1984 power supply repair

[Crazy Ape]

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

It's always good to have a plan-B, even if you don't need to use it.

The patent tells us what sync id for:

A 33-kHz generator 16 is shown coupled the power bus 19 to route a generator signal the various system DC-to-DC convertors. Generator 16 produces a synchronizing signal to operate all of the DC-to-DC convertors in unison to avoid beat frequencies and other such interference. As a result, the system operates more efficiently while the DC-to-DC convertors do not produce annoying buzzing noises that might also interfere with operation of the equipment and/or annoy a system user. A horizontal sync generator 17 is shown coupled to DC-to-DC convertor 14 which operates monitor 15. The monitor's DC-to-DC convertor is operated at the convenient horizontal sync generator frequency to avoid unnecessary and redundant generating circuitry. Interference is not produced by such arrangement because the monitor is remotely located from the system modules.

Well, the buck converter is likely running around 150kHz or there abouts, and being so far away from 33kHz, I'm not sure you'll notice any beat frequency issues.

[connorwk]

Your right its a good idea to have a backup plan.
http://www.ebay.com/itm/Astec-AA7281D-Unisys-5-12V-DC-DC-Adapter-/161081266905?pt=UK_Computing_Servers&hash=item25813112d9
This is mine.
Just found it on ebay. Now its going to be cheaper for the few parts I'm buying from digikey but if they don't work then I guess I shall get this.
Also I sort-of blew up a working one aswell so now I have two dead ones two working ones...
Four working if you count the two in the HD/FD unit. But anyway I'm not toying with any other working ones. All the transistors I could find replacements for come to a total of only about 20 dollars I'd guess with shipping so. And that is for both the broken boards so its not that bad if it all works. The only real issue I have as of now is what to do about the one transistor that is a mystery as of this moment. That 0100 TI thing... I will keep working on the schematic but other then that I could only hope it is still functioning and replace all the other transistors.

[connorwk]

Well scratch me saying I'm missing a bunch of pnps and have extra npns as I mistaken a few npns as pnps so It appears this schematic is more correct then it is wrong. Hopefully this will give me what these no name transistors are.

Oh also question.
Do you think I can use this transistor.
http://www.fairchildsemi.com/ds/2N/2N6517.pdf
In place of this transistor?
http://pdf.datasheetcatalog.com/datasheet/HitachiSemiconductor/mXwvrqw.pdf
I'd like to order everything from digikey at once instead of having to get a few parts seperately from ebay but digikey doesn't supply the old 2sd467's and what I could find on digikey were these 2n6517's. I wanted to know if they were compatible parts as their voltage ratings were much different. I'd very much appreciate it if someone could look at that datasheets and tell me if the 2n6517's would work in place of the old 2sd467's.

EDIT: I found what that strange 0100 part is its the part labeled IC101 on the schematic near the bottom left (if you rotate it clockwise so its readable).
http://patentimages.storage.googleapis.com/pages/US4538073-3.png
It looks like a diode with a line coming out of it between the line and the tip of the triangle. Now being a beginner I do not know what that is or what even to look up to find out what it is. Anyone know what it may be? Its in a standard transistor looking package and that's what it looks like on the schematic so. I have no idea myself.

[Crazy Ape]

Your right its a good idea to have a backup plan.
http://www.ebay.com/itm/Astec-AA7281D-Unisys-5-12V-DC-DC-Adapter-/161081266905?pt=UK_Computing_Servers&hash=item25813112d9
This is mine.
Just found it on ebay. Now its going to be cheaper for the few parts I'm buying from digikey but if they don't work then I guess I shall get this.
Also I sort-of blew up a working one aswell so now I have two dead ones two working ones...
Four working if you count the two in the HD/FD unit. But anyway I'm not toying with any other working ones. All the transistors I could find replacements for come to a total of only about 20 dollars I'd guess with shipping so. And that is for both the broken boards so its not that bad if it all works. The only real issue I have as of now is what to do about the one transistor that is a mystery as of this moment. That 0100 TI thing... I will keep working on the schematic but other then that I could only hope it is still functioning and replace all the other transistors.

Just be aware that the image is quite possibly just a stock image and not the actual item. Someone else on ebay is using the same image.
http://www.ebay.com.au/itm/ASTEC-AA7281D-UNISYS-5-12V-DC-DC-ADAPTER-/360476664474

Also, if you do decide to play with the DC/DC modules from the HD/FD unit, be mindful of the 12v jumper being set.

[Crazy Ape]

Well scratch me saying I'm missing a bunch of pnps and have extra npns as I mistaken a few npns as pnps so It appears this schematic is more correct then it is wrong. Hopefully this will give me what these no name transistors are.

Oh also question.
Do you think I can use this transistor.
http://www.fairchildsemi.com/ds/2N/2N6517.pdf
In place of this transistor?
http://pdf.datasheetcatalog.com/datasheet/HitachiSemiconductor/mXwvrqw.pdf
I'd like to order everything from digikey at once instead of having to get a few parts seperately from ebay but digikey doesn't supply the old 2sd467's and what I could find on digikey were these 2n6517's. I wanted to know if they were compatible parts as their voltage ratings were much different. I'd very much appreciate it if someone could look at that datasheets and tell me if the 2n6517's would work in place of the old 2sd467's.

EDIT: I found what that strange 0100 part is its the part labeled IC101 on the schematic near the bottom left (if you rotate it clockwise so its readable).
http://patentimages.storage.googleapis.com/pages/US4538073-3.png
It looks like a diode with a line coming out of it between the line and the tip of the triangle. Now being a beginner I do not know what that is or what even to look up to find out what it is. Anyone know what it may be? Its in a standard transistor looking package and that's what it looks like on the schematic so. I have no idea myself.

The 2n6517 has some subtle differences when compared to the 2sd467 (2n6517 has higher specifications) but looks like it should work, someone else might like to chime in on this one though and give their thoughts.


IC101 is a thyristor of some description, it looks like an SCR symbol but the zenner like lines in the symbol make it something else. I can't say I've ever seen that exact symbol before, sorry.

[dr_p]

IC101 is a thyristor of some description, it looks like an SCR symbol but the zenner like lines in the symbol make it something else. I can't say I've ever seen that exact symbol before, sorry.

TL431

[Crazy Ape]

IC101 is a thyristor of some description, it looks like an SCR symbol but the zenner like lines in the symbol make it something else. I can't say I've ever seen that exact symbol before, sorry.

TL431

Thanks Dr. P
So it's a 'Programmable Shunt Regulator' (but might not be the same spec as a TL431), that make things interesting since we don't know what it's programmable range is.

Connorwk, for this particular component I recommend you get a voltage reading from one of your known good boards, then we can see what voltage it's set to.

[dfmischler]

Your right its a good idea to have a backup plan.
http://www.ebay.com/itm/Astec-AA7281D-Unisys-5-12V-DC-DC-Adapter-/161081266905?pt=UK_Computing_Servers&hash=item25813112d9
This is mine.
Just found it on ebay. Now its going to be cheaper for the few parts I'm buying from digikey but if they don't work then I guess I shall get this.
Also I sort-of blew up a working one aswell so now I have two dead ones two working ones...

I think this should be plan A.  Don't wait to buy it, or at least make an offer on it.  Ask the seller questions if needed, but it's not like there are 50 of these for sale on ebay.

[connorwk]

dfmischler: The reason that is not plan A is it is cheaper to get the parts for two boards and fix both boards while it is more expensive to buy that ONE and then still have a machine with a broken supply. So I shall try to repair both first then if I fail I'll buy one to get back to where I started I guess.

Dr.p: Thanks for the help!

Crazy Ape:

The 2n6517 has some subtle differences when compared to the 2sd467 (2n6517 has higher specifications) but looks like it should work, someone else might like to chime in on this one though and give their thoughts.

Those are the worries I had as-well but I was unsure if those voltages worked the same way as say the size of a resistor, were you can have a quarter watt resistor or a one watt resistor that they will work the same as long as its more wattage then the circuit needs and they are the same value. Hopefully someone will chime in and confirm if it'll be an appropriate replacement or not.

Thanks Dr. P
So it's a 'Programmable Shunt Regulator' (but might not be the same spec as a TL431), that make things interesting since we don't know what it's programmable range is.

Connorwk, for this particular component I recommend you get a voltage reading from one of your known good boards, then we can see what voltage it's set to.

"The output voltage may be set to any value between VREF (approximately 2.5 volts) and 36 volts with two external resistors."
That came out of the datasheet for the TL431. Are you worried the new part may need different resistor values to give the voltage I will need out of it?
Oh also this is the chip I will need to get as the fairchild TL431 that datasheet is for seems to not be on digikey.
http://www.digikey.com/product-detail/en/TL431AILPR/296-10183-1-ND/404841

[Crazy Ape]

dfmischler: The reason that is not plan A is it is cheaper to get the parts for two boards and fix both boards while it is more expensive to buy that ONE and then still have a machine with a broken supply. So I shall try to repair both first then if I fail I'll buy one to get back to where I started I guess.

Dr.p: Thanks for the help!

Crazy Ape:

The 2n6517 has some subtle differences when compared to the 2sd467 (2n6517 has higher specifications) but looks like it should work, someone else might like to chime in on this one though and give their thoughts.

Those are the worries I had as-well but I was unsure if those voltages worked the same way as say the size of a resistor, were you can have a quarter watt resistor or a one watt resistor that they will work the same as long as its more wattage then the circuit needs and they are the same value. Hopefully someone will chime in and confirm if it'll be an appropriate replacement or not.

Thanks Dr. P
So it's a 'Programmable Shunt Regulator' (but might not be the same spec as a TL431), that make things interesting since we don't know what it's programmable range is.

Connorwk, for this particular component I recommend you get a voltage reading from one of your known good boards, then we can see what voltage it's set to.

"The output voltage may be set to any value between VREF (approximately 2.5 volts) and 36 volts with two external resistors."
That came out of the datasheet for the TL431. Are you worried the new part may need different resistor values to give the voltage I will need out of it?
Oh also this is the chip I will need to get as the fairchild TL431 that datasheet is for seems to not be on digikey.
http://www.digikey.com/product-detail/en/TL431AILPR/296-10183-1-ND/404841

The TL431 is a good guess at the part based on the circuit, I just worry about the 0100 marking, is there anything else written anywhere on the device?
There are Programmable/Adjustable Shunt Regulators other than the TL431, and some support different voltages, setting resistors etc.

[connorwk]

I looked at all 4 boards and these are exactly what I found on all four.
0100         0100         0100           0100
TI8441      TI8441      TI8423        TI8423

The first two came out of the same machine and the last two came out of the same machine so they may still be date codes but they may also just be part numbers and they ended up using a different part. I can open the HD/FD unit and check those two supplies as-well if you'd like not much trouble at all.
Also just for your information I'm pretty sure the board that these supplies plug into set the pin for weither it will supply 12v or 5v so I could take the 12v board and put it in a 5v socket because it is set on the board its plugged into not the power board itself.

[Shock]

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.

It's the first thing I thought of. 

But swap testing if done logically will hopefully prove where the fault is. 
If there is two boards I would next be taking voltages around the board and current at the load and compare the two.
I would have thought this is the next logical step after a physical check.

But one of those big resistors had a dry joint on it and I would have touched up the solder joints or if someone had messed with it before me I would be checking that out instead.

[Andy Watson]

But swap testing if done logically will hopefully prove where the fault is.

I agree, however, the keyword is logically. To logically swap components you have to have some knowledge about their function in the circuit. So far we have a vague schematic from a patent, which may or may not be the right one. Without detailed knowledge, randomly swapping components is equally likely to fry both sets - a theory that appears to have been confirmed within this thread:

Also I sort-of blew up a working one aswell so now I have two dead ones two working ones.

Hang-on a mo.... didn't somebody predict that would happen, about 30 posts ago?

I looked at all 4 boards and these are exactly what I found on all four.
0100         0100         0100           0100
TI8441      TI8441      TI8423        TI8423

I still think the 8441 and 8423 is the date code. 8423 = 23rd week of 1984, that would tie in with date of these power supplies. The 0100 is likely to be the manufacturer's abbreviated part number. TL431 or similar is probably a fair guess. If you can't find the original manufacturer's data you might have to measure its voltage on a working unit.

[dr_p]

@connorwk : I sudgest you read all of this, or at least read this paragraph. PM me if you need more info on SMPS functionality and troubleshooting and want more learning material.

[connorwk]

Shock: Pretty sure no one else have even taken the metal plate off the back of the circuit board before me. I will go around and freshen up all the solder joints on the board to make sure everything is good by the way to make sure there aren't any dry joints.

Andy Watson: Yes you did predict it and a good prediction you've made... But it wasn't from swapping the parts it was from an accident on my behalf... I was soldering on a part from the other board and I soldered on one of the three pins while I bolted it to the heat sink then forgot to solder the other leads in and went to test it instantly hearing a loud pop and a different part on the board blew most likely damaging a bunch of other things as-well. So its indirectly from swapping parts yes but it wouldn't have happened probably if I payed a bit more attention. Now as for measuring voltage on the 0100 part where am I to read the voltage and how shall it help me in determining if the replacement part (http://www.ti.com/lit/ds/symlink/tl431a.pdf) will work? Sorry for the ignorance but I've put the post in beginners for a reason.

Dr.P: I'll be sure to try and read most of it especially that paragraph. Thanks.

Edit: Dr.P this seems to have to do with regulators that turn wall voltage AC down to DC not DC to DC supplies. As this is talking about HV filtering caps and fuses and all the such as which my board has none of its a simple DC to DC converter.
I'm working out the schematic to try and label everything as it is on my board but as it seems that schematic is the EXACT thing I have if anything maybe a resistor or two off. But everything seems the same so you may look at that for reference and tell me where there are HV caps and or fuses but I do not see any on my board or the schematic.

[Andy Watson]

Now as for measuring voltage on the 0100 part where am I to read the voltage and how shall it help me in determining if the replacement part

This from one of your earlier posts:

The output voltage may be set to any value between VREF (approximately 2.5 volts) and 36 volts with two external resistors."
That came out of the datasheet for the TL431. Are you worried the new part may need different resistor values to give the voltage I will need out of it?

Basically, yes.

There used to be a whole family of TL43x type regulators - only the 431 seems to be readily available now. There are only three pins on the device, if you measure across each possible combination of pins it should cover all eventualities! What we want to know is what voltage it's regulating at so that you can calculated the resistor values - assuming it isn't a actually a 431 - and assuming that the original is broke.

[Andy Watson]

I'm working out the schematic to try and label everything as it is on my board

The next useful step would be for you to show us the completed schematic - it might help us give you a more directed response - rather than stabbing in the dark. (Not guaranteeing anything though )

[dr_p]

Edit: Dr.P this seems to have to do with regulators that turn wall voltage AC down to DC not DC to DC supplies. As this is talking about HV filtering caps and fuses and all the such as which my board has none of its a simple DC to DC converter.

it's the same in a mains powered SMPS: after the bridge rectifier and large filtering cap,  it's still a DC-DC converter

[connorwk]

Ok so the schematic for this device would take a lot of work but what I can tell you is the voltage at the programmable shunt regulators reference is 2.5v at the cathode is 20v and the anode goes directly to ground.

[connorwk]

Well I fixed both of the supplies!
For about 20$ for a bunch of the on board parts I ordered from digikey.
I just went and replaced every part I had from digikey except the replacing the odd 100 part since I wasn't sure if the part I found would be an exact replacement I just hoped it was still good... And it was! Well thanks for the help everyone!

[JoeO]

Congratulations.   Nice work!