Agilent 34401A Front Panel Hacking

[mower]

Hello Everyone,
     I was trying to fix some stubborn buttons on my 34401A multimeter. In my frustration, I shorted something I shouldn't have and now the front panel is dead. Taking inspiration from user qu1ck, I am considering a project to replace the front panel with my own custom design. https://www.eevblog.com/forum/repair/hp-34401a-dmm-with-leaking-segments/msg3953096

If I'm able to get a prototype working, I was thinking I could add even more features such as sd card data logging and wireless connectivity.
I have another working 34401A at my disposal to sniff out the serial communications to the front panel.

From my current understanding, the logic levels of the serial communications are 0V and 5V, and floating with respect to earth. Is it safe to connect my knockoff Salee logic analyzer to the main board? Since my laptop would be refrenced to earth, the gnd on the meter would become earth refrenced.

Another question I have is, which power rail would provide the most power to my system, and is it acceptable to use a switch mode power supply for increased efficiency? I see that there is +- 18V, as well as the filament AC rail straight from the transformer, originally for the VFD filaments. Do you think this could provide enough current?

I still have a lot to research on my own but I wanted to start a thread to collect criticism and feedback since I know there are many 34401A experts on this forum.
Thank you for your time,
mower

[iMo]

There are the two isolated inguard and outguard parts on the mainboard (they communicate via optocouplers). Those are floating, and the front panel is floating too, afaik, and the panel's signal voltage levels are shifted (there are 2 variants of the voltage shift). Therefore you have to be careful with your logic analyzer and notebook (both have to be floating too). I doubt you would find more info on that project than in the above link (and the sources for it on the github).
The filament is powered off a floating AC 6V as you may see in the Service Manual (you have to study too). That could be used for powering your new front panel unit when wired properly (after making a DC voltage out of it you would require for your design).

[Kleinstein]

In the 34401 the isolation is between the input section and the GPIB interface part. The dispaly part is connected to the input part.
As long as the DMM is only used with an isolated test voltage it should be OK to connect a logic analyzer, provided the correct local grond is used. Of cause one should not measure mains or other higher voltages while the logic analyzer is connected. A grounded test voltage would also be a problem.

[coromonadalix]

hum  i think a member issued an pcb  substitute  on some thread  not long ago  ... 

Do some searches

[Miti]

Outguard should not be floating, why would it? It contains the IOs and can be connected to a computer. Inguard should be floating.

[qu1ck]

From my current understanding, the logic levels of the serial communications are 0V and 5V, and floating with respect to earth. Is it safe to connect my knockoff Salee logic analyzer to the main board? Since my laptop would be refrenced to earth, the gnd on the meter would become earth refrenced.

Sure, just understand that your meter is no longer floating, don't go measuring mains with it while it's connected to your pc. I would have my meter connected directly by usb to pc while I was developing firmware all the time.

Another question I have is, which power rail would provide the most power to my system, and is it acceptable to use a switch mode power supply for increased efficiency? I see that there is +- 18V, as well as the filament AC rail straight from the transformer, originally for the VFD filaments. Do you think this could provide enough current?

Don't worry about efficiency. You will have to try really hard to exceed the power draw of old 90's era microcontrollers and a vfd display with it's filament heat. I would just use same power rails that standard panel has and stick a linear regulator on it to give you 3.3 or whatever you need. If you want you can use the filament AC, it will provide more than enough power. No need to add switching noise in a precision device when you can avoid it.

[mower]

From my current understanding, the logic levels of the serial communications are 0V and 5V, and floating with respect to earth. Is it safe to connect my knockoff Salee logic analyzer to the main board? Since my laptop would be refrenced to earth, the gnd on the meter would become earth refrenced.

Sure, just understand that your meter is no longer floating, don't go measuring mains with it while it's connected to your pc. I would have my meter connected directly by usb to pc while I was developing firmware all the time.

Another question I have is, which power rail would provide the most power to my system, and is it acceptable to use a switch mode power supply for increased efficiency? I see that there is +- 18V, as well as the filament AC rail straight from the transformer, originally for the VFD filaments. Do you think this could provide enough current?

Don't worry about efficiency. You will have to try really hard to exceed the power draw of old 90's era microcontrollers and a vfd display with it's filament heat. I would just use same power rails that standard panel has and stick a linear regulator on it to give you 3.3 or whatever you need. If you want you can use the filament AC, it will provide more than enough power. No need to add switching noise in a precision device when you can avoid it.

Thank you for the advice!

I was thinking of using a 5.5" OLED, I found one that's based off the same SSD1322 driver used in the 3.12" display. I was thinking of using the bare OLED so that it can be powered by an LDO from the 18V rails rather than the booster circuitry used on OLED breakout boards. This will make the first part of the project just getting the OLED to work. I'll probably buy a 3.12" breakout board just for development and if I get it all working, I'll use the 5.5". This may lead me down some rabbit hole, but I was reading in the SSD1322 datasheet that there is a specific power supply sequence when powering on and off the OLED. Do you think there would be any negative consequences of ignoring the power down sequence? I'm curious how the circuitry on the breakout boards handles this.

5.5" OLED: https://www.buydisplay.com/green-5-5-inch-graphic-oled-display-panel-256x64-ssd1322-parallel-spi
SSD1322 datasheet: http://

My next question is regarding the logic levels of FPDO, the data line from the front panel to the main board. It is driven by an LM339, which from my understanding has an open drain ouput that produces a strong "0" to the negative power rail which in this case is -18V, and a weak "1" to whatever pullup resistor is connected. From the schematic I don't see a pull up resistor, only R613 in series with the output of the LM339 and the MCU on the mainboard. In my new design should I closely follow the original schematic in this regard? Should I go as far as copying the original design and use the -18V rail to power a -13V regulator that powers all 5V logic? I've attached a sub pdf of the schematics.

Thank you all for your help.

[floobydust]

The entire FP board's circuit common is the -18V rail, so that the VFD driver can go that low, as well as to +18V. It does get confusing.

Careful with FPDO, coming out of the LM339 it swings between -18V and open-collector.
R613 23k7 is part of a voltage divider with R506 5k62 (pull-up to +5CV) so main ASIC U501 ends up seeing 5V and 0V swing. You don't want to damage the ASIC.

I had started a replacement Front Panel board, keeping the VFD. I did all the hairy mechanical but lost interest with the chip shortages after I couldn't get parts.

I think OLED display boards with boost-converter or charge-pump would destroy the instrument's noise floor.
The analog front end also runs off the +/-18V, and those regulators don't have a lot of heatsink either.
That 5.5" OLED requires 40-80mA at 15V and guessing 50mA with 50% pixels lit. So it might overheat the Vreg or add too much noise. Something to keep an eye out for.

[qu1ck]

This may lead me down some rabbit hole, but I was reading in the SSD1322 datasheet that there is a specific power supply sequence when powering on and off the OLED. Do you think there would be any negative consequences of ignoring the power down sequence? I'm curious how the circuitry on the breakout boards handles this.

I'm fairly certain everyone ignores these sequences, including standard breakout board manufacturers. Just do a reset by holding corresponding pin down for specified time after it's powered on and you should be good. If you want to be extra cautious just gate oled Vcc until your microcontroller is ready.

[mower]

Quote from: floobydust on Today at 01:47:12 am

Careful with FPDO, coming out of the LM339 it swings between -18V and open-collector.
R613 23k7 is part of a voltage divider with R506 5k62 (pull-up to +5CV) so main ASIC U501 ends up seeing 5V and 0V swing. You don't want to damage the ASIC.

Thank you, I didn't notice R506. I guess in my design I could copy the original design's use of LM339 as level shifters and have 23k7 on FPDO.


Quote from: floobydust on Today at 01:47:12 am

I think OLED display boards with boost-converter or charge-pump would destroy the instrument's noise floor.

This was some of my thinking for using a bare OLED module rather than one attached to a breakout board. I'll still use a breakout board as I develop my own design.


Quote from: floobydust on Today at 01:47:12 am

That 5.5" OLED requires 40-80mA at 15V and guessing 50mA with 50% pixels lit. So it might overheat the Vreg or add too much noise. Something to keep an eye out for.

qu1ck's designed has the OLED powered by 3.3V from an LDO. He said it drew <100mA. Hopefully by not using the booster and just a 15V regulator (or a -3V regulator referenced to -18V) The larger OLED will still draw <100 mA, the datasheet claims 80mA. This is some sus logic and was wondering what you all think of it. Somewhere else I saw someone suggest that the filament supply rail can powers a 10ohm load suggesting it can provide 600mA. If I use that supply rail with a booster, would the noise floor still be affected? Since the power is from an isolated transformer winding.

This may lead me down some rabbit hole, but I was reading in the SSD1322 datasheet that there is a specific power supply sequence when powering on and off the OLED. Do you think there would be any negative consequences of ignoring the power down sequence? I'm curious how the circuitry on the breakout boards handles this.

I'm fairly certain everyone ignores these sequences, including standard breakout board manufacturers. Just do a reset by holding corresponding pin down for specified time after it's powered on and you should be good. If you want to be extra cautious just gate oled Vcc until your microcontroller is ready.

Sounds good to me. Thank you for your reassurance.
I still have a lot to research and learn. I'll order some OLEDs to screw around with. I still have a long way to go before it's worth figuring out a final design. I'm gonna be on edge when I open up a working unit to sniff out the SPI without doing anything stupid and frying the whole thing. What probe points do you recommend? I was thinking of carefully inserting wires with into the ribbon connector on the main board.

Thank you all for your help. Having support from this community is very helpful and encourages me to keep going on with the project.

[Kleinstein]

The VFD filament resistance very likely changes with temperature. So if the 10 ohm are measured with a normal DMM and cold, the resitance in normal operation can be quite a bit higher, like 3 x that.

[qu1ck]

In my design I provided a contact point to solder external 12v for the display but ended up not using it. So 100mA is with the boost converter, if you provide Vcc directly it will be much lower.
Datasheet max figures are for all pixels lit, in reality when displaying some text you have ~20% lit. I wouldn't worry about it too much. Measure the power consumption of native front panel for reference, I doubt it will be lower than a redesign with modern micro + oled.