Desktop PSU relay problem

[Mongrel Shark]

I have a 60v 5A cc cv PSU I got from chinabay.

I was playing with a SSTC that was running from it (yes yes very silly of me, I have a variac for that stuff)

Anyhow now the relays in the PSU don’t switch. I'm still in the process of removing the boards, so pics soon.

The op-amps I could get a probe to and the pass transistors all seem to be ok. The most obvious issue is the relays not switching to different secondary as the range voltage is adjusted or if the current demand goes up.

So while I am looking at the relay driving setup is there anything else I should check while I'm in there?



I had just had the TC throwing a good bit of plasma from 30v input and turned the voltage down to 0v to make a change on the topload. When I turned it back up the psu got stuck in the 5v-13v range. Its low on current too, only puts out about 1.5a. There is a telling lack of the relays clicking sound as the voltage is adjusted through the range.

I'd say the TC was sending a lot too much power to the coils in the relays and either melted the wire on the coils or blew the relay coil transistors, maybe an op amp or something behind that. Induced Voltage on the coils would have been minimal while the relays are magnetised, but when they switch off they probably turn into great wireless power receivers... Even a nail can spew a few inches of plasma stream near this particular coil, It lights large fluros a meter away. So anybody’s guess what kind of voltages the relay coils could have seen, but well into the corona discharge is setting stuff on fire ranges I'd say.

In hindsight I probably should have at least used a longer lead to power the approx 30kv Tesla coil. Running it 8 inches from the psu, with the psu on the shelf at topload height was probably asking for a repair job.... This is why I have a variac...

 

[Harald Kapp]

Check whether the relays get the control signal to the coil.

If the yget the signal but don't switch, chances are the relays are defect. however, I'd suppose it's not the coils (as they are not connected to the output and therefore unlikely to "see" the high voltage that may have coupled back from the TC). Chances are the contacts are welded (stuck) in one position due to an electrical arc casued by high voltage.

If they don't get the signal, trace back to where this control signal should come from. Does the PSU measure voltage and current correctly to generate the control signals to the relays?

In hindsight I probably should have at least used a longer lead to power the approx 30kv Tesla coil.

Longer leads alone will not help much. A good power line filter will be of better use to your setup. Although designed for use with AC, these filters can be used with DC, too. Use a good protective earth connection directly from a wall outlet (or any other suitable erath you may have available at your workbench) to teh filter, bypassing the PSU.

 

[Mongrel Shark]

Yesterday got away from me. Just in the process of working out if its the relays or the relay signal now.

Many of the control signals around the op amps seem to be stuck in the 5-13v range or only show 8v of range (eg one signal on what looks like an op amp shows -10 to -18v), but I think this is likely due to the relays not switching. Current and voltage tracking works fine in the 5-13v range. If I was happy with it only producing 5-13v with a 1.5A max there would be nothing wrong with it

Double sided board is a bugger to trace. I'll try and get pics soon. got a couple of part numbers to look up. I think I see the transistors that switch the relays. Will look up those part numbers and probe a bit more to confirm.

Welded relay contacts did occur to me.


Feedback through the wires connecting the TC would be minimal. Maybe a few 30v spikes. The PSU's output buffer cap would soak most of it up no problem. The problem would have been wireless power. Any metal object (like relay contacts) within range can and will spew plasma. Coils get coupled with the TC and develop huge voltages. Even a short length of wire can be half a turn or an antenna or something, and develop serious plasma breakout a few feet from the coil, especially if there is another object it can arc to that is capacitively coupled on a different phase... I've had the screws in my wooden desk smoking before I moved the HV station to a plastic table. My secondary is about 1400 turns, so if a couple of thousand turns of relay coil coupled with it there could have been hundreds of thousands of volts there.... I had an inch of brush plasma coming from my fingertips (don't try that at home kids, plasma burns suck) just before the PSU died.... Any sensitive electronics should be kept at least 5 metres from this TC when running. A smaller TC has killed USB hdd's from 10 metres, the motor coils couple and blow the crap out of the drivers. Its worse if the HDD is connected by a long USB "antenna" cable to maximise the damage. Was silly of my to fire it up on the repair bench...

If a coil within range of the TC is even halfway saturated by dc then the TC will have little effect, but if those relay coils are driven by AC it could be a very different story. The 50hz makes a great carrier signal... AC relay coils could explain some diodes I couldn't work out.

 

[Mongrel Shark]

Chances are the contacts are welded (stuck) in one position due to an electrical arc casued by high voltage.

Yep. That was the problem. I welded all of them. I found the one that was supposed to switch coils around 12v and it wasn't switching. Took me ages to trace it all out the board is packed pretty tight.
[board notes for those that zoom in] I had another relay (small dc only) and a transistor (G1815 npn) out at the time of the pic. They where tested and working so went back in. Although now I wonder if the relay may have been welded too, and testing cracked the weld.
General overview of the circuit shows a 7812 on a heatsink top left. This is referenced to what looks like its own output in some clever wizards trick (I didn't investigate in detail) so it puts out 24v as a 12v+- setup for the 6 op amps on the 2 IC's (one duel one quad both Ti knock offs) The other heatsink has a TIP41c for a final voltage limiter. The 3 24v relays are flip flops that change the secondary config on the Transformer as the voltage and current demands change. There is also a bank of big npn power transistors (not shown) hooked up to current shunts for the CC limiting on the ground side. The output neg is isolated from ground but maintained at the same voltage. My fluke sees 0.0mv between neg and ground. But they are isolated. I thought that was clever. Especially because the ground reference was taken from the transformer. From a distance one thing stands out. The bloody huge little big capacitor on the output. Its massive physically but only 100v 470μf 105c.... Must have been a cost saving thing??

The pull-down signal was there and the coil tested the correct ohms. I removed it and put it on another 30v PSU to test it and confirm its death. Normally it switches at 14.9v across coil, but it took 18v worth of magnetism to bust the weld. After that it went back to normal operation. The current delivery problems I was getting at first was because I had a current limiting transistor shorting its base to earth via the transformer body, caused by my partial disassembly for repair.. Once I found that the CC side of things was all good again.

While the relay was out I used a clip lead to "simulate" the relay by hand. This got more power to the other 2 relays as the voltage was turned up and they snapped free too. At which point the PSU worked with one relay removed and bypassed to its over 12v output state. Although I wouldn't rate it for 5A at low voltage that way, Would probably burn the voltage regulator trying to step all the extra voltage down..

The fixed relay is soldered back in and its working as it was. I put 3 more relays on my shopping list as the welding possibly shortened their contacts life a bit.

Now on to improving it while I'm in there.

A good power line filter will be of better use to your setup. Although designed for use with AC, these filters can be used with DC, too. Use a good protective earth connection directly from a wall outlet (or any other suitable erath you may have available at your workbench) to teh filter, bypassing the PSU.

The filter is a good idea, but I'm not sure it will work in that configuration. That setup should filter a lot of the power line kickback from the TC transistor switching off. This does throw a spike of up to 30v over supply back down the input. If driving from a car battery a 1000-5000μf cap soaks it up pretty good. If that was the only noise I was worried about. This would be an ideal fix.
The problem I see is that the TC likely (its a matter of some debate) puts a counter sine wave on the earth line in my house. If I was closer to ground (I live on the 1st floor) it would be less of a problem, but if I run the TC from a large car battery, and use a non-grounded floating scope that’s referenced to the car battery. There is a strong sine on the ground pins in all my house wiring. The ground signal is 180 deg out of phase with the topload/transmission sine. It seems to be a capacitive effect. Pretty much every conductive object gets affected in proximity to the TC, either with the transmission phase or the ground/capacitive phase. The only exception I have found is large AH batterys. Or if I go out onto the lawn and drive a pin down for a low induction earth connection. I'm pretty sure its not bad scope referencing. As I can power loads between neutral and ground at any power outlet in any room in my house if a larger TC is running... Or even ground and the window frame. Significant loads like DC motors in the 1-5w range, or rectify a few hundred volts into a lage electrolytic cap fairly quicly. The power is coming from the TC and not the mains. I've tested it a lot. The only way to get a clean flat reference voltage is to incorporate something that can source/sink a truckload of power in a short time and have it connected with the least induction possible.

Its a huge can of worms but you have got my curiosity now. So I will have a play at some point and start another thread when I do.

For now I'll stick to what I know works and beef up the output cap.
Pictured is the original 100v 470μf 105c cap. (I figure it is huge because it was cheap.) Next to a pair of 200v 680μf 85c caps I think I will replace it with.


Its a bit of a tempreture de-rating, but I can't see it getting close to 85c and I end up with a lot more farads with these two in parallel. Its the best I can find in my drawer right now. You can see the FWBR got afterthoughted onto the case for heatsinking and wired to the bottom of the pcb. So I got some bonus through holes I can use for the second cap . May have to extend legs a bit.
I think I will add a better cap to my shopping list once I have measured the space and pins accurately. 5mF or more would be nice.

I have done some research and found that these cheap PSU's are prone to a voltage spike on output when the power is toggled. Because this cap drains too fast and an op amp looses power. 1000μf+ seems to have fixed all the documented cases I found online. My units both do this so it will be nice to fix that and provide a slight improvement to the output stability on nasty loads too.

There where a couple of crappy afterthought solders I fixed up. Other than the cheap output cap I am pretty impressed with the whole thing, considering it was only $130AU.