HP-42S battery drain

[Marcovecchio]

I bought a second hand, perfectly working HP-42S, but I just noticed it has a huge battery drain problem. I measured 17mA, which will drain new batteries in hours.

I have a disassembled HP-17BII at hand, and I just measured it's current draw: less than 1mA on and idle, and a peak of 2mA when making calculations.

I know this has been discussed many times, and with several HP calculator models (I had an HP-28S that showed the same behavior many years ago), but I would like to insist a little more before giving up. I opened the calculator, removed the PCB from the case and removed every component I could from the PCB: 3 capacitors, the diode, the IR led, and even the SRAM and crystal! When plugging the calculator to a power source, I get the same 17mA. This is with only the processor still soldered to the PCB.

I then washed the pcb with isopropyl alcohol and a soft brush, in an attempt to remove contaminants that might be shorting the delicate processor pins, but it didn't work, still 17mA. I resoldered everything back into place. Calculator works, and still the same 17mA.

May I consider the processor is damaged forever? Is there anything else I can try? Is it possible that it's locked in a "power waste" state that even a reset won't clear? Is there any "stronger" reset than shorting - and + battery pins with no batteries installed?

And a last question: why this happens so much to HP calculator processors? I have read about a lot of cases like this over the years. How can a processor fail to the point it draws hundreds of times the current it used to, but keep working perfectly?

Thanks for your help!

Eduardo

[Gerson W. Barbosa]

Possibly a faulty tiny discrete component you may have overlooked. Lyuka’s webpage on memory upgrade includes a complete schematic that might be useful:

http://www.finetune.co.jp/~lyuka/interests/calc/hp42s/

Boa sorte!

[jebem]

As Gerson mentioned, Lyuka schematics may help.
On a different approach, I would check for a previous overcloking that would increase the current consumption as well.
Does the self test pass without errors?
Good luck with your project.

[Marcovecchio]

Thanks for the schematic link!

Unfortunately, I triple checked the components, I removed them all. There are not many, in fact.

The board has no mods too, the crystal is the original 32kHz. But there's something on Lyuka's pages that caught my attention: there's a mod regarding display contrast, he mentions a switching transistor that may fail inside the processor. The display on this calculator is not very good: when I increase contrast, it seems the background gets proportionately darker. It's a long shot, but this might be the problem. Later I'll print and study the schematics, and if I have the patience, I'll try to pinpoint the processor pin which is drawing the excessive current, to see if it has something to do with contrast.

Thanks for the help!

Eduardo

[Allen]

Could it be tin whiskers? around/under/within the processor? Have you examined the pins with a microscope?

[Marcovecchio]

I didn't try a microscope, but with a 20x magnifying glass, I can see some debris, and I think my alcohol wash should have at least disturbed or displaced any possible loose metal particle, but since the current remains at exact 17mA, nothing moved during the wash.

But thanks for pointing this out!

Eduardo

[Marcovecchio]

I have been trying to investigate the issue in more depth, and I confirmed that it's indeed the processor drawing the current: I located the processor GND pins, and carefully cut the PCB trace. The current measures 0mA. I reconnected the trace and immediately got the old 17mA.

However, I noticed another very interesting thing: current draw goes to normal (less than 1mA) in some screens of the self test routine! ROM, ROM OK, DISP, BEEP and a few others measure very low current, while the others show the battery draining issue. That didn't ever happen during normal use. Even turned off, it draws 17mA.

Probably the self test disables the defective processor circuitry somehow. My last test will be to navigate thru the menus and options, changing flags and options, in an attempt to find any setting that may lead to the same behavior I saw during self test.

Another possibility: the processor certainly has a switching transistor to drive the IR led. Is there an easy way to test if the IR circuitry is ok, if I don't have an IR printer? This transistor might be damaged, and leaking current. Does that make sense?

Eduardo

Eduardo

[Dave Shaffer]

(07-13-2020 02:55 AM)Marcovecchio Wrote:  Another possibility: the processor certainly has a switching transistor to drive the IR led. Is there an easy way to test if the IR circuitry is ok, if I don't have an IR printer? This transistor might be damaged, and leaking current. Does that make sense?

Do you have an oscilloscope?

If so, you can make a very simple IR detector. You need a photodiode or phototransistor (typically less than $1 here in the US) and a resistor and a few volts from some batteries.

Something like this: https://electronics.stackexchange.com/qu...photodiode

[KF6GPE]

You may be able to use a cheap webcam, too; a lot of them are sensitive to IR, and the cheapest don't have good optical filters.

[GreyUser]

The dim display issue is evident from the lcd voltage. Easy to measure. If the voltage is correct, the transistor(s) are fine. A bad lcd is a common cause of a dim display.

You’ve said nothing about the external 8 KB ram... have you isolated that as a possible current consumer? The unit will just beep continuously with no ram connected.

I/R: Desolder one or both leads of the I/R diode. If the transistor is shorted, current will go to zero as it has to go through the diode.

[Marcovecchio]

Thanks for the tips on the IR detector!

About the dim LCD issue, I don't think this is the case, the LCD is good as far as I can tell, and I tried two different LCDs (borrowed one from the 17bII), and got the same result.

About the RAM and IR led: as I said in the initial post, I desoldered all components, including RAM and IR led, battery drain still happens.

But today I got to a point where I have to give up: I located the GND processor pins. They are two adjacent pins at one of the corners. I physically cut the PCB trace where these GND pins are soldered, and the current got to 0mA, so it's confirmed it's indeed the processor. I resoldered the cut, got back to 17mA. Calculator on or off, contrast high or low, calculating or idle, current is always about 17mA. The only situation where it's less than 1mA, is in some self tests messages, but I couldn't find the reason.

Well, it's sad, but this 42s will go to the parts box. Thank you all for trying to help, I learned a lot about this model in this fixing attempt!

Eduardo

[ijabbott]

Is transplanting the CPU from a 17BII to a 42S a viable option?

[Marcovecchio]

If we had the ROM in a separate chip, that would be viable, but no, the ROM code is inside the main processor. So if I did that, the 42s would be fully turned into a 17bII.

As an experiment, I though about making a mod: to install an external on/off switch, that would cut power to the processor, but not ram. That way I could turn the calculator off, stop the drain, and keep ram contents. During use, the battery drain would keep happening, but that would be way better than it is now. I don't know if this would work or not, but I may try it someday!

Eduardo

[Marcovecchio]

I would like to share what I ended doing to this 42s: I did the switch mod, and it works very well!

I attached some pictures. I had a tiny switch in my parts box, which I fixed to the PCB side with epoxi glue. The trick here is to make the switch control the power of the Saturn processor only, not for the whole machine!

This way, I can turn the switch on, then turn the 42s on using the usual ON key. After usage, turn it off normally with shift+ON, and then, turn the switch off. Power consumption will be minimal, just enough to keep RAM contents.

After turning it on again (using the switch and then ON button) I get a "Machine Reset", indicating the processor restarted, but all memory content is intact. The effect is similar to the "coma mode", or pressing ON, XEQ and +

I think this is the best I can do to this calculator for now. In the future, I think I'll adapt three AAA batteries as an extension to the housing, making it thicker and heavier, but this will be a low priority project for the future.

Eduardo