Repair of Yaskawa Variable Frequency Drive

[jamesmi]

Hello,

I am attempting to repair my first frequency drive and found it has a bad output transistor module. This particular drive has 3 IGBT modules, one for each phase. I replaced the bad module and every thing tested fine until I hooked up a motor and blew another module. Motor was brand new never used and several amps below the rated current of the drive.

Anyway, my question is wheter or not it is normal practice to change all 3 modules if one is bad.

Thanks, James

 

[spuffock]

for what it's worth, I find that using lamps as loads when testing or repairing high power electronics can reveal further faults with small risk of a loud bang.

 

[Styx]

replacing all three modules?

It depends. If you think abt how a 3ph inverter works 2 phase legs (ie two modules in your case) are switching. From what you are saying 1st one module went then u replaced and turned on 2nd went.

What I think happened is you had an overcurrent on one phase killing an upper and lower IGBT in two modules. It comutates and turnes on the lower/upper of that damaged module producing a shoot-through (blowing module due to high energy switching).

you replace that module but the other module still has a damaged IGBT. You start to switch and you cause a shoot-through on that other module.


The best way to stop this is if you have a module fail test all the devices in the other module. IE take a working module and measure the capacitance of the gate-emitter (so you know what it is) also the diode volt drop of the anti-parallel diode (again to compare)

When drive blows remove known damaged module and remove gate leads to others. Test Cge, Vdiode and Ohm Collector-emitter (to test for IGBT blocking)

 

[John Sorensen]

I used to do VFD service for a "major US manufacturer," here are questions I would ask myself: how much do the parts cost? who is paying for them? how important is this VFD? how far do I have to drive to come out here again? If the answers are: about that much; someone else; very; and/or a long way; I'd be replacing everything back to the main logic board. Send the questionable parts back to the factory and let them test them.

*Something* caused those parts to pop, spewing transistor guts all over the heatsink. A failure could also stress the remaining components. Don't forget to factor in down-time.

j.

 

[jamesmi]

wow, thanks for the quick replies.

John Sorensen:
"...about that much"
That was one of the funniest things I've read in a while. Although quite true. It is not my money, I do know how much it costs and although down time is an issue, albeit not major, we try to save money where we can. If it were necessary to change all three modules than so be it and everyone would be happy. I guess it's probably just me who really wants to know if it is necessary or not. I just hate throwing money away mine or not. My boss would probably say screw it and replace them all. I have them on order and will probably just change them all.


Styx:
I had measured the diode drop of the three modules before replacing the first module, I had not measured the gate-emitter capacitance. I found the upper IGBT in the one module to be bad, the lower IGBT tested fine. This is the module I replaced.

After reading your message I checked the Cge for the originally failed module and found that the upper IGBT had 0 capacitance and the lower had the same capacitance as the new module (10nF). Unfortunately I did not check Cge for the second failed module before I powered it up.

I have now checked the second failed module Cge and found to be the same as the first, the upper IGBT appears to be bad and the lower appears to be fine.

I have also checked the remaining module Vdiode and Cge and is almost identical to the brand new module.

Oh, I should also point out that the modules did not fragment, and appear fine cosmetically. The reason the drive was replaced was because of a short circuit fault even with the motor disconnected. This is when I traced the short to the failed module.

I guess my question now would be since the remaining module tested fine should I bother replacing if money is an object? Could it be bad although it appears to be fine. Time is no issue here and I have ordered two more modules just in case. We are fairly small shop and are trying to save money were we can. I was just wondering if it was common practice for a drive repair shop to replace all three modules.


The reason I asked the question in the first place was because we normally send these drives out to be repaired, and this particular drive cost about $2000 to get repaired. The individual modules cost 400 a piece and replacing all 3 would still only run 1200. The problem is were not really satisfied with the repair shops we have been using. To give an example why, this is the first drive I have tried to repair and after taking the drive apart I found the braking transistor emitter wire was not even connected. It was left uninsulated dangling behind one of the control boards. I tested the braking transistor and it is shorted collector to emitter. It seems this would surely have blown the dc-bus fuse had it been connected. This is probably the reason the drive was sent out for repair in the first place. We have the happen fairly frequenty(dc-bus fuse blowing). So did we pay 2000 to have someone disconnect the emitter wire and send it back as fixed?


I didn't mean for this post to be quite so long, but I do appreciate all the help.

James

 

[Styx]

Well my method is the method I use (Cge, Vdiode, Ohms-ce).
I do it because it tells alot abt how the module failed and how to resolve it.
Also some of my drives use dual 6-packs of IGBT's.

The cost is not the driving factor for just replacing both modules, it is that blooming heatsink compound that seems to get everywhere!!!!

What might have happened is the gate driver got damaged and was no longer able to provide negative gate-volts thus a device turned on by the millar cap when the DC-link rose up. or it's rails were low thus not providing enough gate-volts to saturate the IGBT's.

Always check the gateboard outputs after a blow-up

 

[Styx]

oh another thing also that has saved me alot of time.
If you have just replaced a module (so there exists the possiblity that you wired gate-emitter wrong or other damage)

Run the DC link off a separate power-supply with a current limit. I use a Xantrex variable 300V DC power supply with variable current limit up to 40A. Into a short 10A limit will do no damage (unless the module can only take 5A :wink. Great way to test a fresh build

 

[jamesmi]

Well I replaced both the transistors that were failed in this particular drive powered up and ran 5hp motor just fine. I hooked up 25 HP motor and ran for about 7 hours with no problems, but there was no load. Just today I found out the drive that was put in to replace this one has failed also. So the drive I repaired is in service and working just fine (for now anyway).

I do have another question however. I took apart the newly failed drive to find one of the two PCB's is charred. It appears as if something leaked into the top of the drive and shorted out some of the components. This is the gate-drive pcb I believe. The control pcb appears visually fine, but I have no way to test it as of now. Is it possible to get replacement boards for these drives from anywhere? I know the few times we had boards fail in drives the repair shop we normally use told us that it would be cheaper to replace the entire drive so we did. It does seem a shame to replace a $4000 drive for just one bad board.

I was also wondering if anyone could give any other sources for replacement modules and such for yaskawa drives. We also have some allen bradley drives, but mainly yaskawa. The only place I have found that sold the one transistor module I needed for the first drive was GALCO.

Drive Specs

Yaskawa Model #CIMR-G5A4015
Output 26kVA 34Amps
spec 40151A
charred board is "2PCB"

Thanks again for the help.
James

 

[Doug]

what is the part numbers on the igbt?

 

[dano]

I use a repair service that fixes (among others) Yaskawa drives. The usual cost of repair is about 50% of new and a one year warrranty is given on the whole unit, not just the repaired part of the drive.

 

[jamesmi]

The module out of the failed drive was a fuji 2MBI75N-120, but I wasn't really looking for a supplier for that specific module. I was looking more suppliers that offer parts to fix frequency drives in general. We use several different size drives and multiple suppliers for parts would be nice. The only one I found so far that stocks that specific module was GALCO. Any other suggestions for suppliers would be great.

The drive I'm currently looking into repairing needs a new circuit board. I'm not sure where I can get one of these or if I even can get one. Is it possible to get replacement circuit boards from someone other than Yaskawa?

We normally send all our drives out for repair with mixed results. The one size drive we use in four different places fails quite frequently. Although the repair shop we use also warranties the drives for 1 year, they say it is a motor problem and won't fix it for free. My boss would rather we repair all drives ourselves from now on if possible. I was able to repair one of our drives for $800 and would have cost about $2000 to send it out. I was also able to repair it much faster. We sometimes don't get drives back for 2-3 weeks after sending them out. I was able to fix it in less than one week.

James

 

[Doug]

As far as finding parts try just using google search with the part number.
http://www.hoffind.com/index.htm has some surplus drivers.
On repeat failing controllers you might check with the manufacturer of the output device( this case Fuji) see if they have any recommendations.
Also may check with the motor manufacturer. They may recommend using a snubber circuit ( do a google search on snubber circuit).
Maybe the unit has an engineering change for your problem.They most likely won't tell you what it is however if they state there is a modification or change take many pictures of the circuit board(s) before sending it in. Then after the return recheck and see what they have changed.
I repair laboratory and test equipment for an independent company. 95% of the time the manufacturers wont release information or schematics. There reason always the same proprietary information.
I do a lot of reverse engineering on making prints. There are companies that make schematics and resale schematics.
There are very few forums in this area.
Yahoo groups has some for AB. and instrumentation.
http://www.eng-tips.com/gthreadminder.cfm has various engineering help such as motor control http://www.eng-tips.com/gthreadminder.cfm/lev2/11/lev3/47/pid/237.
You are correct most manufacturers want 1/2 the cost of the unit to repair.
I don't always view forums daily sometimes not weekly. If you want to contact me you can send me an e-mail.
dmac1054@graphtronics.net
Doug