HP-29C ACT compatibility

[PANAMATIK]

HP-21/25 ACTs can repair a defective HP-29C

HP used the so called ACT chip (Arithmetic Control Timing) in all Woodstock calculators HP21/22/25/25C/27/29C and later also in the HP-19C/67/91/92/95C/97. Finally they developed a new Chip for the next generation, the "Spice" calculators HP-31/32/33/34C/37/38, which made the ACT chip obsolete. From the first HP-21 to the last HP-97 this chip evolved and got some internal enhancements in several steps, whereas the details are not known. Basically the later ACTs needed to communicate with the PIK I/O chip and had a few more instructions for printer and magnetic card reader handling.

The HP-29C used the later 1820-1596 or 1820-1741 ACTs, whereas the HP-21 and HP-25 used the earlier 1820-1396 1820-1523.

Although I made a couple of repairs of "Woodstock" calculators by replacing their defective ACT chip with the new ACT chip, I never tried to run an HP-29C calculator with an original HP-25 or HP-21 ACT. Yesterday I made this experiment and put an HP-21 and HP-25 ACT into my HP-29C. I tried nearly every function and tested progamming and the "Continuous Memory". It works with both ACTs without any issues. I assume, that the later enhanced ACTs were used in all models by HP, but only the high end calculators with printer and card reader used the enhanced features. The HP-29C however did not make use of them and therefore the earlier ACTs can be taken as an replacement.

I hope you will not transplant an ACT from a working HP-21 into an HP-29C and throw away the HP-21, which would be a shame. But if for some reason you have an HP-21/22/25 ACT as spare, then you could repair an HP-29C with it.

Here a view to my little box with (mostly defective) "Woodstock" calculator chips, the remains of some "Woodstock" repairs.



Bernhard

[Paul Dale]

Very interesting.

I've not got a 29C, I've never held or see one live (although I was the custodian of a 19C for a while). I might be able to locate/purchase a 25 or 21 which could be upgraded.


Is there any chance the newACT would be able to handle the card reader and printer in later calculators? E.g. replacing the ACT in a 67 can be done but the card reader doesn't work afterwards -- what would be required to make it operate again? I can donate at least one sacrifical board for experimenting with.


Pauli

[PANAMATIK]

(06-08-2017 10:16 AM)Paul Dale Wrote:  Is there any chance the newACT would be able to handle the card reader and printer in later calculators? E.g. replacing the ACT in a 67 can be done but the card reader doesn't work afterwards -- what would be required to make it operate again? I can donate at least one sacrifical board for experimenting with.

This is the question to which I do not have a final answer!

There is a chance to get the card reader (HP-67/97) and printing (HP-19C) working with the new ACT. But when I started this project and made the design the goal was just to repair an HP-25, nothing else. Later I myself had the idea of restoring also an HP-67/97 or 19C. But the timing for handling the serial communication line to the card reader and printer via the PIK chip and to the ROMs and display drivers simultaneously was very difficult for the chosen PIC processor. It is not just done with a few program lines. However still I know intuitively that it is possible to accomplish the goal even with the actual newACT hardware.

Going more into detail:

The PIC processor has only one serial SPI interface. I use this resource for sending and receiving serial data from/to the ROM chips and the anode display driver via the bidirectional ISA line. The SPI clock is used to generate the PHI1 signal. I use another peripheral resource of the processor for generating the PHI2 signal by synchronizing a PWM output with the SPI interface with same baud rate, but phase shifted. The baud rate is 180 kHz. For printing or reading the card reader, another serial line, the DATA line, must be driven simultanuously and there is no peripheral resource left for it.

A solution could be to drive all serial lines via normal I/O pins, 180kHz is a little bit too fast for normal I/O pins and would consume much processor time, but it is a way I can see for making the new ACT able to also handle the card reader and printer. It is not a good solution and some display segments would have different brightness as a side effect, because the timing is not accurate. Perhaps this is the reason why I did not follow this approach with too much energy. You see I did at least try some experiments in this direction.

The more promising solution would be to use the SPI interface for normal operation but, when detecting a PIK card reader or print instruction, to switch to the normal I/O pin method. This is definitely difficult to program, but would make the card reader eventually running with the actual newACT hardware.

Of course, a redesign and creating a completely new project with FPGA instead of the PIC processor would be able to do the task, but this would need to consume my engineering resources for months. In another parallel universe I would have done it already, but obviously I admit I couldn't find this parallel universe here, because I believe I'm living in a real world called planet earth.

However, I'm still working on it.

Bernhard

[Paul Dale]

Bernhard, thanks for the explanation, I understand the difficulties involved and an hardware update would be beyond my skills.

Blanking the display during card reading and printing would be reasonable as a way to avoid different brightness LEDs. It seems like 180kHz is just a little too fast for the processor -- 6 instuctions times 180k is approaching the speed of the CPU. That's not a lot to achieve something useful with. It might be possible to buffer the outputs and then push them in a tight loop.

Is there a suitable PIC that could be used instead? E.g. the 16F18856 comes in the same package and it provides an extra SPI port. The pin outs are different enough it wouldn't just drop in. There are a lot of PIC processors available and something might come closer.


Pauli

[biogon]

Very interesting information about the compatibility of the ACT chip.
I have already repaired several HP calculators like; 10,21,22,25,37,55,65,67,97 and 41, and I have the tools and knowledge to do it.
But, my last purchase a HP 29c does not want to work. After cleaning the electrolyte leaks from the batteries and checking capacitors and voltages of the PS, I have concluded that the ACT, at least, is defective.
I refuse, for now, to look for a 21 or 22 to donate the ACT, and I have been looking the waves in the ACT pins. I notice that the initial RST does fine, but the output / O1 on pin 17 has only 1V amplitude, while the / O2 on pin 16 is ok with 18V.
The high threshold on both is the Vss + 6V, but in phase!
I am considering adding a small circuit with transistors to generate / O1 from / O2 and insert it into pin 17 line of the ACT.
If someone has already tried this solution, I would appreciate if can tell me if it could work, or go directly to replace the ACT.
Thanks in advance.

[PANAMATIK]

(05-11-2018 07:20 PM)biogon Wrote:  If someone has already tried this solution, I would appreciate if can tell me if it could work, or go directly to replace the ACT.

This would indeed be a repair solution if /O1 is the only issue. I'm sure nobody has generated /O1 from the /O2 signal, although it would be possible. You need at least a monoflop, or some counter logic with some transistors as level converter. Not too difficult but also not very easy.

In any case you could repair your HP-29C with the "new ACT" chip or the "Woodstock LP" circuit.

Bernhard

[biogon]

In view of the ACT chips in your blue box, it seems that most are AMI. Can we expect the MOSTEK ACT ceramic chips to be more reliable?

[Harald]

(05-18-2018 10:07 AM)biogon Wrote:  In view of the ACT chips in your blue box, it seems that most are AMI. Can we expect the MOSTEK ACT ceramic chips to be more reliable?

First of all we can expect them to be a lot less common!

Cheers,
Harald

[teenix]

(06-08-2017 07:15 PM)PANAMATIK Wrote:  Of course, a redesign and creating a completely new project with FPGA instead of the PIC processor would be able to do the task, but this would need to consume my engineering resources for months. In another parallel universe I would have done it already, but obviously I admit I couldn't find this parallel universe here, because I believe I'm living in a real world called planet earth.

However, I'm still working on it.

Bernhard

I think some PICs now have FPGA embedded in their cores. Not sure as Microchips website is acting up.

cheers

Tony

[jjohnson873]

There aren't any "FPGAs" inside of PIC micro's, per se. However there are Configurable Logic Cells (CLC) that are core independent and are register programmable in many of the new PICs. With Microchip's latest acquisition of Microsemi, there will be true, very low power, FPGAs available.

[biogon]

Finally I decided to buy a cheap and untested HP-21 as donor.
The HP-21 didn´t work at all. But all of the ACT signals were present.
As the HP-29C PS voltages and start circuit were OK...
I removed the 1820-1523 HP-21 ACT and replaced the damaged 1820-1596 HP-29c with it using pin strips.
It was still not working. But all the 1523 ACT and anode driver signals were present.
I removed the HP-21 cathode driver and replaced the HP-29c device with the same reference.
It was still not working...
As experienced in other HP models I decided to remove the PS circuit and replace all the tantalum and electrolytic capacitors.
It was still not working. But sometimes, touching some ACT pins the display shown some numbers and the keyboard responded sometimes.
I found that the sensitiv ACT pins were nº10 (Ia/Is) and nº11 (DATA).
The Ia/Is signal in the oscilloscope had good 6V (1) level pulses, some less than 1V (0) level pulses, but someone had a variable level between 1 and 3 volts. The same in the DATA pin.
I connected a variable limited 50K resistors in both pins to ground (pin 12), and reduced the resistance value just until the 0 pulses were all less than 1 volt. The resistors were 12KOhms.
I tried and tested all the main features of the HP-29C and the result was OK!
But ERROR was displayed in the starting up, cleared with CLR, and the GTO 00 displayed also ERROR.
I decided to assemble the calculator again in order to use the original batteries.
As result no more ERROR displayed switching ON the calculator, and the GTO instructión needs a .00 to work, not like the HP-25. The repaired HP-29C is full functional now, including the cmos memory retention!

Some pictures of the reparation.

   
   
   

In these pictures you can see the Ia/Is and DATA signals before and after the modification grounding them with 12KOhms.