HP 3478A ROM Dump

[jonwilhelmjr]

I recently Bought a used HP 3478A. I got to looking at the schematic and service manual, I got thinking what if i tried to duplicate this schematic.

Well If I were to attempt this project I would need firmware either for an Intel 8039 or modern Micro.

My question has anyone tried to remove dump the ROM. If not.....How would my youngster self dump the rom?

[Kleinstein]

The HP3478 uses quite some custom chips (the input hybrid, the ADC hybrid), that are not that easy to replace. The firmware would be only one hurdle.

With many, expecially more modern µCs there is no good way to extract the firmware. With the old i8039 one may still be lucky and no protection in place.   If there is an easy way to read the ROM, chances are one can find some copy at the XDEVs web site.

[pqass]

There are two MCUs; U501 and U462. 
The first is ROM-less so its program is stored in U502. 
The second has an internal ROM.  I don't know if its readable with a programmer.

See this link (search for 3478) for manuals (incl. schematics) and ROMs.
I'm having trouble downloading the files with Firefox today.  Maybe it's what you're looking for.

[jonwilhelmjr]

Excellent information and Links. Will my efforts be fruitful or am i just seeking out a learning opportunity and a whole lot of long agonizing pain.

[pqass]

Also, there's this head start from member @fenugrec.

[jonwilhelmjr]

I have created a discord channel I am calling OpenDMM

https://discord.gg/4FTczy9D

[Miti]

Will my efforts be fruitful or am i just seeking out a learning opportunity and a whole lot of long agonizing pain.

Yes. 

[Kleinstein]

Trying to copy the 3478 ADC is a bit tricky because of the ADC hybrid, that contains quite a lot of precision resistors and matched switches. It seems to be the same hybrid as in the 3457 and one may find additional information on the hybrid there.
I see little sense in trying to design a hardware replacement for the hybrids and still rely on old µCs and software that is only available as a non autherized binary copy of a commercial product.

A simpler ADC to coply the MS3 ADC of the HP34401. This gets away with less hardware. The ADC uses an ASIC, but a more modern µC should be able to also replace the ASIC and possibl even improve on the performance with a 12 Bit instead of 10 bit µC internal ADC.
I have build a more advanced version with still similar hardware effort (STM32L051 for control, 4053 type swiches at the ADC and LM399 ref., 2 x ORN type resistor arrays). This is less hardware effort than a HP3478 copy and a lot better performance (e.g. 7 digit resolution at 1 PLC). The SW side is however no yet that complete (e.g. currently only 1 PLC and no scaling for the modes with gain) - so more programming, but also the option to add functions as wanted / needed as programming is in C.

[MathWizard]

I know there's other, maybe older and simpler DMM's that have been hacked, and people even used modern MCU to control them. IDK about all the other chip's in there, but something like a 5.5digit Keithley 197, I've seen the full program for, and I think most the parts are swap-able to something obtainable.

I'd say some of the early handheld DMM's would be a fun project too. And for sure, there are open-source DMM projects, even up to a 8.5dight DMM, or was it 7.5, but it was really impressive, and open-source, there's video's of it on youtube.

[jonwilhelmjr]

Thank you the Keithley 197 is a good point. My initial thought was to take the schematic of older DMMs and modernize the schematic with 2020s part numbers. Then create layouts for the schematics. Order the boards, build them, and try to calibrate with quality precision parts.

After more thought, this adventure of mine may become a path of learning the original design intent and then use modern parts to implement the design intent. To build, test, and calibrate with quality precision parts.

On a different note, I am most differently a hardware guy that knows enough software to be dangerous. suggestions on the software/firmware adventure?

[Kleinstein]

The HP meter hide quite a bit of the hardware effort in the hybrids / custom chips.
The Keithley 19x series may be easier, with less / no custom parts.
In all these cases one has copyright problems with the software and a very hard time to adjust things on the software side, with only a binary copy. Doing changes on the ASM level can be really tricky.
The CPUs/µCs are more like old types that can be hard to get and may need external memory and support chips. In this repect the HP3478 is relatively good choice - but the analog parts is tricky.
A modern µC would have a lot easier hardware (internal RAM and Flasch) and could be programmed in C with a modern compiler.  Debugging the actual ADC operation can still be triky as the HW side does not stop and wants to run in real time. The SW part besides the actual ADC control is usually relative simple, at least for a more basic meter. Having control over the software allows more freedom with the hardware too (e.g. choice of latching relays that were not that popular back than or the use of a standard LCD module for display).

To really get an openDMM one would need to also make your own software. It may be easier to fork from an existing more or less open source project than copying an old comercial meter. So one may start with the tricky SW part ready or at least in a working state.
The logical way for a new 5 digit (and lower end 6 digit) meter is now a ready made ADC chip (or even a speciallized DMM chip set) - so relatively little hardware developement.