ACT available

[PANAMATIK]

There will be a replacement for the famous PMOS ACT chip for HP-21, HP-25 woodstock calculators soon available.

I will offer it as a repair kit in January for shipping within germany.

All you have to do: remove the defect ACT and replace it by the new chip.

In most cases of a not functioning woodstock calculator only the ACT Chip is defect, but Power supply, ROM, RAM and Display Drives are still working. The new Chip takes over all signals on the board, communicates with all components and establishes a running Display and Keyboard matrix while emulating the arithmetic logic unit.

To avoid questions:
The chip is suitable for HP-21, HP-22, HP-25, HP-25C. HP-29C devices with PMOS design and HP-27 with NMOS design, not yet tested or worked out for other devices.
It executes either the original ROM Code from the on board ROM Chips or internal ROM Code.
It contains a program interface for updating the firmware or for other purposes.
The RAM Chip is not longer needed after the replacement, because the chip does have enough internal RAM.

If any other component or the power supply on your board is defect, the replacement will not making your calculator work again. In any other case, you will give your beloved calculator a new life and you will see its beautiful LEDs numbers 0.00 light again.

Price will be about 80,- Euro.

Please mail to sales@panamatik.de if you are interested.

[Paul Dale]

This is interesting. I've got 25 with a damaged ACT chip.

It it the same speed as the originals? I'd imagine it would have to be if it is running from the external ROMs but is this also true when running from internal memory?


Pauli

[PANAMATIK]

The speed is the same as the original in both cases, because the state machine for display multiplexing must be running all the time whether reading external ROM or not, and I choose a suitable chip for the only purpose of repairing woodstocks.

But a high speed HP-25 board, with extended features and additional program steps, is also within my thoughts for the future.

[Massimo Gnerucci]

Nice!

So we won't write anymore that sad: "the only source for your fried ACT is another 25/21..."
Are you thinking of something similar for the Continuous Memory models too?

[PANAMATIK]

I don't own a Continuous Memory HP. But I think it could be not much different, even more easy because I expect there are no PMOS signals any more.

[Katie Wasserman]

(11-17-2014 03:40 PM)PANAMATIK Wrote:  All you have to do: remove the defect ACT and replace it by the new chip.

That's very interesting!!! What micro did you find that has a pin configuration compatible with the ACT? I've looked into this possibility before and gave up after finding nothing that would work without some PCB trace modifications.

[PANAMATIK]

I,m sure You would have found a pin compatible chip on your search, if there were any to buy. Of course they are not. The new 'chip' will be a small printed curcuit in the size of the previous ACT chip. It contains a programmable hardware, also some other components are necessary to make it compatible with the PMOS architecture and run the hardware.

Thanks for your interest. I will place more posts when the first 'chips' are available.

Bernhard

[Katie Wasserman]

(11-18-2014 04:29 PM)PANAMATIK Wrote:  I,m sure You would have found a pin compatible chip on your search, if there were any to buy. Of course they are not. The new 'chip' will be a small printed curcuit in the size of the previous ACT chip. It contains a programmable hardware, also some other components are necessary to make it compatible with the PMOS architecture and run the hardware.

Thanks for your interest. I will place more posts when the first 'chips' are available.

Bernhard

Aha! I was mislead by the use of 'chip'. Anyway it sounds like a good project that you've got there and I'm looking forward to seeing the details.

[Brad Barton]

Did I read somewhere that the ACT from the HP-25 was a suitable replacement for the ACT in the HP-97? I have no clue about the various architectures, so this could be a dumb question, but will one of these work in the HP-97?

[PANAMATIK]

The ACT repair kit could be working fine for the HP-97, it contains the same original ACT chip than any HP-25. The repair kit don't use the original RAM chips, because they could be damaged too. Im not sure how much RAM the HP-97 needs, but it should be within the range that I can offer. And there could be machine instructions for printing, that I do not emulate.
I own a HP-97 calculator and could prove it, but (un)fortunately it is working very well, so I don't want to solder. However, I tested HP-21 and HP-25. Also HP-10, HP-22 and HP-27 should be candidates for a direct replacement, but I cannot promise until somebody could prove it. It needs just some weeks to wait until I can deliver the first units.

Bernhard

[Katie Wasserman]

(11-19-2014 11:03 PM)PANAMATIK Wrote:  However, I tested HP-21 and HP-25. Also HP-10, HP-22 and HP-27 should be candidates for a direct replacement, but I cannot promise until somebody could prove it. It needs just some weeks to wait until I can deliver the first units.

The HP-27 uses NMOS technology not PMOS like the 21, 22 and 25 67, 97, etc.. Did you design your "chip" to handle both kinds of logic? That would be quite a challenge I think.

[Brad Barton]

I have both a non-working and a working 97. I broke the first one by plugging the it in without a battery pack by mistake, and that was all she wrote.

I've read that the most likely problem is a blown ACT chip, but I'm not willing to de-solder the one from the working model to test the theory.

[PANAMATIK]

(11-20-2014 02:50 AM)Katie Wasserman Wrote:  The HP-27 uses NMOS technology not PMOS like the 21, 22 and 25 67, 97, etc.. Did you design your "chip" to handle both kinds of logic? That would be quite a challenge I think.

The NMOS ACT replacement for HP-27 should have a different design. Perhaps it could be less challenging to give it the same design as for a CMOS ACT in HP25C HP-29C etc.
Did the 'Cs' suffer the same problem of damaged ACTs when charging without batteries?

I think I need to buy one or two HP25C and HP-27 somehow in the near future.

Bernhard

[brouhaha]

(11-18-2014 03:35 PM)PANAMATIK Wrote:  I don't own a Continuous Memory HP. But I think it could be not much different, even more easy because I expect there are no PMOS signals any more.

No. Only the RAMs are CMOS, and they still use PMOS levels.

[PANAMATIK]

(11-18-2014 09:40 AM)Paul Dale Wrote:  It it the same speed as the originals? I'd imagine it would have to be if it is running from the external ROMs but is this also true when running from internal memory?

Meanwhile I managed to run microcode from internal ROM as an independent parallel task to display and keyboard scanning. It runs much faster, but key debouncing doesn't work anymore, and one short keystroke enters two or three digits. However Debouncing can be done outside of the original firmware and this works fine.
More problems arise, when running a different firmware. As you can imagine, using internal ROM code offers the opportunity to 'upgrade' a HP-21 to a HP-25 etc.
There are subtle differences to become visible when running HP-25 code on a HP-21 hardware or vice versa, i.e. the keyboard columns layout had switched some traces and this results in different calculated jump instructions. The effect: all keys are shifted by one column, but principally this can be corrected and works also fine.

[Paul Dale]

Now this is great news. I wasn't contemplating running firmware on different hardware, although I'm sure some people will. I'd be very happy with a fast HP-25.


Pauli

(11-25-2014 10:42 PM)PANAMATIK Wrote:  Meanwhile I managed to run microcode from internal ROM as an independent parallel task to display and keyboard scanning. It runs much faster, but key debouncing doesn't work anymore, and one short keystroke enters two or three digits. However Debouncing can be done outside of the original firmware and this works fine.
More problems arise, when running a different firmware. As you can imagine, using internal ROM code offers the opportunity to 'upgrade' a HP-21 to a HP-25 etc.
There are subtle differences to become visible when running HP-25 code on a HP-21 hardware or vice versa, i.e. the keyboard columns layout had switched some traces and this results in different calculated jump instructions. The effect: all keys are shifted by one column, but principally this can be corrected and works also fine.

[brouhaha]

(11-25-2014 10:42 PM)PANAMATIK Wrote:  Meanwhile I managed to run microcode from internal ROM as an independent parallel task to display and keyboard scanning. It runs much faster, but key debouncing doesn't work anymore, and one short keystroke enters two or three digits. However Debouncing can be done outside of the original firmware and this works fine.

Note that the printer (19C, 91, 92, 97) and the card reader (67, 97) are very timing dependent, and very likely won't work correctly if you use asynchronous execution.

[PANAMATIK]

(11-27-2014 04:25 AM)brouhaha Wrote:  

(11-25-2014 10:42 PM)PANAMATIK Wrote:  Meanwhile I managed to run microcode from internal ROM as an independent parallel task to display and keyboard scanning. It runs much faster, but key debouncing doesn't work anymore, and one short keystroke enters two or three digits. However Debouncing can be done outside of the original firmware and this works fine.

Note that the printer (19C, 91, 92, 97) and the card reader (67, 97) are very timing dependent, and very likely won't work correctly if you use asynchronous execution.

Thats true. There are still both, original speed synchronous and fast asynchronous, operating modes available, you can decide which to choose by pressing the '0' key at power up.

I included also a reduced currrent sleep mode, which sets the display to only showing one decimal point after one minute without keystrokes. This saves about 50 mA and can be activated by holding the '.' key pressed at power up.

So, without pressing any key it is still like the original.

And - do you like the idea to have a stopwatch in your HP-25/21?

[PANAMATIK]

(11-18-2014 09:40 AM)Paul Dale Wrote:  It it the same speed as the originals? I'd imagine it would have to be if it is running from the external ROMs but is this also true when running from internal memory?

Some details:

When running the Code from external ROM it is not easily possible to achieve exactly the same speed than the original with the same clock speed. This is because the instructions in the original ACT were executed in hardware before the address for the next opcode is output on the ISA Line. In my replacement I need some more cycles for emulating the instruction and cannot wait for completion, because the display digits must be multiplxed smoothly. So I implemented a simple branch prediction, that assumes a linear program flow, but whenever a branch instructions occurs, it needs two cycles to execute. Thats why it is slightly slower.

Bernhard

[PANAMATIK]

I really was awaiting the first ACT circuits in original size, until they arrived yesterday. I got the first samples, and after testing and a full night programming: finally it works!

I'm proud, that I can present the first HP-25 and HP-21 calculators ever, that are running without its original ACT chip.

See pictures below.

   
The new ACT chip

   
replacement location

   
the old and the new chip

   
power up HP-21

   
HP-25

At power up the new ACT shows a HP-25 or HP-21 logo for 2 seconds, depending on which ROM Code is detected. Then does its normal task.

Optionally, I integrated a continuous memory, so any normal HP-25 transforms automatically into an HP-25C. There is also a stopwatch, which appears by the g Prefix key sequence. Next, the sleep mode saves about one third of current consumption. But you can deactivate all of these new features, even the logo at startup, then you cannot distinguish your new calcuator from your original calculator.

I bought a HP-25C this week and will check, whether the replacement does work also in this calculator. Unfortunately I see no chance to find a cheap defective HP27 or HP29C somewhere.

Bernhard