Repair of HP3437A

[Atomillo]

I recently acquired an HP3437A system voltmeter through eBay. Not because of need, but because of cheap (and curiosity).
Interesting piece of equipment. Aside from the mediocre specs (nowadays even regading speed, its main selling point back in the day), it seems to be a model with relatively few info out there.

Of course, it was non functional. When configured for 1 reading in the 10V range it seems to work. However, increasing the number of readings or changing the range, the displays shows 9999 and the ignore trigger LED lights up.
So far, I've just checked the test points in the analog board and everything seems to be okay. I will start going through the manual. At a first quick reading I'm suspecting the optocouplers but nothing concrete so far.

I attach a few pictures. More will follow soon.

A few interesting tidbits: both input connectors (input and rear) are triaxial connectors (I wonder, if I sold those would I recover the 60 euros spent in the unit?  ). Also very nice is the way the boards unfold and make testing a relatively simple matter.

[Atomillo]

Well, the meters work perfectly, I was just being dumb by trying to use it without reading the manual.
Selecting any NBROFREADINGS will make it so that it takes those readings after a trigger and then locks up.
Of course, I decided to check the function of the key only after measuring all the testpoints and seeing everything perfect.

Don't really what to do now. I originally bought it looking for a fun repair.
Guess I will change the triax connectors and leave it at that 

[wn1fju]

The 3437A is indeed a curious piece.  Its chief utility is the speed of measurement.  I have seen it called out in the performance/calibration procedures for HP function generators where, for instance, it samples a triangle wave and is used to verify the linearity (nothing that can't be done today with a good digitizing oscilloscope).

I obtained two of these units for next to nothing years ago.  I sacrificed one unit for parts.  Indeed, if you work on HP equipment from the 1980's that all use the same front panel leaf-spring switches, you will find that the switches are heat-staked to the front panel boards.  If you need to replace a switch, you have to find one that is new and has not been previously heat-staked.  At least on my 3437's, the switches are not heat-staked; they are simply held to the board with a bracket.  This makes the 3437A an ideal source of "new" switches. 

[Atomillo]

My idea was to use the thing only as a sample and hold (since it that regard it has decents characteristics) by drilling new holes in the back (S/H Output, S/H Reset)

[precaud]

Deane Jensen turned me on to the 3437A back in the early 80's, it's a cool little digitizer. With his help I wrote software for the HP 9845B that optimized its use. It was a bit of a sleeper back then, marketed as a systems voltmeter without much application support. The local DOE labs used a ton of 'em. From the front panel it's of limited use. But over the bus at or below its maximum GPIB xfer rate (~5k sa/sec IIRC), it was as good or better than any other 12-bit digitizer and just as fast. But it also has >1MHz analog BW and 1uS sample intervals. For a while it was the only 12-bit 1Msa/sec digitizer available, and the only way to exceed the 100kHz high-freq limits of the FFT analyzers of the day. Some interesting analog design features in it. I still have it but haven't powered it up in recent years.

[HalFoster]

They are good sources for parts - especially the front switches. They are not heat staked and can be removed very easily - very often needed for repair of other HP gear.

Hal

[Atomillo]

I've decided to keep it.
For the moment, the only modification is that I removed the triaxial connector on the front for a BNC so that I can use it (unfortunately, I scratched a bit the front panel paint. Isn't a big scratch but enough that everytime I see it I will remember it.)

GPIB (sorry, HPIB) really seems to be the way to use its full potential.

[precaud]

IIRC, the 3437A could transfer data in either BCD (slow) or binary (2 bytes/sample). With the latter you have to take care of the range scaling and other things. If you want to play around with it on the bus, I can dig up and send you some code (written in HP BASIC, which you can convert to whatever) that handles the binary data.

[Atomillo]

Many thanks, but I must refuse!
Just a few weeks ago I managed to build a really simple USB-GPIB adapter with great difficulty.
I imagine that working at such high speeds must be complex and I doubt I could pull it off.
Still thanks again for your kind offer.

[TerraHertz]

I've decided to keep it.
For the moment, the only modification is that I removed the triaxial connector on the front for a BNC so that I can use it

That was probably a mistake. Cheap triax connectors can be found on Aliexpress:
    F:\__Equip_info\__Hewlett_Packard\___4061A_semiconductor_test_system\Finding_triax_cables

And there's a reason they used triax.

(unfortunately, I scratched a bit the front panel paint. Isn't a big scratch but enough that everytime I see it I will remember it.)

Nice shiny smooth ring spanners are your front-panel non-scratching friends.

[precaud]

Just a few weeks ago I managed to build a really simple USB-GPIB adapter with great difficulty.
I imagine that working at such high speeds must be complex and I doubt I could pull it off.

Handling 10kB/sec is not at all "high speed"... even GPIB interfaces managed by a 4MHz Z-80 cpu could handle it easily...

[Atomillo]

I've decided to keep it.
For the moment, the only modification is that I removed the triaxial connector on the front for a BNC so that I can use it

That was probably a mistake. Cheap triax connectors can be found on Aliexpress:
    F:\__Equip_info\__Hewlett_Packard\___4061A_semiconductor_test_system\Finding_triax_cables

And there's a reason they used triax.

(unfortunately, I scratched a bit the front panel paint. Isn't a big scratch but enough that everytime I see it I will remember it.)

Nice shiny smooth ring spanners are your front-panel non-scratching friends.

Hi:

I thought the triax was a weird addition because the guard terminal was not connected anywhere.
I could not find the document you mention, could you send it to me or tell me if there is anything I missed?

Many thanks