HP97 The journey begins

[Harald]

Good to see you are making progress Tony!

Cheers,
Harald

[ThomasF]

Following your progress Tony with great interest!
Hope it all works out great!

I don't own a 97, but soon I hope to ...

Cheers,
Thomas

[teenix]

Hi all,

More progress.

A bug appeared shortly after last post which was a bit weird where the display would start flickering with a jumbled mix of numbers for no reason that I could see. This is one of those moments where the shoulders slump while a myriad of possible problems float through the brain.

I thought the processor was resetting for some reason and tried to nail down if it was my code causing the problem, but all seemed ok.

I enabled some special test code, that is in all of the CPU modules, which sends the HP microcode program counter value and any internal registers that I need back to the PC so I can see actual HP program flow. This enabled me to see that the HP microcode thought the RUN/PRGM switch was being switched back and forth really quickly, which mechanically is impossible.

While playing around, I realized that just moving my hand near the switch caused the problem but only in RUN mode. In PRGM mode, the input pin is grounded, so the only sensible reason for this is a floating input to the processor causing it to oscillate, probably at mains frequency while my hand was nearby. I traced the circuit to a resistor that holds this pin in a Logic 1 state until the PRGM switch is selected an lo and behold, I forgot to solder one of the pins. Lots of small parts on this board, so easy to do I suppose.

So far the calculator seems to be working ok. I have tested the program storage and the Continuous Memory. I entered a small program the other day and is still in memory when switched on. I added the clock chip and that seems to be working as well.

I had to make a calculation for a software delay I had to implement in my code, so for its own internals, the 97 did its first job

cheers

Tony

[teenix]

Hi all,

It is getting time to test the card reader, but when I went to insert the ribbon cable that connects the reader board to the main CPU board it looks like age has intervened and some of the tracks have cracked and lifted from the plastic and it looks non repairable.

I cannot find replacements of the same type, and similar types are rather expensive and then there's postage to add, so I made my own which should do for awhile.

cheers

Tony

[teenix]

Hi all,

I successfully wrote to a card and then read it back this morning so it seems the CRC simulator code I have for the 67 works ok in the 97 :-)

It is a bit difficult to test the 97 due to the way it is constructed, but getting there.

The next stage will be adding the separate PIK processor and motor controller circuits. This will be interesting and will have to be enabled bit by bit with the print head disconnected until I know for sure the two processors are working properly together.

I have to minimize the printer handling because I don't want to damage the head ribbon. The new 97 CPU board has an easy release socket that connects to the printer ribbon socket so I don't have to keep pulling the ribbon out every time I need to separate the printer.

cheers

Tony

[teenix]

Hi all,

A while back I posted about a method to repair the coupler on the card readers.

I have repaired a few now and they are still working.

Not having access to a lathe, one of the problems was drilling an accurate hole in one end to fit the worm gear shaft into. If it is not reasonably accurate, the shaft will vibrate in use. I found a pretty good way to do this with just a simple drill press. It sounds a bit long winded but doesn't take much time in practice. Once set up you can make 3 or 4 in a few minutes.

Fit a 5/64 drill bit in the chuck.
Clamp a piece of 10-12mm thick (or so) timber on the drill press table. This must remain fixed in place for the entire process.
Adjust the table so that the lower part of the chuck is about 30mm above the wood and the hole in the table is centered under the drill bit.
Drill the 5/64 hole straight through the wood.
You might need to move the rotating drill up and down a few times to make sure the hole is clean.
Turn off the drill and keep opening the chuck until the drill can be removed without upsetting the timber.

It will be a tight fit, but push the drill bit up through the hole from the bottom so that the cutting tip protrudes above the top face of the timber by about 20mm. The drill will be rigid for this exposed length. Dont tap the drill upwards, push on the end of the drill with the side of some pliers. Your finger tip will be too soft. (and it will hurt) If the table or wood moves, even a tiny bit, you will have to start again.

Remove the sheath and inner copper wire from a 50mm length of the RG58 coax . Place the translucent inner core into the chuck until the end is flush with the chuck jaws and close the chuck just finger tight so the cable does not crush or distort the inner hole.
Start the drill and wind the press down over the exposed drill bit and bore a hole about 5mm into the core.
You should find that the hole is nicely centered in the coax.

Using a box cutter blade or similar, trim the drilled hole end to make sure the worm gear end reaches into the end of the hole that was drilled.
Trim the other end so that the coupler is the same length as the original metal sleeve.

Without glue assemble the worm gear onto the coupler.
Push the coupler all the way onto the motor shaft. Afterwards, you should find that the shaft will have a bit of sideways play - this is ok.
Refit the motor and "lightly" tighten the 2 screws. If your coupler is too long it will force the motor shaft sideways - not good.

The shaft will probably appear tight, and is because the worm shaft and coupler need to be pushed slightly towards the motor. This can be done by gently pushing on the shaft threads while in situ. Push just enough so that there is a small back and forth movement detected. If this cannot be done, the coupler was trimmed a bit too long. Disassemble and trim a bit more.

If good, disassemble the motor again and remove the coupler from the motor shaft and worm gear.

For final assembly, I used a small amount of super glue on the worm gear end to secure it to the coupler. Not too much or it might dribble into the inner hole in the worm gear or out over the side of the coupler. When this has been applied and immediately after pushing the coupler on, put it on a flat surface and roll it back and forth using the length of a finger only with light pressure. This should make sure it is all aligned properly. Leave it sit for an hour or so - hopefully not stuck to the surface.

Refit the coupler and worm gear to the motor shaft as mentioned before. It will appear loose again but is ok.
Refit the motor and tighten 2 the screws.

The shaft may now appear tight again, so move the coupler towards the motor as described above. The sideways movement should be there but not too much.
Reassemble the calculator.
Maybe leave it aside for some time to allow the glue to cure properly.

Job done, unless you want some spares.

cheers

Tony

[rprosperi]

Tony,

Your message got cut-off and was not completed, though the (clever) idea in your plan does come across.

[teenix]

(08-04-2021 01:44 PM)rprosperi Wrote:  Tony,

Your message got cut-off and was not completed, though the (clever) idea in your plan does come across.

I made some boo boo's when I read it back :-)

cheers

Tony

[teenix]

Hi all,

Further to the previous notes on making a card reader coupler, I added some info and images to the Classic Notes PDF on how to make one, available from here.

Wouldn't you know it, I found a way to make the HP-97 and printer work as a single processor set up. It will be a bit of a back track, but well worth it from a complexity view. It is much easier to manage one processor rather than two especially for re-flashing, inter communications, error handling etc.

I thought I had retired, but my boss wants me back flying - if I can get through the Covid lock downs that is. If so that will slow things again

cheers

Tony

[rprosperi]

Though totally unexpected, that's great news Tony! How did you come to figure that out, I'd guess that once committed to the 2 board design, there would be little opportunity to discover the main board could do it all?

[teenix]

(08-12-2021 12:59 PM)rprosperi Wrote:  Though totally unexpected, that's great news Tony! How did you come to figure that out, I'd guess that once committed to the 2 board design, there would be little opportunity to discover the main board could do it all?

Just popped into my head during the night. I wish things like that would happen before I commit to a circuit board though - getting a little expensive. Such is life :-)

cheers

Tony

[Dreato]

I feel I compelled to say something. I ordered Tony's HP67 motherboard last year, and he diligently fought shipping issues and I finally received it a couple months ago. I also ordered his HP67 CM board, and that arrived a month ago. I had issues understanding the accompanying software, And Tony was very responsive to help me understand. He even found a couple bugs which he quickly updated.

Bottom line, kudos to Tony. Now that I'm leaving him alone, he can get back to that HP97 Motherboard, which I also want to order!

Best Regards to all, and Stay Safe,

Dave

[teenix]

Hi all,

Still making some progress although a bit slow as my mother passed away recently and I had to stop for awhile.

I almost got to the end with the single processor design and realized that I just could not get it to work properly. With regards to the printer working with the display, it would almost work but not quite the same as the real 97.

I was forced to go back to a two processor design where the 2nd processor functions only as a PIK emulator.

The fully populated board is shown in the attachment. I still have to test the new printer software and communications between the two processors but getting there.

Getting there.

cheers

Tony

[Massimo Gnerucci]

(09-09-2021 06:20 AM)teenix Wrote:  Still making some progress although a bit slow as my mother passed away recently and I had to stop for awhile.

Sorry to read that. Condolences Tony.

[Harald]

Hi Tony,

I am very sorry to hear your mother passed away.

Harald

[albertofenini]

Hi Tony
I'm so sorry for your loss, please accept my deepest condolences
Take care my friend

[teenix]

Thanks gentlemen for the kind words, much appreciated.

I have begun work on the 97 again. I fitted the board to the calculator and made sure all the wiring fits properly. The printer ribbon cable is off by 0.050 inches when it lines up with the removable socket assembly but that won't matter, there is some room to spare.

I next want to enable some test code to verify the motor and print head output is working properly with the 2 processors before hooking the heads up to power. I have enabled some protection features which will shut the PIK processor down in the event that something happens that shouldn't. Hopefully this will make sure the print heads and motor are always de-energized if this occurs.

cheers

Tony

[teenix]

Hi all,

Boy what a couple of days.

The printer was giving me some grief in the code somewhere. When two processors are interacting, finding the bug becomes quite a task. The CPU board is working outside of the 97 so there is less fiddling with the calculator. My finger is on the external power switch when the motor is operated with what might be unexpected results.

The logic levels for the HOM and OPS were back to front and this was relatively easy to fix.

I have the oscilloscope hooked up the the head driver with the head disconnected while I test. The CRO was telling me the data was not right, but after a few hours I figured out the problem and the print head data began to look like it should. The print data decoding section of the code has been reworked considerably and now is much less complicated. It does rely on timing between the two processors though.

Multiple commands can be sent to the PIK chip for a single print line. The code was working fine with a single command, but when multiple commands were sent, multiple print lines were appearing on the same line without a carriage return. I had to make sure the print lines were narrow so that the head would not butt up against the far side of the printer casing.

The calculator CPU board defaults to TRACE mode when outside the calculator, so pressing CLx prints...

CLx
0.00

The printer printed CLx 0.00 all on the one line.

It took a whole day to figure out this problem. The code has been re-modelled quite a bit for the dual processor setup but it simulated ok, so the problem had to be a timing issue between the 2 processors. Very hard to find. Anyway, after putting it all away, then starting again, over and over as a new idea popped into my head, I started to get some proper results which led to the final solution. Yes, it was a timing issue where the HP microcode PIK command 1120 (Is Printer Home?) was testing for a home condition before the head had time to move enough when printing started. This fault tells the Microcode to start the next print line even though the first one is not actually complete.

Tomorrow, I will connect the print head up and see how the printing goes. The CPU board has a removable connector so the ribbon does not need physical disconnecting. The software allows for print speed and intensity adjustments so I will be seeing how they work as well. The speed setting can be critical if the head moves too fast because it can impact the printer casing before the print line finishes.

cheers

Tony

[aurelio]

Hi Tony
I read only now this post, I'm so sorry for your loss, my condolences
I know what it means.
Aurelio

[teenix]

Hi all,

The printer seems functional now. The paper feed works as well.

Also, here is an image of the CPU board in place.

cheers

Tony