One of our plants has numerous 100HP and 300HP ABB VFD's (both ACS600 and 800 Series). Our plant power is 460/277V. The drives operate off the 460V line. They run cooling gas blowers for our batch processing equipment. These cooling blowers normally run 1 to 2 hours followed by 6 hours off. We have been experiencing violent VFD failures randomly for about 2 years. Some VFD's are running when they fail. Others have failed while they were waiting for the next run period. Most, if not all of the failures seem to be associated with the VFD DC Bus. The last few that I have personally seen evidently arced through air from the positive to the negative bus leg. The VFD front end MOV's, and the output IGBT's are usually not affected. We have had the power company and an independent power quality engineer monitor our plant power at the incoming switchgear as well as at individual VFD locations for weeks at a time. They have seen nothing abnormal on their instruments even when a VFD has failed during the monitoring period. All grounding has been checked and verified to be good. We have replaced the normal VFD cabinet air filters with high efficiency filters than filter down to about a micron, and change them regularly. Has anyone ever heard of this type of problem before?
Eng-Tips is the largest engineering community on the Internet
Intelligent Work Forums for Engineering Professionals
-
Congratulations waross on being selected by the Eng-Tips community for having the most helpful posts in the forums last week. Way to Go!
Multiple ABB AC Drive Failures 1
- Thread starter 45bob
- Start date
Skogsgurra
Electrical
Yes. Fans can actually be driven by air that goes up a chimney due to thermal lift or any other reason - like pressure differences in the plant. If the fan is running faster than the setpoint, it will regenerate and take DC bus voltage above normal levels. I don't know why the capacitors don't explode - perhaps they do in some cases, but flash-overs occur, definitely.
The reason can also be that your deceleration ramp is much faster than the fan can run down if not braked. Try to make the ramp a lot longer or connect braking resistors.
The fact that drives seem to have failed while not running is a complication. Do not understand that one.
Also, there have been cases where high capacitance resonates with the motors which results in high voltages. But that usually requires PFC capacitors connected to the motor and I cannot imagine that you have that - large filter capacitors perhaps, but no PFC capacitors.
I would check these:
Ramp. Make it as long as is practical.
Wind driven? Watch the fans and see if they ever run above setpoint. Or if they run when not started.
PFC or filter capacitors on motor side?
Does a brake resistor help?
Gunnar Englund
The reason can also be that your deceleration ramp is much faster than the fan can run down if not braked. Try to make the ramp a lot longer or connect braking resistors.
The fact that drives seem to have failed while not running is a complication. Do not understand that one.
Also, there have been cases where high capacitance resonates with the motors which results in high voltages. But that usually requires PFC capacitors connected to the motor and I cannot imagine that you have that - large filter capacitors perhaps, but no PFC capacitors.
I would check these:
Ramp. Make it as long as is practical.
Wind driven? Watch the fans and see if they ever run above setpoint. Or if they run when not started.
PFC or filter capacitors on motor side?
Does a brake resistor help?
Gunnar Englund
-
1
- #3
What product is being batch dried?
I have seen similar things occur when particulate matter builds up. Even if not normally conductive, humidity can bring on enough conductivity to cause a tiny current that rapidly (read instantly) drops resistance so the mains can bridge the path, then KAPOW!. It seems to happen in DC situations a little more frequently.
So check for dust/powder/moisture.
I have seen similar things occur when particulate matter builds up. Even if not normally conductive, humidity can bring on enough conductivity to cause a tiny current that rapidly (read instantly) drops resistance so the mains can bridge the path, then KAPOW!. It seems to happen in DC situations a little more frequently.
So check for dust/powder/moisture.
Check to see if the ABB drives have a motor winding heater or motor pre-fluxing feature that is turned on. Both of these features apply a low amount of current into the motor windings when "ofF" in order to either keep the windings warm from the inside out to prevent condensation, or to have a more instant-on capability. If they are turned on, AND you have a windmilling effect, you may end up with regeneration even when you think the VFD is off. I'm not sure if ABB supplies these features, but others do.
"Our virtues and our failings are inseparable, like force and matter. When they separate, man is no more."
Nikola Tesla
"Our virtues and our failings are inseparable, like force and matter. When they separate, man is no more."
Nikola Tesla
- Thread starter
- #6
The environment that the VFD's operate in does have a certain amount of graphite dust in the air some of the time. It is not excessive and it is not enough that you can even see it in the air. But, you can see a very light coating on the floor at times. We have concentrated on keeping this cleaned up as good as possible. Unfortunately, the ABB design requires cooling air to be pulled into the entire Drive Module, and it passes over all the live internal components, as well as the IGBT heat sinks. As I have indicated, we have put high efficiency filters on the VFD enclosures because we were getting some graphite in them in the beginning. We felt that with the new filters, that we had eliminated graphite dust as a possible cause. FYI, some of these VFD's have failed within a week or so of installation and look virtually brand new inside.
Certain other manufacturers have their heat sinks extending out the back of the VFD enclosure with a cooling blower there. Outside air does not get into the electronics portion of the cabinet. We've begun to buy this sealed type of VFD for all new projects.
Certain other manufacturers have their heat sinks extending out the back of the VFD enclosure with a cooling blower there. Outside air does not get into the electronics portion of the cabinet. We've begun to buy this sealed type of VFD for all new projects.
You might consider putting the VFD in a NEMA box and use an air-to-air heat exchanger if graphite is thought to be the problem. Kooltronics is one supplier.
Well dang! If you asked me what the worst dusts would be I would say 1) Metal dust; 2) Graphite dust....
One of my experiences was the inside of refrigerated rail car control panels. Sealed, closed, latched. I place a control board inside the panel, unfortunately it was horizontal. The metallic dust given off by the compressor and fan contactors slowly built up on my board until.. well never mind. It was ugly. My psychiatrist sez I should keep it bottled up.
The dust problem does align with units that have energized busses and are on standby doing nothing. Especially if the fans are not windmilling as mentioned by skogs.
One of my experiences was the inside of refrigerated rail car control panels. Sealed, closed, latched. I place a control board inside the panel, unfortunately it was horizontal. The metallic dust given off by the compressor and fan contactors slowly built up on my board until.. well never mind. It was ugly. My psychiatrist sez I should keep it bottled up.
The dust problem does align with units that have energized busses and are on standby doing nothing. Especially if the fans are not windmilling as mentioned by skogs.
Drivesrock
Electrical
Sounds to me like something from the mains. Is the supply neutral earthed? Do you have input line reactors? (If not, then you should). Do you have any large (ish) DOL loads on the same supply or power factor correction equipment being switched?
- Thread starter
- #11
Yes, the supply neutral is solidly grounded. The VFD's are all equipped with 3% line reactors. No power factor correction equipment in the plant. Remember, when we have had power monitoring equipment online, we saw no line disturbances at the time of the VFD failure. Furthermore, most times, none of the front end components are affected (MOV's, Diodes, or SCR's). We don't see any evidence (at the moment) that it comes in through the line.
- Thread starter
- #12
Here's a plan. Put a DC voltage monitor on the DC bus. Watch particularly for voltage variations when the drive is powered up but not running the motor.
I've been thru this kind of thing a number of times and each time the problem was, in fact, the incoming power quality but the recording equipment is not fast enough to catch the short transients.
I suspect you will find that the DC bus voltage jumps up occasionally and that is only due to incoming power. If it jumps up high enough--well you know, it goes bang.
As an ABB rep, I find the 600 and 800 series drives to be very rugged. When a situation like yours comes up, something powerful is the cause. Graphite powder is always a problem but you seem to have taken effective steps to clean up the air. I'd bet on power problems.
I've been thru this kind of thing a number of times and each time the problem was, in fact, the incoming power quality but the recording equipment is not fast enough to catch the short transients.
I suspect you will find that the DC bus voltage jumps up occasionally and that is only due to incoming power. If it jumps up high enough--well you know, it goes bang.
As an ABB rep, I find the 600 and 800 series drives to be very rugged. When a situation like yours comes up, something powerful is the cause. Graphite powder is always a problem but you seem to have taken effective steps to clean up the air. I'd bet on power problems.
The most straight forward way to check/eliminate voltage spikes as the source of the failure is to connect some Transcient Voltage Suppressors (TVS) across all three phases of the AC power entry to the VFD. I would suggest two 1.5KE440C (bidirectional) in series across phases (six diodes total). This is really cheap [$1 US ea.].
They'll blow if there are spikes with enough energy or stop the damaging spikes if they are short. Or VFD's will continue to pop if voltage spikes are not the problem.
They'll blow if there are spikes with enough energy or stop the damaging spikes if they are short. Or VFD's will continue to pop if voltage spikes are not the problem.
Hello 45bob
Just thought that I would add my 2 cents worth to the good advice that you have recieved above.
I would doubt that the problem is due to mains bourne tarnsients etc as this is more likely to cause component failure than a breakdown of an air insulation gap. The breakdown voltage of the rectifiers will be well below the breakdown voltage of the air between the terminations.
If the problem was due to mains bourne transients, I would expect to see failurs of the input rectifiers and MOVs. If the problem was due to regeneration, I would expect to see problems with the output IGBTs.
To get a breakdown in the air insulation, generally means that there is some form of contamination or ionization. This can be due to foreign matter or insects etc getting where they should not.
I suspec that you are getting a very fine build up of graphite on some of the insulators. When there is sufficient graphite, a small conductive path forms and there is a small discharge across the graphite path. This of course blow the graphite path clear and it may be difficult to find evidnece of it, but it creates ionized air and this will allow the insulating air to breakdown at a much lower voltage and cause the arc between the DC terminals.
I like the idea of using a swab on some of the insulating surfaces to see if there is any sign of grahite there (before a breakdown!) You may need to coat all exposed terminals with an insulation layer, or put physical insulation barriers in place to reduce the problem The best solution is of course to prevent the contamination in the first place. - I would also check the level of contamination on the pcbs as this may be the next point to blow.
Best regards,
Mark Empson
Just thought that I would add my 2 cents worth to the good advice that you have recieved above.
I would doubt that the problem is due to mains bourne tarnsients etc as this is more likely to cause component failure than a breakdown of an air insulation gap. The breakdown voltage of the rectifiers will be well below the breakdown voltage of the air between the terminations.
If the problem was due to mains bourne transients, I would expect to see failurs of the input rectifiers and MOVs. If the problem was due to regeneration, I would expect to see problems with the output IGBTs.
To get a breakdown in the air insulation, generally means that there is some form of contamination or ionization. This can be due to foreign matter or insects etc getting where they should not.
I suspec that you are getting a very fine build up of graphite on some of the insulators. When there is sufficient graphite, a small conductive path forms and there is a small discharge across the graphite path. This of course blow the graphite path clear and it may be difficult to find evidnece of it, but it creates ionized air and this will allow the insulating air to breakdown at a much lower voltage and cause the arc between the DC terminals.
I like the idea of using a swab on some of the insulating surfaces to see if there is any sign of grahite there (before a breakdown!) You may need to coat all exposed terminals with an insulation layer, or put physical insulation barriers in place to reduce the problem The best solution is of course to prevent the contamination in the first place. - I would also check the level of contamination on the pcbs as this may be the next point to blow.
Best regards,
Mark Empson
- Thread starter
- #18
First, let me say thanks to all who have taken the time and effort to help me with my problem. I am very impressed with the quality and quantity of advice on this website.
I am only one of several engineers at my company who are trying to solve the Drive failure dilema. Before I posted my question, I had my theory as to the source problem. Now after hearing from all of you, I would like to tell you all the conclusion that I have come to and why I have come to it.
I once saw a Sherlock Holmes (Basil Rathbone) movie in which Holmes is explaining to Dr. Watson how it was that he was able to solve the mystery. He explained to Watson that once you have eliminated all the things that it cannot be, that whatever you are left with, however seemily unlikely, has to be the answer.
In looking at the failures, I asked myself, what are the things that could cause current to bridge a large gap? The only two things that I could think of were, high enough voltage, or ... a conductive path.
Next, I thought, where could a high voltage come from? Only one of two places. From the line, where it should be blowing the heck out the MOV's and rectifiers (and other equipment being fed from the same source); or regenerative voltage from the motor, but that's easily ruled out from the cases where the motor isn't even running.
So, what am I left with? A conductive path! Even though we have taken some substantial steps (high efficiency filters) to try to keep the Drives contamination free, there has been some evidence that some "dirt" still gets through.
By eliminating what I conclude that it cannot be (voltage), I believe that it HAS to be contamination. Possibly very fine graphite dust that gets pulled into the air intake grilles (located at the bottom of the Drive door), which either slowly deposits itself on the Drive components, or which occasionaly is conductive enough to be the path directly through air.
That's my "Sherlock Holmes theory". If anyone thinks that my logic is flawed, please point it out to me.
Once again, allow me thanks all of you... Skogsqurra, Itsmoked, Jraef, Sreid, Drivesrock, DickDV, and Marke for your thoughtful and solid technical advice.
I am only one of several engineers at my company who are trying to solve the Drive failure dilema. Before I posted my question, I had my theory as to the source problem. Now after hearing from all of you, I would like to tell you all the conclusion that I have come to and why I have come to it.
I once saw a Sherlock Holmes (Basil Rathbone) movie in which Holmes is explaining to Dr. Watson how it was that he was able to solve the mystery. He explained to Watson that once you have eliminated all the things that it cannot be, that whatever you are left with, however seemily unlikely, has to be the answer.
In looking at the failures, I asked myself, what are the things that could cause current to bridge a large gap? The only two things that I could think of were, high enough voltage, or ... a conductive path.
Next, I thought, where could a high voltage come from? Only one of two places. From the line, where it should be blowing the heck out the MOV's and rectifiers (and other equipment being fed from the same source); or regenerative voltage from the motor, but that's easily ruled out from the cases where the motor isn't even running.
So, what am I left with? A conductive path! Even though we have taken some substantial steps (high efficiency filters) to try to keep the Drives contamination free, there has been some evidence that some "dirt" still gets through.
By eliminating what I conclude that it cannot be (voltage), I believe that it HAS to be contamination. Possibly very fine graphite dust that gets pulled into the air intake grilles (located at the bottom of the Drive door), which either slowly deposits itself on the Drive components, or which occasionaly is conductive enough to be the path directly through air.
That's my "Sherlock Holmes theory". If anyone thinks that my logic is flawed, please point it out to me.
Once again, allow me thanks all of you... Skogsqurra, Itsmoked, Jraef, Sreid, Drivesrock, DickDV, and Marke for your thoughtful and solid technical advice.
Skogsgurra
Electrical
That mr Holmes analogy is very good. I used it tonight when solving a problem in a paper mill (details below).
But. You may have left out some facts (like train leaving from Waterloo station at that time, not Victoria station - to paraphrase mr Holmes). One such fact is that there are many other possible paths where graphite dust can and does accumulate. Paths that probably accumulate more dust faster and that also have similarly high voltage stress - not DC but still high peak voltages. And therefore also should have flash-overs.
Another fact that is easy to forget is that IGBTs and capacitors can withstand a very high voltage before breaking down. The working volts and the maximum transient volts are very different ratings. The latter rating often not available, but (often) surprisingly high.
I have had this kind of problem in large drives and small drives of different brands. We still do not know exactly what happened in these drives. Common failure mode is flash-over in DC link or from DC+ or DC- to ground. Clean environment in all cases. Dirt in inverter in one case.
We are all (I think) very interested in the outcome of your case. Keep us informed. Please.
-----------------------------------------------------------
And now for tonight's tale: Paper rewinder. 6000 ft/min 20 ft wide 150 N/ft web tension. Brake generator unwind, Center roll wind. Web tension with Bofors KIS transducer (strain gauge) and tension controller working on BG current limit. Transducer changed for a new "identical" one. Problem: Heavy oscillation. Everything had been checked when I arrived. And I also checked everything - and it was OK. So, something had happened to the tension transducer part. Everything else could be ruled out (mr Holmes).
Tension signal was said to have same filtering as before. Well - not quite. Old filter was specified as a time constant (900 ms) and new filter as corner frequency. So 900 ms seemed to be equal to about 1 Hz - it was thought. And 1 Hz was chosen.
Snag: Filter corner frequency is NOT same as time constant inverse. Filter angle frequency is. So we changed filter corner frequency to (1/0.9)/(2*PI), which is around 0.2 Hz. And then up and running again. Machine had been out of production for two days...
This should probably be in a thread of its own - but.
Gunnar Englund
But. You may have left out some facts (like train leaving from Waterloo station at that time, not Victoria station - to paraphrase mr Holmes). One such fact is that there are many other possible paths where graphite dust can and does accumulate. Paths that probably accumulate more dust faster and that also have similarly high voltage stress - not DC but still high peak voltages. And therefore also should have flash-overs.
Another fact that is easy to forget is that IGBTs and capacitors can withstand a very high voltage before breaking down. The working volts and the maximum transient volts are very different ratings. The latter rating often not available, but (often) surprisingly high.
I have had this kind of problem in large drives and small drives of different brands. We still do not know exactly what happened in these drives. Common failure mode is flash-over in DC link or from DC+ or DC- to ground. Clean environment in all cases. Dirt in inverter in one case.
We are all (I think) very interested in the outcome of your case. Keep us informed. Please.
-----------------------------------------------------------
And now for tonight's tale: Paper rewinder. 6000 ft/min 20 ft wide 150 N/ft web tension. Brake generator unwind, Center roll wind. Web tension with Bofors KIS transducer (strain gauge) and tension controller working on BG current limit. Transducer changed for a new "identical" one. Problem: Heavy oscillation. Everything had been checked when I arrived. And I also checked everything - and it was OK. So, something had happened to the tension transducer part. Everything else could be ruled out (mr Holmes).
Tension signal was said to have same filtering as before. Well - not quite. Old filter was specified as a time constant (900 ms) and new filter as corner frequency. So 900 ms seemed to be equal to about 1 Hz - it was thought. And 1 Hz was chosen.
Snag: Filter corner frequency is NOT same as time constant inverse. Filter angle frequency is. So we changed filter corner frequency to (1/0.9)/(2*PI), which is around 0.2 Hz. And then up and running again. Machine had been out of production for two days...
This should probably be in a thread of its own - but.
Gunnar Englund
Like I said its the graphite.... If you see no tracking then it is a case of a fan coming on and lifting a cloud that goes off like a flash bulb. Same as a flash tube. The whole capacitor bank dumps thru the fault.
Where's my star?? hehehehe
Where's my star?? hehehehe
- Status
Similar threads