My Mark-8 Minicomputer

Status: Fully operational.  This article is complete.

Table of Contents:

Background

Pre-construction

PC Boards 

PC Assembly

Modifications

Backplane

Front Panel Assembly

Enclosure Assembly

Getting it Running

Happy Anniversary!

What's Next?

Links

Acknowledgements

 

Background

The year was 1974.  I don't remember the exact month anymore, but someone showed me an article in Radio-Electronics magazine about building a computer.  That's nice, I remember thinking, but what would I use a computer for?  I was in 9th grade and had my goals set on being an electronics technician and a ham radio operator.  In fact, I had just been accepted by the county vo-tech to go there in 10th grade to study just that.  I went to vo-tech for three years, and during my junior year, something started to change.  Yes, I still wanted to be an electronics technician and a ham radio operator, but I was discovering there was more out there than amplifiers and transmitters.  When we started studying digital electronics, I knew I had found what I had been looking for, because I saw immediately how digital electronics could be applied to create the building blocks of a computer.  I still wasn't sure what I could use a computer for, but I soon started wanting one.

Early in my Senior year, I participated in the co-op program, where I got to spend a few hours everyday working like an apprentice in a local electronics business.  My first position was in a television repair shop, where I did well, but the owner of the shop had to lay me off after just a few months due to economic conditions.  Before long, another opportunity came my way at a company called Syntonic Technology.  I started there working part-time in the shipping department and part-time testing PC boards for an electronic terminal called Wiltek.  The work on Wiltek soon grew to where I was doing chip-level troubleshooting on all parts of the system, and no longer doing any of the shipping.  I graduated from high school, turned down an almost certain appointment to The Naval Academy and instead went to Penn State for Electrical Engineering.  I continued working at Syntonic, and this is where I ran into a real crazy guy who went by the name Joe Pietz.  It was Joe who inspired me to explore advancing my digital electronics career into that of computers.  Joe actually bought one of the first Cromemco microcomputers in 1977 and kept it at Syntonic.  Cromemco was one of the companies trying to capitalize on the enormous success of the Altair, the first microcomputer available in kit form.  Joe wrote Z80 assembly language programs at night in the bar, that actually ran with little or no modifications the next day.  He let me use the computer, and, while I didn't know assembly language, I could write programs in Fortran.  By now, I had a lot of ideas for what I could use a computer for, but now I was really aching to have one of my own.  Unfortunately, I didn't have $5000 to pony up for one like crazy Joe did.

Then along came SD Sales with their Z80 starter kit.  This was a single board Z80-based microcomputer that had everything anyone would need to get their own computer running.  All parts were included for $275!  This was something I could afford, so I wrote the check and in a few days, I had my very own computer!  Parts, that is.  I have a lot of experience with electronics, Heathkits, and the like, so this was no problem.  I assembled the kit and had the computer running in a couple of days.  This was so cool at the time, that I didn't even mind learning Z80 machine language to program it.  Over the course of several years, this computer was modified several times and appears as you see it now in this photo.  You can still see the basic Z80 starter kit board buried in the enclosure.

As you can see, it still works, and I use it occasionally to program EPROMs.  But even though I had my own computer now, I was still wanting the capability of saving programs on disk, and interacting with the computer through a keyboard and video monitor, like I had on Joe Pietz's Cromemco.

In 1978 I started collecting the pieces of an S-100 computer system in the hopes of someday getting them all to work together.  I found a Z80 CPU board kit, which I assembled, and a used Wangco model 82 5 1/4" floppy drive.  I bought a 64K memory card with only 32K installed, and an Artec 12 slot motherboard and card cage.  Along with a homebuilt power supply, and other pieces scrounged from who knows where anymore, I managed to get the system running sometime in 1979.  Over the next 10 years, the system morphed from a mess of wires and boards kludged together to the nice system you see here (it's been pulled apart to diagnose and correct an intermittent boot problem):

This computer is still running remarkably well, although I have started to experience degradation of the floppy disks.  To combat that effect, I run disk tests every year or two on all the floppies, and I replace any that report errors even after reformatting.  Of course, all the files are indexed and I maintain at least three copies of every file on a combination of both 5 1/4" and 8" disks.  I hope to write more about this system in the future.  Anyway, back to the Mark-8.

For quite a few years now, I have been trying to find an Altair 8800 or 8800b, or an Imsai 8080.  I have seen these on occasion on Ebay, and they usually go for over $1000, but I keep looking.  But then one day I saw an ad for the PC Boards and rare IC's for the Mark-8.  I instantly recalled my high school days and the article from Radio-Electronics.  Was it still possible to build one of these?  From the looks of the Ebay ad, it seemed possible.  No, it wasn't an Altair, but I learned that the original Mark-8 is in the Smithsonian and there are perhaps only 5 or 6 Mark-8's now in operation in the world.  The challenge was made.  I decided I would set out to build one, and I purchased the "bare-bones" kit of parts from the Ebay auction.

 

Pre-construction

For about $150, here's what came from Obtronix via the Ebay auction:

Additional parts came from my fairly-well stocked junk-box, Unicorn Electronics, Jameco Electronics and Mouser Electronics.  Orders were placed on-line and arrived within 5 days.  Here's the pile of components with everything:

 

PC Boards

The PC boards look as if they were made by a commercial PC board manufacturer.  There are, however, a number of flaws in the PC boards.  Most of these are pointed out in the Mark-8 hints document.  Somewhere I read that these boards were manufactured from the original tape-ups of the Mark-8 boards.  It would have been nice if the flaws were corrected, though, before the new runs were made.  One of the most grievous flaws is the lack of pads on one side of the board or another, never on any of the IC's, but often missing on the passive components.  This made assembly of the boards a bit more difficult.

While studying the documentation, I learned that the six PC boards were originally interconnected with 41 bare wires strung from board to board along the bottom.  Once the wires were installed and soldered together, the boards were permanently connected.  This was a complaint of many Mark-8 owners, as it made troubleshooting difficult.  I decided to modify my Mark-8 to eliminate this issue.  I discovered that the pad pattern along the bottom of the boards was the same as the pattern along the top and sides where the Molex connectors go.  This meant that I could put similar connectors on the bottom of the the boards, then create a backplane that the boards would plug into, thus eliminating the permanent connection.  Boards could then be removed as needed.  I also saw this as an opportunity to eliminate the Memory bus along the top of the board, and consolidate and improve the switch bank connections to the Input Mux board and the Address Latch board.  These improvements are elaborated on in later sections.

Here is where I ran into another problem with the PC boards.  The holes for the Molex connectors are too small and the pins don't fit through, as you can see in this photo:

This is one of the Output ports, which was supposed to have the Molex connectors even in the original.  So out comes the drill press and the 0.060 bit.

After drilling, the connector fits nicely.

But now there's another problem - drilling the holes larger to accommodate the connector pins drills away the pad, too.  Initially, I thought this would be a major problem, but things work out okay in the end, as you will see later.

Here's a shot of all the PC boards, re-drilled and ready for construction.  It took about 3 hours to drill out all the connector holes, and add the holes for the PC modifications.

 

PC Assembly

Aside from the issues with the missing pads and the connector hole size, PC Assembly was straightforward.  I always start PC board assembly by inserting and soldering the shortest components first and ending with the largest/tallest/heaviest components last.  You can see in the photos that I used sockets for all IC's.  This adds to the cost, but I don't have desoldering capabilities for IC's.  You can also see evidence of some of the modifications.  The modifications are discussed in detail in the next section.  Here is a photo of the Mark-8 construction zone.

I typically use two different soldering irons for much of my electronic assembly: A small, 25W fine tip for small components and the component side of the Molex connectors, and a hotter, larger, chisel-shaped tip that speeds soldering of IC sockets and large components.  For very large point-to-point connections, I use my 100/140W soldering gun that's been in my tool box for over 30 years (longer than the Mark-8 has been around.)

As I mentioned earlier, I expected to have some problems with the Molex connectors, since I had to drill the holes larger.  I soon discovered that many of the connectors could not be soldered on the solder side of the board unless there were runs going to them.  But most connectors soldered rather well on the component side, using the fine-tipped soldering iron, as you can see in this photo:

As the boards were assembled, several tests were made to help ensure that they would work as expected.  First, a visual inspection was made of all IC socket soldering using a 8X magnifier.  Solder bridges can happen rather easily on boards like these that lack a solder mask.  Then I ran an electrical shorts/continuity check.  +5, 0V and -9V (on boards that use it) were checked to be sure they ran to all the proper locations.  The Logic Databook was used to find the power pins on the IC's, as the included schematics don't show these connections.  The power lines were also tested to be sure they don't run to places they are not supposed to run to.

The boards were constructed in no particular order.  All known PC board issues were accommodated.  The following photos show the completed boards.

Output Board

Input/Mux Board

Address Latch Board

Memory board (with 256 bytes installed)

CPU Board

LED Display Board

 

Modifications

A number of modifications were made to the Mark-8.  Listed here, in no particular order, are the modifications:

1.  The 41 bus wires at the bottom of the boards were replaced with a backplane.  All PC boards plug into the backplane.  To facilitate this, Molex connectors, of the type used for the IO connections and connections to the front panel switches, were soldered along the bottom of each board.  These connectors can be seen very clearly on the pictures above.  Details of the backplane are in a later section.

2.  The memory output bus, which traversed the top of the Memory, Input Mux and LED PC boards as an 8 wire harness, was replaced by 8 pins added to the 41 at the bottom of the card.  All three boards were modified similar to the photo of the Input Mux board shown here:

3.  The Input port connectors on the Input Mux board were increased to 10 pins each.  The two extra pins were used for a 0V ground reference (one per port) and an interrupt input (one per port).  A spare NAND gate in IC3 was used to combine the two interrupt inputs and feeds the EINT pin (pin 36) on the buss.  The EINT inputs include a noise suppression filter.  This was done in anticipation of doing something useful with the input ports.  The board may need additional modifications if both input ports and their interrupt inputs are used simultaneously.  These modifications can be seen in the photo above.

4.  The switch register or interrupt instruction port connector was also increased to 10 pins.  The two new pins are used to provide 0V and a current-limited 5V to the switch register.  This means that the switch register switch assembly can now be built as a simple assembly with one 10 wire cable coming from it and going to the 10 pin connector on the Input Mux board.

5.  In the original design, the address latch board had a single wire soldered to it called "write out" that then soldered to the memory board "R/W In" connection.  In designing a backplane oriented system, this signal had to be made through the buss, rather than from board to board.  Pin 17 on the buss was unused, so I routed this signal, which I named FPRW*, on this unused pin.  Both the Address latch board and the Memory board were modified for this signal.

6.  The address latch board has all the control register switch connections, except the JAM* signal, which routes as a single wire to the Input Mux board.  To simplify switch assembly wiring, I combined the switch connections to the Address latch board into two connectors, and brought the JAM* signal in here on an unused pin as well.  I also added the 0V ground reference signal to another  unused pin.  To get the JAM* signal to the Input Mux board where it is needed, I connected the line to another unused buss pin, pin 2.  Then I picked the signal off the buss from the same buss pin on the Input Mux board and routed it to the appropriate logic input.

7.  In the original design, only two of the four output ports had an output strobe readily available, and none of the ports were setup to provide the necessary 0V ground reference.  I increased all four 8 pin connectors to 10 pins and added the 0V grounds as well as buffered strobes.  The buffered strobes required the addition of one IC, but unfortunately, the photo of the Output board above was taken before the new IC was added.

8.  This is probably very trivial, but I substituted 74LS75 Quad Latch IC's for the 7475 IC's.  I did not have enough 7475's and when I tried to purchase them, the best price I could get was nearly $5.00 per chip.  (OUCH!)  Since the pinouts were the same, and they are used only as output ports, I figured the deviation could not cause too many problems.  In fact, it should save a little power as well as almost $30.00.

That's it for the modifications from the original design.  As you will see in the backplane section below, I included one spare slot.  This is in anticipation of constructing a non-volatile memory board for storing programs in either EPROM or battery-backed CMOS RAM.  I will publish the design of this board in these pages after it is completely operational.

 

Backplane

The backplane was designed to eliminate the need to permanently hand-wire all six boards together.  It has the added advantage that there is one spare slot for future additions, and it carries 10 additional signal lines: 8 memory data out lines, the FPWR* signal and the JAM* signal.  Adding 10 lines to the existing 41 (only 39 were actually used in the original) gives 49 lines.  I rounded this up to an even 50 and used the remaining line as an extra 0V ground reference line.  I designed the board using the software provided for free by PCB123.  While not as robust as PC design software I have used in the past, it does an adequate job and free sure beats $4000.00!  Here is a photo of the PC board as it arrived from the PC manufacturer.  All holes, except the mounting holes are plated-through.  There is no solder mask or silk-screen.  The material is regular double-sided 0.063 FR4.  The board is exactly 6.00 by 9.00 inches.

Included on the board is a two-pin connector (0.1" centers) for the front-panel power LED along with a current-limiting resistor.  This is run off the +5V power.  There are also provisions for connecting both the +5V and the -9V power.  For my Mark-8, I found two wall-transformer-type regulated switching power supplies, with outputs on standard 5.5 x 2.1 mm DC power jacks, with center pin positive.  The backplane is designed to accommodate two PC mounted DC power jacks, one for each supply.  Each supply input has a 50 uF filter capacitor.  For the future, I added a place to put a 4 position terminal block that could be used to provide power instead of the DC power jacks, or it could be used as a source for add-on circuits.  Finally, note that the buss pins 9 to 16 (using the original counting scheme) are broken between the CPU/InputMux boards and the rest of the buss.  This is a requirement for the original boards.  But if someone else were to use my backplane, and figured out a way around this quirk, there are pads provided to reconnect the broken buss lines.

Assembly of this board was boring (pun intended) with so many (350) pins to solder.  But during assembly, I discovered a problem.  When I laid out the board, I got the power input filter capacitors too close to the CPU board.  Fortunately, the DC power jacks did not interfere.  In the photo here you can see the clearance problem:

I resolved this problem by drilling another hole in the board for one of the capacitor's leads, then rotated the capacitor about 90 degrees, creating the necessary clearance.

One more problem I have encountered with the backplane, which is not fatal, but may tarnish the idea of the backplane: once the boards are installed, they are very difficult to remove.  The Molex connectors really clamp down on the headers of the backplane.  I may be overstating this problem, however, as I have yet to install the backplane in the enclosure with the backplane secured by all its mounting holes.  Perhaps then it won't be so hard to pull a board out.  I will update this problem as it evolves.  Anyway, here is a photo of the completed 50 pin Backplane:

 

Front Panel Assembly

Years ago I had ambitions to construct an Altair-type front panel for my S-100 computer described briefly in the background section above.  Plans called for about 20 miniature SPDT toggle switches and a handful of SPDT momentary action switches.  I picked these switches up somewhere and was saving them for the front panel project, but it never got off the ground.  So, when the Mark-8 project came along, I figured I could use those switches for this, even though they are not exactly authentic, and the momentary switches are actually pushbutton switches rather than toggle.

The front panel supplied with the kit was not drilled to accommodate individual switches.  From the documentation, it appears that some type of paddle switch was used originally, and they were attached to a sub-panel of some sort, which was then attached to the main front panel with some form of adhesive.  Since details were not clear, I had to devise my own mounting method.  Since I had already consolidated the switch wiring on the PC boards, it made creating individual switch subassemblies easy.  The Interrupt switch register is completely independent of the Control switch bank, and each subassembly connects to only one PC board.  I was at a loss for material to use for the subpanels; I had no metal that would look right and wood was out of the question.  In searching for some sort of plastic, I thought of the idea of using phenolic perf board.  At least it would work temporarily until I could come up with something a bit more permanent and more appealing..  Also, it had the advantage of being pre-drilled, which made layout and construction easier.  I used 28AWG flat ribbon cable to construct the wiring harnesses, which I terminated in Molex headers.  Since the headers aren't really made for wire-applied use, the wires are simply tack-soldered to the pins.  This might become a source of failure in the future.  Also note the use of cable clamps as strain-relief on the ribbon cables.  28AWG wire is too easy to break with a determined tug.  Details of the Front Panel Assembly can be seen in the next two photos:

In addition to the front panel switch assemblies, the main power switch and the power LED were wired as assemblies with connectors.  My commitment to the use of connectors increases cost, but it also ensures I can take it apart to fix it if the need arises.  The completed front panel assembly doesn't look too bad with the phenolic switch mounting panels.  Perhaps I won't have to replace them.  Labels for the switches would wait until after I got it working.

 

Enclosure Assembly

The original Mark-8 is shown in a metal enclosure.  Not having the proper metalworking tools, I immediately turned my focus to constructing a wooden enclosure.  I was significantly into the design of the wooden enclosure, and had already purchased material, when I discovered that the original enclosure was pre-fabricated.  However, I didn't check to see if it was still available, as I thought my idea for a wooden cabinet would make it look just as nice.

By this point in the construction, I knew how everything was going to mount - the backplane, the AC wiring components and the front panel.  Unfortunately, I still could not do any power-on testing because I could not get all the boards plugged into the backplane at one time.  Completion of the enclosure was a must.

A table saw or a radial arm saw would have made cutting the wood much easier.  Instead, I made the cuts as carefully as possible with a saber saw.  The results were only satisfactory, and if I had to do it again, I would have found a better way.  After cutting, I laid out the holes in the wood and drilled them.  Many of the holes had to be counter-sunk to accommodate the length screws I had, and to get the heads below the surface to prevent scratching other things this would site on.  The next two photos show the wooden pieces after drilling, but prior to staining.

After the pieces were stained, I waited 24 hours for the stain to dry.  It seemed like an eternity.  But then I started assembly.  First the sides were assembled to the bottom.  Then the front panel was assembled to the enclosure.  In the following photos, you can see how I re-used the wooden PC support rails that came with the Mark-8 kit as reinforcing bars for the sides.  The short piece in the middle will be explained later.

The second photo shows the addition of the AC power assembly.  Since the 5 and 9 volt power supplies are actually wall-transformer-type power supplies, it made sense to use the duplex receptacle.  I wired these to the power switch via the two-position in-line connector and an IEC-style AC power receptacle.  A standard IEC power cord will plug in here.  Later photos will show the addition of the third-wire ground from the power receptacle to the duplex box and to the front panel.

Now the short wooden strip:  I discovered that if I could limit the depth to which the Molex header pins traveled into the Molex receptacles on the PC boards, as the PC boards were inserted, I could reduce their grip on the header pins, thus solving my PC board extraction problem.  The wooden strips were just about the right height and I already had one on the left side.  I just had to add a short version on the right.  By adding or deleting washers under the backplane, I can control the backplane height in respect to the wooden strips, and thus attain optimum header to receptacle mating.  Too many washers - the boards go too far and are more difficult to remove; too few washers and the boards don't make good contact and they pop out.

 

These two photos show the completed enclosure construction, awaiting the PC boards.  Here the AC power cord is connected and the first power-on tests were performed to check proper power distribution on the backplane.  I also checked that there were no backplane shorts and the backplane mounting standoffs were still insulated.

 

Getting it Running

Next, the PC boards were inserted.  This was the first time I attempted to insert all six PC boards together.  Problem #1: These boards take some force to get them down onto the backplane headers.  I started thinking that maybe the backplane idea wasn't so good, but I got all of them inserted.  The boards became easier to insert and remove after several insertion/removal cycles over the next several days, so I'm back to enjoying the pluggable backplane.

Now is the time for the smoke-test.  I ran over all the tests I had performed in the past, looking for shorts, ensuring power was on the right pins, etc.  As recommended in the documentation, I removed the CPU chip for this first test.  Then I threw the power switch.  Uneventful.  A few of the front panel LEDs came on, but other than that, it seemed okay.  Okay until I tried loading various addresses into the address latch.  I could not load anything but zeros into A12, A13, CC0 and CC1.  Address increment via the Examine switch was working, so I loaded 1's into the 12 lower-order address, then hit Examine.  All 12 low order bits incremented to zeros, but A12 did not come up.  Troubleshooting the Address Latch board is going to be fun, I thought.  In my stack, it's behind the LED Display board, and the 74193 chips are at the backplane end of the board - nearly impossible to get a scope probe on.  But wait - these chips get their direct-load inputs from the buss, and output directly to the buss.  And I can see the chip LOAD* signal on IC7 pin 8.  In a few minutes with a logic probe, I confirmed that the input buss was correct and the LOAD* signal was coming out of IC7.  I pulled the board and confirmed with a meter that IC7 was indeed connected to IC11, pin 11, the chip load input.  That was enough evidence to convince me to change IC11.  Once again, I turned it on and ran the address load test.  But I got the same results!  Can I have another bad 74193?  This was my only spare (except for some 'LS193's.)  Next, I swapped IC10 and IC11; this time, the test showed problems with A8, A9, A10 and A11.  This is really strange behavior, so I decided to carefully check all the wiring on the board, solder connections, etc.  As I was doing this, I discovered that the Increment signal from IC5, pin 6 goes to IC8 pin 5; the cascade output of IC8, pin 12 goes to IC9 pin 5.  So good so far.  BUT, IC9 pin 12 DOES NOT go to IC10 pin 5, rather to IC11 pin 5, and IC11 pin 12 goes to IC 10 pin 5, contrary to the schematic.  A little more checking revealed that this increment line is correct, and IC11 actually drives the A8 - A11 lines and IC10 drives the A12, A13, CC0 and CC1 lines.  The schematic is wrong and I replaced the wrong chip!  Then when I swapped the chips, I swapped a good chip for another good chip, leaving the bad chip on the board.  In the end, I replaced IC10 with my spare 74193, and the address latch was fixed.  All bits now load 0's and 1's properly, and all four chips increment when the examine switch is pressed.

My next test was very successful.  I loaded the address 000 000 into the address latch, then wrote many combinations of bits into this address:  All 1's, all 0's, alternating 1's and 0's, etc.  This confirmed that the data bus, from the switch register, to the InputMux board, through the CPU board and out to the memory board, was good.  It also confirmed that the new memory out buss I created to replace the wiring harness was good, as this buss runs from the memory to the LED display board, displaying the output of the addressed memory cell on the MemData LEDs.  At this point, I also had confirmation that the Examine and Deposit function were working and the JAM switch was wired correctly.  At this point, I looked at the READY line, buss pin 8 and determined it was correct (low) for the RUN/STEP switch in the STEP position. Also, I could see a narrow positive pulse on the READY line when I pressed the Single Step button.

Next I inserted the CPU chip and turned the computer on.  I entered the simplest of all programs, the one everyone enters first to check their Mark-8 computers:  @000 000  JMP 000 000.  I verified the program, set 005 in the Interrupt register, pressed the Interrupt button, moved the JAM switch to the INT position, and pressed STEP.  Nothing happened.  At this point, I wasn't entirely sure of the start procedure.  I tried it again.  Still nothing.  I wasn't sure what I should see happening, either.  I tried again.  This was getting old.  I started asking questions about what should be happening.  I got out my oscilloscope - a 35 year old 10MHz single-trace Heathkit I built as a kid - and started poking around.  Fortunately, my CPU board, like most other Mark-8's I have seen, is at the very back of the stack and all IC pins are accessible.  Along with my logic probe, I began verifying all the static logic levels on the CPU board, where I assumed there was a problem.  As many of the signals as possible, I traced them out to a buss pin.  I was hoping to find a bad level somewhere, but nothing was found.  I deduced the purpose of the CC0 and CC1 bits, figured out the logic behind the SL0 and SL1 signals, and determined how the A12 and A13 lines help control I/O functions.  Then I made a discovery: I don't remember exactly how I got here, but I found a broken solder connection on the SL1 line where it solders to the backplane connector.  To the naked eye, it looked good, but my meter told me otherwise.  With my 8X magnifier, I could see the hairline crack.  After repairing the connection, I re-entered the test program and tried running it again.  Again, nothing happened!  Surely this was it.  But it wasn't.

Additional troubleshooting was futile.  I was now to the point where my old scope just wasn't capable enough.  I needed a dual-trace scope so I could start comparing signals and looking at the relative timing of the various signals.  I could get the CPU to run, at least I thought it was running, because I could see repetitive signals on the output of IC26, the state decoder, when I moved the RUN/STEP switch to RUN.  But I could not discern if the signals were correct, and I didn't know which ones were supposed to do what.  To my good fortune, I was discussing this dilemma with a good friend of mine, Dave Costanza.  He said he had a dual trace scope I could borrow.  (The scope you see in the following photos is Dave's.)  This scope turned out to be one of two critical items.  Now besides the dual-trace scope, I needed the 8008 data sheet as well.  By this time, I was convinced there was something wrong in the state decoder area, or the logic following it, but I could not see it.  After several hours of searching Ebay and the Internet, I could not find the 8008 manual or data sheet.  Then I recalled an earlier dialog I had with Bryan Blackburn where he said he had a bunch of 8008 and Mark-8 stuff collected into a CD-ROM.  I inquired, and luckily, one of the documents he had on the CD-ROM was the 8008 User's Manual.  Bryan requests $20.00 for a copy of the CD-ROM, and I made quick payment via Paypal.  In a few days, the CD-ROM arrived by mail.  The first thing I did was print out the 8008 manual - it had exactly what I needed - the states that correspond to the S0, S1 and S2 output pins on the CPU.  Armed with the dual-trace scope and new knowledge of the state decoder and logic, I started troubleshooting again.  I have been at this for nearly two weeks now.  Here's a photo of the troubleshooting in progress:

The scope, channel 1 (on top) shows the T1* signal from IC26, pin 6.  It is apparent that the CPU is running, in fact, it is repeating a 5T cycle over and over.  So, I should see a T2 state, and also T3, T4 and T5 state.  Here are two photos, one showing the probe connections and the other showing the same T1 trace, plus the T2 signal (bottom trace) from IC26 pin 4.

Next I moved the channel 2 probe to IC26, pin 7.  But what's this?  Where's the T3 state?  Surely it cannot be this glitch?

Next I checked T4 and T5.  Here are photos of their signals:

I also looked at IC26, pin 9, and surprise, surprise - it looked like what I expected to see for the T3 state!  So, I deduced that my processor was looping through T1, T2, TWAIT, T4, T5 instead of T1, T2, T3, T4 and T5.  This would quite correctly hose the CPU board logic.  But what could cause this?  I looked at the RDY line - it was rock solid high.  Perhaps the INT line?  No way - it was low and not moving.  Could it be a bad CPU chip?  Many years ago, I acquired a i8008-1 CPU chip.  I no longer remember where I got it, how much it cost, or even if it was good.  I had nothing to loose for trying, so out came the "kit" CPU chip and in went my "souvenir" CPU chip:

 

Happy Anniversary Mark-8!

That was it!  My Mark-8 is alive!  Happy Birthday & Happy 30th Anniversary!

Both CPU single-step and CPU full-speed operation checked out fine.  But there's one more twist.  I entered the register increment and write to port 0 program from the Radio-Electronics article, but it did not work.  The program seemed to be running, the 7475 latches on the Display board were being strobed, but nothing was coming up on the Port 0 LEDs.  Immediately I suspected the 7475's - they were two very old chips from my stash and the marking were all but worn off.  Perhaps they weren't 7475's after all?  Perhaps they were just bad?  I replaced them with two spare 74LS75 chips (remember, I wouldn't buy 7475's for nearly $5.00 each) and the problem was fixed.  Here is a photo of my Mark-8 happily running the register increment test program:

The rest was relatively easy.  I applied labels to the front panel and rubber feet to the bottom, then screwed the top on.

PROJECT COMPLETE!

 

What's Next?

This is a list of possibilities for my Mark-8:

Plus, I'll be turning my attention back to my S-100 system again.  While I was buying parts for the Mark-8, I also bought a large assortment of replacement parts (mostly IC's) for the S-100.  It would be a nice project to test all them.  Then I have two spare 8" drives that are supposed to be bad - I'll check them out, perhaps I can fix them.  Also, I want to rewrite my monitor program and BIOS to permit two terminals - the built-in terminal (actually on a plug-in card) and a regular serial terminal (I have an HP 700/92).

 

Links

Electronics Parts Companies I used:

www.jameco.com
www.mouser.com
www.unicornelectronics.com
Obtronix@yahoo.com (No web site - see Acknowledgements)

 

Acknowledgements

1.  God and Country (no explanations needed)
2.  My wife Barb, and my kids, Rob and Elyse, for tolerating my seemingly endless hours in the "dungeon"
3.  Steve G. of Obtronix  for providing the bare-bones kit to get started (This was found on Ebay).  Steve also replaced the bad 8008 CPU - Thanks, Steve!
4.  Bryan Blackburn for helping pull together the documentation, and making the CD full of info available
5.  Dave Costanza for loaning me his oscilloscope