HP-65 1333A04841 restoration it doesn't work anymore ...

[albertofenini]

Hi everyone, I’m restoring this unit but I definitely need some advice …
The unit was not in terrible conditions, but it had some oxidation on the card reader, on the keyboard pcb and on the cpu board
Gummy wheel to be done, but that was no surprise

After an initial cleaning it was working, although had this strange behavior: it never turned on at first try if it has been turned off since a while
You had to turn it off and then on again and it was working, after that you could turn it on and off with no problems
But if it was left off for a while, again, it needed a double try.
Also some times it was showing erratic behavior like in the video where after turning on it was showing a kind of a countdown
But at this stage it never showed erratic digits or partially lit segments

I cleaned the card reader pcb and the keyboard, and I did the gummy wheel with two glued o-ring
but at that time I didn’t want to bother the cpu board

The unit was reading and writing with no problems, I almost copied an entire standard pac

But, as you can hear in the videos, the noise of the card reader was not regular, it was like something was not centered,
As the noise was almost like a wave … nevertheless it was reading a writing without failures
I thought that after all gluing the o'ring could have caused some eccentricity

However, I didn’t like it, and this evening I tried to change the gummy wheel, at first with a small section of a silicone tubing
and later with two not glued o-rings

After I did this the unit was not reading anymore, although the card reader was now working smoothly as it should have been since the beginning

Then I swapped then this CPU board with a working one and I discovered that this configuration was reading with no
problems at all (though I didn’t test writing) while this CPU in another working chassis was still not reading

At this point I treated this CPU with water and vinegar and rinsed with isopropyl and let it dry well as I thought
that may be the oxidation on the CPU board I had seen was causing this problem

Sadly after that I haven't been able to get it working again, it only shows random digits totally meaningless as it can been seen in the last pictures
and it now needs several tries to turn on

What can I try or what can I check ?

Is reworking the board an option ?

I Have a set of spare ROMs, is this issue more a ROM or a Hybrid one ?

Thanks in advance for any help and take care !!

Pictures and videos can be seen here : HP-65 1333A04841

[Maximilian Hohmann]

Hello!

I wouldn't have worried much about the sound of the card reader. This is very much like my Ti-59 which I got in 1979 and which still works today. With the same modulated sound.

My guess why it would not read and write after the gummy wheel change is that one of the connecting cables to the card reader board broke during the disassemly and reassembly. I would have started troubleshooting form there.
This alone does of course not explain the erractic display that it shows now. Still the chances are high that this is caused by a faulty solder point or PCB trace rather than a defective ROM or something similar. Some weeks ago I was able to fix a broken Casio (a rare-ish AL-8S with DATE functions) by „beeping“ every single PCB trace until I found the one that was corroded through, invisible under the protective green paint.

Don't you have several other HP-65s by now? You could interchange boards to be sure, but that of course always carries a risk of causing extra damage.

Good luck
Max

[teenix]

A slim chance maybe, but the hybrid casing may not be sealed properly after all these years and moisture got in after washing.

Faulty power supply caps

cheers

Tony

[albertofenini]

(03-17-2022 11:37 PM)Maximilian Hohmann Wrote:  Hello!

I wouldn't have worried much about the sound of the card reader. This is very much like my Ti-59 which I got in 1979 and which still works today. With the same modulated sound.

This is what I will do in the future ... if it works don't touch it, lesson learned

My guess why it would not read and write after the gummy wheel change is that one of the connecting cables to the card reader board broke during the disassemly and reassembly. I would have started troubleshooting form there.

That was not the case because another CPU board in that chassis proved reading so I knew the card reader assembly
was working and connected properly

This alone does of course not explain the erractic display that it shows now. Still the chances are high that this is caused by a faulty solder point or PCB trace rather than a defective ROM or something similar. Some weeks ago I was able to fix a broken Casio (a rare-ish AL-8S with DATE functions) by „beeping“ every single PCB trace until I found the one that was corroded through, invisible under the protective green paint.

It already had some issues, not turning on at the first try with no reason and showing erratic
behavior and it really had some oxidation to remove sooner or later ...

Don't you have several other HP-65s by now? You could interchange boards to be sure, but that of course always carries a risk of causing extra damage.

Good luck

thanks and take care !!

Max

[albertofenini]

Hi Tony

(03-18-2022 12:10 AM)teenix Wrote:  A slim chance maybe, but the hybrid casing may not be sealed properly after all these years and moisture got in after washing.

that could definitely be the case, consider that cleaning the card reader and the pcb I have started to
use hot water and vinegar, as it is more efficient and I have not taken into consideration that this may have caused a problem with the sealing of the hybrid,
what are the chances the it will dry and get back to work ?

Faulty power supply caps

these I can try to replace, I also would like to check some voltages and signals,
like Ph1 and Ph2, can I keep the board powered or I'm making things even worst ?

cheers

thanks and take care !!!

Tony

[teenix]

If moisture got into the hybrid then the vinegar solution may have ended up on the exposed chip surfaces but I don't know what this will do to them. Hopefully this did not occur.

Another thing that might happen is that if the inter-board connections (vias) were corroded and the vinegar wash has cleaned that off, they might not make proper connections. As it is a multi-layer board the inner layers may have broken or increased resistance connections also. Hard to trace.

There is an initialization circuit inside the hybrid that resets the calculator chips at power on (Init line - pin 36). If this is faulty it may start up erratically. I imagine it stays low for a short time at switch on then goes high.

If it all appears to be working (when it does) then maybe it will be ok to run while testing as long as nothing gets warm or hot.

cheers

Tony

[albertofenini]

Hi Tony
Thanks for getting back to me
I have tried few minutes ago after letting the board to dry on a heater all day but
It behaves like yesterday

I have ordered a lcr t7 that should arrive tomorrow and I will be able to check
all the passive components

I would avoid de-soldering the hybrid as this version has the crimped pins and they usually detach from the body

Once I will have checked the components I can try to replace the roms and see what happens

I wanted to power it and see if the hybrid is generating a good ph1 and ph2

Also I would like to check the correctness of Vgg Vss and Vbat

Can I do that or I can fry something at this stage ?

Thanks again and take care

[teenix]

(03-18-2022 09:35 PM)albertofenini Wrote:  I wanted to power it and see if the hybrid is generating a good ph1 and ph2

Also I would like to check the correctness of Vgg Vss and Vbat

Can I do that or I can fry something at this stage ?

Thanks again and take care

I guess you have to try or you won't know. It won't take long to test if everything is connected before switching on.

cheers

Tony

[GreyUser]

The basic 65 information can be found here: https://www.hpmuseum.org/cgi-sys/cgiwrap...i?read=389

The HP-65 was documented by Tony Duell and his hand drawn schematics are available on the HPCC cd-rom.

[rprosperi]

(03-19-2022 05:53 AM)GreyUser Wrote:  The HP-65 was documented by Tony Duell and his hand drawn schematics are available on the HPCC cd-rom.

Tony's Schematics are available online now, individually or the full collection:

http://www.hpcc.org/cdroms/schematics5.0/index.html

[albertofenini]

Hi everyone,

after two days of sitting on a heater the CPU board is now (mal)functioning again

It doesn't turn on at the first try, but after a couple of tries it stays on and performs calculations
also small programs have been written and tested

We have checked the main voltages and with a Batt of 3.9 V we have measured the following
Vgg - 12.56, Vss (no card reader connected) 6.4, Vbsw 3,844 and Vcc 7.96

Also, we have measured Ph1 in, Ph2 in, Ph1 out and Ph2 out and these are also ok
as you can see in the updated photo album

However, it doesn't read nor write cards

We have then swapped this CPU board with a working one S/N 1333A20064 and this assembly
(1333A04841 card reader with 1333A20064 cpu board) is reading just fine, RA and RB are also
in the updated photo album

We may say the card reader (at least for reading) is working and the wheel correctly sized and aligned

We have then put back the original CPU board, and surprisingly the RA and RB curves are also fine
but still the unit gives error at the end of the card reading operation

As a counter test, this CPU 1333A04841 with the other unit card reader 1333A20064
is not reading as well, so is safe to say that this is a problem specifically with the CPU board 1333A04841

What could it be and what can we further check ?

Also, since it still does not turn on at the first try, is it worth to change all the caps ?

Is there a more precise method to test if the gummy wheel is correctly aligned
(like measuring the current drained from the motor ?)

The updated album is here :updated album HP-65 1333A04841

Thanks to everyone and take care, have a good weekend !

[teenix]

In one read oscilloscope trace the pulse widths look longer. I know the data clocking is somewhat controlled by having the RA RB data bits interlaced, but I don't know how the width of the pulses affect the data.

The 65 relies on correct timing for cards due to the method of data transfer to and from memory.

I don't know what generates a read error, because there are no circuit details inside the hybrid, but I suspect it must come from the CRC chip and may be due to a program memory under or overflow during reads. Too many or not enough bits read for example.

There could also be a set amount of clock cycles allowed to read each change of a data bit, and if these are exceeded an error occurs. (Just guessing)

Can you write to a card with the faulty CPU board. If so, does this card read ok.

The "Init" signal from the hybrid is there to reset the circuits after switch on. There is a test point for this available. Maybe monitoring this at switch on you will be able to see the change of state in relation to the power coming on. It should change to the operational state after the power supply has stabilized. I imagine the operational state is a Logic HI.

If it is not too difficult, changing the tantalum caps shouldn't hurt. There is a 2.2uF cap connected to the "InitC" pin on the hybrid. (Pin 36) This might need changing as it is responsible for the timing for the reset pulse.

cheers

Tony

[albertofenini]

Hi Tony,

thanks for getting back,
we tried to write a card and we weren't able to do so, but we didn't capture any trace, will do that later

We can change caps but with caps of the same era from another board, we'll start with the Init one
and see if something change, will post more later today

Thanks for help !
Take care

(03-19-2022 07:39 PM)teenix Wrote:  In one read oscilloscope trace the pulse widths look longer. I know the data clocking is somewhat controlled by having the RA RB data bits interlaced, but I don't know how the width of the pulses affect the data.

The 65 relies on correct timing for cards due to the method of data transfer to and from memory.

I don't know what generates a read error, because there are no circuit details inside the hybrid, but I suspect it must come from the CRC chip and may be due to a program memory under or overflow during reads. Too many or not enough bits read for example.

There could also be a set amount of clock cycles allowed to read each change of a data bit, and if these are exceeded an error occurs. (Just guessing)

Can you write to a card with the faulty CPU board. If so, does this card read ok.

The "Init" signal from the hybrid is there to reset the circuits after switch on. There is a test point for this available. Maybe monitoring this at switch on you will be able to see the change of state in relation to the power coming on. It should change to the operational state after the power supply has stabilized. I imagine the operational state is a Logic HI.

If it is not too difficult, changing the tantalum caps shouldn't hurt. There is a 2.2uF cap connected to the "InitC" pin on the hybrid. (Pin 36) This might need changing as it is responsible for the timing for the reset pulse.

cheers

Tony

[albertofenini]

Hi Tony, Hi everyone

we did some more tests ...

we tried to write a card, just after turning on the unit
the updated photo album shows a sequence of WA and WB for the 1333A04841 assembly

also writing a card after keying a small program failed, the recorded cards shows content that
has nothing to do with the original program

after reading a card, like any from the standard pac, all the memory steps are equal to 35 01

we have located the cap connected to pin 36 of the hybrid, is a 3.3 microFarad,
we have replaced it and specs can be seen for the old and the replacement one

sometimes the unit seems improved, sometimes still needs two tries to turn on

however, it still doesn't read

even if the card reader has proven to be working with another CPU, we have tried
all the possible adjustments with the eccentric but still nothing happens

also, we have monitored the Init Status at the testing point for both units,
they are quite different and both can be seen in the updated photo album

updated photo album is here : updated album for HP-65 1333A04841

what's funny is that after the first cleaning, it has red and written tens of cards ....

thank again for all the support and take care !

[teenix]

I attach the reset signal at switch on from my oscilloscope.

Top trace - reset, about 45mS

Bottom trace is VCC rise.

cheers

Tony

[albertofenini]

Hi Tony

What do you think of the Init trace of the non working unit
compared to the other ?

Does it gives you any idea about the behavior?

Take care !!

[teenix]

The trace I posted is what I would expect, held low while the power supply stabilizes and then when the 3u3 capacitor charges sufficiently, the output goes HI allowing the logic devices to start working.

The output should stay HI from then on.

On other Classic models this was from a clock driver and reset IC, (this may be internal in the 65 hybrid), in the 97 it is a simple transistor/resistor/capacitor circuit, Spice have internal logic resets.

Are you sure the trace was on the correct test point? I see yours oscillating which shouldn't happen.

If the Y axis (voltage) is the same for each model you tested, they are quite different from each other, even from the one I posted.

cheers

Tony

[albertofenini]

Hi Tony

Yes the scale is the same for both units
and the testing point was the fourth from bottom up of the testing points at the left side of the board
That should be the Init
Tomorrow will try to combine it with the Vcc like you did and see what comes out
Thanks again and take care !

[albertofenini]

Hi Tony, Hi everyone

few more updates on this unit 1333A04841

We have traced Vcc and Init on both units, screenshots are available in the updated photo album
they definitely look the same to us

We noticed that the unit 1333A04841 when it doesn't light up at the first try and if it's not switched
off for the second try, few seconds after the first try shows a blinking 0.00

We tried to compare RA and RB, we have been more accurate in taking the traces
screenshot are included, they really looks similar between unit 1333A20064 (working) and 1333A04841 (not working),
still the second doesn't read

However, in few occasions, the unit 1333A04841 has completed reading a card, sometimes a steady 0.00
has shown up a little too early and other times the screen remained blank, in both cases however the
execution of whatsoever was read on the card was erratic, two videos are included in the updated photo
album

Hope this can help ...

Thanks to everyone, take care !!

Updated photo album is here : Updated album for 1333A04841