• Please review our updated Terms and Rules here

PDP-9 at the RICM

I fixed the problem in the core memory where it was picking up bit-7 in the upper 4k of the 8k stack. There were two sense amplifier flipchips that were misbehaving. The processor passes instruction test #1, but not #2. It is failing in a section of instruction test #2 that I don't fully understand. I think that the failure is actually about 10 instructions past where it halts. I think that it does the wrong thing but does not halt, and on the second pass it halts because of something that failed on the first pass. Debugging that is the project for Saturday.
 
After studying the source code for Instruction Test #2 I found that many sections of the tests depend on subsequent tests to clear memory locations. In our case there was a dropped bit on a DZM instruction so it didn't clear the right memory location. After I fixed the instruction the diagnostic ran Ok for more than 30 minutes.

I tried to boot the many versions of the ADSS operating system that we have on DECtape. Some would display the header and prompt, but would not respond to keyboard commands. I was many Parity and Timing errors when booting, so maybe the TC02 DECtape controller of the TU55 DECtape drives need some attention.
 
I ran diags on the TC02 DECtape controller. Most of the functions work OK, but it looks like the END bit in Status Register B is not being transferred to the accumulator. That should not be difficult to fix.
 
I tried formatting a DECtape on the PDP-9. The tape moved just a few inches and stopped with the Timing error light on. I will study the Maintenance manual and schematics this week to see what turns the TIM light on.
 
I found that you can't have the WTMK switch on when you start the DECtape formatting program, you have to turn it on when it asks. It wrote one pass of the DECtape and ran off the end of the reel. I investigated the clock that controls the write timing. If the clock is too slow the image of the DECtape will not fit on the physical tape. The clock is only enabled when the formatting program is running, so I got the 'scope ready to look at the clock signal when it is enabled. Now the serial console port has a stuck bit and won't accept a "Y" to start formatting. I will fix that Wednesday.
 
I am back to working on the PDP-9. The DECtape diagnostic ran for about 30 minutes without errors, and then said that it was looking for block 1101 which doesn't exist.

The issue with wrong block counts has nothing to do with the TC02 DECtape controller or the TU55 drives. It looks like the DECtape diagnostic is getting corrupted when it is running. The instructions that copy the literal value for the number of blocks (1100) has some extra bits so the diagnostic loads and instruction instead of a literal and uses the wrong number of blocks.

Memory location 221 should contain 506471 and actually contained 506671. Memory location 223 should contain 206472 and actually contained 206672. Memory location 221 and 223 both picked up bit 10. I fixed the contents of 221 & 223, reran the diagnostic, and the picked up bits returned. So it looks like we have another memory problem to fix.
 
We tried the DECtape diagnostics with the PDP-9 just powered on. It ran for about 20 and then went into space. We tried the Instruction Test #1, but it didn't load correctly. We tried the Basic Memory Checkerboard, and it didn't run correctly. It halted at an address that was not within the diagnostic.

Saturday we will check the core memory voltages and timing adjustments.
 
I am making some progress on the PDP-9 at the RICM. After adjusting lots of voltages and delays in the core memory subsystem it will run the Memory Address Test, but only with the processor running at about 10% of the normal speed. I ran the diagnostic for several hours at the lower speed without errors. Afterwards it ran the diagnostic at full processor speed, about 1 MHz. So some adjustment is not happy when the core memory is cold, and is changing when the core memory warms up. The PDP-9 core memory design is very different from the PDP-8. It doesn't have a thermister in the core stack to adjust the core memory voltage. It should be loads of fun to fiddle with all of the memory adjustments to get the core memory to work both when cold and warm.
 
It boots ADSS again!

We had problems loading diagnostics from paper tape. The paper tape format is 6 bits per stripe that are assembled into an 18-bit word by the paper tape controller and then transferred to core. Most times we would see the low-order 6 bits from one word would be in the high-order 6 bits of the next word. Our guess was that it was missing or getting an extra strobe from the paper tape feed hole. We found that the +10VDC was missing from the little card cage in the paper tape reader. After studying the schematics we found that the +10VDC and -15VDC are supplied through the data cable from the I/O controller chassis. Margin switch #3 and its associated fuses supply the power to many of the I/O cables, and the +10VDC fuse was blown. After replacing the fuse we were able to successfully run the core memory and instruction diagnostics. I imagine that the LINC-8, and maybe the straight-8, have the same paper tape reader and get the power the same way.

We tried booting ADSS from DECtape using the paper tape bootloader. We got lots of parity and other DECtape errors. We tried several different DECtapes, and found one that booted on the first try. So the RICM's PDP-9 is running again after lots of debugging and repairs. I need to reformat the DECtapes on the PDP-9, and then write the DECtape contents using my PDP-8/e so we have more bootable versions of ADSS.

We still need to fix several G850 motor controller FlipChips so we can get the third TU55 DECtape working. If you run ADSS on an 8k PDP-9 you can't have dynamically assigned I/O devices for the compiler so you need 3c DECtapes to compile and run programs. I have spare SCRs and Unijunction transistors for the G850, but so far have been unable to diagnose and repair one.

We also need to rewind a sensor coil for the crankshaft of the paper tape punch, and rebuild the second power supply that feeds +30VDC to the paper tape punch.

I have been collecting the FlipChips for the 34H graphics option so we can play Spacewar! on the PDP-9. I think that I have everything except for the W681 Scope Intensifier FlipChip. If I can't find a W681 I will make one. We also need to add the EAE option to run Spacewar!. We should have all of the necessary Flipchips to do that.

A long range goal is to get UNIX V0 running on this machine. It needs EAE for the filesystem, so we will need that. We could make a disk emulator like the LCM did for their PDP-7, and write a new driver for the emulated disk. I have also been thinking of writing a system driver for the TC02 DECtape controller and using that as the system device. It would not be nearly as fast as an emulated disk, but it would need special I/O hardware either. Comments?
 
You must log in or register to reply here.
Back
Top