Fixing a power supply for Chameleon Labs 7720

[leigh]

Howdy,

The power board on my Chameleon Labs 7720 decided it was hungry, and that it was going to eat all the fuses that I feed to it. As soon as I power it up, it pops the internal fuse (1A). I'm talking about the internal power board here, the one that's powered by a 24VAC input on the back. (It's supplied by the included "line lump" AC transformer.)

So I've got to get in there and figure out what's gone south. Continuity testing with a Fluke DMM hasn't found any obvious faults to ground. However, if there's a bad filtering cap in there that's shorting (the manufacturer says they've seen the ceramic ones go bad first), it might only short while powered up with the full supply voltage, yet still look like an open circuit to a DMM... right?

And, while I'm in there, I figured I might as well make any easy improvements to the supply. This older post suggests pimping it out like the "bugle power supply" (PDF of that design here). Anyone had any luck with that? Even if PCBs were still available for the Bugle power supply, it wouldn't be as easy as dropping one in there, since the Chameleon power board has both +/- 15V and +/- 12V outputs. Not that that's what was being suggested in that older post, but totally replacing a faulty power supply is a tempting idea.

Leigh

 

[PRR]

> Continuity testing

What, the beeper? That's for quick-sorting trailer lights.

Put the Fluke in OHMS and get actual numbers.

24V at >1A is <24 Ohms, so you are looking for readings under 24.

Don't feed supplies unlimited fuses, it hurts their tummies. Replace the fuse with a 100 Ohm 10 Watt resistor. Now it can't pull over 0.24 Amps, you can leave it powered-up until you find where the smoke is getting out or you find a place which should be voltage but is dead-zero instead.

Do NOT start dreaming of "pimping" until you get the base system (which worked well until now) *working right*. Pimping while fixing is a path to the scrap-bucket. (Seen that done a lot on DIY forums.) For a mere +/-15V +/-12V supply, it may be reasonable to separately build a from-scratch pimped supply.

 

[JohnRoberts]

A common fuse eating fault for a power supply is a shorted rectifier diode, less common is a shorted capacitor, or transformer.

A VOM can help find a bad diode, or cap.

JR

 

[mjrippe]

Have you disconnected the power supply board from the circuit?  If not, the fault may be a power coupling cap on the compressor board.

 

[leigh]

> Continuity testing

What, the beeper? That's for quick-sorting trailer lights.

Put the Fluke in OHMS and get actual numbers.

Well put! (Although this Fluke combines its beeper mode with a Lo-Ohms mode, so there are numbers available, but point taken.) In any case, the next idea is exactly what I needed:

24V at >1A is <24 Ohms, so you are looking for readings under 24.

Don't feed supplies unlimited fuses, it hurts their tummies. Replace the fuse with a 100 Ohm 10 Watt resistor. Now it can't pull over 0.24 Amps, you can leave it powered-up until you find where the smoke is getting out or you find a place which should be voltage but is dead-zero instead.

V = I*R to the rescue! (Funny how that's so often the case, yet how often I forget to use it.) I had been thinking of powering it up with one of those dim-bulb "poor man's variac" setups, since it was immediately blowing fuses, so all I had to test was a cold circuit. But subbing in a heavy resistor for the fuse is an easier way. Thank you! I'll see what I can suss out using that approach.

 

[JohnRoberts]

I know I am repeating myself, but power supply faults are generally pretty simple.

If your fancy Fluke VOM has a diode scale all the better "to check for bad diodes with".

The variac/current limited smoke/heat test, IMO is for after you have checked all the obvious suspects.

Sorry, you are free to spend your own time as you wish, just saying what I would do..

JR

 

[PRR]

> one of those dim-bulb

Yes; but on the 120v side of the wart it has to be a very low-Watt incandescent, and on the 24V side you need a 24V lamp which may not be readily available. (You could get two 1156 tail-light lamps, 12V 0.4A, and series them. But 1156 has been going out of style.)

John is right that direct short-testing is less exciting/dangerous than smoke-testing. However until you know your testers *well*, getting readings in-circuit is confusing. You are "looking at a diode" but in-circuit there is a sneak-path through the main caps, the transformer, and possibly those teeny-caps we find for RF/TVI reduction. You can test any part out-of-circuit by lifting all but one leg; but that's increased damage to PCB. IMHO, at that point you may as well shot-gun: replace all electro caps and diodes. And since it failed, yet you love it, use higher-current diodes than the penny-pinched designer had to use.

The 24V AC through a large resistor "is" the same as a diode tester but testing ALL of it at once.

 

[mjrippe]

Forgot to mention, check the voltage rating of the caps.  I opened a dead Chameleon unit that ran on 24v and the caps were only rated 25v!  25v caps would be fine for your 15v rails of course, but make sure they didn't cheap out and use 16v caps.

 

[leigh]

OK, got to poke around a bit more today.

First did cold circuit testing, looking for too-low resistances to ground. Didn't see anything even close to being as low as 24 ohms.

The diodes in-circuit (cold circuit) are measuing a 0.55 V drop, with a 1V+ continually rising measurement in reverse (due to the caps they are connected to charging, I believe)... in any case both diodes measure about the same. (And yes, I did say "both" diodes - the power board does not use a full-wave recifier. Is that necessarily a sign of a cheap design? I don't take it as a great sign.)

So then I fired it up with a 680 ohm, 10 watt resistor (the only high-wattage resistor I had on hand) in place of the fuse, and I am seeing voltage everywhere I should. The voltages are quite a bit lower than they should be, and I'm not clear if having the resistor in place of the fuse would cause that (I can see how it would limit current, but can't figure what the extra resistor in place of the fuse would act as a voltage divider with.)

Coming right off the diodes, I'm seeing VDC of +7.3 on one side and -5.9 on the other. Then, at the other end of the filtering ladder (3 big caps worth), the VDC is +6.6 and -5.1. After that, it hits a pair of 7815/7915 regulators, and leaves the power board at around +5.8/-4.9.

And, in answer to a previous question, I do have the power board output disconnected from the rest of the compressor circuit right now. It was still blowing fuses like that.

Still working on this...

Leigh

 

[PRR]

> "both" diodes

It is a voltage doubler. By far the goodest way to get +/-DCV out of a single AC winding. A fine plan for small current.

24VAC ought to peak-out at 36V both sides. Which is suspicious close to the 35V rating on those regulators.

With the way-large resistor, it won't peak-charge, it will sag badly. Which is what you want for smoke-free testing. You are getting around 7V out, at maybe 12mA, each side, which is not unreasonable.

You need to get that voltage up and see if a diode, capacitor, or regulator is breaking-over at higher voltage.

 

[JohnRoberts]

OK, got to poke around a bit more today.

First did cold circuit testing, looking for too-low resistances to ground. Didn't see anything even close to being as low as 24 ohms.

Many PS faults are resistive, like a backwards opamp, might measure high impedance but draw lots of current when powered up.

The diodes in-circuit (cold circuit) are measuing a 0.55 V drop, with a 1V+ continually rising measurement in reverse (due to the caps they are connected to charging, I believe)... in any case both diodes measure about the same. (And yes, I did say "both" diodes - the power board does not use a full-wave recifier. Is that necessarily a sign of a cheap design? I don't take it as a great sign.)

Good diodes are always good...

So then I fired it up with a 680 ohm, 10 watt resistor (the only high-wattage resistor I had on hand) in place of the fuse, and I am seeing voltage everywhere I should. The voltages are quite a bit lower than they should be, and I'm not clear if having the resistor in place of the fuse would cause that (I can see how it would limit current, but can't figure what the extra resistor in place of the fuse would act as a voltage divider with.)

If the fuse is in series with transformer primary, the resistor you added forms a divider with the transformer primary impedance.

Coming right off the diodes, I'm seeing VDC of +7.3 on one side and -5.9 on the other. Then, at the other end of the filtering ladder (3 big caps worth), the VDC is +6.6 and -5.1. After that, it hits a pair of 7815/7915 regulators, and leaves the power board at around +5.8/-4.9.

That looks reasonable.

And, in answer to a previous question, I do have the power board output disconnected from the rest of the compressor circuit right now. It was still blowing fuses like that.

Still working on this...

Leigh

Is anything getting warm?

With no load connected I understand the voltage drops across the regulators because they have active circuitry inside. I am not sure about the almost volt drop across the "filtering ladder"? What exactly is that, an RC ?

JR

 

[leigh]

It is a voltage doubler. By far the goodest way to get +/-DCV out of a single AC winding. A fine plan for small current.

My misgivings stand corrected. A "bipolar supply from a non-centertapped transformer" schematic can be found here (all the way at the bottom of the page).

Went out and got a couple more 10W resistors in smaller values (than the 680 ohm one I had here). Testing again with them next...

 

[leigh]

Retested again with a 100 ohm resistor, then a 50 ohm resistor, in place of the fuse. Voltages are up to normal everywhere. Then tried again with the fuse, and it pops immediately.

I'm using these 1A fuses from Radio Shack... is there a chance I should be using slow-blow fuses instead? I asked the manufacturer what the replacement fuse should be (since it's not spec'd anywhere in the compressor's manual), and they said 1A but did not specify slow-blow.

 

[PRR]

Get fuses from another source.

I have seen 220r resistors in a 220K bubble-pack. That time the wrong-color caught my eye. With fuses it is much harder to tell 0.1A from 1A by eye.

 

[leigh]

And, to circle back around to John's questions:

Is anything getting warm?

With no load connected I understand the voltage drops across the regulators because they have active circuitry inside. I am not sure about the almost volt drop across the "filtering ladder"? What exactly is that, an RC ?

I haven't checked for anything getting warm yet... no smoke in any case.

Yes, by "filtering ladder" I meant several RCs in series. Three big electrolytics, each stage separated by a 51 ohm, larger wattage resistor.

The power board puts out both +/- 15 and +/- 12 volt supplies. It looks like the first two RC filtering stages after the diodes are shared, and then the two different-voltage sides each get their own private 3rd RC stage. The 15 volt final output is handled by the above-mentioned 7*15 regulators, and the 12 volt final output is handled by a pair of transistors.

This might all be a bit easier with a schematic, but the manufacturer won't release even just the power board schematic.

 

[leigh]

Get fuses from another source.

I have seen 220r resistors in a 220K bubble-pack. That time the wrong-color caught my eye. With fuses it is much harder to tell 0.1A from 1A by eye.

I can verify that these fuses do indeed have "1A" stamped on their metal caps. Are you saying I should not trust that?

 

[JohnRoberts]

Trust but verify... time to test the fuses.  8)

If there are 51 ohm resistors in series with the unregulated PS you can directly compute current consumption using ohms law (I=E/R).

Since 51 ohms would  drop 51 Volts at 1A  It does not seem likely that an amp plus could ever be drawn through that network.

High current draw must be elsewhere, or fuses are faulty.

What happens if you put a fuse in series with your power resistors?

testing a fuse is not as simple... they do not have a sharp threshold, but a promised fail time at 1x rated current, with shorter times for higher current overloads.

If it is a 1A fuse figure some way to test that it does not fail at 500mA.

JR

 

[PRR]

51 Ohms is a BIG clue. You can't have a half-Amp past that point!! It also means the operating current must be much-much less than a half-Amp.

24V AC through 25 Ohms, 0.96A.... a "1A" fuse should survive that for hours.

> Are you saying I should not trust that?

Who at Radio Shack knows enough to verify fuses? The counter clerk? The store manager? Both are hired NOT for their electronics know-how. I've worked in the warehouse and the electronics skills are even lower, and they aren't supposed to touch the stuff. There's a guy in Texas who checks-off lists of crates that came off the ship. There's folks in Asia who *should* be watching that the right fuse-wire goes with the right cap-stamps in the right bubble-card, but the manager is thinking about his motorbike and his girlfriend and the actual workers don't get paid enough to care.

 

[keefaz]

I'm using these 1A fuses from Radio Shack... is there a chance I should be using slow-blow fuses instead? I asked the manufacturer what the replacement fuse should be (since it's not spec'd anywhere in the compressor's manual), and they said 1A but did not specify slow-blow.

Is the power transformer a torroidal type ?

 

[leigh]

51 Ohms is a BIG clue. You can't have a half-Amp past that point!! It also means the operating current must be much-much less than a half-Amp.

24V AC through 25 Ohms, 0.96A.... a "1A" fuse should survive that for hours.

Yes, indeed - although, on each side (+/-) there are a couple caps to ground before the first 51 ohm resistor (a 1000uF and also a 0.1uF ceramic). Manufacturer claims they've seen the ceramic fail a few times.

Also the power circuit splits to somewhere else after the fuse, besides the junction of the two rectified diodes, via a wire tacked on to the underside of the circuit board. I need to understand where that goes better... thought it was just to power the lamp for the meter (which is lighting up fine when I have that power resistor in there, so it's not shorted), but maybe I better double check my understanding of that.