HP-19C Repair Odysee

[PANAMATIK]

(03-08-2021 09:32 AM)teenix Wrote:  Hi all,

I looked into a PIK chip replacement circuit just at an ad-hoc level and just parts including PCB should be around the $15 mark give or take. It is a bit of a hassle because most of the PIK signals are 0-6V and require level shifting to suit a modern processor.

The main thing is getting it going. One problem is the time involved and how this relates to demand which doesn't seem great from what I've seen so justification becomes an issue too.

The other problem is that this is a controller for hardware which is pretty much irreplaceable. There is a real chance of causing damage during development. Burnt print heads, motor drivers, stripped gears or a damaged printer mechanism are examples if the software crashed for some unforeseeable reason at a critical point. This is much more relevant than interfacing to the card reader or display.

Unfortunately, all of the printer hardware is totally reliant on software for correct operation, (assuming the driver circuits are ok) so it is quite a risk to take, a problem which I am still grappling with for the 97 CPU board replacement.

cheers

Tony

The HP-19C printer module is more suited for development than HP-97, because its mechanic cannot get blocked. Also I would take my partially damaged printhead, which has still 2 pixel lines left for developing. I see two different ways to go:

1.) Development of a new HP-19C processor board with integrated keyboard scan and printer pixel output in one chip, could be a PIC processor. This would spare the complicated communication through ISA and DATA line between ACT and PIK. This solution is not suited for HP-97
2.) Replacement of the PIK chip as exact as possible, suitable for HP-19C and HP-97. I assume it would need a FPGA solution for correct timing and it is more difficult to develop (not to mention alle the level shifting problems).

I'm not sure which way to go at the moment.

Bernhard

[teenix]

(03-08-2021 10:47 AM)PANAMATIK Wrote:  The HP-19C printer module is more suited for development than HP-97, because its mechanic cannot get blocked. Also I would take my partially damaged printhead, which has still 2 pixel lines left for developing. I see two different ways to go:

1.) Development of a new HP-19C processor board with integrated keyboard scan and printer pixel output in one chip, could be a PIC processor. This would spare the complicated communication through ISA and DATA line between ACT and PIK. This solution is not suited for HP-97
2.) Replacement of the PIK chip as exact as possible, suitable for HP-19C and HP-97. I assume it would need a FPGA solution for correct timing and it is more difficult to develop (not to mention alle the level shifting problems).

I'm not sure which way to go at the moment.

Bernhard

The 19C certainly seems a better option for printer interfacing. I can leave the heads disconnected for most testing. I might have to add a temporary switch to detect the print carriage far left movement to open circuit the motor as a safe guard. It would have been nice if HP designed the 97 carriage screw setup the same as the 19C but perhaps was too late to change the design.

I don't think FPGA is required, I already have working PIC code for the 97 which I know prints correctly formatted data while performing the other required functions. I am sure it could be easily ported to a PIK chip scenario even with the bus interface requirements. The downside is risking damage to the CPU board which you will be familiar with from developing your other great products.

I also have two thoughts with the 97 CPU board. I can go with interfacing to the original printer board, or add new motor and head drivers. The original printer board seems quite dependable and is made of available parts, (although fixing the encapsulated PWM module innards would be tricky). I have a motor driver chip suitable for battery systems and am going with that at the moment as this gives a solderless changeover, but both methods have their ups and downs, a bit like my job as a pilot ;-)

cheers

Tony

[Jonas Sandstedt]

Hello,

Regarding printheads for HP-19c and HP-97, when I measured my HP-19c printhead the resistance was about 14 ohms for each resistor. When reading Sharp manuals, such as

https://sharppocketcomputers.com/4HK7JnF...manual.pdf

it seems as if the Sharp printheads have about the same resistance (see page 10).
Sharp also has a small flexcable that runs into a connector (see 26th page).
I guess the same may be the case for the Sharp printer CE-126P and calculator EL-550, as well as other similar calculators with thermal printers such as Panasonic, Texas and Canon of the early 1980's. Has anyone tried to use such a non-HP calculator or printer as a printhead donor for HP-19c or HP-97? There might be technical aspects that I have overlooked, it would anyway be interesting to know if it could be even remotely possible, since that could open for new replacement opportunities.

best regards
Jonas

[PANAMATIK]

I can't find the dimensions of this Sharp printhead in the manual. For a donor of the HP-19C printhead, the ceramic part of the printhead should have similar dimensions. Also it would be nice to find a cheap thermo printer, which is available today.

Bernhard

[Jonas Sandstedt]

I agree, but it seems as if thermal printers of today do not apply a travelling printhead, but have a fixed printhead (as in PC-100 from TI). Even if the non-HP printhead fits, it is of course dubious if the existing flat cable would be useable or if it had to be joined with an original, or connected by means of soldering. Perhaps the Sharp connector could be used and connected to the HP PCB? If I stumble on a donor non-HP printhead, I will let you know.

Here are some of the cheapest I have found today:

https://www.ebay.de/itm/144017424440?_tr...%7Ciid%3A1

https://www.ebay.de/itm/144021023236?_tr...%7Ciid%3A1


best regards
Jonas

[Kostas Kritsilas]

I had a thought that the printheads from current thermal strip printers used on POS (point of sale) terminals may be worth looking into. As far as I have been able to determine, they are just a row of print elements all the way across the thermal paper, not a moving print head. However, on my 82143A printer, the replacement paper is the same as the thermal strip printers (2 1/4" wide, or so).

The alternative is to look at the printer mechanisms from the 82143/A, 82163 (the HP-IL printer) or the HP 82240/A/B. If these printers have replacement heads and/or mechanisms, that could be another way to go.

[jjohnson873]

(04-27-2021 11:47 AM)Jonas Sandstedt Wrote:  Hello,

Regarding printheads for HP-19c and HP-97, when I measured my HP-19c printhead the resistance was about 14 ohms for each resistor. When reading Sharp manuals, such as

https://sharppocketcomputers.com/4HK7JnF...manual.pdf

What a great find, Jonas! Nice to see a detailed manual on the full Sharp printer/cassette. And I applaud the quest to find a suitable replacement for the HP-19c printhead. Now I should dive into this endeavour and try a similar search. Thanks again for posting a pdf of the service manual.
~ Jim J. ~

[PANAMATIK]

The development of a PIK chip replacement has reached a first milestone.

What you see in the two images below is my totally corroded HP-19C mainboard, which had all ICs damaged, but now brought to life again. The ROM and RAM chips and the ACT processor are removed, the also totally corroded LED display is replaced by a HP-25 LED display (my replacement),which is pin compatible, but has only 12 digits instead of 13 (I could have also used an original HP-25 display). Only the two display driver chips are used, which I took from an HP-22 board, but any display drivers from any Woodstock would do, and of course also the original HP-19C display drivers if they were still OK

The new ACT HP-19E (the lower 22-pin IC) is programmed as an HP-19C emulation, which runs the original HP-19C ROM code and drives the display and makes calculations. It communicates with the PIK chip, to read keystrokes and output print lines.

The new PIK prototype (Printer Interface Keyboard) (28-pin Socket using the same PIC16F1518 hardware during development, but programmed as "new PIK") is already able to scan the keyboard and send the keystrokes to the HP-19E ACT. This is half the way done to the final goal. But the other half is more stony: controlling the printer motor and precisely timing the printer output.





The concept for repairing an HP-19C mainboard with defective RAM, ROM, ACT or PIK will be: replacing the ACT and the PIK chip without any other modification. This concept will not allow to replace only one of them, because implementing of the original communication between the ACT and PIK was too hard to implement with the PIC processor.

Another concept would be: replacing the complete board including the display drivers, which is still under investigation. It would allow to display alphanumeric characters.

The next step will be: to drive and control the printer motor.

Bernhard

[teenix]

After playing around with the PIK chip myself a bit and realizing some (but not all) of its complexities, I have to say awesome work, well done :-)

cheers

Tony

[PANAMATIK]

(05-03-2021 02:05 PM)teenix Wrote:  After playing around with the PIK chip myself a bit and realizing some (but not all) of its complexities, I have to say awesome work, well done :-)

Thanks Tony!

I want to show here one of these complexities, which I had to deal with today and to overcome. It is also an example of how crazy the HP-19C machine is constructed. Obviously IMO as the result of the idea to use the existing ACT chip and the PIK chip for the HP-19C, which were originally designed for the "Woodstock" calculators (ACT) and the HP-97 (PIK) by their engineers, and do not fit very well for this machine.

As mentioned the PIK chip constantly reads the 8x4 keyboard matrix (every 5-6ms) and determines the keycode. In case a button was pressed, it fills a 16 byte buffer with the keycodes, which will be read by the ACT processor and processed. But there is no way to read the information from the PIK chip, whether a button is still hold down. There are two buttons, SST and R/S, which show the actual program step, while pressed. How does this work?

The hardware designers decided to concurrently let do the ACT chip the keyboard scan as well. For this purpose they only connected ONE of the four keyboard rows to the ACT keyboard matrix inputs. This is the row where the buttons SST and R/S and 5 other buttons are located. For any other buttons it is impossible to detect, if they are still pressed. And indeed by analyzing the HP-19C ROM code, I found that only in case of the SST or R/S button the status flag S15 is tested, which tells the ACT that one of the keyboard inputs are at low level.

A crazy enable signal, switched by three transistors, makes it possible that the other ACT key matrix inputs can be used to detect the state of the PRGM/RUN and MAN/TRACE/NORM switch and doesn't interfere with the S15 flag.

Today in the morning I got the key matrix running and was able to enter keystrokes, which were detected by the PIK, and I could perform the first calculation: I added two numbers. But I could not yet test whether a key was hold down. Also the PRGM/RUN switch was not yet to detect. I needed the whole day to achieve this. Now the HP-19E ACT in conjunction with the newPIK can detect the state of the PRGM/RUN switch, and detect whether the SST and R/S key is still pressed, and I was able to enter my first program.

Code:


01  25 14 01 LBL 1
02        01 1
03        41 + 
04     16 64 f PAUSE
05     14 01 GTO 1

And it works! Also when using the SST key to see the program step, before it is executed.

Bernhard

[jjohnson873]

(05-04-2021 08:25 PM)PANAMATIK Wrote:  Today in the morning I got the key matrix running and was able to enter keystrokes, which were detected by the PIK, and I could perform the first calculation: I added two numbers. But I could not yet test whether a key was hold down. Also the PRGM/RUN switch was not yet to detect. I needed the whole day to achieve this. Now the HP-19E ACT in conjunction with the newPIK can detect the state of the PRGM/RUN switch, and detect whether the SST and R/S key is still pressed, and I was able to enter my first program.

Code:


01  25 14 01 LBL 1
02        01 1
03        41 + 
04     16 64 f PAUSE
05     14 01 GTO 1

And it works! Also when using the SST key to see the program step, before it is executed.

Bernhard

Congratulations for persistence, Bernhard!

[PANAMATIK]

My first attempt to understand and control the printer Motor.

In the photo below you see an original HP-19C mainboard and printer board, opened for oscilloscope measurements. To get access with the probes I had to put the keyboard aside and connect via flat ribbon cable. It is a fully working HP-19C hardware, however without printhead and flex cable (see post #1 to learn why the printhead is missing), but the motor moves correctly.



In the above image to the right you see a fully working HP-19C, which I hopefully don't have to open for this research. In the background you see another printer board with defective motor control. With this board the motor moves twice as fast as intended and prints the digits much too wide. The number PI is printed partially reverse on the way back, which you see below.





The oscilloscope screenshot shows the FWD (Forward) Signal and the REV (Reverse) signal of the PIK chip with correct motor control. The forward speed is regulated and needs about 1 s for printing, the reverse signal causes the motor to move full speed to get to the HOME position in about 400 ms. Notice that the HP-19C motor shaft always moves in one direction, the REV signal is just used to accelerate. As the PIK chip was originally designed for the HP-97 hardware, which moves the motor bidirektional and uses the REV signal normally, the HP-19C hardware uses the REV signal just for going full speed. Another crazy design detail.

Bernhard

[[kby]]

There should be a pot that adjusts this. There is on the HP-10.-kby

(05-06-2021 07:24 PM)PANAMATIK Wrote:  My first attempt to understand and control the printer Motor.

In the photo below you see an original HP-19C mainboard and printer board, opened for oscilloscope measurements. To get access with the probes I had to put the keyboard aside and connect via flat ribbon cable. It is a fully working HP-19C hardware, however without printhead and flex cable (see post #1 to learn why the printhead is missing), but the motor moves correctly.



In the above image to the right you see a fully working HP-19C, which I hopefully don't have to open for this research. In the background you see another printer board with defective motor control. With this board the motor moves twice as fast as intended and prints the digits much too wide. The number PI is printed partially reverse on the way back, which you see below.





The oscilloscope screenshot shows the FWD (Forward) Signal and the REV (Reverse) signal of the PIK chip with correct motor control. The forward speed is regulated and needs about 1 s for printing, the reverse signal causes the motor to move full speed to get to the HOME position in about 400 ms. Notice that the HP-19C motor shaft always moves in one direction, the REV signal is just used to accelerate. As the PIK chip was originally designed for the HP-97 hardware, which moves the motor bidirektional and uses the REV signal normally, the HP-19C hardware uses the REV signal just for going full speed. Another crazy design detail.

Bernhard

[PANAMATIK]

(05-09-2021 12:25 AM)[kby] Wrote:  There should be a pot that adjusts this. There is on the HP-10.

This would be too easy. The 50k poti P2 is adjusted to about 10k, I measured it, turning it doesn't change anything. Also I have replaced the U11 LM339 comparator IC, but this was not successful too. It could be a defective Network U17 or the PIK chip itself, which regulates the motor speed.



The keyboard of my "totally corroded HP-19C" was not too bad allover, but some keys were stuck. Therefore I cut a trapezoidal opening into the clear tape to get access to the contacts under the dome. Cleaning with a soft cotton swab with diluted alcohol made the keys work again.



There is also a trick to simplify combining the two PCB boards when assembling the HP-19C, which is said to be very difficult. Remove the LED display, then first stick the horizontal contacts by looking through the hole to the left, for getting their actual position, then moving the short distance to the holes. The following step is, also stick the vertical pin contacts, which is now rather easy.





Bernhard

[teenix]

Looking at Tony Duel's circuit, the 50K pot should adjust the motor speed, and the 200K pot should adjust the intensity.

cheers

Tony

[PANAMATIK]

(05-09-2021 02:16 PM)teenix Wrote:  Looking at Tony Duel's circuit, the 50K pot should adjust the motor speed, and the 200K pot should adjust the intensity.

Exactly! But only if the circuit works properly :-(

Bernhard

[teenix]

(05-09-2021 02:34 PM)PANAMATIK Wrote:  

(05-09-2021 02:16 PM)teenix Wrote:  Looking at Tony Duel's circuit, the 50K pot should adjust the motor speed, and the 200K pot should adjust the intensity.

Exactly! But only if the circuit works properly :-(

Bernhard

Too true :-)

I'm pretty sure the HP-97 has full motor speed for reversing the carriage.

The PIC chip can gate the PWM signal from its VM pin so that the motor is controlled by the PWM or steady voltage for full speed.

The FWD output must also have circuitry to detect when the OPS switch is activated to stop the printer when paper has run out even though this is also done in 97 hardware. This is tested when PWM is present and when not, although I don't know if the 19C has this.

cheers

Tony

[PANAMATIK]

(05-09-2021 05:56 PM)teenix Wrote:  Too true :-)

I'm pretty sure the HP-97 has full motor speed for reversing the carriage.

The PIK chip can gate the PWM signal from its VM pin so that the motor is controlled by the PWM or steady voltage for full speed.

The FWD output must also have circuitry to detect when the OPS switch is activated to stop the printer when paper has run out even though this is also done in 97 hardware. This is tested when PWM is present and when not, although I don't know if the 19C has this.

Hi Tony,

You are right. The HP-97 indeed has a complete H-Bridge for controlling the motor in both directions. For short print lines the HP-97 skips blank characters and returns earlier, this is much faster than the HP-19C, which has to go always the full distance to the left and right (see HP-97 service manual schematics). It seems the PIK generates a PWM signal only on the FWD signal, not REV signal, because it always goes full speed backwards and doesn't print bidirectional.

Just gating the FWD PWM signal, instead of adjusting the duty cycle, would be an easy way to control the speed. The FWD PWM duty cycle of my defective HP-19C at full speed is about 75% on my scope.

The HP-19C has no OOPS, (OOPS! Out Of Paper Switch).

I'm currently programming the time-shifted output of the print characters, which let them appear in "italic style", which has the intention to switch the current of the printhead pixels not all at the same time. The PIK of HP-19C and HP-97 has 62 printable characters, 26 upper case letters, 1 lower case letter e (for showing e^x), 10 numbers and 25 special characters + - * etc. (the HP-97 PIK may have different special characters) . I read the exact print matrix with a magnifier glass from the paper printout and programmed it already in my newPIK.

Because the HP-19C doesn't use all 26 letters in the ROM code, I couldn't assign the 6-bit codes for letters J,K,Q,U,V,W. Perhaps analyzing the HP-97 ROM will reveal it.

The timing for the printhead output is relaxed, because the pattern must be changed about every ms while printing, this is easy for a todays micro. A print line has a maximum of 22 characters, each has 5 print columns and 1 blank column between the characters, time delayed output by 7 rows, printed in 1 s, needs to output a new print pattern every 1000/22/6/7 = 1,08 ms.

Bernhard

[teenix]

Interesting :-)

My method has a specially generated character "ROM" inside the controller which was easy to create.

When required, the preconfigured data is placed in a print buffer and then output to the heads with PWM as determined by the battery voltage. The characters are automatically in italic and properly spaced. This way the code for actual printing is quite small.

The PIC output to the heads is shown in this post.

https://www.hpmuseum.org/forum/thread-12...#pid130620

cheers

Tony

[PANAMATIK]

(05-10-2021 12:13 AM)teenix Wrote:  The PIC output to the heads is shown in this post....

Interesting!

I thought to control the print intensity by the pulse length of the pixel. But PWM is more elegant or even necessary for avoiding overheating.

Bernhard