My HP 5342a acquisition

[GRFixedGear]

Bought "for parts or repair" for 101usd from ebay but the clock source switch is shown in the external position. Do I power it on when it arrives or follow the eevblog mantra: don't power it on,  take it apart?

[Kim Christensen]

Anything with that description I'd open up before powering it on. Who knows what's inside.

[srb1954]

Before powering on provide an suitable external 10MHz clock source. Otherwise, open it up first and check if there is an internal oscillator installed - a vertical plug-in module towards the rear of the unit. If there is an internal clock module installed switch to INT CLK before powering on.

Without either an internal clock module or an external reference clock the internal microprocessor can't run and the whole unit will appear dead. This may be the reason that the unit was sold for parts or repair if there was no suitable clock source being fed to the internal micro.

[GRFixedGear]

Thanks for the advice.  Checking the complement of cards installed was my first thought.  I have ordered an OCXO to install since it's marked as a plain-jane unit without any options installed. So even if it's missing the tcxo I won't be completely hosed.

I'll take pictures and bring you all along for the journey when it gets here end of this coming week.

Jason

[GRFixedGear]

Well we should have some fun with this one.

Opened it up and looked at all the cards which all appear to be in the right spot. Checked the power wafer and its good for 120v supply where I live. Fuse is good. Flipped the clock source over to internal. So time to apply some angry pixies.

Nutin on the screen. No smoke, no fire, but no digits.

Got something going on the power section as evidenced by a nearly constant red LED on A21 and faint flickers from the green one on A20. Gonna work through the troubleshooting flow chart but since the supplies are restarting nearly constantly its not helping (+5a, +5d, -5, +15 and -15 all bouncing in time with the supply restarting). Actually saw some constant voltages on the connections to the oscillator slot, but havent had time to confirm whether theyre proper or not. Tried pulling most of the cards out which didn't seem to affect that characteristic. Ohm readings from all those to chassis ground shows +5d at ~15ohms, but everything else was K's if not M's.

Thats where we're at so far but would take any advice people have to give.

Thanks
Jason

[GRFixedGear]

It it must have been scared by the vaunted knowledge available here because now its come up just fine. Certainly I didn't fix anything. But its at least alive!

[GRFixedGear]

But can't replicate that powerup, so its defniitely flakey

[GRFixedGear]

Visual inspection of A21 shows a discoloration of CR4 (I think? anybody got a set of readable component location diagrams?!), and to a lesser degree two diodes "west" of CR4 (CR1 and 3? theyre respectively tied to the +5v and -5.2v test points above pins 1 and 5 on the edge connector).

What is supposed to be the +24VDC out, related to that CR4, is reading +30.2V with respect to chassis and the adjacent pin. The voltage into the recifier from the over transformer into CR2 reads 25VAC RMS on a fluke 179. Still chasing backwards from there since that seems high to produce 24V or 27V depending on where on the schematics you read what is expected.

25Vrms * sqrt(2) - 2 * 0.7v gives me almost 34V out of the bridge rectifier. Even if I go with the "max" voltage drop in the brdige im at 33.something V. no wonder the 30v zener is getting hot. But literally checked all the connections from the voltage selector block and how I have the wafer set and it sure seems right! Output of the oven transformer is completely even. 12.7V on each leg from the center, 25.4V end to end. That suggests to me that the transformer is fine but leaves me at a loss to explain why the voltage appears to be so much higher than its supposed to be.

Q8 (Q? its a 78L12!!!!!) also appears shot as its just passing through 28v instead of regulating down to 12V for the constant power output to the oven controller. Curiously it failed to save the 215Ohm resistor feeding it (facepalm)

[GRFixedGear]

Advice requested here.
I can ebay order a set of extender cards to start going through things. Alternatively I could tack look wires onto test points and component for the same purpose. The set of extenders www.ebay.com/itm/223253140858 is 130usd plus shipping. Can tack and label a lot of wires for that much $$$. Are the cards worth it?

Seems an isolation transformer is in my near future... or should I instead invest in a value priced 7kv isolated scope probe instead?

To try to tame down the 24V that is running too hot out of the oven transformer, do I try some inline dropper resistors on the 120vac mains feed? Or suck it up and get a value priced variac?

Anybody with history with these beasts seen something like this before?
Thanks!
Jason

[GRFixedGear]

Well apparently my last attempt at an updated timed out and got wiped out, so lets give it a go again.
I am making the baseless assumption that the Oven transformer is supposed to have the same number of windings on both sides of the primary, and that if it did the resistance of each side of the primary would be the same.
My measurements show 213.4ohms on the "top" side of the primary (as seen on the schematic of the power section of the 5342a) and 193.9ohms on the bottom side of the primary.
I'm taking this to mean that there are some shorted windings on the primary of the oven transformer, leading to a higher than desired output on the secondary and the root of my problems.
I THINK this transformer https://www.digikey.com/en/products/detail/triad-magnetics/F-362XP/5032161 should be compatible with the footprint on the motherboard of the 5342a and in the ballpark for the power requirements of the oven and the always-on portions of A21 board.
Unless someone can give me other advice I will try to confirm that its footprint compatible (and properly sized physically and electrically) and order it and some of the other parts that would have been stressed on A21 by this condition: bridge diodes, 30v zener, 78L12.

Caveats: I'm not magnetic guru so I can't correlate the ~5% loss in primary windings with the 25% increase in Vrms output from that transformer but.... for all I know BOTH sides of the primary have shorted but to a different degree.

Thats the plan.

[RolandK]

If you have a shorted winding on a transformer you have a very low ohm load. As the transformer is havily loaded it is getting warm or hot and the output voltage is lower. If you have the correct fuse installed it will blow.
One shorted winding is enough.
But: defect Transformers are not very common.

I had a case where i connected the earth to the screw of a toroidal transformer. This gave a 1 winding short. The transformer was buzzing loud and i switched it of at once. After understanding my fault and removing the wire everything was fine. When someone else in the forum had the same on a photo, complaning about buzzing he rejected my advice.

My advice:
- First replace the 78L12 and check that it works correct.
- Check the power consumption, it should be below the value given in the manual.
- Check all voltages of the power supply for their DC value (often in manual / schematic) and AC value (about some mV).

Next are the connection, sometimes they just want to be reseated to remove some corrosion.

[GRFixedGear]

Thanks, Roland

I had already re-seated all the cards a few times to break any corrosion free.

I appreciate your experience with shorts in transformers. I have no idea how this unit was abused before it came to me. For all I know somebody put 5 fuses in it trying to figure out this problem and some short in the transformer ended up burning up the very fine wire on the primary side for the shorted bits. All I do know is that with A19, 20, and 21 out of the unit that transformer by itself was not acting in accordance with the test values in the service manual, so out she came.
The output of the transformer stayed way higher than it needs to be even under a 100 ohm load (0.33 amps way higher than designed for).
Got a 10VA footprint compatible model with the right output. I will be trying it "remote" before I solder it in permanently to avoid any more heat stress on the motherboard after the removal.

Parts on order and we'll see what happens. I lose the 100vac and 220vac capability without the center tap on the special HP transformer but those are unobtanium so we'll live with it.

Coming are:
- oven transformer replacment
- full bridge for A21
- 30v zener for A21
- 12v regulators (2 of them on A21, plus a few extra)

[shakalnokturn]

Is there any chance the voltage selector wafer is marked erroneously and you think you're wired for 120V while really set at 100V?
That could explain secondary being too high.

True that top and bottom primary windings should have the same number of turns, if I'm reading the schematic correctly for 120V setting both windings are paralleled.
Same number of turns doesn't necessarily mean same DC resistance because of inner vs outer position on the bobbin requiring different lengths of wire for a same number of turns.

It seems that for 100V setting only the lower portion of the top winding is used so rather unconventionally I'd expect about 100 Ohms at the plug for 120V setting vs. 178 Ohms for 100V setting.

[GRFixedGear]

Is there any chance the voltage selector wafer is marked erroneously and you think you're wired for 120V while really set at 100V?
That could explain secondary being too high.

I can double check that again but I'm 99.2% certain that was not the case. I rung out the connections on the back of the power selector and found that B (incoming power) was not directly connected to A (orange wire going to that center tap on the top half of the primary). Further, I was able to measure the 120V on the actual input terminals to both parts of the primary on transformer, where the center tap was a wire leading up into the top part of the transformer. If it had been configured wrong, I would have expected a higher voltage at the "floating" portion of that top primary segment.

Same number of turns doesn't necessarily mean same DC resistance because of inner vs outer position on the bobbin requiring different lengths of wire for a same number of turns.

This seems incredibly obvious after I read it but was not something I had considered. Thanks for this bit of teaching. I will file that away for future application.

Thanks, folks. Keep the advice coming!

[GRFixedGear]

I can double check that again but I'm 99.2% certain that was not the case.

Indeed the wafer and the wiring on the back of that power selector appears to be wired as described on the schematic.

Jason

[GRFixedGear]

Just in case I know have a 6627a quad bi-polar-capable power supply on order. In conjunction with my current power supply which could serve as the 5Vdig supply even if we cant get the back end working we have a fighting chance of getting the front-end working!

[GRFixedGear]

Some good news for us all.... that transformer I ordered is a drop-in replacement for 120/240 operation. Pins drop straight into pcb holes, mechanical attachment holes line up perfectly with the screw holes in the board. Some simple #8 nuts on top of the transformer frame and its attached just like the original. Triad F-362XP transformer for the win

Bad news for me... it acts the same as the original HP one and I damaged it enough taking it out that I can't go back.

So. With a replaced bridge, 30v zener, and 78L12 on the A21 assembly it seems much happier. BUT. I still need to put some external load on the "24V" rail to keep the zener from getting overstressed especially when the OXCO has warmed up and the load on that 24V unregulated supply drops dramatically. Even with the load I see ~1.8V ripple at 120hz lining up with the cycles of the bridge rectifier. And without the OXCO load or an external 200ohm load it settles out above 30V and the zener gets hot hot hot.

I did check the other 78L12 which gets fed from 15V from A20 and its happy. I haven't checked any of the A21 circuits related to the +/-15 or +5v/-5.2 analog supplies yet.

I did tune the pot to tweak timing of the switcher controller. It was running at 22khz as I found the device but having it down to close to 20khz now (19.9khz).
A21 TP2 and 3 look good with complementary waveforms.

Main switcher output from A19 to A20 looks kinda like I expect it to but had trouble getting a screen shot of that from the scope (yes, i'm running the device on an I.T. so I can probe it with the scope and not blow anything up)

Even with all that, I keep getting shutdowns.
Looking at the 5Vd signal coming from A20 into A21 shows that it rises up but doesn't get above ~3V before it plummets back down so I'm looking for the reason behind that.

A19 opto isolator is a subject of some concern right now. I cant tell if my failure to get a blocked diode measurement in one direction is based on the resistor network around it or a problem with the LED side of that device. To be fair, I havent gotten a good reading on the output side of that opto yet either.

Thats what I know so far.

[GRFixedGear]

Please advise on this specific question. Page 8-99 of the service manual (deep in the power supply troubleshooting section) talks about probing A19TP4 vs TPG for ground of the scope probes. I see no numeric TP markings on the schematic or board for A19. Please confirm they're actually trying to look at A21TP4 and A21TP2 in the figures on this page, with probe ground still connected to Chassis (or the lower leg of C15, C20, or C16 more practically)?

If they're NOT talking about A21TP4, where ARE they trying to indicate?

Thanks
Jason

[GRFixedGear]

Having run through all the checks for A21 and A19 in the service manual I started having a look at A20. Measuring diode drops gave me a big surprise on the 5Vd section. The diodes read short both directions. Removed and tested diodes and they're fine. Tried the small ceramic cap, C10, as the next most easy thing to try. Still the shorts. Broke out the kelvin clips and the 3457a and started doing some 4-wire resistance measurements around that part of the circuit. I seem to have the smallest reading across the terminals of C11... One of the big quad terminal electrolytics there in in the 5Vd section. I dont want to contemplate replacing those bastards right now so think I'm gonna go sleep on that finding. Yes, plural... if one of them failed I don't think I'm going to trust the partner right next to it. Visually those caps seem fine. I don't see any deposits on the board.

The journey continues!
Thanks, all!

[GRFixedGear]

https://www.digikey.com/en/products/detail/vishay-sprague/604D172G010GJ7/5611853

oofta

$86 per but you gotta buy 20

[GRFixedGear]

Removed C11 and now I have no more shorts on the board. Measuring 0.05ohms dc on the cap itself. "She's dead, Jim"

[Kim Christensen]

$86 per but you gotta buy 20

At that price,    I'd be finding a much cheaper substitute and make it fit.

[GRFixedGear]

I'd be finding a much cheaper substitute and make it fit.

Yup. Did that exact thing. Put 2 individual 1000uF, 35V caps (Nichicon, 10000 hours at 125c) in the two footprints and some jumper wires to span the distance and complete the circuit.

Cue the frankenstein memes. ITS ALIVE!!!

 

I did go back and tune the recommended voltage drop on the overcurrent signal per the service manual.
Still to come: testing both impedances on the low range and also testing the high range input. And sorting out what to do with my resistor load solution to the zener overheating problem.

[GRFixedGear]

sorting out what to do with my resistor load solution to the zener overheating problem.

If I model the voltage from the transformer and bridge rectifier as 34V, with a source impedance between 10 and 2 ohms, the best case I end up with 220mA with the single 30v 5W zener which is significantly above the 158mA regulation current listed for the device.

If I stick with simple zeners, it appears putting four 7.5V 5W parts in series can handle this situation. 7.5V zeners list 630mA capacity which would handle a much more aggressive source impedance, even with the lower aggregate zener resistance from that series combination (6ohms vs 8 for the single 30v zener).

Is this a decent kludge to fix this problem?  What else should I consider in this case?

Thoughts?
Jason

[GRFixedGear]

For the purposes of buttoning it all up and running it with the top shield on the card cage, I put in parallel 30v zeners and endeavored through the full "check out the instrument" routine in Table 8-5 of the service manual.
I got all the way through Step 5 without any hiccups.

Important discovery... be very very careful using NanoVNA's as a signal source. But trying to use the NanoVNA as the source for the 1GHz signal did not go so well.
I decided that before I plugged that amazon special into my carefully resuscitated 5342a, I was going to see what its output level was on my old (donated to me) 435A power meter... and boy howdy was it a good thing I did. The output was way hotter than expected but fortunately the same donation that included the power meter included some N type attenuators to bring the level down to the desired level for the checkout procedure.

Even then.... something about the output of the NanoVNA was not making the 5342a happy. In steps the TinySA and its output capability. Still needed to tune the output level on the power meter and match with a attenuator value I had in the drawer, but after that step 6 went well.

I'm still not necessarily seeing the test mode 1 output with a sign and value directly following it. So watch this space to see if that is really a problem.

Thanks for following along,
Jason