Fried my power supply by doing a bad thing

[jrmichler]

TL;DR Summary

Tried to use it as a battery charger to measure current and voltage. Hooked it up backwards - OOPS.

I have a Multicomp 0-30V 0-5 amp power supply. I did some tests using it as a battery charger. It worked very well the first few times, then I accidently hooked it up backwards and killed it. There was light sparking when this happened. No fuses blew. It powers up normally, and I can program in any voltage and current limit. When I turn on the output, it immediately goes into current limit at any voltage over about 0.1 volts.

I looked inside and saw multiple circuit boards with SMDs. I am not going to try to reverse engineer this thing. Does anybody know if this problem is likely some easy to replace component (no SMDs)?

And a side question. That big toroidal thing that looks like a transformer has six pairs of visible wires coming out. I can understand one pair for input, another pair for output, but what could be the reason for another four pairs of wires?

 

[dlgoff]

... but what could be the reason for another four pairs of wires? ...

Maybe to get four other voltage sources?

 

[Baluncore]

I can understand one pair for input, another pair for output, but what could be the reason for another four pairs of wires?

That does look like a power-frequency main transformer. Maybe the output is isolated from the front panel controls. It might use synchronous rectifiers. Maybe it has voltage taps that are changed for different AC inputs.
If the transformer was part of an inverter, it would have several feedback windings to control the drive circuit.

 

[Tom.G]

If you are lucky, it is just a reverse protection diode (or maybe a Zener) directly across the output terminals that is shorted.

If unlucky, the series voltage regulator transistors are fried... or perhaps the rectifier that supplies the output voltage.

If either case, they would be the physically larger components on the circuit boards.

 

[Rive]

When I turn on the output, it immediately goes into current limit at any voltage over about 0.1 volts.

It's likely sensing it right and there is some kind of short on the output terminals. That would mean only a few components to check (and replace).
Capacitors, diodes, only a few pieces.
You can try to check it on the terminals with a DMM (while the PSU is off and disconnected, of course).

That big toroidal thing that looks like a transformer has six pairs of visible wires coming out. I can understand one pair for input, another pair for output, but what could be the reason for another four pairs of wires?

I cannot tell for sure, but based on that big heatsink (right near a fan) it has some kind of dissipative regulation. For that kind of PSUs the dissipation depends on the difference between the input and output voltage of the regulator (and, of course: the current).

To limit the dissipation, it's a common practice to switch between different input voltages. So, let's say: for 5V output it'll work from 10V coil, and for 15V output it'll pick a 20V coil.
Also, it's common to have a separate coil to feed the internal electronics.

 

[Tom.G]

Another possibility is the current sense resistor for current limiting may have lost its 'magic smoke' and is open from the overcurrent. A common value is 0.1Ohm.

 

[Baluncore]

Another possibility is the current sense resistor for current limiting may have lost its 'magic smoke' and is open from the overcurrent. A common value is 0.1Ohm.

That could explain why it can be programmed, but current cannot be drawn.

No fuses blew. It powers up normally, and I can program in any voltage and current limit. When I turn on the output, it immediately goes into current limit at any voltage over about 0.1 volts.

If that is without any external load, then the output stage or over-current protection has become a short circuit across the output.

 

[jrmichler]

Thanks, people. I'll look into it and let you know. Might not be today, though.

 

[Guineafowl]

On mine, it was a fried LM324:

https://www.physicsforums.com/threads/linear-dc-power-supply-repair.923696/

 

[hutchphd]

Also never underestimate a really good visual inspection, particularly if sparks or magic smoke has been involved.

 

[DaveE]

Maybe you could open it up a bit and post some good images of the PCBs. Particularly the one that the big output wires go to, and the ones with big devices on them, that's most likely the problem area. Device P/Ns would also be very helpful.

First thing is a good visual inspection. Then look to see if it has a big diode directly across the output to protect against exactly this sort of thing, it will probably be fried.

 

[dlgoff]

On mine, it was a fried LM324:

https://www.physicsforums.com/threads/linear-dc-power-supply-repair.923696/

Dang. Very sad; memories of Jim Hardy.

 

[jrmichler]

Maybe you could open it up a bit and post some good images of the PCBs. Particularly the one that the big output wires go to, and the ones with big devices on them, that's most likely the problem area. Device P/Ns would also be very helpful.

First thing is a good visual inspection. Then look to see if it has a big diode directly across the output to protect against exactly this sort of thing, it will probably be fried.

The only sparks were between the leads and the battery. No visible or odor of magic smoke. Photo of output board. The three bolts with nuts are the outputs, the brown and black wire pair is apparently the power to the outputs.

And a photo of the other end of that blue and brown wire pair:

Immediately to the left of the wire pair connection to the board is what looks like a diode. It has a gray stripe at the upper end, and a label on the board (not visible in photo) D19.

That big heat sink on the right side background of the OP has four NJW0281G, which Digikey lists as Bipolar (BJT) Transistor NPN 250 V 15 A 30MHz 150 W Through Hole TO-3P-3L.

I took some ohmmeter readings. There are three output terminals.
The black one is labeled -
The green one labeled GND
The red on labeled +

Resistance from black to GND = OL in both directions (overload, infinite ohms)
Resistance from red to GND = OL in both directions
Resistance from black to red = 0.3 ohms in both directions
Resistance from both black and red to accessible end of diode = 0.3 ohms in both directions.
Both directions means that I switched the DMM probes.

It looks like I should remove that circuit board and visually confirm that the diode is connected across the blue and brown wires. If so, it's fried. So what should the specifications of a replacement diode be if I cannot find a part number?

 

[DaveE]

It looks like I should remove that circuit board and visually confirm that the diode is connected across the blue and brown wires. If so, it's fried. So what should the specifications of a replacement diode be if I cannot find a part number?

Not necessarily. Since it's connected to an unknown circuit, you don't really know what you are measuring. Without a schematic, you really need to disconnect it to know for sure. OTOH, yes, it's likely. So, if you'll probably have to replace it anyway, cutting a lead isn't that expensive. If you do that try to cut it so that you can solder it back together if we are wrong.

It has a label on the package, I can see part of it. But, yes, we can pick a common replacement. That's not a very sensitive application.

Also, 0.3 ohms is small enough that you have to be careful about the measurement, like the lead resistance and such.

 

[Averagesupernova]

3 ohms across the output is a very good first clue. It should not be there. Disconnect things until you narrow down what is shorted.

 

[DaveE]

OK, if we continue the speculation that it's a shorted diode that needs replacing, then here's what I would choose:

1) The biggest diode that fits on the PCB without difficulty. It looks like maybe a DO-201 or DO-27 package to me. Those have a body of ~5mm dia. x ~9mm. Lead diameter is also important so it fits in the PCB easily. The point here is that you want it to tolerate as much abuse (power/energy) as it can while it tries to protect the circuitry upstream. Size matters for energy absorption.

2) Absolutely must have a reverse voltage rating above the maximum PS output. I wouldn't use anything below about 40V for a 30V PS, but I'd be more comfortable with 60V. Higher that that is fine but of little cost or benefit.

3) The highest forward current rating you can easily get. This probably means a Schottky diode. Something like:
SB2060 - 60V, 20A, DO-201
SB10150 - 150V, 10A, DO-201
STR10100 - 100V, 10A, DO-201

4) Speed doesn't really matter.

The selection tables at sites like Digikey or Mouser are your friend in this regard.

If it is just a reverse diode, you can disconnect it and test the PS without it to see if that is the only problem. It has no function in normal operation.

It also could be a transient suppression zener, something like 1N4755A, but I strongly doubt it. They won't work as well at reverse polarity protection, by a lot, and transient voltage protection would be better located upstream of filter elements (L's and C's at the output) with a smaller device.

 

[dlgoff]

@DaveE
I would add: make sure the diode has a little room between it and the board to let a little air circulate under it.

 

[Averagesupernova]

@DaveE
I would add: make sure the diode has a little room between it and the board to let a little air circulate under it.

I suspect that diode is only used once and it did the job it was supposed to do. It is not likely to be conducting in normal use.

 

[jrmichler]

The 0.3 ohm I listed above is the lead resistance. I got the diode out. It measures near zero ohms in both directions. It's a 1N5408. Digikey lists it as obsolete, and a 1N5408G as a drop in replacement. Cost is $0.47 each, with minimum order of one.

Drat. I can't think of anything else to order at the same time. Oh, well, such is life. Maybe I'll buy a couple extra for the next time I do something stupid inadvertently erroneously.

Thanks, everybody. I would not have figured this out by myself.

 

[DaveE]

Did you turn on the PS? There may be other stuff to fix too.

 

[jrmichler]

So I went out to the shop, plugged in all the connectors, and turned it on. No smoke. The measured output voltage matched the display voltage, and it properly powered a test load (a 12V incandescent bulb).

 

[berkeman]

Digikey lists it as obsolete, and a 1N5408G as a drop in replacement. Cost is $0.47 each, with minimum order of one.

Digikey's minimum order is $25, no? Or have they gone the Amazon route?

Very glad that the troops helped you to figure this out.

 

[Averagesupernova]

If the diode did what was intended nothing else should have been damaged. As long as it wasn't blown open the rest should be protected.

 

[jrmichler]

Maybe to get four other voltage sources?

As I turned up the voltage, I heard a relay clicking at about 9 and 16 volts, and (I think) at another higher voltage. I did not take notes.

Digikey's minimum order is $25, no?

A little hard to read, but $8.00 with tax and shipping for 5 diodes.

 

[DaveE]

Cheaper than a new PS.

 

[Averagesupernova]

I heard by the grapevine digikey no longer has the $25 minimum. I haven't ordered from them for a while. There's one way to find out.

 

[berkeman]

I heard by the grapevine digikey no longer has the $25 minimum. I haven't ordered from them for a while. There's one way to find out.

Yeah, based on JR's invoice, that is no longer a requirement. Good to know!

 

[jrmichler]

The new diode is soldered in, the power supply reassembled, and the power supply tested with everything working properly. No more battery charging with this power supply....

 

[Baluncore]

That is good to know.

A particularly difficult job I have found, is the design of a simple reliable battery charger, that is both accurate and immune to a reverse connected battery. Without a processor to check things are OK before making the connection, there are just too many possible modes of failure.

 

[Tom.G]

There used to be chargers that would tell you almost instantly if you connected a battery backwards.

The rectifier was a few metal plates coated with Selenium, a high current would of course overheat the rectifier.

The aroma was rather strong and reminiscent of rotten eggs... not something you would soon forget.

https://www.google.com/search?q=selenium+rectifier+smell

 

[dlgoff]

There used to be chargers that would tell you almost instantly if you connected a battery backwards.

The rectifier was a few metal plates coated with Selenium, a high current would of course overheat the rectifier.

The aroma was rather strong and reminiscent of rotten eggs... not something you would soon forget.

https://www.google.com/search?q=selenium+rectifier+smell

IIRC they were used in the high voltage supplies for TV picture tubes.
From: https://en.wikipedia.org/wiki/Selenium_rectifier#:~:text=A selenium rectifier is a,rectifiers in the late 1960s.

 

[Averagesupernova]

I seem to recall seeing those in old tube radios. Rectifier for the plate supply. Maybe I'm thinking of something else. I recall working on a record player that used just two tubes they each had 35 volt filaments. Wired in series and burned up the remainder in a power resistor. Must have had a selenium rectifier for the plate supply.

 

[dlgoff]

35 volt filaments.

Some old sets I remember working on had 117 volt filaments.

 

[Tom.G]

IIRC they were used in the high voltage supplies for TV picture tubes.

Nah... A Selenium plate is only good for 20V or so.

The TV high voltage rectifer was the 1B3 vacuum tube, peak plate voltage of 26,000V and a 1.25V filament. The filament was also fed from a winding on the flyback transformer, which of course was floating at the rectified HV DC.

1B3 data sheet at:
https://www.google.com/url?sa=t&rct.../1/1G3GT.pdf&usg=AOvVaw2nEe3LzCQJ3fW69uU8LwQN

 

[Baluncore]

IIRC they were used in the high voltage supplies for TV picture tubes.

High-voltage stick rectifiers were made from silicon rectifier pellets, stacked end to end. I don't believe HV "stick rectifiers" were ever made from selenium diodes.

Selenium was used in some lower voltage "metal rectifiers" which needed a stack of oxide coated cooling fins.