Change output voltage of SMPS

[ajp8868]

Hi All,

Kaihui sell this power supply: https://m.alibaba.com/product/60421519548/2018-KAIHUI-switching-ac-dc-led.html

Which is the 12/24v version of a power supply I need (12v 4a, 28v 2.1a). It's the exact same layout and I would guess they just use a different set of resistors with the reg ic?

My question is, how easy would it be to get this to output 28v on that rail? Could someone help me do it once it arrives?

Thanks!

[magic]

This PSU looks too simple for independent regulation of both output rails. It probably only regulates one, and the other rail is determined by 2:1 turns ratio of the secondary windings on the transformer.

It might be possible to mod 24V up to 28V, but then 12V will become 14V, most likely.

Another issue is that the same transformer usually has yet another "secondary" winding located on the primary side, which provides low voltage power to the switching controller. This voltage will also increase by 16%, which would probably be harmless, but it's something to keep in mind when playing with those things.

A hypothetical 28V output version of this PSU would likely use a different transformer, particularly if the lower rail must be 12V exactly.

They say that they can produce other voltage versions, maybe it's an option if you are ordering more than a few units.

[ajp8868]

Ah thank you for that detailed explanation, makes perfect sense.

They do a 12/28 version but it's OEM (I'm guessing they only produced it for Chauvet).

Do you think it would be possible to get a transformer with the correct ratio to get the voltages I need..?

[magic]

I suppose it would be hard, you have no idea who makes this transformer and what's inside and it could be more or less a custom order for this particular PSU manufacturer.

Perhaps 13V/26V would be close enough?

I suppose your problem is finding replacement for a broken PSU in some commercial product?

[Bud]

I've seen a trickier voltage divider feedback loop, where both output voltages were fed to the resistor divider in a proportion, so changes in any of the rails resulted in adjustment to restore the balance.

[ajp8868]

To be honest it might be close enough! Though it might cause a problem with the led if it's trying to draw more current than it can handle?

Yep it's exactly that, I can get a replacement through the manufacturer, but it's more than the light's worth so I'm happy to try other routes before biting the bullet.

I really wanted to repair the one I've got, but it's blown up a couple of tracks and surface mount resistors so I can't even read them to get replacements. So at this point I'd be spending more on components lol

[magic]

Post pics of the disaster, no harm in trying.

I've seen a trickier voltage divider feedback loop, where both output voltages were fed to the resistor divider in a proportion, so changes in any of the rails resulted in adjustment to restore the balance.

It doesn't matter. OP want to change the ratio of output voltages, and this ratio is solely determined by turns ratio. (OK, I'm ignoring voltage drops of secondary rectifying diodes, but the principle stands: increase one rail and the other increases too).

For better regulation, the PSU would need two transformers and two independent controllers, or some sort of "buck" converter on the secondary for the lower rail like in ATX PSUs.

edit
Any chance that the 12V rail is barely used at all? If so, it could work with any 28V PSU and some step down converter to power the logic or whatever else needs 12V.

[ajp8868]

It might work but it would need a few amps at least as it runs about 6 stepper motors and the main board.

But I guess if I can get a powerful enough 28v psu and a buck that can handle it, it should work

How can I calculate what wattage 28v psu I'd need?

[tunk]

Calculate the total wattage needed (assuming ~80% efficiency of the converter),
and then divide it by 28V to get current. Or use the PSU in your first post, adjust
it to 28V and use 2-3 high current diodes to drop the voltage to ~12V.

[ajp8868]

Thank you!

So damage report: there's a couple tracks blown which I've soldered a couple and replaced one with a wire (not currently connected at one end)

It blew up a 0.15ohm 2w resistor Which was on the high voltage side of the optocoupler. I didn't have any of that value so I've used 1.5ohm - will that work?

It blew up a couple tracks and resistors on the smaller PCB which has those 5 pins on the high voltage side. I thought this was the voltage regulation circuit but I'm not sure?

It just blows the fuse instantly when plugged in but I can't find any obvious shorts, I'm running out of ideas to fix it!

Could the transformer cause a short?

[wasedadoc]

A resistor as low as 0.15 Ohm could be a current sensing resistor in the emitter or source of the switching device.  If so, replacing with a resistor 10 times the value is going to severely reduce the power that the supply can deliver.

[ajp8868]

A resistor as low as 0.15 Ohm could be a current sensing resistor in the emitter or source of the switching device.  If so, replacing with a resistor 10 times the value is going to severely reduce the power that the supply can deliver.

Ah ok, best change that then. If it's only current sensing, is the 2w one in place actually overkill or is there a reason for it?

[magic]

This looks like the primary switching FET is shorted. It often happens when they fail that all three terminals get shorted together and then high voltage passes through the transformer (very low resistance), through the shorted MOSFET and blows up the current sense resistor (0.15Ω) and often also the gate drive circuit - some resistors, maybe a diode and probably the UC3843 chip too.

UC3843 is a very common and well known controller, you can easily find datasheets, application notes and any other information. Also replacement chips, if necessary. This PSU should be repairable, but there are many components which may have been exposed to high voltage and destroyed, and failing to identify and repair just a few could lead to another bang and wasted time.

Don't try to turn it on to avoid further damage. Make sure all caps are discharged and see if you can detect a short between the high voltage rails on the primary with DMM. Remove the 1.5Ω resistor - it's wrong value anyway. With this removed, test for drain-source short in the MOSFET. If it appears shorted in absence of the 0.15/1.5 resistor, it's 99% sure the MOSFET is shorted internally - desolder it and test out of circuit. It shouldn't have a short between gate and source, and if you short gate to source, it shouldn't conduct from drain to source.

If bad FET is confirmed, test all resistors on the UC3843 board. Check for internal shorts in UC3843, like FET drive pin to GND or VCC.

[ajp8868]

Magic, I think you are spot on. I've already replaced the MOSFET as it was indeed shorted.

I stupidly plugged it in after replacing the known bad capacitor (but before realising the MOSFET was bad) and it went bang again. I think this is what blew the 3843.

With not being able to identify at least one resistor on the board with the 3843, is it even worth going through it all? I would have liked to save it from the scrap heap, but it might also be useful for spares - well, whatever's not fried!

Potentially if I could somehow identify that resistor and replace everything found to be bad on that board, I think it would work. But without having a big enough supply of spare parts, I think I'm out of luck/budget!

Thanks very much for your advice, I appreciate it. Trying to learn as I go, but even after so many years there's still new challenges!

[magic]

The blown resistor probably just connected UC3843 with the MOSFET so it could be replaced with some ballpark value.

To repair this, you would need to produce schematic of this small board and figure out which parts are doing what. This gives some idea what their typical value might be and whether they were at risk of damage.

When I dealt with a UC384x PSU with this exact same kind of damage, the blown parts were:
- the FET
- the current sense resistor (your 0.15Ω)
- resistors and a diode which connect the FET to the chip
- the chip, it was visibly cracked and smoked in my case
- some transistor in the chip's low voltage power supply

I replaced all of that and also tested most other components (it was a valuable PSU so I was quite paranoid) and it came back to life.

[ajp8868]

Those are the exact same symptoms, right down to the mosfet being cracked! I've got this so far, but am I going in the right direction? The resistors coming from pin 6 are the ones I need to identify - that are connected to the gate of the mosfet.

How far do i need to go with the schematic?

[magic]

Yes, I think the blown stuff in the corner is resistors coming from pin 6 to the gate. There is also a diode there, all these components may be blown. See how they are connected, measure all resistors, test the diode, read all markings that are still readable.

There is a 1kΩ resistor which seems to have arced to a nearby capacitor. Both are connected to the main board, what these connections are?

Possibly current sense (pin 3) may have received something nasty too, see where that signal comes from (ultimately - from the 0.15Ω resistor) and what components are along the way.

Check the chip for shorts, particularly from pin 6 to power supplies or between VCC and GND.

Check if the short circuit across high voltage rails on the main board is still there (if not, and 0.15Ω is removed, temporarily replace it with a jumper). Remove the bad FET. Confirm that the short goes away.

[ajp8868]

going to continue tomorrow so I'll put more detail in about those pins - this is what I have so far. Meant to attach this to the previous post..!

[ajp8868]

Tested all of those and I'm pretty certain all the components in the bottom right of the small board are blown - diode showing 0.07 drop both ways, and resistors are showing massive values.

I can't detect any shorts on the main board or the 3843, so I'm hoping we're good on that side of things.

One thing I'm wondering is if the transformer could be what's causing the fuses to blow, could this be a failure point if high enough voltage got to the wrong bit..?

Is there enough info in the diagram above to guess what the destroyed resistor will be? The one that has part marking left is 250 (I'm not certain on the 5, but definitely 2#0), how do I work out the other?

[drvtech]

To prevent further damage when testing blown up SMPSs one useful technique is to provide an external, current limited supply to the controller chip (UC 3843 in your case). That way you can verify that it is oscillating and driving the gate of your fet before putting high voltage on it. You can check the frequency and see it come on and off at its uvlo voltage (16V from memory?)

Edit to add: that resistor is likely to be 22 ohm (220).  Also I wouldn't put money on the UC3843 having survived. Enough voltage has been applied to those resistors to blow them up and, as the only path for that current was through the output of the chip, it will have been damaged. My suggested test with a bench supply powering the chip will demonstrate that. Just for the avoidance of doubt, that test is done with NO input supply (a.c.) connected. Although if you didn't realise that you shouldn't be anywhere near the innards of an SMPS!

[ajp8868]

Thanks I will test that, but I'd guess I'll need a scope for showing that pwm output? I'm going to buy one at some point but I don't have one yet..

With regards to the resistor, are you talking about the completely destroyed one or the one on the right with partial marking?

[magic]

It looks like R6 on your schematic is the 2x0 in the corner and R2 is the completely charred one.

The 2x0 could be 220 or 270, which is 22/27 ohm. The black one may be a few ohms. Was 4.7 on my PSU.
These aren't very critical resistors, it should still work with them a little off.

The diode is obviously dead.

I also doubt that the chip is still good. Considering its cost, I would just order a replacement together with the FET (or even two or three of each).

[ajp8868]

Thanks Magic. I'll use some donor boards for the resistors in that case, I have some that are very close to those values and plenty of diodes.

I'll let you know how I get on after replacing those parts and testing for shorts.

[magic]

Before powering up with the FET installed it's good to have confidence that the chip is working correctly and it won't turn on the FET permanently or for too long. If that happens, you are back to square one.

The chip must be good. The current sense circuit (pin  3) must be good.

It would also be nice if the path from the optocoupler to pin 2 is functional, otherwise overvoltage may occur on the secondary.

[ajp8868]

I'll replace as much as I can and checking each individual part as I go.

Also definitely replacing the chip as per your advice!

I've replaced the FET already and no shorts there anymore. I have a spare in case it goes, but I'm certainly hoping it doesn't.