Hp 6236B low 6V current and ammeter problems

[de_light]

I purchased a used HP 6236B power supply and it has some issues with it. Thanks to Garrettm for guiding me through some basic steps but he has suggested a post in the repair forum now. I am a beginner. It was advertised as working and tested but the seller has kindly refunded me so I have this item for parts, for postage only, essentially.

I have read the manual available at http://reza.net/files/hp-6236B.pdf
I have performed some basic diagnostics: the voltage outputs are accurate on +6V, +20V and -20V. Tracking is fine. Current on the +20V and -20V outputs has been calibrated at 0.555 Amps as per the manual


Problems:

1. Low 6V current output

2. Ammeter reading too high, even with power off - ??fades over time


Low 6V current output

6V output is between 1.039 and 1.321A depending on the position of R46. At full counter-clockwise setting, it is at 1.321A. 

I have fully loaded all the outputs as described in the manual. I have used resistors in series (e.g. 18+12 Ohm for the 20V outputs) connected via alligator clips and leads to the respective outputs. I do not have a rheostat available. The maximum output is still around 1.3-1.4A even with the 6V output and COM shorted with leads.

The beginner in me wonders why the maximum output is almost exactly half of what it should be?

2. Ammeter reading off

The ammeter always reads between 0.1A and 0.25-0.3A even with the power supply off. I have followed garrettm's instructions and removed the meter and plastic cover and have manually reset the needle to zero yet the problem persists. The black disc will not move the needle either. Initially, when I turned the meter off, it sat at around 0.3A. Then, over a period of some hours, it slowly moves to sit on zero once more. If you transiently turn on the power and after a few minutes, turn it off....it will still sit around 0.3A for another few hours.

Pictures are mainly of interest for renovation purposes. I still have some way to go - there is a lot of weird paint spatter that I managed to remove most of on the ammeter. The voltmeter and other bits needs sandpaper and plastic polish but it is looking better than it did. Looked like the actual case hadn't been opened up in twenty years. I put the crimped grounding in place and also replaced a 0.1uF RIFA X capacitor behind the transformer.

[garrettm]

That diode looks pretty suspect if you ask me.



I'd look at the schematics and see if you can find the correct part and swap that mess out of there. Could be related to the low current output on the 6 volt rail.

[garrettm]

For reference.





Notice the lack of wacky diodes bodged together.

[de_light]

Holy smokes.....what on earth?! That is crazy. I've removed it. It was soldered down really quite hard and I ended up lifting some of the connectors around the through-hole a little.
It's like something out of 'Frankenstein'. I've no idea what I'm looking at!

It appears to be two diodes. The first is black with a grey band at the end:  My ESR tester states it is a diode with an Ir of 4 nA and an Uf of 667mV (C-37pF)

The second which is attached to the nut/bolt= Ir = 5.8 microAmps Uf=553mV C=0pF

Schematics suggest this is supposed to be a 7.5V Zener diode part number 1N4353B

Do you think https://www.jaycar.com.au/7-5v-1n4737-1-watt-zener-diode/p/ZR1407 would be a suitable alternative?






Pic of the voltmeter cleaned off. Started with 240 grit sandpaper, then 600 wet and dry, then 1000, then Autosol acrylic polish by hand. Didn't take more than fifteen minutes.

Interestingly....note how the ammeter has dropped right down.

[duak]

I've got some similar hp & Lambda supplies with analog meters and I've found that the pointers are susceptible to static charge buildup.  Just dragging a cloth across the meter window is enough to move the pointer.

[garrettm]

Holy smokes! Those displays look great. I probably need to polish mine too, one has a scratch and the other has a patch of white haze from being rubbed.

As for the zener, I couldn't find any info on it. But looking at the package size of the zener on my board, it looks like a 5 watt part.

I'd argue the On Semi 1N5343B would be the closest replacement.

https://www.mouser.com/datasheet/2/308/1N5333B-D-1801430.pdf

Unfortunately, Jaycar doesn't seem to carry that part. You can get it on Mouser here in the US, but you will need to check distributors for Australia.

[garrettm]

I did a little more looking and the 1N5343B is avaliable on eBay from multiple international sellers (mainly China). So that could be an option if you can't find a local distributor.

[de_light]

Ordered!

[de_light]

I've added some more photos of the board to see if there are any other dodgy components that others can spot whilst I wait for that Zener diode to arrive. Apologies in advance for some of the shadowing.

[de_light]

Unfortunately, replacing the Zener didn't change any of the values.

[WattsThat]

Have you measured the level of ac ripple present on the 6 volt output (under load)? The main filter cap, C27, 5600uf is quite possibly 40 years old. Have you verified the three bias/supply voltages? (-6.2,-12,4 and +7.5).

The other obvious culprit could be the current sense resistor, R48. It’s not likely for a power wire wound to drift too far up but since it’s 0.25 ohms, it’s possible. The only sensible way to test without a four wire ohmmeter is to measure the voltage drop across it with a known load.

[de_light]

Have you measured the level of ac ripple present on the 6 volt output (under load)? The main filter cap, C27, 5600uf is quite possibly 40 years old. Have you verified the three bias/supply voltages? (-6.2,-12,4 and +7.5).

The other obvious culprit could be the current sense resistor, R48. It’s not likely for a power wire wound to drift too far up but since it’s 0.25 ohms, it’s possible. The only sensible way to test without a four wire ohmmeter is to measure the voltage drop across it with a known load.

1.  +7.5V bias, TP1 to   = 7.2V OK
2. -6.2V reference     = -6.02V OK
3. -12.4V bias            =   -9.82V  BAD

I went on to check Q12 by removing it and testing it as per https://vetco.net/blog/test-a-transistor-with-a-multimeter/2017-05-04-12-25-37-07. Seemed to be OK.

There isn't an open between pin 1 and 3 of Z1, it reads 6.5kOhm ( I haven't removed the board from the case so may not be getting this right).

I was getting some odd readings when I tried to test Q11 so I tried removing it and got some even odder results. Which pin is the emitter on this package? From what I read, the collector was ground attached to the metal plating so when I removed it, I think I couldn't really test. Might be wrong here.

I may go ahead and replace this transistor as it seems the likely culprit - I doubt Z1 is easily found! Could you suggest an alternative package that I could use instead of this one that is a bit more available these days?

C27 is coming up on my ESR meter as 7765microfarads with an ESR of 0.31 ohms.

[de_light]

Would a TIP42C TO-220 PNP transistor be a suitable alternative for Q11?

[WattsThat]

Q11 is 2N3740 which is in a TO-66 package, a TO-220 packaged device is not a good idea as it must be heatsinked. The 2N3740 is still widely available, in stock at Mouser albeit expensive. The pinout of the TO-66 is the same as it’s big brother, the TO-3, which is shown on page 7-1 or on its datasheet, all over the web. Yes, the outer metal can is the collector.

I would not rely on a resistance measure of Z1 while in circuit and I wouldn’t remove to test, it’s a simple matter of measuring the voltage on pin 2 of U4 which should be 50% of the actual -12.4 bias voltage.

That ESR measurement may be okay, sounds high to me but I’d rely on the AC ripple voltage. Only if the ripple is out of spec would I care about the ESR. Ultimately, only if the ripple is high, the cap gets replaced, not before. But then I’m not a recapper kind of person. Except for C34. That sucker is going in the trash because it says RIFA on it. Just google it on this board. They’re little time bombs.

[de_light]

Q11 is 2N3740 which is in a TO-66 package, a TO-220 packaged device is not a good idea as it must be heatsinked. The 2N3740 is still widely available, in stock at Mouser albeit expensive. The pinout of the TO-66 is the same as it’s big brother, the TO-3, which is shown on page 7-1 or on its datasheet, all over the web. Yes, the outer metal can is the collector.

I would not rely on a resistance measure of Z1 while in circuit and I wouldn’t remove to test, it’s a simple matter of measuring the voltage on pin 2 of U4 which should be 50% of the actual -12.4 bias voltage.

That ESR measurement may be okay, sounds high to me but I’d rely on the AC ripple voltage. Only if the ripple is out of spec would I care about the ESR. Ultimately, only if the ripple is high, the cap gets replaced, not before. But then I’m not a recapper kind of person. Except for C34. That sucker is going in the trash because it says RIFA on it. Just google it on this board. They’re little time bombs.

Thanks for your reply.

1. C34 has already been replaced! I read all about the RIFA time bomb problem 
2. I don't have an oscilloscope so I think I will just go ahead and replace the electrolytics. I've actually removed them all last night and tested them individually - a lot of them seem to have an ESR that sounds too high to me as well. There are a couple that seem okay, but I may as well just replace the lot.
3. In terms of replacing the transistor, it adds up to being pretty expensive if I get a 2N3740 + shipping. I've read a TO-220 is a direct replacement for the TO-66 series so I could heatsink it and see what happens. Is a TIP42C an OK replacement on specs? It seems to be OK from the datasheet: the Ft is 3Mhz vs the 2N3740 which is 4Hz, and the hFe is 15-75 vs 30/100 - would this be a much a of a problem?
4. Looking for a replacement cap for C33 which seems to be in the reference/bias circuit - rated as 490uF / 85V. What would be a decent replacement cap for this? Would a 470uF /100V be acceptable to use, or would a 1000uF/100V be preferable?

[WattsThat]

1. Good on the RIFA. The last thing you need is for one of those things exploding. It’s a mess and it reallly stinks. Like a skunk.

2. You do not need a scope. Your DMM on AC volts gives you everything you need to know about ripple. The frequency is a don’t care, you couldn’t change if you had to. Well, not exactly true since it should 2x mains due to the full wave arrangement, it would be 1x if a diode was open but then it would have really, really bad ripple!

3. You can certainly try the TIP, it will at least confirm if it’s the problem. You can just bodge it in for a trial, don’t waste time making it pretty just test to see if the voltage comes up. Turn it with your meter connected to the -12.4 to measure it without wasting time, don’t leave it on any longer than necessary to let the voltage stabilize the measurement and turn it off. That way you limit any heating that should be minimal. If that brought the voltage up, great, then you do it right and heat sink it. If nothing changes, put the old part back. All that said, in that application, you should be able to tell if the device is bad by examining the three terminal voltages while in circuit when it’s on, it’s just a basic emitter follower/pass regulator.

4. There is no harm in going with 1000uf, 100v if it fits. The 470uf would certainly work since electrolytic tolerances can be wide, i.e. +80/-20%. To make it fit, you’ve also have the common values of 560 and 680uf. What ever works for you, it’s not critical so long as you meet the old minimum.

Not to be critical but you’ll learn that the “shotgun” repair methodology, that is just pulling and exchanging components because you don’t know how to troubleshoot and/or you cannot find anything obviously defective - is a recipe for failure. The more you do, the higher the probability you will damage a trace, pull a plate-thru or install a component incorrectly.

You need to spend some time to understand how the circuit works in order to troubleshoot. The HP service manuals always have a description of how things work and this supply is no exception. Section 4-35.

Couple that with the a basic understanding of the passive and active devices and you have a starting point to analyze the collection of parts in front of you and look for the voltage that isn’t what it should be. That’s where you start. If you pull a part and it checks okay, put it back it and make sure things still behave the same way as before. Don’t go changing six parts at once and then wonder why it still doesn’t work. Troubleshooting is an investigation where you apply your knowledge of what should be until you find something that is not as it should be and then determining why it’s wrong. Just throwing parts a problem is like being a blind squirrel. Yeah, once in awhile they find a nut but it’s just luck, not skill.

Some closing thoughts:

1) If you think a scope would expand your knowledge, one of those $20 eBay STM32 2.4 inch LCD things is all you need to troubleshoot a linear power supply. See the videos below.

2) I have a functional 6236B on my bench, if you want some voltages from a functional unit, ask away.

3) You might find this series of videos helpful, its a fair amount of content in the six episode playlist with the last episode using the information presented in the first five to explain some good troubleshooting ideas for linear supplies.

https://m.youtube.com/watch?v=9iPHTWFDXn0&list=PLL_nf1OmixTTKhw2oVBtCZUYoMecz-YW5&index=2&t=726s

[de_light]

Hey, thanks heaps for this reply. Great tips. I'll have a read and re-think my strategy - I don't want to do the shot-gun approach. I'll get on to watching that video and take it from there.

[garrettm]

eBay has a bunch of NOS or used 2N3740 for around 7-10 USD. That is pretty expensive for a single part, considering most components are only around 50 cents a piece. But it is a direct replacement.

https://www.ebay.com/sch/i.html?_from=R40&_nkw=2N3740&_sacat=0&_sop=15

Worst case, I've got one of these in my junk bin--just need to unsolder it from a board. I'd be willing to sell it to you for 2 USD + whatever shipping comes out to be, likely 3 USD using USPS.



The problem with using other package types is that it will be a pain to heatsink correctly. You will likely need to cut a plate of aluminum to make an adapter. Depending on your skill level and tools avaliable, it might be cheaper to just buy the metal can part over the newer plastic packages--especially if you consider your time in making it all work.

Have you measured the forward voltage of the base-emitter and base-collector yet? For a PNP BJT, connect the black lead to the base and the red lead to the collector or emitter using your DMM. In "diode" mode you should see 0.5 to 0.7 volts forward voltage. A bad transistor will likely be shorted (<0.5 volts, usually 0.1 V--although "leaky" junctions can appear okay with this test).

I just fixed an HP 214B pulse generator that had a faulty PNP output transistor for the trigger out. The little guy was shorted base to collector and base to emitter, basically turning it into an ohmic junction. HP used some sort of custom (i.e. unobtainum) TO-18 part that I luckily had in my junk box. It pays to keep the boards out of junked gear.

Here's the faulty transistor, Q20 PNP high-speed switching type. The manual lists 1853-0218 for the part #, but a 1853-0357 was used in the unit.

[garrettm]

Here's a pic with the faulty part in situ (1853-0357).



And here's with the new PNP (1853-0218).



Finally, here's the trigger output with the new part installed, 6dB attenuator with scope set to 50ohms input. Trigger amplitude and wave shape look to be in spec.

[de_light]

I watched that series on the linear power supply - he's great! Will be watching more of his stuff. 

eBay has a bunch of NOS or used 2N3740 for around 7-10 USD. That is pretty expensive for a single part, considering most components are only around 50 cents a piece. But it is a direct replacement.

https://www.ebay.com/sch/i.html?_from=R40&_nkw=2N3740&_sacat=0&_sop=15

Worst case, I've got one of these in my junk bin--just need to unsolder it from a board. I'd be willing to sell it to you for 2 USD + whatever shipping comes out to be, likely 3 USD using USPS.

Have you measured the forward voltage of the base-emitter and base-collector yet? For a PNP BJT, connect the black lead to the base and the red lead to the collector or emitter using your DMM. In "diode" mode you should see 0.5 to 0.7 volts forward voltage. A bad transistor will likely be shorted (<0.5 volts, usually 0.1 V--although "leaky" junctions can appear okay with this test).

Thank you so much for that kind offer. I may have to take you up on it as the eBay shipping is extortionate. I've tested the voltage, between base and collector it is 0.049V and base-emitter is 0.155V - so it's looking like it is shorted!

[WattsThat]

If those readings are out of circuit on your dmm’s diode range, yes, it’s shorted.

That’s odd though because the low -12.4 bias voltage would be more likely caused by an open device, not shorted.

[de_light]

Sorry to mislead - I actually do think it is open. I've removed it from circuit and tested it according to this website
https://vetco.net/blog/test-a-transistor-with-a-multimeter/2017-05-04-12-25-37-07

I've got:

Base (+ lead) to Emitter = OL
Base(+ lead) to Collector = OL
Emitter(+ lead) to Base = OL (FAULT?)
Collector(+ lead) to Base = 0.549V (OK)
Collector(+ lead) to Emitter = 0L

So this would correlate with an open Emitter to Base?

[jh15]

Some meters have a very low voltage so as to NOT energize semiconductor junctions to better troubleshoot in circuit items. Make sure your meter is set to test junctions,

Or a plain old analog vom.
edit: seeing you have a good collector to base voltage, shows your meter is measuring OK.

[WattsThat]

Agreed, base to emitter is open. Fits with the symptoms.

[de_light]

Thanks. I'm trying to understand the schematic to work out why an open B-E would cause a low voltage on this rail (beginner). I've read 4.35 of the manual that details the operation of the reference/bias supply and looked at the schematic.

So, going from the ground to the -12.4V rail are two in-series 10K resistors. In an ideal world, that means that there would be 12.4V going across 20kOhm of resistance which results in 620uA of current. Putting 620uA of current across a 10K resistor on it's own gives 6.2V. This is compared to the reference voltage coming from VR1. Both of these enter the U4 op-amp.

The U4 output controls the operation of a Q12 driver (NPN) so current then flows out of the emitter to the base of Q11. The manual then states ' Driver Q12 controls Q11 by shunting part of the base bias supplied by R68' (I don't understand this part).

So, what I don't understand is how this actually works in practice. If the voltage drop across the series resistor (voltage divider?) pair is low (and therefore the output to the -12.4V rail is low), what is the ideal operation of the U4 op-amp? I would have thought that if it increased the base current going to Q11 via Q12, this would divert current from the E-C of Q11 and further reduce the -12.4V output?