Power Supply 60v 250A Repair

[CrazyTiger]

Hi,
Its a long holiday in my country, since I was stuck with my Wavetek 9100 Repair, I decided to take a look on a beast of a power supply I acquired some time ago.
This is a TDK Lambda ESS 60 250. Its basically a 250A "bench" Variable Power supply. I got this unit defective with previous repair attempts by someone.

It requires 3 Phase supply 415V So before powering up, I opened up each section, checked for shorts, lose contacts. I found a 33 ohm 25W Wire wound resistor open (marked in red). Here is the block diagram for reference:



These resistors are used to soft start the PSU, in which the resistor is connected to capacitor in series to control inrush current, after a given time the relay engages and bypasses the resistor. I found it strange a big resistor like that to be open, so I checked the capacitors, they were good, the inverters were also not shorting. Anyway I removed the connection to the inverter just in case, replaced the 33ohm, and powered up. I got 300V on one cap and nothing on the other. Checking back the replaced resistor, the resistor is dead 

What else can be an issue here?
The huge coil, in between?

[CrazyTiger]

That resistor looks like it's part of a soft start circuit, the relay contacts across it are intended to short it out after some pre-determined period of time, if it's gone open then the PSU will never start..

It *could* be that the resistor has just failed but on a 15KW supply, I'd be *exceptionally* wary of doing anything more than testing every component I could find before replacing anything.

I'd definitely not be poking about in there if I was at all unsure

For this I have extensively checked everything to be sure, in fact I have removed the entire load section (4x inverters) before replacing the resistor. But even in this condition the resistor is dying even without the inverter sections. This indicates a very high current draw, but I don't see anything that draw huge amount of current, and in that condition, the breaker should have tripped.

PS: Due to public holiday I could not get a quality 25W Resistor, I used a cheap Chinese made one also rated 25W, any chance they cant take high inrush current. A quality Vishay/Ohmite one is rated for 550V, but I am not sure about the chinese one.

[NiHaoMike]

Try an incandescent light bulb instead of the resistor. If it stays lit with no load, there's a short somewhere.

[bdunham7]

I don't think the differential mode choke would be an issue, but you can just test its DCR and leakage just to be sure. 

Assuming that it is C2 that is not getting charged up, you might have to reconsider whether it might be bad.  How did you test it?  Can you swap C1 and C2 relatively easily?  I know those things are huge and scary!  Are they connected with cables and screws or are they mounted to the board?

A grossly substandard resistor might be internally vaporized by the current surge, but that doesn't explain what happened to the original.  The peak current would be 10A at the very most.

One other thing to consider is the rectifiers.  One shorted rectifier would put one phase of AC across a huge cap and that would certainly burn your resistor, but might not trip the breaker because the resistor limits the current.

[free_electron]

One other thing to consider is the rectifiers.  One shorted rectifier would put one phase of AC across a huge cap and that would certainly burn your resistor, but might not trip the breaker because the resistor limits the current.

This. of one of the diodes in the bridge is shorted this will happen.

[CrazyTiger]

I don't think the differential mode choke would be an issue, but you can just test its DCR and leakage just to be sure. 

Assuming that it is C2 that is not getting charged up, you might have to reconsider whether it might be bad.  How did you test it?  Can you swap C1 and C2 relatively easily?  I know those things are huge and scary!  Are they connected with cables and screws or are they mounted to the board?

A grossly substandard resistor might be internally vaporized by the current surge, but that doesn't explain what happened to the original.  The peak current would be 10A at the very most.

One other thing to consider is the rectifiers.  One shorted rectifier would put one phase of AC across a huge cap and that would certainly burn your resistor, but might not trip the breaker because the resistor limits the current.

1. I checked the capacitance and ESR of the capacitor, ESR was low, and capacitance was well within tolerance, even used an external power supply to charge the capacitor out of circuit till 100VDC, to see how it charges and discharges, its holding the charge good, I can swap the caps, when I reinstall it back.
2. Regarding the rectifier, I tested all 6 diode, Forward voltage is about= 0.48V each and no leakage

[bdunham7]

Something may be failing under full load.  The previous suggestion of putting a light bulb in place of the resistor might be a good diagnostic idea here.  But be careful--these PSUs have enough energy to simultaneously cook and skin an alligator!

[coromonadalix]

your green squares,   the common connection   is  what i can call  the ground of the psu for many sections, could be called virtual ground   etc ...   

 if the resistor is changed and doesn't blow up    you will be good to take measurements   but keep this ground line in mind, it can play you some tricks


do you have a link for a full schematic ?

[BrokenYugo]

I'd be looking harder at the rectifier if you just checked with a DMM in diode mode. Do you have access to a megaohm meter or other current limited source of voltage around 500 to 1000V to check for proper reverse behavior? It is possible for a diode to fail halfway like that, if it is in a circuit where there isn't enough current to instantly destroy it to a shorted or blown open state.

[CrazyTiger]

Update: I have a fluke 1587 Insulation tester, will be using it to test the rectifier again. Maybe like most of you commented the rectifier might be half-dead (zombie-rectifier?) and a multimeter diode tester don't have enough juice to properly test it.

The light bulb is a good idea, will try that if the rectifier measures good.

[CrazyTiger]

Update: I have a fluke 1587 Insulation tester, will be using it to test the rectifier again. Maybe like most of you commented the rectifier might be half-dead (zombie-rectifier?) and a multimeter diode tester don't have enough juice to properly test it.

The light bulb is a good idea, will try that if the rectifier measures good.

The rectifier has been tested with my insulation meter at 1000V, all the diodes reverse voltage reading is about 50MΩ, hence there is no leakage, forward reads 0MΩ. The rectifier is good.
I have also tested the common mode choke for any sort of leakage to the body, or to the opposite winding, bott yielded 500MΩ, so no problem with choke too.

Next up will be light bulb test.

[strawberry]

if one of bridge rectifier diode is shorted = short circuit on mains supply
400V/30A and half-dead (zombie-rectifier?) = I can bet, this powersupply wont be repaired 

[bdunham7]

if one of bridge rectifier diode is shorted = short circuit on mains supply

Excellent point.  I recall having shorted rectifiers take out caps, but apparently not bridges, or at least directly mains-connected ones.

[trobbins]

Another diagnostic option is to feed DCV in to one phase and neutral, and between phases, and confirm nominal dc current paths, and no dc current paths where it shouldn't flow due to the other bridge diodes and status of the in-rush relay.

[fzabkar]

In addition to a light bulb, it might be an idea to disconnect 2 of the 3 phases at a time. This would narrow down a faulty rectifier under load.

[antenna]

Try swapping the caps around and see if the other resistor blows.  Being you only charged it to 100v, it could still be arcing over when it gets closer to 300v.  Sure, normally caps turn into a really smelly confetti when that happens, but just maybe?

[srb1954]

In addition to a light bulb, it might be an idea to disconnect 2 of the 3 phases at a time. This would narrow down a faulty rectifier under load.

But this could create more stress on the diodes in the remaining connected phase as they will be carrying the full surge current for up to half a cycle compared to the lesser average current and shorter conduction time that would occur if all phases are active.

In this case it would be wise to also use a series resistor on the mains supply as well as a light bulb. The light bulb will initially have a low resistance until it is warmed up and will not limit the initial peak surge current sufficiently. The light bulb really only provides protection against prolonged excess currents as its resistance increases considerably when the filament warms up. The extra fixed resistor, sized to limit the expected peak current to an individual diode's maximum surge rating, would help protect the diodes against the shorter term excess currents.

[edpalmer42]

So, at this point, all you're trying to do is charge the input capacitors?  All the following circuitry has been disconnected?  And you're trying to figure out why the original and then the replacement resistor blew?

Here are some random thoughts.........

A 35 ohm, 25W resistor has a full load current of ~1.2A @ 30V.  Bdunham7, you said you expected a peak current of 10A.  Where did you come up with that number?  I'd expect a much higher number, but I have no experience with 3 phase.  In any case, I'm not surprised that a 25W Chinese resistor wouldn't be happy with an 8X overload current spike and a, possibly, over 10X voltage overload.  What could possibly go wrong?     Note that the DF60AA120 3 phase bridge rectifier has a surge current rating of over 900A.  Tiger, can you find a part number and datasheet for the original resistor?  Does it have a surge rating?

Why did the original resistor blow?  Maybe just age?  How old is this thing?  It sounds like those resistors are living a very hard life.  If you have the room, I'd recommend replacing both resistors with higher wattage units.  Note that they have to be *rated* for high wattage, but they don't have to *dissipate* that power.  The surge comes and goes so quickly that heating is minimal.  So, for example, you can use a resistor that's designed to be mounted on a heatsink, but not use a heatsink - or maybe just a small one.  Many years ago I used a similar circuit to soft-start a 300W projector bulb.  A high quality 20 ohm 10W resistor died a quick death.  A 25W resistor was robust enough to take the surge, but heat wasn't an issue.

Does relay K1 operate?  If not, the resistors are being stressed even more because they have to survive the complete capacitor charging cycle.  Now, heat could become an issue.

If all the output circuitry has been disconnected (don't forget the A600 Phase Loss board) you can slowly ramp the voltage up - either AC or DC - and see how the circuit responds.

If you decide to use a light bulb, remember that the bulb has to be rated for the supply voltage.  So, if you don't happen to have any 400V light bulbs kicking around, put two 220V bulbs in series.  When I powered up a 33V@33A supply, I had to use a 100W bulb before it would start, but that was with all the circuitry active.  If all you're doing is charging the input capacitors, it doesn't really matter what you use.  The lamps will flash on and then slowly go out.

[strawberry]

soft-start relay could be faulty(high contact resistance/ or open) or not activated at all
if there is no shorts after that resistor
soft-start relay could be operational even if there is excessive capacitor leakage above ~300VDC

230V3f = 560VDC

[CrazyTiger]

The instrument was manufactured at 2004, so its about 18 y.o. It was used in a harsh environment and Malaysian weather is 80% humid. So anything and everything will go wrong.

Today I tested it connecting back everything. I Injected 60V DC at the rectifier DC side to see if there is any loading effect. It did not load anything. I am waiting for parts now to further diagnose. I have also ordered a new rectifier, might as well change it to see it. I also thought about changing capacitor but that cornell dubilier cap is about 150 usd per piece.

Regarding Soft-Start Relay, till date the relay is yet to engage, but this is likely due the control circuit have a feedback to detect the presence of 300V to turn on the Relay, the control circuit is using 4x UC3825 which I believe without feedback is at shutdown condition. The relay is turned on by a comparator where on end is a simple RC delay circuit, the other end is 5V from UC3825 Vref, which is not present since the UC3825 is in shutdown mode.

What could possibly go wrong?     Note that the DF60AA120 3 phase bridge rectifier has a surge current rating of over 900A.  Tiger, can you find a part number and datasheet for the original resistor?  Does it have a surge rating?

The resistor that was found inside is Onics AH-25W 33R, i could not find a datasheet for it.
Except a facebook site https://www.facebook.com/onics.resistors/
Looks like the resistor was made in india. Although the block diagram states 35ohm and capacitor value of 4.7mF, in the unit I found the resistor installed was 33 Ohm and capacitor installed was 6.0mF.

I have ordered both 33 ohm and 35 ohm resistor from Element14.
https://my.element14.com/vishay/rh02535r00fe02/wirewound-resistor-35-ohm-25w/dp/1369529?ost=35ohm+25w&cfm=true.

I do have a question how does neutral work in this circuit. Why is neutral tied in between both the caps, what is the purpose of this?

[Jay_Diddy_B]

Hi,

Let me explain what is happening in this design.

Here is a simplified block diagram:



The input is 415 3-phase.
415 is the line - line voltage.

The Line neutral voltage = 415 / sqrt 3 = 240V

The peak voltage is 240 x SQRT 2 = 340V

Rectifier topology

Because the neutral is connected the rectifier is a three-phase half wave rectifier. Capacitors C1 will be charged to 340V by D1-D3. C2 is charged by D4-D6 also to 340V. R1A and R1B are to limit the inrush current to around 10A.

This requires a 5-wire connection to the line, 3 phases, N and GND.

Inverter Loads

Inverters INV #1 and INV #2 are connected to C1 and INV #3 and INV #4 are connected to C2. This limits the voltage on each inverter input to 340V.

Mostly likely Fault

Because R1B is burning up, it suggests that INV #3 or #4 has a shorted input.

Jay_Diddy_B

[bdunham7]

A 35 ohm, 25W resistor has a full load current of ~1.2A @ 30V.  Bdunham7, you said you expected a peak current of 10A.  Where did you come up with that number?  I'd expect a much higher number, but I have no experience with 3 phase.  In any case, I'm not surprised that a 25W Chinese resistor wouldn't be happy with an 8X overload current spike and a, possibly, over 10X voltage overload.  What could possibly go wrong?     Note that the DF60AA120 3 phase bridge rectifier has a surge current rating of over 900A.  Tiger, can you find a part number and datasheet for the original resistor?  Does it have a surge rating?

The peak current in the resistor would be no higher than the peak voltage available over its resistance.  The P-N voltage here is ~230VAC and thus the peak voltage is about 350V.  350/35 = 10.  Unless the resistor is a total POS, any reasonable wirewound power resistor will typically tolerate peaks like that at least for a while, even though they aren't always specified for it.  Still, yeah, those are big capacitors and those resistors probably see a pretty big thump each turn on.

[bdunham7]

Mostly likely Fault

Because R1B is burning up, it suggests that INV #3 or #4 has a shorted input.

OP stated that he disconnected the inverters.  Perhaps he could clarify?

[edpalmer42]

Onics web site appears to be https://powerresistor.in/ , but there's more info at http://resistoronline.com/ .  There's no 'AH' resistors, but there are 'AHR' resistors http://resistoronline.com/pdf/new%20design%20AHR.pdf .  This states that overload spec is 5X rating for 5 sec.  The Vishay datasheet for the resistors you ordered says the same thing.  So even with those surge ratings, those resistors are too small for the job unless Lambda talked to the factory and got their signoff on the design.  But perhaps the replacement Chinese resistor wasn't rated for a 5X overload.

If the driver, or the relay itself as suggested by strawberry, failed in the soft-start circuit, I'd almost guarantee that the resistor would fail.  While you're waiting for parts, remove and test that relay.

Bdunham7, thanks for the explanation.  That was way too obvious.     What's unknown at this point is how much more of a surge would be caused by the rest of the circuitry.  It seems odd to me that they would limit the surge to 10 amps when the input breaker is 40 or 50 amps.

Jay_Diddy_B suggested that Inverter #3 or #4 could have a shorted input.  Tiger, do you know if it's possible to run each inverter by itself?  Some of those Lambda supplies are built up from seperate blocks so, for example, it might be possible to remove Inverters 2,3, and 4.  Does each inverter have it's own internal fuse?  If so, what's the value of that fuse?  Does it have it's own input capacitors?

One shocking fact about this thread that nobody has mentioned is the absolutely criminal lack of pictures.  We demand pictures!!    

[bdunham7]

Onics web site appears to be https://powerresistor.in/ , but there's more info at http://resistoronline.com/ .  There's no 'AH' resistors, but there are 'AHR' resistors http://resistoronline.com/pdf/new%20design%20AHR.pdf .  This states that overload spec is 5X rating for 5 sec.  The Vishay datasheet for the resistors you ordered says the same thing.  So even with those surge ratings, those resistors are too small for the job unless Lambda talked to the factory and got their signoff on the design.  But perhaps the replacement Chinese resistor wasn't rated for a 5X overload.

This topic has been debated before.  If you insist upon specific datasheet approval for something like a 100X (power) surge for 1ms, you'll have a much harder time finding a part.  But typical wirewound power resistors will shrug that off, and if you need proof you can characterize it yourself or consult the manufacturer.  5X (power) for 5 seconds would be similar to 25X (power) for 1 second and so on, but of course you might want a bit more margin as you go down that route.  And of course 25,000,000,000X (power) for one nanosecond might be problematic even if you weren't over the voltage limits, so there will be limits to the principle.  5X 25W x 5 seconds is 625J.  Charging that capacitor to 350V will dissipate about 280J in the resistor.  Some WW resistors could experience limited life or resistance change as a result of that, but I wouldn't expect them to blow up on the spot.