HP 34401a DMM with leaking segments

[bitseeker]

That's a unique idea. Let me see what I have or could locate that would be neat to send to you.

[coromonadalix]

Maybe send some email of your project at Xdevs  for your / substitution replacement display it could be an good resource

[qu1ck]

xdevs submission process looks way too cumbersome. Uploading docs to ftp is so last century.

I did submit my project to hackaday tho. Maybe they will feature it.

[coromonadalix]

Now on HackaDay

[bitseeker]

Congrats, qu1ck!

https://hackaday.com/2018/08/29/faded-beauty-dmm-gets-an-oled-makeover/

[Zucca]

Congrats, well done!

[HighVoltage]

How nice to see it on hackaday.
Congratulations.

[coromonadalix]

Have you seen this display ?  RS232 interface   would it be possible to adapt / interface it would the physical size fit

http://noritake-vfd.com/gu256x64d-7000bx.aspx

pdf
https://www.eevblog.com/forum/repair/fluke-8846a-repair-shenanigans/?action=dlattach;attach=511772

[qu1ck]

You can interface it (using parallel, rs232 is too slow) but it's too big. I think you can fit at most 135x35mm display without taking dremel to your dmm.
Also it costs half the dmm

[justanothername]

So.
I spent the day trying to get the display working, but I must give up for now, I'm completely stuck.
Installing the platformio IDE and the tools in windows was not easy, i must say this seems to be a linux-only project. I really would apreciate just a .hex file to upload with JFLASH but there is nothing on github.
So I don't really know if I did everything right, since there is a tool needed in Windows to get the driver working (Zadig) and there are many driver options to choose from. I was able to flash the bootloader onto the MCU by soldering SWO wires onto it. And I was able to find the correct settings in Zadig to upload it at least once. The status now is that it displays the bargraph, but nothing else (see attached picture).
I was not sure how to connect to the PCB to the display board of the 34401a, but at least clock and int seemed clear. I tried crossing out MISO and MOSI (DI and DO) but no reaction.  The problem now is that i don't know if it just can't decode the display data or if it the MCU somehow crashes after the bargraph is drawn.
So:
What is the correct connection? Is there a hex or bin that works when i just flash it by SWO? how should I set the boot jumper (I can see by looking the picture that position F is soldered)? How can i assure that the program is working at least?
That was quite a frustrating journey until now

[coromonadalix]

Qu1ck  told me the pinouts used  are in parallel of the falt ribbon cable

here's the answers he gave me :

I took power from floating 5v, output of U553.
Your other option is to take it from input of that regulator, in that case you are less likely to feed any switching noise from oled screen (it has a built in boost converter) into multimeter, but I highly doubt that matters in any way. My measurements did not show any noticeable impact.
But if you take it from input you will be on the limits of lm1117 power dissipation with the polygon I have for it on the board. You can stick small piece of metal as heatsink on it or better yet rip out the boost converter from the screen and feed higher voltage into the panel directly. That will limit current consumption of the front panel converter (my board) and eliminate any switching noise. But I didn't bother, total current consumption of my board + oled screen is 60-70 mA. If you feed 5v into lm1117 it will be just fine. Just don't take it from +18v becase a) it's used in sensitive analog stuff and b) voltage drop for the regulator will be too high.

FPINT is not currently used


Send him an email if you need some help,   in a month i'll order the lcd to try this project for myself too, i have a nice 34401 to play with. 

But he told me his project is not tested with the newest pcb version having the 75518 vfd driver ... i have 2x 34401a with this pcb

[justanothername]

Qu1ck  told me the pinouts used  are in parallel of the falt ribbon cable

For testing purposes i was powering the PCB with a laboratory power supply. Powering was ok, I could even see the resonator working at 8MHz.
I checked all the solder joints end even probed with the scope on the MCU pins, the signals were present. I managed to use the jlink with SWD in platformio and flashed the program without bootloader. Aside from the bargraph beeing drawn, no data was displayed, just  as before. The last thing before I reassembled the unit was that I tried every possible permutation of the cable connection.
I used these contacts at the flat ribbon connector:
GND (1)
IGFPDI (4) - tried at MISO and MOSI and INT
IGFPSCK (6) - connected always to SCK
IGFPDO (2) - tried at MISO and MOSI and INT
IGFPINT(10) - tried at MISO and MOSI and INT
As I can see in the service manual, there is also a version of the display board with an 80C51 microcontroller. Idk if I can determine the used components without desoldering the VFD, but maybe there are different protocols used in later versions. For now I will leave the VFD soldered in, as it is fully functional, but a little dim.
For the powering situation, another way is to use -13V as VCC and -18V as VDD (gnd). Then use the outputs of the comparator U607 as inputs to the OLED-board. Both (-18V and -13V) should then be fed through some ferrites.
Since the OLED panel got boost converters on it, it may be necessary to add some additional shielding. Maybe i will try to improve the EMI radiations of the converter. The usual methods would be to increase output capacitances, to add shunt capacitors from in to out, or to change in/out capacitors to lower-ESR types.
But first things first, the next goal for me is to actually see something on the display.

[coromonadalix]

For the front panel, if you look on the back pcb and see two dip rows of soldered pins  + the ribbons and vfd pins ...  youll know that you have the nec unobtanium display .....

The newest version has only the vfd rows pins and the ribbon pins  and pn: 66512 sticker on it, my other pcb vfd doesn't have the sticker ??

sn : us3610xxxx
sn:  3146-A75xxxx

Both have the firmware : 10-05-02  ??
Version: 06-04-01 and lower = old pcb
Version: 06-04-02 and higher = new pcb

Once again,  qu1ck wrote  not tested with the newest pcb version

VCC  (2)       floating 5 volts  output of U553,  i would not dare to poke on the +/- 18v lines they are for the measurements sections ... it could affect the meter ?
GND (1)
IGFPDI (4)     - should be MOSI
IGFPSCK (6)  - should be SCK
IGFPDO (2)   - should be MISO
IGFPINT(10)  - should be INT     but  not used

We see clairly (picture attached)  he used 5 pins on his pcb, not the 6th (int)

[coromonadalix]

Oh   i think we need the adafruit gfx librairies to compile it ??  The only thing i have on hand is the stm32 blue pills boards,  they are arduino IDE compatibles ... 

I have received from Qu1ck one of his pcb,  but don't have time to assemble it right now,  too many projects on hand, and i go on vacation january 10

There is an firmware.bin (attached)  who is already compiled  .. try this one

The other firmware_debug.bin (attached)   spit out on serial  some data ??

He wrote :  the eeprom is not needed for now ??

[justanothername]

The newest version has only the vfd rows pins and the ribbon pins  and pn: 66512 sticker on it

This seems like my version (attached pic).
I can confirm that I tried the configuration you described and it does not work. Do you know if the panels are interchangeable across versions? If not, then it is obvious that there are protocol differences.
For the firmware: I use the STM32F103CB with double flash size, so I recompiled the project for this controller. However, it should be no problem to flash a firmware compiled for C8 straight onto the CB.
As for the adafruit gfx libraries, you can load them with Platformio within VSIDE, but you need to manually delete SPITFT as stated in his howto.md.
here is the modification to platformio.ini for my board:

Code: [Select]

[env:release]
platform = ststm32
framework = arduino
board = genericSTM32F103CB
board_build.variant = custom
debug_tool = jlink
upload_protocol = jlink
firmware attached.

[coromonadalix]

the pcb are not interchangeable if you dont downgrade the firmware version prior to thoses i've wrote,  come from a Keysight service note ...

[qu1ck]

justanothername
Congrats, you appear to be the first person aside me who built this even though I already sent out a few boards over the last months.

coromonadalix already gave you most of the information and everything they said is accurate. You have already found a way to compile and flash the firmware but just for posterity, prebuilt binaries are available on github, the link is just not intuitive to find for people unfamiliar with github interface:
https://github.com/openscopeproject/HP34401a-OLED-FW/releases

Your display initializes with bargraph, which means firmware works fine. STM32F103CB and STM32F103C8 are exactly the same, even though c8 is supposed to have half the flash, in practice they work fine and can be flashed with more than 64k binaries. There are no guarantees, of course, but neither I nor many other blue pill owners ever saw a c8 chip with less than 128k actual flash.
You definitely have new front panel, which I never tested because I don't have it. Lets try to make this work together and if needed I can make adjustments to firmware to make it compatible with new meters.

First flash the debug firmware (to compile it just add #define DEBUG in config.h or compile with debug target with 'platformio run -t debug'). That firmware prints out lots of information into usb serial so just post here the info that it prints out when meter is just turned on and idling on voltage range. There should be steady stream of updates to front panel.

Second thing you can do is grab a 5-10 second capture of clock and data lines with logic analyzer if you have one (again with meter idling on voltage range).

[justanothername]

You definitely have new front panel, which I never tested because I don't have it. Lets try to make this work together and if needed I can make adjustments to firmware to make it compatible with new meters.

Wonderful! I'll get back to you later next week since I'm off for new years eve and propably have to get sober first.
PS: I have 9 PCBs left that I don't need. If anyone needs one, this is the deal: you send me a PM, I send you my adress, you send me a postcard, I'll send you a PCB back in a letter.

[coromonadalix]

@Qu1ck    could the code be ported to the arduino ide's ??

I do know the blue pill must have the right boot loader ??   could it be viable or doable

[qu1ck]

@Qu1ck    could the code be ported to the arduino ide's ??

Yes, it should be doable since arduino IDE for bluepill uses same framework as I do. But frankly I have no interest to spend time on this, you'll have to figure it out.
My advice: ditch the arduino IDE like the glorified trash notepad with a compile button that it is. Any of the platformio IDEs are miles ahead and completely superseed arduino IDE. I use Atom based one because it came out first but VSCode is even better.

[justanothername]

After a long time a quick update.
I now got myself one of these cheap replacement VFDs from aliexpress. If someone needs the oled display i built, please send me a private message, maybe we can figure something out.

[qu1ck]

Just wanted to note, I'm still open to help you or anyone else with newer meter and some free time to resolve incompatibility with current firmware. Since I only have an older meter I can't do it myself. I would like to see logic analyzer dumps of SPI bus to start.

[justanothername]

Just wanted to note, I'm still open to help you or anyone

Thank you for the offer! The thing is: I got paranoid. I've been watching how the voltage reading of the meter was changing when the case was unscrewed. Then I thought about the extra two clocked devices (cpu @ 8mhz and oled @ i dont know) I was about to put into the housing.
I know it is bullshit and will not affect the precision, but I did shit my pants and went with the replacement vfd.
I'm very sorry.

[coromonadalix]

I do have recent hardware front pcb's in my two 34401a meters, but i dont have the logic analyser thing ....     never used such tools

[qu1ck]

Get yourself cheapo analyzer, any hobbyist should have one in their tool kit.
https://www.ebay.com/itm/USB-SALEAE-24M-8CH-Logic-Analyzer-24M-8-Channel-FPGA-debugging-tool-With-buffer/173828458153?hash=item2878fbc6a9:g:E6kAAOSw56Fc0-yt

It's supported by open source software sigrok/pulseview: https://sigrok.org/wiki/Downloads