ABB ACS800 PRE-CHARGE RESISTOR FAIL 1

[gobigorgohome]

Has anyone else experienced pre-charge resistor fail in ABB ACS800 VFD's. Our particular model is 400 A 575 Volt. The pre-charge resistor is 3.3 ohms 65 watts and is always left in the circuit, under the assumption that the half wave rectifier diode that is in series with it will not be forward biased when the switched rectifiers (scr's I think) are on. Unless that is the control board does something unpredictable and turns off the front end rectifiers when there is DC load on the DC bus and then cooks the resistor.

Of course, we are the "only" users to have experienced this failure 4 times now.

Thanks for reading.

 

[Skogsgurra]

Have had that problem in other inverters. But not (yet) in an ACS800.

Did anything unusual happen to the grid at the same time? Any recordings? And, yes, "that has never happened to anyone else..." should not be used to tell the customer "...and probably not on your site either"


If it has happened, the supplier should be eager to learn what, how, why. And not try to blame something else than the design. I am really getting ired with the attitude that ABB has developed over the last years. They were not like that ten years ago.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[sibeen]

Guunar, surely you're on board with the new and improved management practices that most major manufacturing companies now adhere to.

About 12 years ago, whilst I was working for one of the major UPS manufacturers, an edict came down from upon high that the company was to no longer issue Field Service Bulletins (FSB) as they made the company look bad. It was the CEO who made the call, a bloke who was of course from an accounting background.

Of course it caused no end of grief for years to come, but at least the company didn't look bad.

 

[ScottyUK]

I wonder how much repeat business your former employer has lost as a result of that decision? It might not show for 10-15 years but I have a long memory for companies who don't support products we buy. I have an equally good memory for those that do support us.

 

[sibeen]

Scotty, cough Chloride cough. They are now owned by Emerson, and from I hear the customer is treated even worse.

 

[Skogsgurra]

sibeen: "but at least the company didn't look bad"
For how long?


Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[gobigorgohome]

Heh he, yeah, I hear ya. So far though, I haven't found evidence of other users with the problem. You know, you start typing your problem in the search box and it autofills with your issue.

It's in a BIG corporate's BIG plant, and there is some real consideration going on to "find" the problem. We already have fluke 1750's on the AC sides, in multiple drives. We are trying to decide if we want to start monitoring the DC side but that is an expensive task since were looking at 800VDC and need all sorts of exotic measuring instruments to cope with that.

The 1750's are real good at AC systems, but are not very helpful on the DC side. We have a YOKOGAWA DL850 that has super recording capability, but the isolated differential probes that I would like to use are 2K a pop plus 6K per pair of input cards. ugh. And, since you can't predict the next failure location, we would be best to monitor multiple units, cause you can be sure murphy will blow up the drive right next to the one that has all the test equipment on it.

 

[Skogsgurra]

Sorry that I am so far away. I would fix it in a jiffy.

But I can give you a few tips:

1. Use an ordinary recorder with a decent band-width. You do not need to trig on the mains voltage, so any recorder that samples at least at 1 kSa/s will do.
2. Measure grid voltage with voltage dividers and fused test leads against ground. You probably already have fused test leads. So that cost is zilch. If you do not hae them, you should.
3. Measure DC link voltage the same way. But show DC+ minus DC- (differential measurement).
4. Trig on any convenient signal. Like the "general failure", DC link low or whatever.
5. Start logging and wait for next failure.
Note: if you have the drives lose to each other, you can connect as many DC links as the recorder allows. The grid only needs to be supervised once.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Skogsgurra]

Oops - "Note: if you have the drives lose to each other"

shall read "Note: if you have the drives close to each other"

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[mikekilroy]

Might be useful to look at actual temperature of that little resistor too. Is it possible it is getting a little voltage all the time and is near max temp normally? That it just takes a small 'event' to cook it? If warm, perhaps scoping voltage across that diode might tell why it is being turned on?

http://www.KilroyWasHere<dot>com

 

[gobigorgohome]

I work at an electrical testing and engineering company, and we can and will find out why, unfortunately (at least for the end customer) it takes time and money. Often more than the equipment costs outright to replace.

 

[gobigorgohome]

Oh, didn't catch your post before I reposted. So that may not make a ton of sense.

What is your choice for voltage dividers? We have used some Northstar VD Series (VD60's) on some medium voltage applications, but they are big, bulky, hazardous to touch potentials. I considered using sensing resistors from High Resistance Grounding applications, but they are not continuously rated. I also considered HV resistor components that you can buy individually, but I'm certain the safety guy won't let me in the plant gate with those.

As for resistors running warm, the third party information that I have says no. But I can't verify that.

 

[Skogsgurra]

800 V is NOT High Voltage.

Those security guys have no idea about real threats and go mostly by paragraphs. They have hindered real work to be done so many times that I avoid talking to them. Security Theatre.

Standard resistors (talking small .4 W metal film) have 250 rating). Use ten of them and you have a voltage divider that easily (and legally) withstands 2.5 kV.

OK, I do not use such resistors, although you can find them in many VFDs (to measure DC link voltage!). I use 2 W resistors that are rated a lot higher and have used them when measuring on Dynos at the Scania main plant in Södertälje as well as on Alstom locomotives (I chose a few of the better known customers, but there are lots and lots of others)with no objection or danger to humans or Equipment.

The only problem with resistive dividers is that you lose high speed step response, which can be a problem when you look at PWM edges and the ringing. You need something like 30 ns step response then. But that is also possible if you make the foot resistor 50 ohms and use the scope's 50 ohm internal impedance. But we are not talking high speed here - only milliseconds or fractions of it.

I also use a home-brew recorder that is so cheap that I can leave as many as needed on a plant for weeks. One has to be practical - not only P. C. and better safe than sorry. I have worked in the field for around 50 years and never caused an accident of any kind. And, mind you, I ain't dead either.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[gobigorgohome]

Okay, now you got me curious, I have seen some PC's running Linux as data recorders, but they needed quite a bit of additional hardware for A/D conversion.

I have picked up some arduino boards and SD card shields to take a crack at a home brew version just for sh*** and giggles, but never really considered using anything like that for a megabucks customer.

 

[Skogsgurra]

Arduino is* too slow for our use. We use a custom designed board with a fast PIC with a built-in 10 channel, 12 bit A/D and programmed in assembler. I've done work for ABB with it. And for Siemens and other large companies. To honor ABB's ARGUS (or rather blink at them), I baptized it almost the same.

*Or was, there are new boards out now. Haven't checked if they are good enough. We sample 50 000 Sa/s so we get 5 kSa/s on each channel.

Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.