1984 power supply repair

[Andy Watson]

Perhaps Dave ought to do a video on fault-finding vs witch hunting. Randomly swapping components without determining the cause of the fault is quite likely going generate two sets of faulty components. It would be a far simpler task to trace out the circuit before replacing every component on the board. Tracing out the circuit would, at a minimum, be an education and it might just give you some clue as to how to track down the fault and as to what caused it.

[connorwk]

I'm not really replacing all the parts on the board just the 12~ transistors (about 5 of the transistors are just 2n4401s) and the lm393p chip plus a few diodes.
Building a whole other circuit may be easier but then this computer isn't quite original and also there are other pinouts on this boards of which I don't know what they do since I haven't looked into them yet so this board may do a bit more then just 36v to 5v.

[Paul Price]

Check the connectors for corrosion. You can determine the voltages output by this supply with your DVM  at the connector of the the working power supply board.

Rather than replace all these transistors just measure their base-emitter, base-collector, collector-emitter junction resistances and look for shorts and open junctions. You can also compare the collector-emitter resistances of the suspect transistors to the working board.

It's just a hunch, but I would bet giving you better than even odds that replacing all these transistors would accomplish nothing.

[Crazy Ape]

I do believe the original schematic would make fault finding easier, particularly if you get sick of randomly replacing parts and hoping for the best.

[connorwk]

Well if you have and or can find the original schematic then great! But this is from 1984 and these computers are already pretty rare as it is the power boards in the are probably just as rare if not even rarer. I have in a previous post given model numbers and such so if you would like to look for it that would be great but its about 20 bucks for the transistors for two boards so I think I will get them but the only problem is I can find most of the transistors but some of them being from the 80s are unknown parts.
If someone would be willing to look at a schematic and tell me whats probably wrong I could I guess draw one up with the known parts and put part numbers in for the unknown parts but that's about all I could do.
Also could anyone tell me what these transistors may be.
0100           D467
TI8423             B
Those are two transistors I cant find datasheets for. Also there is another transistor that shows two diodes anodes out and cathodes connected on the center pin having the part number of C10P048. I couldn't find a datasheet on it but I assume this would work in place of it?
http://www.digikey.com/scripts/DKSearch/dksus.dll?Detail&itemSeq=135187593&uq=635125971238011117

[Zad]

Wouldn't he need to apply a load to the circuit to test the output voltage outside of the computer?

This.

Some SMPSUs will fail to start if the main rail doesn't have a load on it.

[Andy Watson]

Also could anyone tell me what these transistors may be.
0100           D467
TI8423             B

Try  2S D 467, and are you sure that is TI8 ? Could it be TIP 42 B ?
For the amount of time you've already spent on this you could have traced-out the circuit.

[Crazy Ape]

Well if you have and or can find the original schematic then great! But this is from 1984 and these computers are already pretty rare as it is the power boards in the are probably just as rare if not even rarer. I have in a previous post given model numbers and such so if you would like to look for it that would be great but its about 20 bucks for the transistors for two boards so I think I will get them but the only problem is I can find most of the transistors but some of them being from the 80s are unknown parts.
If someone would be willing to look at a schematic and tell me whats probably wrong I could I

Yeah, old stuff can be a pain to repair sometimes.

OK, I think I've tracked the schematic down, you'll have to eyeball it and see if it's a match.

This looks like the patent for the supply, complete with schematics and mode of operation.
http://www.google.com/patents/US4538073

Original Scan:
http://bitsavers.informatik.uni-stuttgart.de/pdf/convergent/patents/4538073_Modular_power_supply_system_May83.pdf

As for any strange part numbers, post them here and someone might be able to help out with what they are. Most parts will have modern day replacements that are cheaper.

[Paul Moir]

The double diode is likely a pair of schottky diodes for rectification coming out of the transformer.  Anything with a reasonable voltage and current spec should be just fine.

However, checking these parts in circuit is simple.  Just put on your diode check mode and measure across them.  You should find a ~.6v drop on the transistors (both legs) and 0.3ish volts for the schottky diodes.   If they're not shorted they're likely good.

Another thing; if it has no output, the problem is likely in the high voltage side (if the schottkys aren't shorted).  It's exposed to the mains and therefore susceptible to spikes, etc.   On a simple switchmode like this I'll run through the high side semiconductors with the diode check, which should take less than 5 minutes.   Then I'll check the resistors starting with the higher wattage ones because they're more stressed.  These check in circuit just fine; all your looking for is something close to their value.  They fail open circuit, so anything less than their rating indicates they're good.  If the fuse hasn't blown, likely the driver transistors are fine.    I wouldn't even begin to bother with the low voltage side except to have a visual look at the capacitors.

When you find the problem and fix it, initially power it with a series light bulb to protect the new parts while you make sure it's now functioning normally.  Often blown transformer driver transistors are a symptom of some other problem.

Replacing parts willie nilly is a waste of time.  Find the problem and fix it.

[Crazy Ape]

The double diode is likely a pair of schottky diodes for rectification coming out of the transformer.  Anything with a reasonable voltage and current spec should be just fine.

However, checking these parts in circuit is simple.  Just put on your diode check mode and measure across them.  You should find a ~.6v drop on the transistors (both legs) and 0.3ish volts for the schottky diodes.   If they're not shorted they're likely good.

Another thing; if it has no output, the problem is likely in the high voltage side (if the schottkys aren't shorted).  It's exposed to the mains and therefore susceptible to spikes, etc.   On a simple switchmode like this I'll run through the high side semiconductors with the diode check, which should take less than 5 minutes.   Then I'll check the resistors starting with the higher wattage ones because they're more stressed.  These check in circuit just fine; all your looking for is something close to their value.  They fail open circuit, so anything less than their rating indicates they're good.  If the fuse hasn't blown, likely the driver transistors are fine.    I wouldn't even begin to bother with the low voltage side except to have a visual look at the capacitors.

When you find the problem and fix it, initially power it with a series light bulb to protect the new parts while you make sure it's now functioning normally.  Often blown transformer driver transistors are a symptom of some other problem.

Replacing parts willie nilly is a waste of time.  Find the problem and fix it.

It's actually a problem in a DC to DC converter module (36v input), so no mains in this case.

I think this is the circuit diagram of the module, though not confirmed yet.
http://patentimages.storage.googleapis.com/pages/US4538073-3.png

[connorwk]

Andy Watson:
I am 1000% sure it says.
0100
TI8423
It has clear printing and I have good vision so I am very sure.
If you think it is 2S D 467 then I guess that should be the part I use. Since digikey doesn't have that specific part would this do as a replacement
http://www.fairchildsemi.com/ds/2N/2N6517.pdf
for this part?
http://pdf.datasheetcatalog.com/datasheet/HitachiSemiconductor/mXwvrqw.pdf

Zad: I put the supply IN the computer to test it so I am very sure it has a load when I test it. Also I know it supplys power even without a load so it doesn't matter anyway.

Crazy Ape: Thanks I will look it over It seems right considering all the computers electronics where made by Convergent and that's who the patent was issued to. Although some things seem off on it so I will have to do a close look over. Thanks very much though hopefully it is it the dates seem right so it may as-well be it.

Paul Moir/Crazy Ape: Crazy Ape is right this has nothing to do with mains power just DC to DC also Paul I know replacing parts seems like a bad idea but this is posted in the beginners section for a reason. I am a beginner and don't know where to start testing and the next best thing I could do is just replace parts. Thanks for the advice now that I know this I can start testing parts as you have told me. Only real problem is on this board there are a TON of resistors and a bunch are in parallel so testing in circuit may be a bit of an issue. I will do the best I can.

I appreciate all the help with this. Thanks!

[Andy Watson]

Andy Watson:
I am 1000% sure it says.
0100
TI8423

In that case I would think that the 8423 is a date code.

If you think it is 2S D 467 then I guess that should be the part I use. Since digikey doesn't have that specific part would this do as a replacement

Now that you know it's an NPN transistor why don't you check it with your meter - as somebody has suggested further up the thread. If you must replace it, have a look on ebay - it appears to be awash with them.

[Crazy Ape]

Take note of Paul's hint with the light bulb, only in this case you would use three 12v globes in series, and place them in the 36v input line to the DC/DC converter module.

Seriously, if you don't do this and you miss a bad part during your replacement, you have a good chance of seeing you new parts fry.

The theory is, that if there's a short in the circuit, the bulb will light full brightness and take the load rather than the circuit.

[connorwk]

Andy Watson: Actually your probably right I just looked at another board and it has this.
0100
TI8441
So I guess this transistor is just a strange cheap made transistor that I most likely won't find out what it is unless that schematic is right and it tells me what it is.
I guess I will check the npn with my meter then but what does "Epitaxial" mean does it make it a special npn or is it just some other way its made and it works the same as a regular npn?

Crazy Ape: Ok I shall be sure to do that then and hopefully I won't see anything fry.

[Crazy Ape]

Burroughs B25 = Convergent Technologies NGEN workstation.

[connorwk]

Yea pretty much.
Just has Burroughs stickers and labels on it.

EDIT:
Nice going Crazy Ape this appears to be the exact schematic for this power board!
Also that is why I cant just make a separate board for this because these boards can be used as a 12v source as-well when a select pin is active. So fixing this board would be best.
Now to go through the schematic and find out what is what.
Thanks again!

The only issue with this schematic... Well all the part labels are incorrect between the board and the schematic itself... So now I need to track down and relabel everything. It may not be the final revision either so some things may be different on the patent then on the board too... Hmm...

[Crazy Ape]

Yea pretty much.
Just has Burroughs stickers and labels on it.

EDIT:
Nice going Crazy Ape this appears to be the exact schematic for this power board!
Also that is why I cant just make a separate board for this because these boards can be used as a 12v source as-well when a select pin is active. So fixing this board would be best.
Now to go through the schematic and find out what is what.
Thanks again!

The only issue with this schematic... Well all the part labels are incorrect between the board and the schematic itself... So now I need to track down and relabel everything. It may not be the final revision either so some things may be different on the patent then on the board too... Hmm...

I was half expecting the component labels to be a mess, but the actual circuit itself should be very similar, if not the same as the patent.
Now you can at least work out if the mystery transistors are PNP or NPN.

This might help you remember which is which in the diagram:
PNP = Point iN Please.
NPN = No Point iN.

[connorwk]

Hmm... The schematic seems a bit more different then I thought... I know I have 5 PNPs on my board and in the schematic I only see two PNPs... Also there are 4 NPNs on the schematic while on the board there can be 4 with the know npns and mystery parts but it just seems odd that there are 3 missing pnp transistors on the schematics... Some of the schematics seem right and its odd... I guess this is a different rev so I may have to make a new schematic using the old one as a starting point.

[Crazy Ape]

Who many amps does the board need to supply?

There is an easier way.

[connorwk]

I am unsure of how many amps the computer needs but there are an absolute ton of dip chips one multiple boards all being powered so probably a good amount. If your suggestion is to make a different board and throw it in there I will have to decline it as I like to keep these old machines as original as I can so I would rather spend 50 dollars for all new parts to put on this board then to make up a 10 dollar board and slap it in there. Just my personal opinion.

[dfmischler]

OK.  But what is your plan if one of the transformers is shot?

[Crazy Ape]

I am unsure of how many amps the computer needs but there are an absolute ton of dip chips one multiple boards all being powered so probably a good amount. If your suggestion is to make a different board and throw it in there I will have to decline it as I like to keep these old machines as original as I can so I would rather spend 50 dollars for all new parts to put on this board then to make up a 10 dollar board and slap it in there. Just my personal opinion.

That's fair enough, I like to keep things original myself.

If I had to cheat, I'd grab an adjustable buck converter, replace the adjustment pot with two fixed resistors (set for 12v / 5v) and an electronic switching scheme to switch them based on the voltage select line. So no, not really making a board from scratch, but certainly not original.

I just looked at the schematics in the patent, that voltage select is a jumper that's either in place or not in place to choose 5v or 12v for the module. It's a manual setting, not under control of any electronics. You don't need the above mod. You could just set the buck converter to 5v or 12v depending on what the dead board is set to and put it to work.

Tech has changed somewhat in the last 30 years 
http://www.aliexpress.com/item/DC-DC-8-40V-to-3-30V-8A-Switching-Buck-Power-Supply-Module-Step-Down-Converter/720982862.html

[connorwk]

dfmischler: If a transformer is shot (which I highly doubt it is) I think I'd be crazy enough to take it out and rewind it.

Crazy Ape: Yes the tech has changed a lot hasn't it. If I had to do anything I'd probably do something along what you have said. The only strange thing about this supply is that there are two other pins on this supply. A power inhibit line and a 33khz sync line. That sync line is so that it can tell if a supply isn't present in the machine and if it isn't present or has failed then the machine will not power on. Thus the other problem with trying to integrate something else. I think I am really stuck with fixing this supply unless you want to design a full new supply with syncing and all.

[Crazy Ape]

dfmischler: If a transformer is shot (which I highly doubt it is) I think I'd be crazy enough to take it out and rewind it.

Crazy Ape: Yes the tech has changed a lot hasn't it. If I had to do anything I'd probably do something along what you have said. The only strange thing about this supply is that there are two other pins on this supply. A power inhibit line and a 33khz sync line. That sync line is so that it can tell if a supply isn't present in the machine and if it isn't present or has failed then the machine will not power on. Thus the other problem with trying to integrate something else. I think I am really stuck with fixing this supply unless you want to design a full new supply with syncing and all.

I edited my above post slightly, since you don't need voltage select, just set the right voltage for the module.

You can likely ignore the sync line, they are all tied together (all the modules) and connected to a 33khz sync generator that is external to the DC/DC converter module. I don't see how this can detect a failed supply in any way. You'll find that since the buck converter uses it's own oscillator, the sync line can be ignored.

Power inhibit could be used to control a FET placed on the output of the buck converter, having the desired effect.

[connorwk]

Alright well I'll try to save this one first then go on to brewing up something custom.