Agilent 34401A

[Keith956]

Anyone familiar with these? I have one I picked up as not working; it powers on and the display lights all segments correctly, then beeps and then the display remains blank.

It's dated March 2006 and hasn't been got at. I've checked the internal power supplies and they all seem OK and there's no sign of anything fried. Alas these things do not look like they were intended to be repaired!

[Radio Wrangler]

First port of call with older HP/Agilent gear is the HP journal. You'll get a good idea of how it works and why things were designed the way they were if there is an HPJ article on it. Then the service manual will be a lot more digestible. The service manual will have fault-finding trees. Maybe you don't have the exact equipment the manual calls up, but you can get by wit whatever you can lay hands on with a gross fault like not working.

Expect TWO microprocessors. One at the nice earthy voltage side of things and one on the floating measurement side of things.

Ummmm Did you check the power supplies on the floaty bit?

DAvid

[Keith956]

Thanks David, I will do some googling for the HP journals.

I have the service manual, but it's only helpful for error codes that are reported during self-test. It doesn't appear that the unit is getting as far as that, or perhaps the communication with the display has a problem (I'm guessing the display does it's own self test and illuminates on startup, then waits to establish communication with the board cpu?)

I have checked the floating side PSUs, the +/- 18v supplies are in spec, the 5V supply looks good... but the reset output pin of the LM2925 is switching at about 4KHz. I first thought this might be the regulator gone, but temporarily disconnecting it from the board showed the reset line was being driven by the main cpu (80C196) which I believe can do that as it its reset pin is bidirectional. And I read somewhere that one of the interrupts causes the cpu to assert the reset pin low; perhaps the cpu stuck in an initialisation loop? Its clock is OK at 12MHz and the RD pin has activity that tens to suggest it's doing something.

Also, apparently the SRAM startup test result is not output to the display, but can be queried via the HPIB... alas I don't have a gpib card to try and talk to it with. It's not proving to be the simple repair I'd hoped!

[Radio Wrangler]

For GPIB, use the sort of USB-GPIB adaptor cable that's now on the go.

Back in the days of GPIB cards for PCs, the couple of chipsets in use had bugs. We at HP learned never to use HP cards, but to use National Instruments ones. The chip manufacturers refused to re-spin their parts to fix the bugs as there wasn't enough turnover to justify it.

David

[factory]

Correct me if I'm wrong but I thought I read something about being able to switch the display off in the manual.

Edit: Yes it does mention being able to disable the display, but it should be enabled again with a power cycle. Also the menu should still work and bring the display on to use the menus.

Should this not be in the test equipment section as the 34401A is quite an old design, note some Agilent branded ones could be older HP ones with new front covers & buttons, seems to happen occasionally.

David

[Radio Wrangler]

It's a better fit in the test equipment section. That's where people would expect to find it. The 34401A is definitely a classic and had a long production run.

So I've moved it as suggested.

David

[ortek_service]

Originally Posted by factory

Should this not be in the test equipment section as the 34401A is quite an old design, note some Agilent branded ones could be older HP ones with new front covers & buttons, seems to happen occasionally.

David

It even survived into the more-recent Keysight name change: https://uk.rs-online.com/web/p/multimeters/0501354/
- But has now been replaced by various newer colour-LCD models.

I seem to recall the original HP ones had all the same colour 4mm sockets, which was a bit odd, until they changed that. And still had lots of 34401's being used at work until being replaced with newer models a few years ago.

[Dekatron]

Have you tried the self test that can be performed at power on?

Hold down the blue shift button and then power it on still keeping the blue shift button down for at least 5 seconds and then relase it, it should now display TESTING.... and either show PASS or FAIL after some 15 seconds or so.

[Dekatron]

Googling "HP 34401a beeps" found one hit at EEVBLOG: https://www.eevblog.com/forum/testge...artup-problem/ with a similar problem (general hits: https://www.google.com/search?client...p+34401a+beeps) maybe EEVBLOG is a good place to start to read?

I own a few of these and all were bought secondhand, but I specifically asked for the power on test to pass before buying any of them and also for the internal error registers to be clean. I had them professionally calibrated and they were all within their tolerances. I had to refurbish the displays on them as they were a little bit dim, but after that they have been working for many years now without a hitch.

[Keith956]

Thanks Martin, yes I have tried the power on self test by holding the shift key. Does nothing alas. I’m going to try querying it via the GPIB when I can get hold of a gpib-usb adapter, as apparently the onboard sram test result can be accessed that way, or so I read.

[Keith956]

I managed to get hold of a NI gpib-usb adapter and downloaded a trial copy of LabView to try and talk to the 34401A.

First tried to talk to another DVM on GPIB and could interact with it fine. But when I try the faulty 34401A, it 'sees' an instrument at primary address 22. But sending it an *IDN? command gives error EABO, a timeout. So it would seem that the earthed logic is not active?

[Dekatron]

Did you check that first link at EEVBLOG that I wrote, it had some repair information that seemed to be common, maybe you could check it in your meter to see if it has the same problem.

[Keith956]

Yes Martin I checked that link; the 12MHz clock is fine as far as I can see as is the earthed 5V supply. Surely the fact that an instrument can be 'seen' at a gpib address means the interface chip is running - although not responding to external commands, which is odd.

[Radio Wrangler]

Probably hung up waiting for something. You've said the earthed side 5v supply is up. But what about the floating side supplies? It could be waiting for a 'measurement done' signal from that side.

David

[Keith956]

Originally Posted by Radio Wrangler

Probably hung up waiting for something. You've said the earthed side 5v supply is up. But what about the floating side supplies? It could be waiting for a 'measurement done' signal from that side.

David

Hi David,

Yes all supplies are measuring correct - the +5V earthed, the +5V floating, and the +/-18V floating. And I've tried swapping the display board with another 'good' 34401A to eliminate that too.

I will try and see if the floating logic is talking to the earthed logic (I believe it's supposed to do so through optoisolators). Might give more clues - I had hoped that the GPIB would be able to tell me an error code rather than being unable to respond at all.

The MCU on the earth side (87C51) controls the beeper which does sound once after the display goes on/off, so it would seem that is doing something.

[Radio Wrangler]

Is the clock oscillator running for the floating side CPU? would be the next port of call.

David

[Keith956]

Yes, both 12MHz oscillators are running. The floating side MCU is periodically taking the reset line low as mentioned earlier. As I understand it, this can happen via an interrupt e.g. on error.

[Keith956]

So in case anyone else has the same problem with a HP34401A...

I spent quite a while looking at the address/data bus on the SRAM and EPROM (pic1) and it appeared that it was getting through some sort of test where the address was being incremented, then in the middle of the sequence it reset and started again. It looked like a 0xFF was being read, which is the reset interrupt vector.

Now someone else had reported a similar fault (display goes blank, reset line active low in a 4.5KHz cycle) and replaced their SRAM chip. As this was a lot easier to test out than burning a new EPROM, a 32k x 8 SRAM was purchased from a Chinese supplier on ebay, the old one snipped out and pads cleaned up (pic2) and replaced (pic3, excuse the poor soldering).

With the board and display back in the case it was time to switch on. The self test passed (pic 4, 5)!

[Radio Wrangler]

Good fix!

David