What's new
What's new

VFD Loud bang.

I like your idea of pull out the VFD and have a look. I'd have probably already done that.

No telling what you may find............. I've seen many things that I would not have believed except they were right there when I looked.

What you find may point out the solution.

BTW, your original post said the sound seemed to come from the motor, not the VFD. So the motor may be worth a look.
 
The max voltage you will get during regen for a 480v drive is as high as 1000v, and yes, you may kill the drive if this continues. The current isnt just fed by the motor, its the vfd during slowdown. Once it switches from regen to dc injection, the peak dc voltages will fall from the dc bus brake voltage (800vdc approx) to 600 volts (415vac *1.41).

The motor is savable. Wash and scrub out wherever is arcing and pour a coat of polyurethane or regular clear 5 min epoxy in the winding.

How far apart the wires are doesnt matter with regard to the thin layer of carbon and copper metal oxides deposited everywhere.
 
Any short loud enough to be considered 'loud' over a spinning lathe will leave blast marks and pitted copper, as well as destroy the drive. Guarantee this isn't happening.

A 240V VFD will have a DC bus running about 330VDC normally, lifting up to ~400VDC under braking conditions, and the drive likely uses components rated for 450VDC continuous, that will handle 600VDC for a few milliseconds happily. Double these figures for a 480V drive.

Braking creates a slowly (in electrical terms) rising DC bus voltage. The motor doesn't really see anything different electrically during braking vs acceleration at a given speed, just the sign of the slip changes.

I agree that you:

  • aren't using the lathe in the way it was designed - the original three-phase native setup will not have had electric braking, though it might have had electrically released mechanical brakes, or just a straight mechanical brake.
  • are probably finding that something in the gearbox is not meant to see that sudden a torque change.
 
We have not been told what voltage the machine runs on. I would think that if it was 480, that would have been mentioned, but..............

The mention that the noise seems to come from the motor is of interest and tends to support a mechanical cause, but does not prove it.
 
It is not definitely not mechanical. It is also fairly random. I'm running 3 phase 415 volts so I am up there. By the way, it is roughly 1000v per mm for free air. There is also a smell when it happens. Not sure if it is ozone or a burned component but the VFD and motor still work. Maybe there is a MOV in the VFD that is dying. May pull the VFD out and have a look.

415v means we should assume a 480v rated vfd and so the dc brake voltage will be in the 550vac*1.4 territory. Unless they make auto sensing vfds that can distinguish between 415 and 480v nominal or its a euro specific default programming.
 
Even configured for 415VAC, almost all drives I've seen aim for a peak DC link voltage of ~800VDC, or about 565V AC equivalent.

No 415V industrial equipment should have clearance less than ~3mm, and creepage is normally about double that. The terminals in the motor should be spaced by at least that amount, and the windings themselves are insulated - you need to look at the punch-through voltages for that, not the clearance/creepage.


The motor electrically simply doesn't see any higher stress under braking than under acceleration. The VFD isn't tripping, so it's not a DC link overvoltage. You've almost certainly got a mechanical issue.
 
  • Like
Reactions: JST
I slowed down the acceleration and deceleration and the lathe has worked fine. I then changed the motor pulley from 100 to 200 mm (3 belt) to get more speed. While testing and winding the pot up and down, I got the bang again. Much more inertia with this big pulley. .

Finally pulled the drive out and opened it up.

There it is. These caps are all in parallel. The drive ran fine even with the dead cap. My guess, on slowdown, the back emf from the motor caused the spark in the cap across the cap fillaments.

Took out 6 caps 220uf and replaced with 2 x 680uf, reassembled and reinstalled. So far, working perfectly and the motor even sounds happier.
 

Attachments

  • Drive.JPG
    Drive.JPG
    47.3 KB · Views: 10
Caps often burst with a loud bang.
But just once......

The capacitor may be protecting the rest of the system by arcing internally. I have not seen very many capacitors that fail like that. It is actually how they are supposed to fail, but usually, they finish the job by blasting out some of the metal foil inside, which causes more damage when it shorts other things inside the VFD case.

Goes to show how even good descriptions of the problem can be misleading. You had BOTH a legitimate loud bang, and presumably the sound of arcing in the failed capacitor.

Now the question is what caused the failure. Could be age, really the effect of heating over time.

By the way, I am not too confident of the replacement using a single capacitor to replace a bank of 4. The reason for the bank of 4 is to get to the net "ripple current" rating needed for the power level of that drive.

Ripple current causes heating, and heating is what generally causes capacitor failure.

You may have another failure, happening sooner, if the 680 uF capacitor does not have a really high ripple current rating, equal to the entire bank of 4 that were originally in there.
 
I beg to differ. Of the 4 spare VFDs I have, all use singe capacitors of the order 560 to 680 uf, so my replacement will be fine.

> Ripple current causes heating, and heating is what generally causes capacitor failure.

Over voltage and age are the causes of capacitors failure.

The ripple is a function of load and capacitance value, not the number of capacitors. ESR is what causes the heating.
The ESR many be lower with 4 caps but accurately measuring ESR is difficult. Unless a capacitor is specifically noted for low ESR, they are all much in the same basket.

What is ESR in a capacitor?
Equivalent series resistance (ESR), also known as internal resistance, is a value representing the loss of useful energy.

ESR and heat is a function of current not ripple.
 
Not so fast.....

Ripple current is the flow of current in and out of the capacitor as the voltage changes across it. There can be no voltage change without current flow (charge added to or taken from the capacitor), that's just how they work.

Yes, the ESR directly affects the ripple current rating, since heating is current squared x resistance. But, so does surface area, which dissipates heat, etc. Because of the square factor, ripple current is the major issue with heating.

Heating is the primary factor in lifetime. Yes there are other factors, but heating tends to cause shorter life. If heating is too much, life will be "very" short ;):eek::angry:. Each manufacturer has a formula for lifetime vs ripple current, voltage, etc. Ripple current is a "proxy" for heating.

Number of capacitors definitely affects life vs ripple current. With one large capacitor, all the ripple current goes through that capacitor, and so through it's ESR.

With several capacitors of lower value adding to the same total uF, the ESR may or may not be different per capacitor, or even for the total, but the surface area total is a lot more, allowing better heat removal. The net ripple current rating is the sum of all the parallel capacitors.

You may have noticed that capacitors for SMPS often tend to be long and relatively thin. That maximizes surface area, and minimizes the length of the heat flow path, for good cooling. "Dissipation" is not the issue, "heating" is. If you can keep them cool, their ratings increase.

When you replaced the 4 capacitors with a single one, the ripple current rating of that one had to be equal to the sum of the 4 that were in there before. If it is not, the lifetime may be considerably shorter.

If the ESR of the one is the same as each of the others, the dissipation will be 16 times higher, since the ripple current is 4x higher.... the one capacitor has to handle the current that 4 did before, and it has to dissipate 16x the heat.

Presumably the designers of the thing did the calculations, and used 4 capacitors because they were needed. Your "redesign" may or may not be sufficient.

I'd be betting there may be an issue unless you did the design work and know the one can do what is needed. Or unless the load on the VFD is sufficiently lower than the design maximum load that the ripple current will be much lower than the original setup was designed for.
 
I have a BSc in Electrical Engineering. I know what I'm talking about. In this case, 1 or 4 capacitors has no practical difference at all. Biggest design issue with capacitors is cost/availability and physical size. Never once in 45 years of electronics have I come across a design for heat dissipation for capacitors.
 
I have a BSc in Electrical Engineering. I know what I'm talking about. In this case, 1 or 4 capacitors has no practical difference at all. Biggest design issue with capacitors is cost/availability and physical size. Never once in 45 years of electronics have I come across a design for heat dissipation for capacitors.
Suit yourself.

I've been designing successful power electronics, including VFDs, for over 40 years, and I have not yet heard that ripple current is nonexistent, nor that parallel capacitors do nothing. In fact, the exact opposite is what is generally accepted.

I have noted also that the folks who make capacitors appear to violently disagree with you. They think heat dissipation is definitely a thing for capacitors, as is paralleling them for increased ripple current rating. Likewise the folks who design SMPS, who often ensure airflow over the capacitors, and commonly use a number of smaller value capacitors instead of one larger value unit.

I have even seen special designs with actual heatsinking for the capacitors. That is quite rare, but in some cases may be needed when reliability, and size, and/or the environmental considerations require it (generally military or aerospace products).

The folks who designed that VFD apparently agreed. It was not that they could not buy capacitors with the voltage and total capacitance rating in one package, it was almost certainly that the probable lifetime with a single pair was insufficient due to ripple current and resultant heating. That would lead to early failure.
 
Suit yourself.

I've been designing successful power electronics, including VFDs, for over 40 years, and I have not yet heard that ripple current is nonexistent, nor that parallel capacitors do nothing. In fact, the exact opposite is what is generally accepted.

I have noted also that the folks who make capacitors appear to violently disagree with you. They think heat dissipation is definitely a thing for capacitors, as is paralleling them for increased ripple current rating. Likewise the folks who design SMPS, who often ensure airflow over the capacitors, and commonly use a number of smaller value capacitors instead of one larger value unit.

I have even seen special designs with actual heatsinking for the capacitors. That is quite rare, but in some cases may be needed when reliability, and size, and/or the environmental considerations require it (generally military or aerospace products).

The folks who designed that VFD apparently agreed. It was not that they could not buy capacitors with the voltage and total capacitance rating in one package, it was almost certainly that the probable lifetime with a single pair was insufficient due to ripple current and resultant heating. That would lead to early failure.
Book Learnin' can be a dangerous thing (especially when it's ony the course syllabus) - I've worked on a lot af AC-DC convertors and large amplifiers (e.g. servo amplifiers moving 60-foot dishes horizon to horizon in a few seconds), capacitor ESR is a big issue when you suddenly have to dump a few hundred amps into a big motor, and yes, they used banks of paralleled capacitors and programmed maintenance included measuring individual ESRs every 6 months or so and measuring temperatures under load when exercising the dish motion systems.

I myself "hacked" a VFD input stage (voltage doubler to the bus capacitors so it thinks it has 480v in when it's on a 240v supply) - this seriously increases ripple voltage and current AND increases heat dissipation, I have to change them every few years to allow for the reduction in their capacitance values thanks to thermal degradation...

Marsheng, have you designed / worked on motion control systems, or motor drive electronics?
 
I started with switch mode power supplies 45 years ago.

You can argue all the merits of ESR's and capacitors banks, but in this application with a 4 kw 3 phase VFD, having multiple caps or single does not matter.

If single is so bad, then what does Eurotherm, Allan Bradley, Huanyang, NFLixin all use large single capacitors? My guess, in this case, it was easier to fit smaller caps in the space available.

My favourite saying, "Rules are for fools and a guide for the wise." Where you are working is voltage doubling and or other exceptionally high loads, yes your argument holds, but in this case, VFDs under normal circumstances are expected to last for 10+ years and are designed accordingly.

Of the 20 or so VFDs I have running, this is my first cap fault and my guess is, my deceleration was too fast and dumped a high voltage on the caps. .
 
Of the 20 or so VFDs I have running, this is my first cap fault and my guess is, my deceleration was too fast and dumped a high voltage on the caps. .
Or its just defective. If you dont have a brake resistor then the nominally 340vdc can get as high as the trip threshold.

I suppose its not unheard of but a 450v rated capacitor used at 340v for a long time, then briefly held at 450 volts, would it really have enough internal leakage to explode?

So i think that one that exploded was internally leaking and the others were not. If they all had a leakage problem above the nominal 350vdc, they would all just warm up a bit during deceleration, but not enough to explode any of them.
 
how many is an option based on specs.

What the ripple current rating is...... vs the ripple current it will actually see.......that is "rather important", and is disregarded at your peril of early failure.

You can do as you like. I only suggested a check of specs against what it will see, not predicting instant failure. Time will tell......
 
Suit yourself.

I've been designing successful power electronics, including VFDs, for over 40 years, and I have not yet heard that ripple current is nonexistent, nor that parallel capacitors do nothing. In fact, the exact opposite is what is generally accepted.

I have noted also that the folks who make capacitors appear to violently disagree with you. They think heat dissipation is definitely a thing for capacitors, as is paralleling them for increased ripple current rating. Likewise the folks who design SMPS, who often ensure airflow over the capacitors, and commonly use a number of smaller value capacitors instead of one larger value unit.

I have even seen special designs with actual heatsinking for the capacitors. That is quite rare, but in some cases may be needed when reliability, and size, and/or the environmental considerations require it (generally military or aerospace products).

The folks who designed that VFD apparently agreed. It was not that they could not buy capacitors with the voltage and total capacitance rating in one package, it was almost certainly that the probable lifetime with a single pair was insufficient due to ripple current and resultant heating. That would lead to early failure.
Yeah - I'm on of them EEs too...........

Back in the 90's I got my fire hose dose on capacitors on VFDs. They were blowing up on our machines on a regular basis - one vendor's drive. When I say blowing up, the cabinet doors were buckling. Some of these drives had close to probably 60 capacitor cans in it. I got to rebuild more than my fair share of the capacitor banks.

And yes - it was all about ripple current and temperature, but in our case there were also installation issues. It was a poor design and having good connections between the caps and bus bars was tough, and there were also stresses in the bus bars after assembly. Eventually we figured out how to make it all better. We did however a few years later change out all the caps on one machine for a cost around a million dollars.
 








 
Back
Top