capturing voltage spike with fluke 289?

[coppercone2]

###patient stabilized, diode no longer burning with capacitor replacement and variac input###


So I have a diode that keeps blowing up and I am not sure if its over current or over voltage. I want to hook it up to a Fluke 289 to measure it. The common mode on the circuit is high and the nearby circuits are rather high voltage and it makes me nervous using any other equipment around it.

The diode is a regulator diode followed by a switching transformer that provides 3 isolated rails on a welder for IGBT control. Circuit is Clamp diode + zener diode + 100nF capacitor to power IGBT driver in isolation.

Not too sure what mode I should use to try to capture a spike.. normally I would use a scope meter but mine is totally SOL. Will doing AC peak detect or DC peak detect make more sense? The diode goes short circuit very quickly. Not really even sure if a DMM is applicable to this situation.. Anyway I want to learn if I can use this meter to trouble shoot real dodgy shit like this with greater safety. It does not need to be too accurate, its just for repair diagnostics. Weird noise and over-heat flux smoke within 2 seconds of power up and it reads dead short. I wanted to get it right first try because I am tired of resoldering a diode on every power up.

[BeBuLamar]

If the diode shorted I think you have over voltage. Make DC voltage measurement across the diode using peak min/max mode.

[coppercone2]

Ok, that was my first guess, but its good to be sure because I am sure that the less I have to deal with this the better. I though DC might have some kind of digital filter on it or something that might cause a problem.

[bdunham7]

What diode is it and what voltage does the circuit run at normally?  I would expect that the first step would be to put in higher voltage diode, and then when you've worked your way up to a 1N4007 (or whatever) then you aren't going to want to connect a DMM to a circuit that is blowing a 1kV rated diode.  So you need high voltage probe.  I don't know if the DMM is going to catch your spikes either---the F289 peak detection mode is good for 250µs.

[BeBuLamar]

Ok, that was my first guess, but its good to be sure because I am sure that the less I have to deal with this the better. I though DC might have some kind of digital filter on it or something that might cause a problem.

No it can capture a peak voltage of 250microsecond. Make sure it's peak mode and not regular Min/Max because that can only capture 100 milisecond values. You should also put the meter in manual ranging and set the range for the voltage you expect to see. Othewise it locks into the range when you switch Peak mode on and that could be a range too low to catch the peak.
If set the meter for DC voltage measurement. Go into menue and select Peak put the probes into a 120VAC outlet I would get a Max of about 170V an Average of 0 and a Min of -170V.

[coppercone2]

actually this might be necessary, it looks like when I replaced the IGBT the diodes are blowing one by one. I replaced glass diodes that were blown with ceramic diodes that were rusty and I think maybe the rusty diodes are just bad and replacing the IGBT made the diodes stop blowing but the last shady diode is still in there, in an hour when it has some time to rest ima replace the last diode with a digikey glass diode and maybe it will work. if it keeps blowing the new diodes Ima try the meter out

so basically 3 diodes were replaced, the same one kept blowing, when I replace the IGBT the other diodes start blowing, which were shitty ceramic. I replaced 2/3 and they both work, so the last old diode is (blown) in circuit, and the front panel display changed to show 0V instead of 243V. So when the third diode is replaced it will be interesting to see what happens. perhaps it will work or they will all blow up and destroy the welder but I am gonna risk it. I thought ceramic diodes would be better then glass but who the hell knows what kind of shady bs I got on ebay. I also put the 1.3W diodes instead of the 1W diodes. I don't know what was there originally because I just found a fragment of glass stuck to adhesive

[coppercone2]

you know what fuck millers bullshit power distribution 'topology' I don't even want to trouble shoot, I am gonna power those things from batteries and see what happens, fuck those multi phase bullshits or whatever they are. cascade failure waiting to happen

[coppercone2]

and my leading theory is that the poorly labeled transformer has internal leakage that is shorting the diode. I think the oscillator sounds like its running correctly and I recall once the diodes blew when it was not connected to load (I was still not sure at the time). I think the other signs of arcing on the PCB can be explained by punch through occuring at the transformer between primary and secondary. I suppose I should desolder and test it, but I hope I can disable that entire section and run it on batteries because I am sick of undoing all those lugs and bolts

but I will do one vmax test for fun, i got those dioddes on binding posts right now so its easier to replace lol

[coppercone2]

well OK, the max voltage (I set the meter for 500V range) on the PEAK mode says 30V, and it drops rapidly. Is this too much maybe? For a no feedback supply I almost think this is normal but I am not sure.

fuck it, I am just gonna put a bigger wattage diode first before more fucking around, I forgot my reasoning behind choosing the 1W diode. I think the fragments of glass I recovered brought to mind a 20V diode that I knew about but I guess 2/4 digits (really 1 digit because the last one was a little ambigous) is a shot in the dark based on manual voltage.

[BeBuLamar]

30V across the diode in the forward or reverse direction? If it's reverse direction it's not too high for most diodes. If it's in forward direction then a lot of current is going thru it. Is the bad diode shorted or open?

[coppercone2]

the diode is a regulator and it failed shorted. I am measuring directly across the regulator diode

Again it might be too fast a pulse but it also might just be the wrong diode.  I figured it was 1Watt but I am thinking that maybe IGBT driver supply needs more then 1 watt. The parts that were there disintegrated while I was trying to remove the glue and the size of various diodes is very similar. Thinking about it, the new diode i put is maybe little smaller then the old diode.

I am kind of thinking based on my IGBT driver designs that I kinda messed around with a little, 1W is a little low for a supply, and I noticed when I put 1.3W instead of 1.0W diodes, they lasted a little longer. I figure somthing this big wants like 2W at least. They had glass diodes so I don't think it goes above 3W but I bought some 5W diodes to wire in on posts if I have to incase it was some weird unobtanium diodes there. The 5W might actually fit on the PCB ok.

[bdunham7]

OK, it is a zener diode?  Then the issue is most certainly too much current, so you should measure that.  It would take an extreme transient to damage a zener, although your welder is one of the few places where you just might find one. Perhaps the current-limiting resistor or circuit has shifted in some way with age so that a marginal design has now become faulty.  There are also 5W zeners, 1N5363B--just note that higher wattage parts still generate just as much heat as smaller ones so in a confined space a higher rating won't always eliminate overheating.

[coppercone2]

It was like potted so it can probably handle more steady state current and the pcb was a little discolored but I killed out most of the sections that were bad and fixed it up. Yeah now I’m remembering the diode I got as a replacement seemed a bit small compared to the disintegrating ones. Probably should measure current, just has a trollex connector so it’s a hassle … but I suppose it’s time to get medieval on it

[coppercone2]

well I tried a 5W zener and it smoked that too.

The thing is, I don't see any resistors on the output of the transformer. Is it possible that there is a current limiting resistor on the input to the 3 winding transformer that powers the zener diodes that shorted out and its just not current limiting? I am not used to switching supplies, I assume they might current limit the whole transformer at the input rather then current limit the output of each transformer winding.

If that resistor was gone, it would smoke any diode, since its connected a ultra low impedance it looks like (I think its rectified mains on the input).

Since it smoked the little diode when the load was disconnected, and it smoked the big diode when the load was connected, I assume it is a good indicator that its not a downstream but up stream problem. Must be one of those fucking silicone coated rubber resistors. Hard to read those bastards when they put varnish on top.

time to work on 'bishop'. Yay I get to try to read a label on some gummy shit covering some burned shit that is again covered with some other kind of gummy shit. What the fuck happened to ceramics god damn, electronics turned into gumby. when is miller electric releasing their gynoids?

[coppercone2]

well nope, I desoldered those resistors and they check out OK. It seems like every fucking part near the transformer is OK. I don't know how it keeps blowing diodes,  I replaced all 3 with 5W ones that bearly fit on the circuit.

I wonder if one of the resistors that was a little charred nearby has carbon inside that is arcing over at high voltages but measuring within spec on DMM. I don't know if I should start removing parts and testing them 1 by 1 on a HV power supply or what.

Have you guys seen stuff that measured totally fine within like 1% or 5% and had a arcing problem?  

I am starting to think there is some kind of voltage spike going on that the DMM can't capture. I just don't know what will happen if I ground one of the isolated IGBT driver supplies with a oscilloscope.. I am almost thinking about just getting some more shit for acetylene and forgetting this whole TIG welding thing because this seems way too fucking fragile. I suppose I can desolder some parts and test with HV but I am seriously losing hope here.

Could there be something wrong with the transformer? I read that the cores can crack or something like this, it looks fine cosmetically and the windings seem fine.

[coppercone2]

And BTW the middle zener diode fried this time, I replaced all 3 1W diodes with 5W diodes. Its random which diode breaks. There are two resistors infront of the transformer that are within band codes, and some beefy diodes that all measure OK. I think this has something to do with the 800V and soot or whatever.. I really feel like I would have found a faulty part by now, from experience. This feels like a new problem .

When I hooked up the PCB that its connected to without IGBT it measured about 3mA consumption per 20V rail. The ebay pics of the board show little diodes, I think they were infact 1W diodes originally. I will desolder some power 150Khm resistors that meausred fine and measure the PCB with a HV power supply, in an area that I cleaned up from soot using rubberized abrasives. Maybe its not clean for the HV found in the instrument, but I will say on some other power electronics I worked on, mildly scraping the soot off with a pick (where it still looked shitty) was enough to break the carbon short, I feel like my job was over kill already but who knows.

[bdunham7]

Maybe take the time to sketch out the circuit as best you can and post it so we can have a look. 

[coppercone2]

Ok, the plastic film under the PCB has two holes that are blown right through it (fairly thick plastic sheet) under where the diode is. I wonder what voltage level it took to make those holes (and crack the input rectifier brick).

I think there is some kind of problem with the substrate to be honest. I see similar pictures online near the 150K resistors being burnt.

If it punched 2 holes in the plastic film under the welder I assume that might be enough to fuck up the substrate?

[coppercone2]

ok reverse engineering is gonna take a while because it means i need to remove the backplane board and undo a bunch of lug connectors and shit. If i try that right now I will end up smashing it into the wall, the sheer cost of the object is the only thing keeping it alive right now. There is some serious bullshit going on here because there are like no success stories about this welding machine, they replaced the PCB with a different version too, there is probobly some insidious shit going on. I want to smash it with a sledge lol

I will try to see where I can disconnect that whole section and plug it into 3x 20V linear supplies. I am not working on it any more until I can prove its not two problems interacting with each other on different circit boards

[bdunham7]

If it punched 2 holes in the plastic film under the welder I assume that might be enough to fuck up the substrate?

 

I said:

It would take an extreme transient to damage a zener,

And I think that might just qualify as 'extreme'...

[coppercone2]

I have an idea too, to use a LCR meter to measure the resistance from the last transistor in the circuit to the input of the transformer to see what kind of impedance that has. Maybe it will find lack of impedance. But I really think it might be some unwanted path that developed from HV stuff. I think maybe those big boy capacitors have a path to the diodes some how

I don't know how to find it besides depopulating the whole board and testing it with like 1kV to find leakage

[BeBuLamar]

You are talking smoke then it's not overvoltage but rather over currennt. You said the diode shorted buy if it smokes it would be open afterward not shorted. If you have 30V reverse bias it's no problem because diodes are rated 50V at least. However, if you measured 30V in the forward bias direction that's a lot of current is going thru the diode as normally you will have only about 0.7V in the forward bias direction.

[coppercone2]

The zener diode is reversed biased I think, if that is the notation used to describe a diode making a positive voltage rail. So the voltage peak max on the 289 DMM is 30V and it drops down to 20V. I noticed this because the diode I was measuring was NOT the one that burned up at the time.

So primary : 3x secondary with 1x diode on each. At start up one of the diodes will at random burn up and then the other two sit at 20V while the big cylinder inductor on the corner of the PCB makes a buzzing noise. I think that the shorted diode on the secondary acts as a current limiter because of the big load dropping the primary voltage or something lol?.

When the test is done one of the diodes is always reading a dead short, as soon as I clip it from the circuit the MUR120 reads the normal voltage drop again (the circuit has a MUR120 and a 20V zener together with a 100nF capacitor. The capacitor and MUR120 are fine. The diode remains short the month after I clip it off from PCB (I have a pile of them now).

On the primary side of the transformer there is a 1500Ohm resistor and some other resistor that reads in specification. Upstream there is a IGBT or FET with a diode that I also replaced. Upstream of that is a IGBT or SCR (bigger package) that is next to 4x 150Kohm resistors that have burn marks on the PCB but measure fine when desoldered. I cleaned up this area using abrasive when they were all desoldered.

I have not powered up the PCB without either shorted or fresh diodes because I am worried it will destroy the IGBT control board attached by the cable harness.

There was some browning from the big capacitors on the varnish that was peeled off.. dmm does not detect short to any adjacent path.

Each of the loads for the 20V rails was measured independantly with a power supply and they draw less then 10mA, but the IGBT is not connected to the board. I replaced the big IGBT and checked all the big power diodes nearby and there are no problems there. At one time I forgot to connect the load cable to the PCB with fresh diodes and the diodes also burned up with no load connected.

Since all the resistors nearby read fine, I think the only explanation is that there is some leakage going into the circuit? I am going to find the output of the last driver thing on the PCB and see if I can measure AC impedance to the input of the transformer through the whole circuit path... if I know the resistance then I can measure the AC voltage on the primary with a scope that is properly connected on power up and see if there is excessive voltage there considering the impedance is known, rather then trying to measure the isolated secondary, and make the assumption that the diodes should be running at 1-3 W (since the 5W diode is clearly much bigger then what was there originally). I can use the HV scope probe for this.

[coppercone2]

Hmm there is a LEM HY-25P on the PCB, looks like a current sensor. I wonder if this is bad can it have some monster inrush or something that takes out the circuits some how

Could this be a inrush control relay issue? I wonder if they use that technology, there is a relay here for some reason on the main board

I wonder if it could be that the filter caps are too much a load because of low ESR that they pull too much current because some kind of soft start routine is being ignored. I mean you figure that would be done locally but maybe its causing some sort of problem. I held the board under light for a while and I just don't see any thing suspicious

I wonder what would happen if i just add series resistance to the transformer primary lol

What I am saying is that maybe the design is not short circuit proof so they make slow start to protect it and when that goes out there are problems. Maybe its to keep heat down or efficency up?

I could remove the filter caps and see what happens. Or its possible one of the cards has a leakage issue that is frying it from the other end. the two cards are a probelm. But the 20V rails are totally isolated, so I don't think its the cards because I suspect stuff would be blazing hot/smoking on them if that were the case. I wonder what will happen if I disconnect the load and desolder the 20V isolated filter caps that are possibly a dead short on fast start up.....

[coppercone2]

well I started to trace out the transformer primary, it looks like there is no series resistor anywhere.

It goes from semikron igbt brick thing to 35V capacitor to heat sinked diode to 1mH inductor that goes to one side of the transformer

the transformer has a 10nF foil across it along with a flyback diode

the other side goes to a IRF640 middle pin (so its like a low side driver? , makes sense if capacitor is on high side), then the source of that goes to 2 0.5 ohm resistors that are in spec, which are in parallel to provide 0.25ohms of ESR and that goes to other semikron igbt pin. the middle pin is also connected to some transistor by 7.5 kOhm.

the coil resistance is 0.219 Ohm, the resistors are 0.25Ohm, and it has all the traces, and the diode is in series with it so it does some voltage drop.

So the ESR of the switched loop is 0.5 ohm + trace + diode + IGBT.

and the transformer is really low value, with kelvin I measure 12 miliohms (not sure if this makes sense, but i guess its high frequency)

so what does the 0.5 ohm esr (seems to be by design) and 12 miliohm transformer winding say? i get 50 amps. ??

im not exactly sure whats wrong but i imagine if this is being chopped at the wrong frequency and it depends on that inductor for resistance maybe something really bad is happening? i am gonna draw it on paper later to see if i missed some other path or something lol

actually the transformer is connected to common on one side (on a pcb) through the inductor 1mH and the series resistance is 0.25 ohm + igbt + transformer + inductor and the diode is like recycling? or whatever, and there are some passives branching off through it. (diode and the non common side are connected directly to SKiiP module).