        my desk and the PDP8/S was the result. It was also the first time at DEC that
        any one person had designed a computer single handedly", Saul B. Dinman
        It was the least expensive general purpose computer made with second
        generation (discrete transistor) technology, and it was the smallest machine to be mass
        produced.

Prototype: Click here for information and pictures of the prototype PDP-8/S. It was built entirely
        on the Blue Laboratory Panels you see below. (H901s).

Standard configuration:  CPU with 4K of memory, plus PT08 110 baud current
        loop teletype interface and teletype.  Both a rack-mount and
        table-top versions were sold (both 9" high by 19" wide by 21"
        deep).  The rack mount version included slides so it could be
        pulled out for maintenance.

Expandability:  The CPU supported the standard PDP-8 negibus, but I/O
        bandwidth was 1/5 that of the PDP-8.  Thus, most, but not all
        PDP-8 peripherals could be used.  A few DEC peripherals such as
        the DF32 came with special options such as interleaving to slow
        them down for compatibility with the PDP-8/S.

Peripherals:  At the time of introduction, the following negibus
        peripherals were offered.

        -- Type 750 high speed paper tape reader and control ($3500).
        -- Type 75A high speed paper tape punch and control ($4000).
        -- Type 138 analog to digital converter ($4500).
        -- Type 139 analog Multiplexer ($3300).
        -- Type 30N precision CRT display ($13,400).
        -- Type 34D oscilloscope display ($3600).
        -- Type 370 high speed light pen ($1625)
        -- Type 350 incremental (Cal Comp) plotter and control ($8,900 up).
        -- Type 451 card reader and control ($14,900).
        -- Type 451B fast card reader and control ($25,600).
        -- Type 450 card punch control for IBM Type 523 punch.
        -- Type 64 (later 645) Mohawk line printer and control ($28,900).
        -- Type 250 (RM08) serial magnetic drum (256K words for $43,600).
        -- Type 552 DECtape control (for type 555 DECtape drives, $9500).
        -- Type 555 dual DECtape transport, $7400).
        -- Type 57A magnetic tape control with IBM type 729 drive ($15,200).
        -- Type 580 magnetic tape system with one transport ($19,700).

        By 1966, the following peripherals had been added to the line:

        -- Type AA01A three-channel digital to analog converter.
        -- Type CR01C card reader control.
        -- Type TC01 DECtape control for up to 8 TU55 transports.
        -- Type 251 drum (8-256 tracks, 8 sectors/track, 128 words/sector).
        -- Type 645 line printer control.
        -- Type 680 data communications system (allows 64 teletypes). ... ya right!

        By 1967, the following peripherals had been added to the line:

        -- Type AF01 analog to digital converter and Multiplexer.
        -- Type AX08 parallel digital input port,oscilloscope interface,A/D,clock,etc
        -- Type 338 Programmed Buffered Display (vector graphics).

        By 1968, the following new peripheral had been added:

        -- Type DF32 fixed head disk system (32K to 256K words).
        -- Type BE01 OEM version of the TC01 (no blinking lights).
        -- Type BE03 dual TU55 drive for the TC01 or BE01.

        Finally, as DEC abandoned the negibus, they introduced the
        DW08B negibus to posibus converter so newer posibus
        peripherals could be used on older negibus machines, and the
        DW08A posibus to negibus converter to allow use of old
        peripherals on new machines.
 
 

Survival:  These minis were produced until the TTL IC came out which brought the cost of CPUs down
        and the speed increased. These machine were rather slow and low cost so PDP-8/S systems were quickly
        discarded as newer machines became available for comparable prices;
        thus, they are less common today than the Classic PDP-8, even
        though comparable numbers were made. There were over 3000 Classic 8 model computers
        made and many of them were kept and are still running today.
        This is the only desktop model known to exist. Further, only 6 other PDP-8/S remain today!
 

Ahhhhhhhhhhhhhhhhhhhhhhhhhhhhhh, the desktop PDP-8/S. A little dusty but well worth picking up. This 8/s should be running in a few days. It is in mint condition but I have not powered it up yet as I want to condition the supplies first.
 
 

inside the beast.

Serial # 566.
 

Another angle showing the core memory and power supply in the back.

The core memory stack. The white boards are the supporting hardware.

This was an *extremely* lucky find. This unit has been sitting idle for over 25 years! I got 1400 spare flip chips with it.
 

Update: I just finished pulling apart the power supplies and found the capacitors to be in mint condition. I tested them on my LCR meter and found them well within their ranges (both C and Q)!!!!

The power supply - 1 transformer, 1 full wave bridge, 6 capacitors. NO REGULATION!?!??!?!?
A regulator was used to supply a memory reference voltage.

Check back next Wednesday to see if this mini is running FOCAL!
 

IT'S TUESDAY AND THE DESKTOP 8/S IS RUNNING!
 
 

First trick was to check the caps and "condition" the supply. Voltages were perfect and I ran the supply for 5 minutes and set up a scope to look for spikes. No spikes and one happy power supply!
 
 

First, pulled off the capacitor pack. Checked them out on an LCR and they tested as good as new! No leakage coming out of them either!

Yuck.. Those *yellow* boards are supposed to be white! This mini is dirty!

Pulled off the front panel to start cleaning!

While pulling all the Flip Chips to clean and inspect I decided to take some pictures of the core - I will never pull this module out again!
Serial #597

The entire core (all the boards have been washed and cleaned and now look new).

The entire stack uses 24 Flip Chip slots in the 8/S.

I am sooooooo happy I closely inspected every board. Can you see what's wrong with this picture?

A lock washer was stuck between the power resistor and a diode... What a mess if I had power the unit up without tearing it down.

The front panel and sides removed for cleaning. Notice all the power supply wires are disconnected so the supply could be tested safely.
 
 

The 8/S cleaned and reassembled.
 

Cleaning Time - 16 hours!
 
 

The 8/s had quite a bit wrong with it (unlike my first one).

#1) Bad contacts on the backplane resulted in a brutal cleaning attempt with many different contact cleaners.
#2) The main R401 clock had a bad solder contact forcing the 8/S to do nothing but light up the "RUN" light when switches were pressed.
#3) Two accumulator R202 Flip Chips had bad contents causing *most* registers to store rather "random" numbers. (remember this mini is bit-serial)

The 8/S was now reading core memory - the RIM loader was still resident!

#4) A bad solder contact on a transistor on the R602 Pulse amplifier in the C/W (Clear and Write) memory section caused the 8/S to never complete a memory cycle (only when writing). This was easily found as the PAUSE light remained lit when a deposit was attempted.

The PDP-8/S uses RC clocks (that's right - no crystals)... so the memory system must be tuned time to time.. Oh, BTW: It also does not have PS regulators so you have to tune a memory reference voltage supply. I always use my Tek720 DSO for restoring old minis as it is very portable and can be hung anywhere while working inside a rack or cabinet.
 

The core memory system tuning area - R401 is a RC clock. A701 is the memory reference voltage. The memory is a very simple coincident current system with only two adjustments. The master current adjustment - A702. The strobe adjustment is through R401. The processor and memory operate on different clocks. Every time the processor requests memory access its clock stops; when the memory cycle is complete, the processor starts.
 

The R401 - memory clock

The A702 - voltage reference

(for interest sake): The R401 above the core memory (green module) is the actual CPU RC clock.
 
 
 

First, the memory reference voltage must be set.

-12.8V is within range. The core memory reference voltage changes from -9.6V to -13.3V depending on the temperature!

Now we can set the memory cycle time...

I had to extend a board to gain access to the memory cycle signal. I cannot access the backplane on this desktop model easily.

6.344uS cycle time is within specs. It will vary +/- 100nS depending on temperature. I can make the 8/S memory system work from 3uS to 9uS but 6.3uS is the best setting for zero problems.

Core memory is tuned!!!
 
 
 

#5) an OPR instruction was a IOT?? - sounds like a weak transistor? It was, The R151 (binary to octal converter) had a bad transistor quickly replaced!

8/S is now single stepping through programs :-)
 

#6) PDP-8/S barfed when trying to run programs "full speed" - another weak transistor - in all places??? the INCR R111 had a transistor switching between 0 and .7 volts. Not good. The 8/S would run a program for awhile and all of a sudden the PC would go into hyper space.....
Transistor replaced and the PDP-8/S is....................
 

RUNNING!
 
 
 

PDP-8/S running a program that tests all core locations and major instructions :-)

Notice the brightest light at the bottom right side... "RUN" :-)
 
 

Happy 8/S counting through memory locations; reading and writing.
 
 
 
 

I wonder, if I mate these two will I get another one????
 

Total time on hardware - 2.5 hours and what a mess!
 

Next step - PT08 - the teletype interface.