SOLVED: HP 3458A repair? ideas needed. Does someone have ROM-images?

[mzzj]

The old trusty 3458A failed.

Has had problems with "ram test low error" for a while if powered down but this has been "fixed" by resetting the unit after power-up. But now the 3458A has failed totally.

Now the meter is non-functional:
-No relay clicking on power on
-Display with all pixels lit
-Beeps on key presses, resetting the unit makes no other effect than "beep" sound.

Power supply voltages look good.

Anyone with similar experience or ideas where to start?

This is old unit with date codes from -89 so can expect pretty much everything.
EPROM gone bad?? does anyone have ROM images for 3458A?
Wont need latest or greatest version

[tautech]

http://www.ko4bb.com/getsimple/index.php?id=manuals&dir=01_ROM_Images_and_Drivers

[Dr. Frank]

Hello,
the first error relates to dead batteries inside the nvSRAMs (DALLAS).
This has been discussed in several forums, how to save the content of the cal-RAM, and reprogram it into new parts.
As these batteries last between 10 and 20 years, a 1989 instrument should have been repaired long time ago.

The second error is not common; it's the first time, that I heard of that.
Most probably, it's located on the A5 outguard controller (main processor), as display and hardware are not controlled correctly.
Well possible, that one EPROM lost its content.

Please visit xdevs site, he also hosts all FW versions: https://xdevs.com/fix/hp3458a/

For a first attempt, you may update your six EPROMs inside your unit to the latest FW 8.
FW 9 is also available, but that's not necessary.



Frank

[mzzj]

Hello,
the first error relates to dead batteries inside the nvSRAMs (DALLAS).
This has been discussed in several forums, how to save the content of the cal-RAM, and reprogram it into new parts.
As these batteries last between 10 and 20 years, a 1989 should have been repaired long time ago.

The second error is not common; it's the first time, that I heard of that.
Most probably, it's located on the A5 outguard controller (main processor), as display and hardware are not controlled correctly.
Well possible, that one EPROM lost its content.

Please visit xdevs site, he also hosts all FW versions: https://xdevs.com/fix/hp3458a/

For a first attempt, you may update your six EPROMs inside your unit to the latest FW 8.
FW 9 is also available, but that's not necessary.


Frank

Agreed, should have replaced the nvRAM chips long time ago.
I was even fully aware of fact that the nvRAM's are failing at some point but the unit was located at our second lab at the other side of the country and I had always "more important" things than preventive mainteinance going on.

Thanks, I had been trough Xdevs site but failed to see the ROM links 

[TiN]

Since NVRAMs are used as SRAM also, sick chips can cause cpu to hang and then FP MCU (located on A6 board) will just lit all segments and beep on any keypress, as it can't talk to main processor and get upset. I saw this behaviour when I was playing with SRAMs.

Firmware dumps are right there : https://xdevs.com/fix/hp3458a/#firmware
And tool to read CALROM (both for Win and Raspberry Pi linux-gpib) is down in "Calibration" section.

Likely your cal data is lost, but I would try reading it anyway just to make sure. Once you replace NV SRAMs (2 x DS1230Y) meter should be able to boot (if no other faults). If it will throw errors CAL DATA LOST and such, well, you know what to do.

[mzzj]

Since NVRAMs are used as SRAM also, sick chips can cause cpu to hang and then FP MCU (located on A6 board) will just lit all segments and beep on any keypress, as it can't talk to main processor and get upset. I saw this behaviour when I was playing with SRAMs.

This* was one of my theories also but I havent got replacements yet to test the theory.
* (FUBAR sram causing the cpu to hang before being able to print out any error messages)

Good to hear that someone is familiar with the problem. 

I'll keep my fingers crossed that the cal ram would hold the data until I get things running again.  Calibration was going to expire soon in any case so the meter is going to cal lab  in any case.

[TiN]

Be sure to have meter run at least two weeks non-stop prior sending for cal, to get reference stabilized again. Also worth to check ADC stability per Service Note 18. I had details covered in my article.

[mzzj]

Replaced the ds1235 ram  with M48Z35-70, no change for better. Waiting for the rom-chips and prommer to try fresh firmware.
This is starting to get intreresting. 

[alanambrose]

>>> no change for better

That's a little odd - I would expect a rom error if the eproms were dodgy. Maybe the DS1220Ys?

This may help - and I would replace them all the parts on that list while you're there:

http://anagram.net/nuts/3458A/3458a refurb parts.xlsx

...and put the SRAMs in sockets if they don't already have them.

Alan

[mzzj]

>>> no change for better

That's a little odd - I would expect a rom error if the eproms were dodgy. Maybe the DS1220Ys?

This may help - and I would replace them all the parts on that list while you're there:

http://anagram.net/nuts/3458A/3458a refurb parts.xlsx

...and put the SRAMs in sockets if they don't already have them.

Alan

AFAIK DS1220 is only for calibration constants and it would not render the meter to nearly dead state?

Tested so far:
-All oscillators running, clock signals marked in schematics look ok.
-Main mcu data and address buses look alive, (front panel reset causes activity)
-Display/sentry board mcu U400 alive, pin17 grounded shows character test display.
-HD63A50 Serial link between main mcu board and display/sentry board looks alive
-

[ManateeMafia]

Do you have a GPIB adapter? This may tell you if the digital board is responsive and the problem may only be in the display board.

I don't know if a bad display board could hang the rest of the meter but if you have activity on the address and data lines that could be a plus.

Bad cal ram should not cause your problems. Normally it would just throw errors but keep on working.

[TiN]

Or you can read your current ROMs and compare to dumps.

[mzzj]

Do you have a GPIB adapter? This may tell you if the digital board is responsive and the problem may only be in the display board.

GPIB adapter is on the way, along with TL866 prommer. I  have access to those at local HAM/radio club also but I was busy last week and didn't make it to the club meeting.

Main MCU board and display board look like standard components mostly without exotic custom IC's.
But there is couple of buggers that can make repair way more difficult if not impossible: 4pcs programmable Holtek  GAL chips and the display controller 80c51 MCU  with internal ROM.

--
TiN: didn't see ROM FW version is 5.4 available on web so cant make comparisions

[TiN]

Can you upload your 5.4 fw? I can take a look on it as well.

[ManateeMafia]

If you haven't done so already, you can download the following utilities...

http://www.ke5fx.com/gpib/readme.htm

There is a utility to backup the calram. It may help if you are interested in preserving the old constants.

[alanambrose]

Starting with the easy stuff - see if both power supply boards are working properly and change the calram sram chips. If you can recover the cal data, that would be great. What's the date code on the ds1220s? And yes, I do think that dead calram would bring the meter to its knees.

A.

[mzzj]

Can you upload your 5.4 fw? I can take a look on it as well.

I will if there is some interest for it once I get the programmer.

[mzzj]

Starting with the easy stuff - see if both power supply boards are working properly and change the calram sram chips. If you can recover the cal data, that would be great. What's the date code on the ds1220s? And yes, I do think that dead calram would bring the meter to its knees.

A.

This was my 1. step.

calram (DS1220)  datecode 9012. I'll try replacing the calram because I dont have anything better  to try at the moment.

[Dr. Frank]

For test purposes, you may replace all three nv-RAMs by their SRAM counterparts, i.e. 6116 for the cal-RAM, and 2x 62256 for the others.
Maybe you already installed the opt. 001 (extended memory), then you may use 2 of these.
Frank

[mzzj]

For test purposes, you may replace all three nv-RAMs by their SRAM counterparts, i.e. 6116 for the cal-RAM, and 2x 62256 for the others.
Maybe you already installed the opt. 001 (extended memory), then you may use 2 of these.
Frank

Replaced now all the NVRAMS with ST alternatives, no luck or change in behaviour. 
Hooked the original DS1220 calram to external +5v power supply after removing it from the board in a hopes that I am able to still read it once I get prommer to my hands.

[mzzj]

SOLVED: flashed fresh ROM images and meter become alive!

[TiN]

Great and expected news. Now you have load of work to test everything.

How about old image, did u manage to dump ROMs? I'd keep them in case someone with good copy can help confirm validity.

[mzzj]

Great and expected news. Now you have load of work to test everything.

How about old image, did u manage to dump ROMs? I'd keep them in case someone with good copy can help confirm validity.

https://www.dropbox.com/s/006ptzvzzn722ea/hp3458a%20rev5.3roms.rar?dl=0

[mzzj]

Tried to erase the orginal ROM chips just for fun and curiousity.
Found out one of the original 27c512 chips wont erase completely, has row of DF pattern in trough most of the data. Also measured the suspect data io to ground and vcc pins but no shorted protection diodes.

[mzzj]

And as a side note these were ugly fakes:
http://www.ebay.com/itm/171331916013?_trksid=p2060353.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

Or probably not exactly fakes but piss-poor old recycled scrap labeled with new datecodes.
Topside shows datecode 1132 and bottom 9622 

Not a problem for my repair project because I used new ST nvrams from DIGIKEY