Restoration : HP 4262A LCR meter

[TiN]

Time for some non-typical repairs for volt-nut, meaning myself.

Some might remember my help on 1-day repair of sick 34401A for Taiwan student, Sonic. This is 2nd part with other piece of HP gear.

I have newer 4263B, which got fixed and pimped while ago in this thread. But this 4262A is earlier unit, with date codes on ICs from 1976-78. Older than me, yikes.



Superdirty and ugly looking.



Gives just this on power on. No reaction on buttons, knocking or kicking it...



Will see how this goes..

[ManateeMafia]

Signs of corrosion and it looks like mold is growing on power cable at left. I am sure a reseating of boards and components will be necessary.

Good luck.

I am waiting on your next thread: "Restoration: 42U Rack"

[TiN]

Right after "expanding living space" project perhaps...

Quick reseating board did not help. And currently time does not permit proper service (full disassembly with photoshoot, washing parts, replacing capacitors, etc.)

[Twoflower]

I Like to scan trough these pictures.

Some devices look like corrosion took place. As you mentioned the front panel does not work: The ribbon cable on the lower right shows some red-brown colour coming out of the connector... But I guess you're already aware of this.

[Vgkid]

Looking forward to the repair. I see to many trimmers.

[deepskyridge]

I have a 4261A in similar shape, I am currently looking for a extender card before getting into it.

The front panel switches are well known for being intermittent so will probably do a tear-down and thorough cleaning first.

Good Luck
Gary

[Jwalling]

Whoah!  Check out those white ceramic EEPROMs with the raised round windows! They look like portholes on a boat. I bet they're 2708s. And note the processor chip has a voltage written on it. I saw that once in an audio spectrum analyzer (HP 3582A I think) The CPU board was not working correctly. The voltage at the CPU chip was not correct, and there was a pot to adjust it. After I adjusted the pot to match the voltage written on the CPU, it began working. Is there any ID # on the CPU?

The board located at the bottom left has a couple of transistors (round with white bodies and a black top) that I remember seeing from 1970's electronics stuff. Were they epitaxial types?

Note also the massive crystal - looks like a 3.579545MHz color burst unit from an old NTSC TV.
Very cool! 

Jay

[PaulAm]

That processor is HPs proprietary NMOS nanoprocessor developed by HPs Loveland Instrument Division.  It's basically an 8 bit processor optimized for instrument control.  It has an 11 bit program counter, one 8 bit accumulator and 16 8 bit storage registers with a grand total of 42 instructions.  They used it in a number of instruments and GPIB controllers.  In addition to the 4262a, they also used it in the 3225a function generator, among others.  It could directly access 2K of memory and up to 500K with bank switching.  Pretty amazing what they managed to do with that processor.

Yes, that was 2048 bytes of program memory.

That negative voltage written on the chip is the backgate bias and varied from -2 to -5V.  It varies from chip to chip.

Factory cost was under $27 with an ALU and under $20 without according to documentation I've managed to track down.

I have a 4262a, it's been a nice reliable piece of equipment.  Good luck getting that one cleaned up.

[Macbeth]

That processor is HPs proprietary NMOS nanoprocessor developed by HPs Loveland Instrument Division.  It's basically an 8 bit processor optimized for instrument control.  It has an 11 bit program counter, one 8 bit accumulator and 16 8 bit storage registers with a grand total of 42 instructions.  They used it in a number of instruments and GPIB controllers.  In addition to the 4262a, they also used it in the 3225a function generator, among others.  It could directly access 2K of memory and up to 500K with bank switching.  Pretty amazing what they managed to do with that processor.

Interesting. I found the user manual here.  $27 back in 1975 is now $123 with inflation.

[BFX]

  next interesting thread go TiN  

[Cubdriver]

Hmmm...  This thread might prompt me to take the cover off of mine to take a peek inside.  I got it about a year and a half or so ago, and it's been working (apparently), so I haven't been overly inclined to poke at it.  Might be a good opportunity to try the IPA clean of the switches that aren't supposed to click but most of them now do.

It's a nice piece of gear - good luck getting it back up and running.  (As if there's any doubt at all that you won't ultimately succeed  )

-Pat

[Jwalling]

That processor is HPs proprietary NMOS nanoprocessor developed by HPs Loveland Instrument Division.  It's basically an 8 bit processor optimized for instrument control.  It has an 11 bit program counter, one 8 bit accumulator and 16 8 bit storage registers with a grand total of 42 instructions.  They used it in a number of instruments and GPIB controllers.  In addition to the 4262a, they also used it in the 3225a function generator, among others.  It could directly access 2K of memory and up to 500K with bank switching.  Pretty amazing what they managed to do with that processor.

Yes, that was 2048 bytes of program memory.

That negative voltage written on the chip is the backgate bias and varied from -2 to -5V.  It varies from chip to chip.

Factory cost was under $27 with an ALU and under $20 without according to documentation I've managed to track down.

I have a 4262a, it's been a nice reliable piece of equipment.  Good luck getting that one cleaned up.

Very interesting, thanks. Any idea on how the proper bias voltage was determined on each chip?

Jay

[TiN]

Spent a little bit today to take photos..

As always, all hi-res photos are public in upcoming article draft. Just few best ones here:

Old school CPU and ROMs with some rust on them:



This one has -5.0 bias writing on it  .

Nasty-nasty mushrooms on cable insulation:



Also ribbon cable connector running to front panel from controller broke apart when I pulled it out. That will be easy to fix, just with generic 34-pin dual-row 2.54mm pitch header.

Cleaning time for front panel:



I'll let it dry few days. Switches for keys are almost exactly same as ones in HP 8642B, so I know what to do with them...

[Jwalling]

Spent a little bit today to take photos..

Old school CPU and ROMs with some rust on them:

Look to be 2K x 8 mask ROMs. Given the the garbled display and no response to buttons, bit rot may be an issue if it ain't the power supply. Early HP 8112A and 8116A generators with Mostek parts had that problem as well.
PaulAm commented on that here:
https://www.eevblog.com/forum/testgear/lcrrecommendation/msg533671/#msg533671

There's a set of binaries here, search for 4262A:
http://www.ko4bb.com/getsimple/index.php?id=manuals

Looks like the pinout is different from 2716, though. That could get a bit ugly.
Best of luck - will keep watching your progress!
Jay


Jay

[Twoflower]

By looking through the pictures I noticed by pure luck that there might be a diode cracked on the A11 board.

[TiN]

No, that guy is alright. Just a lights effect on photo 

[cncjerry]

The last problem on my 4262a was that the shorting bar wasn't shorting.  The banana jack was corroded causing it to feel tight but it didn't clamp the bar. It also looks like your has a loose cable to the led display.  I had that problem as well.  I replaced the jacks with some high quality parts though I wish I had the originals.  Took a lot of probing to find it.  Nice meter when working.  Mine has the extra thumb wheels on the right hand side.  Let me know if you need and voltages, etc probed on mine.

[Tomorokoshi]

It hurts to see that corrosion. Looks like someone literally used it as a boat anchor.

[TAMHAN]

Hello Tin,
I have but two things to report here, as I was lucky with mine. Bought it "non working" off ebay, but it lived well so far.

a) Error 1 in the self test almost ALWAYS means unhappyness with the DuT. The meter is REALLY Picky about the shorting lead - I can upload you a picture of the one I use.
b) The warm-up time for the unit is insane. On mine, some self tests usually fail for the first 15 mins after power on.

If I can help in any way, chant a devil's song!

Tam

[TiN]

Thanks for hints. Currently I have meter sitting disassembled, waiting for better time when I could work on it for few hours straight.
Cleaned front panel with hot soapy water and light scrubbing, it went well. That painted silkscreen is pretty robust, so it can take some amount of abuse easily.

For b) one dealing with 7.5-digit meters and 3458A on daily basis, I'm used for half a day warmups and pre-heat periods  .

[Cubdriver]

Hello Tin,
I have but two things to report here, as I was lucky with mine. Bought it "non working" off ebay, but it lived well so far.

a) Error 1 in the self test almost ALWAYS means unhappyness with the DuT. The meter is REALLY Picky about the shorting lead - I can upload you a picture of the one I use.
b) The warm-up time for the unit is insane. On mine, some self tests usually fail for the first 15 mins after power on.

If I can help in any way, chant a devil's song!

Tam

Sounds like something must be amiss - mine passes the self tests immediately after power up.

-Pat

[cncjerry]

Mine passes self tests immediately, probably one of the fastest boots ever.  One of the things I like about it as it rarely has any warm-up variance in values.  I would put a meter on the rails to see if they move during warm-up.

Also, I wouldn't be in a big hurry to fix the clicking switches and if you do clean them, I've used a slight amount of WD40 on the spring.  This will also fix the stuck buttons on Tm5006 modules as well.  All you have to do is put a little on a q-tip and touch the plastic where the leaf spring touches.

[TAMHAN]

Well, just in case anyone ever stumbles over it. Here is a picture of the cable I use.

Stolen with pride from an institute which did me rough some years ago...

[TiN]

Had chance to work a little on this sick box.

One of original binding post is busted, and unlikely that I'll be buying HP original replacement, so went with BNCs instead:



Pitch is not same as modern LCRs, but still useful with usual LCR cable probes.

Also replaced rear rusty BNCs with fresh new ones:



And X-cap:



Cleaned chassis, replaced original fungus coax wiring to brand new pieces.
Almost blown transformer by stupid mistake, but was quick enough to escape the damage

Meter power on as before, and quick PSU checks revealed absence of -12V (+188mV instead on it). So CPU does not get his -5.0V love and hence nothing works.
Supply -12V from Keithley 2400 SMU allowed LCR to boot, but fail selftest with messages FAIL 1 and FAIL 2.

Checked A9Q4, replaced A9R6 and A9C5 - still no -12V. Likely TO-3 2N3055 BJT is dead or LM741 opamp. Negative -12V is simple linear supply, tracking from +12V supply by 10K/10K resistors. Both resistors are OK. I don't have any high power BJTs to try, so will have to wait to get new one first. 

Also discovered that 4261A is very similar box, with most of analog boards same as 4262A. PSU A9 board is even labelled same.

[cncjerry]

wow, that cleaned up nice.  peroxide?  Problem with using BNCs is you can't attach the test set to the front.