Hardware part done

 The hardware part of the project is more or less finished (one knob needs reattaching in a way that will make it stay on, and one lamp is missing). The picture on the left shows the mounting plate for the new electronics package, the right picture shows the electronics installed inside the panel (and running). 


An impression of the end result is shown below. The VT100 may become the console terminal for the IBM (yes, it's as bad a mismatch as any, but I don't have any IBM terminals). The laptop below the panel is controlling it. All the lamps, switches and other controls now work flawlessly.
My apologies for the quality of the picture, it's dark here now, and a flash wouldn't bring out the beauty of the lit lamps.

Of course, the panel isn't capable of anything useful until the software has been written. More on this in my next post.

FPGA implementation of model 30

I just found out that Lawrence Wilkinson is working on an FPGA implementation of a 360 model 30 (yes, that would make it an implementation of an implementation of an architecture, not unlike the software implementation of the model 65 that I'd like to create if I can find the right documentation). For the FPGA implementation go here. From what I've seen sofar, he's come amazingly far already.

Electronics v2.0 complete

 I've completed the board with the new electronics package. If you compare it to the prototype (shown below), the new board looks much neater, and will more easily fit inside the panel, leaving enough space for the cables.

The next job is to mount the new board inside the panel.

Board shipped, looking for docs

The PCB has shipped. It should arrive tomorrow or Friday, after which I'll have some work to do again.

In order to have a faithful emulation, I'll need to write an emulator that implements the microarchitecture of the Model 65. To this end, I'll need some documentation. I've posted the following in the IBM Historic Computing forum on LinkedIn:

A friend of mine who was a software engineer at Digital Equipment Corporation gave me a console panel for a System/360 model 65 that they had in their office (the console panel, not the system).

I am restoring the console panel, and I'm creating an interface to a PC to make all the lights and switches work as originally intended.

I'm now looking for field engineering documentation in order to make the emulation as faithful as possible to the original. These are some of the document titles I'm looking for:

System/360 Model 65, 2065 Processing Unit, Field Engineering Maintenance Manual
System/360 Model 65, 2065 Processing Unit, Field Engineering Diagrams Manual

System/360 Model 65 Field Engineering Manual of Instruction
System/360 Model 65 Field Engineering Theory of Operation Manual
System/360 Model 65 CPU and Channels Training, Field Engineering Education Supplementary Course Material

The same material for the model 60, 62 or 67 in stead of the 65 would help too, given the similarities between these machines.

One thing in particular that I'm looking for that would help a lot is a data flow diagram showing the register and bus names. Another thing that would be very useful is a listing of the BCROS microcode.

The functional characteristics manual doesn't offer the level of detail I'd like.

PCB nearly finished

I just received a new set of pictures of the PCB manufacturing. The board is nearly finished now; the copper pads have been tinned, and the silkscreen printing has been applied:

PCB progress...

I have received some more pictures of the PCB board being made:


After etching

After applying the solder mask

 

This is where the first problem with the PCB's shows: the oval pads for the power connectors and connectors to the USB interface look ok on top of the board (they're on the left side of the board in the pictures above), but on the bottom side of the board (where they will be soldered) there's too much solder mask, so there's only a very small rim to solder to. This means I'll have to scrape off some of the solder mask. Fortunately, there aren't many pads like this. I'm not sure where it went wrong, but it's too late to do anything about it.

PCB manufacturing in progress

I just received photos of the PCB after the exposure step; all the holes have been drilled, and the PCB traces are visible:

PCB ordered

The PCB design has been finalized and sent off to a manufacturing company. It could take up to 8 working days to finish. A small setback was that the board as I originally designed it was too large for them, so I had to reduce the length to 43 cm. Of course they only tell you that the dimensions are too large when you try to actually place the order, not when you do the cost calculation on their website.

PCB design

Debugging the veroboards with the pile of wires between them is proving challenging. I've therefore decided to let this be a prototype, and will re-implement the circuit on a printed circuit board. I've redrawn the schematics in kicad, including the actual connectors this time. The autorouter is now very busy getting it all to neatly fit a single 14 by 58 cm pcb. This will make for a much cleaner and thinner electronics package that will fit the panel without sticking out in the back.

Christmas decoration

Even though it's not entirely working yet, at least we have some decoration just in time for Christmas! Some of the lamps don't light while others light all the time. Time for more troubleshooting again...
Some of these problems come in groups: the first 16 lamps on output # 6 don't light up, the first 16 lamps on output # 7 light up all the time, and the first 24 lamps on output # 8 don't light up. These are the last 2, 2 and 3 chips on those outputs, so I'm sure I messed up output board number 2 in a somewhat consistent manner (it almost has to be the connection from one shift register to the next). 172 out of 235 lamps are working perfectly.
Not all switches are completely working either. Five pairs are ganged together (shorted somewhere), and one plainly doesn't work at all.

Electronics OK

The electronics package now works correctly. The inputs didn't work yet, but I found out that the latch signal was active low. I changed the programming again, and that now works too. The outputs still work with the altered programming, because the output latches are edge triggered.
As you may or may not be able to see in the picture above, The electronics are mounted in two layers of pcb's on an aluminium panel that fits behind the blank panel of the IBM console. The pcb's with the connectors are mounted on a second aluminium panel. I now need to sandwich the two aluminium panels together. To this end, I need some long helical ridges wrapped around steel cylinders, which necessitates making a trip to the hardware store to get some (bolts, that is).

Electronics package complete

The electronics package is now complete. So, after an initial inspection for obvious problems (no shorts between +5v and ground for instance), it's time to try it out.

I put 18 LEDs on a board along with a flat cable with an IDC connector. I wrote a simple program that would flash all outputs on and off in a checkers pattern. I hooked it all up and... nothing...

So, that means time for troubleshooting. The first problem I found was a short between the CLOCK signal and the +5v on one of the input boards. A tiny amount of solder bridged two tracks on the veroboard. Then I found that the outputs on the Velleman kit needed pull-up resistors. I added those, found out that the outputs were inverting, changed the test program accordingly, and... still no success... Then I found out that I had crossed the outputs to the CLOCK and LATCH signals. I changed the test program to swap those two signals, and voila, the lights started flashing. Next up is testing and if necessary troubleshooting the inputs.

Btw, who else thinks logic analyzers are a godsend?

Why did I pick this hobby?

Over half the inter-board connections have been made now. 846 soldering points to be precise. Only 498 to go... Remind me why I picked this hobby again?

Topbar design

The 360 console panels had a topbar featuring the IBM logo (not striped yet) and "system 360" logo. Unfortunately, it is missing. A system 370 topbar came with the panel, but it doesn't fit. I recreated the design from some photos, so now all I need to do is find an outfit that can print it on aluminium.

Boards complete

The electronics boards have been completed. The next tedious task will be wiring the eight boards together. Some 700 wires will need to be connected between boards.

Panels rewired

All the anels that need rewiring have been rewired with ribbon cables. The wiring for the main panel was left intact' as it was already wired with ribbon cable, so I adopted the existing wiring into my design, and simply put IDC crimp connectors on the end of those cables.

Schematics

Here are the schematics for the PC interface. As you can see, two of the digital outputs from the K8061 are used to control the shift registers, and the remaining 6 outputs each drive 5 8-bit shift registers for a total of 240 digital outputs.
The 8 digital inputs are each connected to 3 8-bit shift register to record a total of 192 inputs. Five analog inputs are used to record the setting of the five potentiometers, and one analog output is used to drive the voltmeter.

The circuits will be realized on 8 pieces of Veroboard (eurocard size): two for the input shift registers, two for the output shift registers, one for the 220 ohm LED series resistors (using resistor packs), and three for the ribbon cable IDC connectors. The 10K pull-up resistors for the switches are placed on the connector boards.

Cleaned up

All of the panels have been cleaned up. The panel with the dent behind the potentiometer has been straightened out.

The first part of my order of electronic components came in yesterday. I put together the interface kit and it's functioning correctly.

As a test, I've replaced three of the bulbs in the plug-in lamps with orange LEDs. The result looks pretty good. Three done, only 234 to go...

Disassembly

After disassembly, what's left is an empty frame. Unfortunately, all of the panels except for the main panel shown below (register displays and toggle switches) was wired into a single cable harness. To get the panels out, I had to cut the harness, but then, I was going to get rid of it anyway.

I cleaned all small detachable parts (knobs, button faces) by soaking them in mild laundry detergent overnight. Remember how much tape residue was on the buttons? After cleaning them, they're readable again.