More holiday downtime

New Years guests mean everything back in storage for a couple of days.

Testing of disk tool on Alto system Diablo drive highlights remaining issues

ALTO DISK TOOL

I brought the new board, fpga and related objects over to the Alto, hooked it up to the Diablo on that system and read some cartridges for archival purposes. Unfortunately, I was still seeing sectors with errors.

Something is going wrong, leading to errors which we are not experiencing when the Alto reads the same cartridges. Unless the Alto is ignoring errors and never is affected by any incorrectly read sectors, there is something I can improve in my fpga logic.

We did boot up all the other cartridges we had, which all ran well, but that was with the Alto driving the Diablo. The original disk cartridge we received had been cleared by a utility that overwrote the media with a pseudo-random stream of words. This seemed ideal to use to write an archived disk image which had many Alto games.

This was the first time we tried to use the WriteEntireCartridge transaction, which appeared to work. However, when we tried to boot the cartridge on the Alto, we heard it seek and read from two locations but then stop. Clearly the written data was not done well enough to work properly.

My wishful thinking yesterday, that the surface scratching on the cartridge at my house was the sole reason for checksum errors while reading, is proven wrong. I have to keep working until my tool reads my cartridge entirely correctly (and writes sectors properly). 

Validated new board and connectors, ready to be used on the Alto system to archive disk images tomorrow

ALTO DISK TOOL

I set up the testbed and began to check the functionality and signal integrity of the new board and connectors. Seeking was checked to be sure all the address bits were working - but in fact the 4 and 16 lines were not working. I found that the chip pins involved were not well soldered, fixed it and went back to test.

In addition, I have a Digilent Nexys2 FPGA board that was purchased by Al Kossow in order to give him a working disk driver and emulator. It had connectivity problems in the USB connector, with the connector itself bent and some other problems where it was attached. I used my heat gun to remove the connector, bought replacements and will surface mount the new part when it arrives.

The disk drive testing resumed and my seek addresses were all working correctly. Next up is getting the ReadGate and other signals tested. I am emitting '1' to ReadGate which should pull the output of chip U2 side 1 down to ground, thus activating pin 39 on the ribbon cable to signal pin E.

However, when I monitor pin E on the terminator, the line is not going down to ground. I have continuity from the fpga connector pin to chip U2, from the chip output to pin 39 on the connector, and all the way through the cable and drive to the terminator.

I might have a flaw with chip U2 such that it is not working, even though all the connections to and from it are good. I have to do a power-on test and verify the voltage levels at the chip. Watching the scope, I could see that the signal at the terminator changes as I probe the chip pins, indicating I probably have a poor solder contact there.

I decided to remove the chip entirely with my hot air rework gun, then resolder it properly. With that done, I verified good solid behavior, including the Seek, ReadSector and ReadEntireCartridge transactions. I uploaded the contents of the pack, along with the checksum validation array, so that I could check that the data was extracted properly.

The results were just as before, so I now have the production level board tested and ready to use tomorrow on the drive attached to the Alto and the other cartridges we have there.

As I packed up everything for transport, I decided to inspect the disk cartridge surface and the heads. I can see scraping and discoloration at the low cylinder numbers near where I am experiencing almost all of the checksum errors reading that cartridge.

Surface marking that is probably the cause of my errors reading some sectors

Surface marking on the other side of the platter

I will tentatively ascribe the errors to the surface condition, thus hope that my tool works well with undamaged cartridges on the other Diablo drive, the one that came with the Alto. We will know more tomorrow after we read and upload all the content. 

Holiday downtime

Due to the holidays, my test setup is stored away and nothing can be done on the project. Back soon!

Changed emulator board design and sent to fab

ALTO DISK TOOL

My early testing with the Emulator role PCB shows that it is not going to work properly. The signals going from the fpga to the Alto require pull-up and pull-down resistors to give the ribbon cable termination. This should have been placed on the Alto side, but they were not, instead terminating the signals that ran from Diablo to Alto by resistors on the drive side.

I need to expand the PCB to add a pair of resistors per outgoing signal, a total of 22 added resistors. Good thing I only ordered one board for $66, of course with shipping and tax on top, as it is wasted. Time to get to the designing (and to order more components).

I had the 1401 restoration team meeting yesterday and spent quite a bit of time chasing down a flaw that resulted in card reader error stops. Greatly improved now, so that programs which use the read then print instruction can get through hundreds of cards before stopping. It had been failing on the third card pretty regularly.

Thursday morning I changed the emulator card design to add in the termination resistors and decided to send out for a new card through the same fast turnaround service I used previously. Another $100 and should be about two weeks to receive it.

Construction progress on PCBs and cabling, plus more testing

ALTO DISK TOOL

Parts are on order to complete the remaining driver and emulator boards - shipped today but not likely to get here before Xmas due to the heavy volume at all the delivery services. I did put on some of the components I still had - capacitors, IDC connectors and driver chips, for instance - before putting the boards aside.

Testing out the new driver board involved checking a number of items, as listed in yesterday's post. Step 1 was already complete.  

1. Verify that SelectUnit1 is switchable and not permanently on from the fpga
2. Verify that the terminator has the pull up and pull down resistors installed for all the signal lines I am driving, but none on the lines returning from the Diablo to my fpga.
3. Monitor signals incoming to the PCB to validate their correctness
4. Monitor signals at the terminator when my PCB is driving the Diablo to validate their correctness.

Steps 2 and 3 were relatively straightforward, as there is easy access to the pins needed to monitor those circuits. Step 4 is more challenging and will require hooking on micrograbbers and multiple cycles of removing and installing the terminator. 

I found several of my incoming signals had termination on the sending end, as well as on my board. I clipped off the resistors on those signals. Everything else checked on with termination resistors, allowing me to complete step 2 above.

Step 3 will be done by a simple power up and attaching the scope probes to the tops of the three level shifter mini-boards across the top of my PCB. 

I took some time to put the IDC connector on the ribbon cable that attaches to the Alto disk controller card on its other end. This cable is now attached to the emulator board. 

Powered on new driver PCB and tested its correct functioning, also mostly build emulator role board

ALTO DISK TOOL

Today I worked on the connector from the ATX PC power supply to the driver board of the disk tool, or alternatively a direct wired cable from the power supply to the board. I went with a modification of the cable I had been connecting via wire nuts, but soldered everything to a 4 pin connector. It has an indicator light, the load power resistor and the connector to fit to my PCB.

Power is good, so I proceeded to power on testing. I had mixed results from the first test, other than it passed the magic smoke and hot component tests. I could set up and accomplish seeks, although this worked with or without the SelectUnit1 line activated which is NOT what should happen.

The drive should ignore all signals unless SelectUnit1 is pulled to ground and while it is ignoring incoming signals, it should not drive any return signals to the PCB. Instead, I see status, sector marks, sector numbers and other signals driven regardless of my Select slide switch. 

Seeks worked, but no response to a ReadSector or ReadEntireCartridge transaction. Time to do a few things before I proceed:

1. Verify that SelectUnit1 is switchable and not permanently on from the fpga
2. Verify that the terminator has the pull up and pull down resistors installed for all the signal lines I am driving, but none on the lines returning from the Diablo to my fpga.
3. Monitor signals incoming to the PCB to validate their correctness
4. Monitor signals at the terminator when my PCB is driving the Diablo to validate their correctness.

Step 1 yielded the insight that I had changed the logic in the fpga to hold SelectUnit1 on, thus the behavior was what should be expected. Eventually, this will be switched off except for the duration of transactions, but during testing it is easier if I can observe ReadData, ReadClock, SectorMark and other signals all the time.

I began looking at the terminator resistors and cross checking it all to the schematics and design of my board. I didn't complete this before my shipment of the emulator board PCB blank arrived.

In addition, my emulator role PCB arrived today and I began to assemble components on it. It is time consuming due to the small size of the parts. The resistors and capacitors are size 0805, one of which is held in tweezer tips in the photo below, on lined writing pad paper with one line visible for scale.

0805 size component on lined paper, in tweezer tips

I ran out of the header strips I need to mount the last two components, each a level shifting mini-board that connects 3.3V and 5V sides for four circuits each. Once these come in, I can test out the wiring of the board prior to firing it up to test the disk emulator role.

As you can see, the disk driver and disk emulator boards are fairly similar in overall appearance, but are quite different in details. They are electronically keyed, using pins 20 and 21 of the fpga, so that when the board is connected, the fpga logic knows whether it is operating in emulator or driver mode by which of the pins are grounded.

disk emulator role board, assembled but for two level shifter miniboards