HP/Agilent/Keysight Switching Power Supply Controller Repair

[ZGoode]

Howdy Y'all,
I'm working on repairing a broken HP 4277A LCZ meter I picked up a few weeks ago and am 99% sure I have narrowed down the problem to this sub-board on the main power supply assembly.  The part number in the service manual is 1813-0255 and there is practically zero information on the internet about the sub-assembly.  It consists of a 16 pin DIP, 3 SOT-23 components, and a few jumpers and capacitors.  I have identified the DIP as a NEC uPC1042C.  I am pretty sure about the SOT23's but not 100%.  The SOT23's are marked as A3, Y24, and L5.  I believe these three parts to be a 1SS181 diode, 2SA1461 PNP, and a 2SC1623-L5 NPN respectively.  It also appears to be using some kind of ceramic substrate material instead of something like FR-4, something I have actually not seen before.
Has anyone here replaced this component before or worked on a piece of equipment with a faulty one of these?  From what I can find is there are several pieces of equipment from this era of HP equipment that use this subassembly.

[Swake]

That is a power supply switcher chip. Search for application notes, with a bit of luck you'll find something very equivalent to the module.

Similar ceramic boards can be found in philips tv's from the same era. Somewhat fragile, can't solder stuff on it as far as i remember. Philips used to have resistors 'printed' as a layer on those boards then coated with black paint.

[coromonadalix]

yep  reworking on theses are almost impossible

[ZGoode]

So far I have been able to successfully desolder the parts without visually damaging anything (might be damage I'm not able to see).  I'm planning on X-raying it sometime this week to get a better idea of what I am working with here.  The one good bit of news I can see so far is that all of the SMD parts on the board appear to be off the shelf, with the only proprietary part being the ceramic board itself

[ZGoode]

So I made some progress on the internal imaging of the board today.  I took some images using the Keyence VR and SAM.  I'm planning on doing another pass with the SAM to see if I can clear up the image a bit.

[ZGoode]

I went ahead and redid some imaging today on the SAM with a resolution of 5um.  This should hopefully make it possible for me to reverse-engineer the board now.

[Swake]

Beautiful, I want an x-ray machine now too 

The rectangles you see are the resistances, the small lines are laser cut marks to fine tune their value.

[ZGoode]

Believe it or not, those images were not taken with an X-Ray machine.  They were taken using a SAM (Scanning Acoustic Microscope).  I would have used an X-Ray, but mine is temporarily offline while I do some maintenance on it.  I'm lucky to work in a research lab which specializes in NDT, primarily acoustic methods, so a lot of ultrasonics and microphones.
Yeah, I was assuming those were fine tuned resistors.  While I should be able to measure them just fine with my 34401A and a Kelvin lead setup, I'm curious what kind of tolerance they need to hold.  When I go about with my replicated board, I'll probably use 1% tolerance resistors, although if the price isn't that much higher I might bump them down to 0.1%

[ZGoode]

The rectangles you see are the resistances, the small lines are laser cut marks to fine tune their value.

So with these "printed resistors" on the board, I'm curious if there is any other reason a ceramic one sided board was used here instead of a more standard FR-4.  Are those resistors the only reason ceramic is used here?

[Swake]

This is a 'simple' power supply. The chip has a max 100kHz oscillator. Looks like there is nothing fancy that needs particularly high tolerances. The application notes has all E12 standard resistors. 10% tolerance is going to be good enough.

Ceramic was used as carrier in the early days of surface mount components. I think it was more to test out the new manufacturing process than anything else. Gradually increasing manufacturing difficulty while putting volume in the market and testing for reliability. All of these where small boards as far as I remember, that means it doesn't need big machines to be manufactured. Those printed resistors where just another cost reduction probably.

[ZGoode]

I removed the soldermask from the board which ended up making reverse engineering a lot easier.  There are 29 resistance sections built into the substrate.  The 5 smaller capacitors are 10nF (unsure of voltage rating) and the 2 larger ones are 100nF (also unsure of voltage rating).  Based on the 4wire measurements I got this should be fairly easy to reproduce.  I still have not been able to track down suitable modern alternatives for the transistors, but the diode pack is still produced.  And the UPC1042C is still readily available on eBay.

For the PCB design, I tried to keep the layout similar to the original to make it easier to verify if the connections are correct

[coromonadalix]

wow  kudos for your work,   it means you'll be able to work an pcb substitute ...

[Harry_22]

... and am 99% sure I have narrowed down the problem to this sub-board on the main power supply assembly.

Hi!

Of course everything you do here is very interesting.
But very often the malfunction occurs in the remaining 1% of which we are confident.
You could check sub-module with a datasheet on NEC uPC1042C. And if there is a problem then most likely this is it.

Have you found any faulty components?

[ZGoode]

You are probably right that the UPC1042C is the likely culprit here, but the ceramic module is a pain to work on, so that's why I wanted to make a PCB replacement.  The transistors and capacitors test fine, so it most likely is the UPC1042C

[Harry_22]

Work can be parallelized. Restore the ceramic module and make a new one.
Used NEC controllers are still available on Ali for pennies.

[ZGoode]

Well, I have finally received my boards from JLCPCB and I can confirm that my reverse engineered design does work.  I got the power supply to turn on and successfully calibrated the rails within the specs listed in the manual.  Within the week I will clean up the schematics and give a list fo the values I ordered on DigiKey for anyone who might need this in the future.