Astec Power Supply Issue... I am stumped!

[vanderk]

I have been working on two Astec Kaypro power supplies (AA12450 and AA12350) on and off for the better part of a year or so. I simply cannot get them working! Both units have been totally recapped, resistors checked, and I even replaced all the transistors to include the power transistor. Also replaced were all the diodes and bridge rectifier. I have replaced almost everything except for chokes. All resistors were good and I see no shorts on the common mode, control and power transformer (T1, T2, and T3).

Here is the issue with both units (they both exhibit the same problem):

- While slowly increasing voltage on a variac, the crowbar engages at around 40-50VAC. I do have B+. There is a high pitch chirp coming from the capacitors, and I can also hear the crowbar clicking.
- I see -12vdc when the crow bar engages and, as intended, no output on the 5VDC and 12VDC points.
- Lowering the variac voltage to just before the crowbar kicks in, I see approximately -12VDC, 6VDC, and 14VDC.

I am completely stumped and refuse to give up on this! I was careful when replacing components and used only exact replacements and close values.

Any assistance would be greatly appreciated!

Thank you,

Chris

 

[Chuck(G)]

What are you using for a load on the +5V (where the crowbar SCR usually is?).

Small Astec switchers generally use a simply Zener+SCR crowbar on the +5. If you don't have a load on the +5 line, the crowbar will fire as soon as the voltage exceeds V[sub]Z[/sub]. Use a load that's about 10% of the rated +5 rail output current.

 

[vanderk]

Thank you for your reply! Thinking it was a load issue, I installed the power supplies in a working Kaypro (swapping out the good power supply). The Kaypro refused to start but rather the power and drive LEDs pulsated with the crowbar kicking in. I know I should have used resistors for a load and tested before installing; but I was sure that I had resolved the problem. There are not many components left to replace. I can check the resistor values again. I have triple checked all replaced components with a known working power supply--all match.

I find it odd that both power supplies exhibit the same problem. These were pulls and there was damage to some of the diodes. I tested the transformers for shorts but maybe there is a problem with them?

Did I mention how much I hate working on switched mode power supplies?

 

[Dwight Elvey]

You need to power it with an external supply on the 5v and usually the +12V. Don't keep connecting it to AC. Do you have a schematic? How does it start the voltage to the regulator circuit?
Dwight

 

[Fletch]

Thank you for your reply! Thinking it was a load issue, I installed the power supplies in a working Kaypro (swapping out the good power supply). The Kaypro refused to start but rather the power and drive LEDs pulsated with the crowbar kicking in. I know I should have used resistors for a load and tested before installing; but I was sure that I had resolved the problem. There are not many components left to replace. I can check the resistor values again. I have triple checked all replaced components with a known working power supply--all match.

I find it odd that both power supplies exhibit the same problem. These were pulls and there was damage to some of the diodes. I tested the transformers for shorts but maybe there is a problem with them?

Did I mention how much I hate working on switched mode power supplies?

I was about to ask what you were using as a load. I have two thoughts.

1) Have you resoldered the output connector? These seem to have problems with cracked solder/dry joints.
2) have you looked for a short on the 5v rail?

 

[Chuck(G)]

I don't know precisely which Astec supply your Kaypro uses--or if there are any schematics for it. I do recall a similar problem with a non-Astec power supply that drove me nearly to tears.

It wasn't the overvoltage crowbar that was kicking in, but the overcurrent protection. Turned out to be a current sense resistor that aged too much and triggered the overcurrent protection on very light loads. Not saying that it's what's going on with yours, but something to investigate.

 

[Hugo Holden]

If you can post the schematic indicating what you have replaced I might be able to make some suggestions about a test sequence to get to the bottom of it. If you don't have the schematic, likely you will need to copy it out from what is there anyway to end up with a successful diagnosis and repair. Global parts replacements can often lead to more trouble than good. The repair always needs to be targeted after data is acquired from testing and a theory formulated about what is going wrong. Ideally you would also have an isolated scope for tests in the live side of the psu (I use a Tek powerscout 222ps for this ).

I'll give an example of a fault in one psu I fixed against all the odds some years back: The unit would shut down on occasion for no apparent cause. It was determined after extensive testing by others that none of the resistors, capacitors, transistors or IC's were faulty, at that point the pcb itself was getting blamed.

The fault turned out to be in a ferrite cored transformer, driving the output transistor on the primary side. The glue had separated between the two ferrite halves to make a very small air gap and the inductance of the transformer had dropped. This was undetected by more than 5 technicians who had worked on the board before me. I found it because I scoped the drive waveform at the transistor base and noticed an unusual waveform.

So it is quite interesting what faults can still be present when most of the parts have been replaced, so you have to have a very open mind about what could be wrong and as noted set up the supply with suitable dummy loads.

One interesting thing is that both supplies are doing it, which suggests the remaining components that have not been replaced on each board, one has exactly the same fault as the other (Seems very unlikely) or of the components that have been replaced, one on each board is for some reason an incorrect or unsuitable part, rendering the same fault to both boards, which seems more likely.

 

[Dwight Elvey]

You can do any trouble shooting unless you can look at what the regulation circuit is doing. You can't do that unless you can see the values on each stage. That is real hard to do while it is tripping every time you tune it on.
Use a bench supply. You can put it in and out of the normal operating range with a bench supply and find what is failing. Stop replacing parts that are likely not even the failure.
If you don't have a schematic, trace it out and draw one.
Dwight

 

[Fletch]

One interesting thing is that both supplies are doing it, which suggests the remaining components that have not been replaced on each board, one has exactly the same fault as the other (Seems very unlikely) or of the components that have been replaced, one on each board is for some reason an incorrect or unsuitable part, rendering the same fault to both boards, which seems more likely.

Is the working PSU you have also an Astec? Could you use it as a visual reference to check that components on the faulty boards are in the right place, have the right values and polarity?

 

[Dwight Elvey]

Automotive light bulbs make good loads for computers. Once stabilized they are almost constant current loads. This means a 12v bulb that draws .15 amps will draw about .15 amps at 5V. They also provide less resistance until they stabilize, similar to a large capacitive load.
Dwight

 

[vanderk]

Thank you everyone for your guidance. Unfortunately, I do not have an external PS for low DC voltages. I have just about everything on my bench, but this is something I will need to build or purchase. I may have some spare car bulbs I can use for loads. I presently don't have the correct value resistors for dummy loads. The working Kaypro power supply does not go into protection mode without a load.

I did verify all components with a working Astec PS from a Kaypro; all new components match up. I conducted an exhaustive search for schematics for the A12350 and AA12450, but they are unobtainium; however, I did find some information on the Astec AA12090 and it has similar components but a different layout. There was some troubleshooting information that I followed to no avail. With no schematic, I became desperate and followed the "shot gun" approach and started replacing (carefully) suspect components.

To an earlier question, I did install both units in a working Kaypro but the crowbar kicked in (click, click, click, click...)

Despite seeing voltages on the secondaries, perhaps there is a hidden issue with the transformer as was suggested by Hugo? I have been tempted to pull the transformer from a spare, working PS and swap it out.

I have already put so much time into these power supplies, but I have contemplated tossing them in the trash and cutting my losses. I've worked on countless pieces of equipment from HV tube gear to modern HF transceivers, and these SMPS have proven to be the most challenging items that have been on my bench! Go figure! :-x :-(

 

[Chuck(G)]

Are you sure it's the overvoltage crowbar? What happens if you lift a leg of the crowbar SCR? Have you checked out the zener connected to the gate? A shorted zener there would certainly create the effect you're describing.

 

[vanderk]

Are you sure it's the overvoltage crowbar? What happens if you lift a leg of the crowbar SCR? Have you checked out the zener connected to the gate? A shorted zener there would certainly create the effect you're describing.

Hi, Chuck, Yes, I did check the Zener. It was fine. Out of curiosity, I did pull the leg of the crowbar. Considering I am totally fed up with these supplies, I applied power, stepped away, and pop! One of the 16V caps on the 12V rail railed.

EVERY component has been either checked or replaced. There must be an issue with the transformer. Unless someone wants these free, I am going to toss them. I could have built a power supply from scratch with all the time I put into both of these supplies. I certainly don't want to waste anyone's time on this forum as well.

I am extremely thankful for the quick responses!

 

[Chuck(G)]

Not every component, I suspect. A shorted transformer primary winding (not completely unknown) would certainly drive the output voltage up too high. I don't know how you would repair this, short of rewinding the thing. Since you've got a good supply, compare the bad supply transformer winding resistance with the good one.

A fully-loaded PSU should never have to pop the overvoltage crowbar.

 

[Fletch]

Hi, Chuck, Yes, I did check the Zener. It was fine. Out of curiosity, I did pull the leg of the crowbar. Considering I am totally fed up with these supplies, I applied power, stepped away, and pop! One of the 16V caps on the 12V rail railed.

EVERY component has been either checked or replaced. There must be an issue with the transformer. Unless someone wants these free, I am going to toss them. I could have built a power supply from scratch with all the time I put into both of these supplies. I certainly don't want to waste anyone's time on this forum as well.

I am extremely thankful for the quick responses!

Where these earmarked for a project Kaypro or are they spares? If it's for a project the BBC Micro community have had a lot of success switching out their failing Astec supplies for modern MeanWell triple output PSUs.

 

[m_thompson]

We had a difficult time repairing a DEC PDP-12 switching power supply. It turned out that the layout of the E-B-C in the replacement transistors was different than in the original transistors. The replacement transistors had the same part number as the original transistors. Once we determined which replacement transistors were in backwards, and replaced them turned 180 degrees, the power supply worked OK.

On another linear power supply that was a difficult repair, we connected a lab power supply through a high value resistor voltage divider to the base of the output transistor. We verified that making changes to the lab supply voltage changed the voltage output when connected to a fixed load. We then worked our way back through the regulator circuitry until we found part of the circuit that didn't behave correctly. This only works on linear power supplies.

 

[ldkraemer]

One of the 16V caps on the 12V rail railed.

After that 16 Volt cap blew, did you reconnect the Crowbar and see if the short was cleared? It would be
an interesting test. I'd check the same cap on the other supply.

Larry

 

[Hugo Holden]

I think that it is important to follow the repair task through to completion. If you give up on them, then the input you have put in so far will be wasted and on top of that it will always bug you that you failed at the task. I have a motto or policy of never giving up on repairs, sometimes that does cost me though.

It really is not a hard task to document the schematic. I have done this before for many types of apparatus, switch-mode psu's and even entire video monitors where the schematic was unobtainium. Once you get into the swing of it, it is quicker than you think. Start with pencil and paper and once the schematic is known, draw it out neatly with a computer drawing program. I use the vintage program Microsoft Picture It.

The schematic of course holds all the clues as to how it is supposed to be working and where tests should be made. Without it you are flying blind.

The main reason why you have not succeeded repairing the supplies so far is you don't have the schematic. Without that for any more than a very simple fault, it is not possible to make many tests, acquire meaningful data and formulate a theory on what could be wrong. So to succeed without the schematic (unless you are lucky with a simple fault) it is a low probability event.

Possibly, if you had copied it out in the first place, you would not be in the frustrating predicament you are in now and feeling like giving up. So I really would encourage you to copy it out and study the particular design before moving on to more testing/fault finding. Also, if you copied it out you could post it, then many others could help with suggestions.

 

[Dwight Elvey]

I agree with Hugo. You should not give up. Not having the needed bench supply is maybe a good reason to get one. You may be surprised at the amount of uses they have. I consider it to be one of my most important pieces of test gear. I consider it only second after my scope ( actually scopes ). My meter is third. Even a cheap kit supply can be useful for this use.
Drawing a schematic and making a layout drawing will allow you to do a much better troubleshooting. What you have tried so far is the typical easter egg hunt. The fact that you have a similar schematic to start from is a great starting point. Posting the schematic, even a pencil picture, here will allow people like Hugo and others that understand such things to show you where to look for things that can go wrong. It will help you to fix your problem and help others to understand the mysteries of switching power supplies. It will not only benefit you and others.
This group is a great asset. You first come with I'll lost. Maybe you were looking for the simple answer. We usually have none to supply you with. We do supply a great resource in trouble shooting abilities but it is a two way street. Doing with a message board is slow. Most only have one or two times a day to respond. Still, the value in learning for you and others is great.
Dwight

 

[Al Kossow]

there is a collection of astec schematics here to get you started, but as Hugo and Dwight have said, you should try to draw out your boards
http://bitsavers.org/pdf/astec/

many of their designs are similar
if you do come up with a drawing, i'd like to archive it to help others in the future