Transient Voltages Causing Drives to Trip on Overvoltage 2

[rockman7892]

I have 2 480V Siemens MM440 VFD's which I have discussed on this forumn previously. I have an issue with both of these drives tripping on a DC Bus Overvoltage when these drives are not running but rather are just sitting idle. The only thing that can cause these drives to trip on overvoltage while sitting idle would be a voltage spike or fluculation on the line side of the drive. I'm assuming that a voltage spike of a certain amount will cause the DC bus link to go high and thus trip the drive on an overvoltage condition.

I have set up a power monitor to try and capture some of these events. However when I see that drive is tripped and then check power meter I see nothing on the power meter. I suspect that since the power meter (Fluke 1735) is only capable of capturing events longer than a half cycle (8.3ms) then it is possible the voltage spikes are happening too quickly for the meter to capture. I have done some research and saw that these voltage spikes are usually very quick and in the order of .5m - 2ms which would be much too quick to capture wih my meter. Would a voltage spike this quick be enough to cause the DC bus to go high and trip the drive?

I would suspect that if these drives were tripping on an overvoltage condition as a result of a transient then I would see other drives througout the plant on the same distribution system trip as well for the same reason. I do not see this happening. Of course this is assuming the transient is coming from the utiltiy or somwhere up in the plant. These drives are by far located at the furtherst point on the distribution system. Does this furthest distance have anything to do with only these drives tripping?

 

[Marke]

If the drives do not have AC line reactors in series with the input, I would recommend that you add them.
The AC line reactors will slow down the charging rate for the DC Bus capacitors and so the transient must be there for longer to boost the DC Bus.
If the transient is long enough for the dc Bus voltage to be elevated to trip level with the line reactors in series, there is more chance of other devices being used to get rid of the energy.

I have also seen this effect where the transient was lifting the three phase network above ground, (not phase to phase) and the output was effectively coupled to ground by the capacitance in the cable and the motor, and the output stage was acting as a rectifier. In this case, the problem was isolated to a faulty earth neutral bonding on the supply transformer and when this was sorted the problem went away. The stimulus for the transient was... HV power factor switching once again.
Transient monitoring between the input phases showed no result, but transient monitoring between an incoming phase and earth showed a very high transient with energy.

The problem with very fast transients is that the voltage monitoring circuit can be fooled into thinking that the average DC Bus has risen when infact there is just a transient superimposed and that is very different.

Best regards,

Mark Empson
L M Photonics Ltd

 

[oldironman]

I have had a similar experience in CA. 4 identical drives would trip out on Bus overvoltage at the same time (apparently) while in idle. By idle, they were turning fans at a low rpm. The correction caps would come on and the drives would trip out. It turns out the drives did not trip at exactly the same time. In the end, talking with Siemens and our UL. rep. we tied the DC buss's together between drive pairs with appropriate fusing. If either bus had any spare capacity in it's caps, it could take on the extra peak and ride through it. In our application the drive pairs were in close proximity to each other in each unit so it was not expensive and got us through the peak. These were MM430's. This may apply to your situation or not.

 

[rockman7892]

Oldironman

That is a very interesting solution that you mentioned regarding tying the DC bus's together. Can you explain more how if either set of bus caps had spare capacity it would ride through the transient? Are the bus DC capacitors the limiting factor to these transients and DC overvoltage conditions? Why is that so?

I'm assuming that we are talking about the capacitor on the front end of the DC bus used to smooth the ripple from the rectifier?

 

[rockman7892]

Without knowing the magnitude or duration of the transient, is there a method for determining reactors or suppressors on line side of drive?

 

[ozmosis]

I'm not sure it would make too much difference but I have been known to be wrong now and again...
Attached are the part numbers from Siemens to order the line commutating choke. (AC Line Reactor to most people)
Check the size of MM4 you have and pick the option. (you will find this part number of the label on top or side.

 

[rockman7892]

Ozmosis

Thanks for the information on the line reactor.

I know typically reactors are used as a current limiting device and others above have mentioned that the reactor will limit the dc bus capacitor charging rate from a result of the surge.

Does this reactor in this situation work by limiting the current to the drive associated with this fast transient or does it somehow work as a filter as well for attenuating the transient voltage magnitude and duration. I want to learn more on how this reactor works in this application. Any information I can find anywhere?

 

[LionelHutz]

The impedance (resistance) of a reactor increases as the frequency increases. A transient is composed of one or more higher frequency waveforms. In the case of a capacitor bank switching, the frequency is typically a ringing which might be, say, 10 to 30 times the line frequency.

So, the reactor presents a higher impedance to the transient (10 to 30 times higher) then it does for 60Hz line power. If you insert a fairly high impedance between the transient and the DC buss capacitors then the it will take longer for the transient to charge the capacitors. Hopefully, the transient has passed before the buss capacitors are charged much higher.

 

[rockman7892]

LionelHutz

The frequency of the large voltage spike in waveform that Gunnar posted looks to be about 1.5ms which would equate to about 10 times line frequency.

So then this 10 time line frequency will cause the impedance of the reactor to increase by 10 times line impedence and will present a higher impedence to the transient thus limiting the current that can pass through to charge the capacitors?

 

[LionelHutz]

Yes.

 

[electricuwe]

At least the waveform shown in Skogsgura's first post looks quite similar to a waveform which is described in the german standard VDE0160 (EN50178) specifying requirements on electronic equipment many years ago.

More than 10 years ago I performed some investigations how drives could be protected against such kind of events.

The basic outcome was that such an event -at least as defined in the standard- injects an amount of energy into the drive that is so high that you cannot cope against it with varistors or with storage in the DC-link.

But I assume activating the brake should work as long as there is some impedance in the AC-line to limit the current.

 

[rockman7892]

I'm still trying to look for a solution to this issue. The other day the two drives tripped at exactly the same time, which almost definitely tells me that there is something on the line side of the drive causing this.

The last two times this happened, it seemed to happen as other equipment was shutting down. There is no PFC switching on this system, but is it at all possible that other motors or other pieces of equipment shutting down could cause some sort of voltage transient that would effect the drive like such?

 

[jraef]

Possible, especially if you have a weak system with automatic tap changers, or one supplied by a local generator that doesn't react fast enough to a major step change in load.


"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376

 

[LionelHutz]

What steps have you taken already to solve the problem?

 

[ozmosis]

I have a screenshot of a site with a similar problem to what was described, on a VFD in the same family.
It was happening at 3.40am on a saturday morning when most loads were off, the voltage was higher, and a PFC cap was switching on/off.
The problem with the drive was that it was not running. The trip state was evident in the morning when the VFD was required to run but as the fault was still active the VFD would not start as it could not clear the fault. A change of logic then cleared the fault and even though the drive would still trip, it would restart anyway and clear the fault on the drive.
Not actually solving the problem but actually drive tripping was not the problem. As I have spent the last 22 years telling people: when a drive trips, it is doing it's job. "Don't shoot the messenger!"

 

[rockman7892]

LionelHutz

Like I mentioned in an earlier post I had place a fluke 1735 meter on the line side of the drive hoping to capture any transients on the system that may have caused the drive to trip. After looking through this post and seeing that most of these transients are typically much quicker than 1/2 cycle (too quick for the fluke)I realized that I would not be able to capture any with this meter. I am now working with the utility and trying to see if they have a high speed type meter that they can lend me to try to capture these events. I am waiting to hear back from them.

The funny thing is, for the 3-4 weeks for which I had the Fluke meter hooked up to the line side of the drive there were no reported incidents of this drive tripping. After realizing the meter would not capture transients I removed the meter, and since then there have been several trips both at night and during the day. I found this to be strange.

This may be a crazy question, but is there any reason why having the fluke meter connected to the line side of the drive may have prevented it from tripping? Maybe by presenting an high impedence or something? I'm sure this is purely coincidal but I figured it would be worth mentioning.

 

[DickDV]

Rockman, have you installed line reactors or brake resistors to deal with this?

Or, has this been nothing but conversation up to this point?

 

[rockman7892]

DickDV

I spoke with Siemens about brake resistors and I was informed that the chopper circuit was only active when the drive was in the "run" mode so therefore the brake resistors would not do anything to eliminate this issue while the drive is sitting idle.

I have not yet installed line reactors. I wanted to see for myself that there was indeed a transient on the line that was causing these issues. I have been waiting on a response from the utiltiy trying to borrow one of their high speed meters which they have used out here in the past.

 

[jraef]

This may be a crazy question, but is there any reason why having the fluke meter connected to the line side of the drive may have prevented it from tripping? Maybe by presenting an high impedence or something? I'm sure this is purely coincidal but I figured it would be worth mentioning.

Hmmm... interesting. Theoretically, if you used clamp-on CTs for your meter (as opposed t flexible current probes such as Rogowsky coils), there is more of an effect on the line passing through them than with regular toroidal CTs. It wouldn't be much, just some extra leakage inductance, but it's an interesting observation. Might be a good argument for adding the line reactors. If just a tiny bit of extra inductance is making a difference, I would think real line reactors would solve the problem, as well as provide other benefits.


"If I had eight hours to chop down a tree, I'd spend six sharpening my axe." -- Abraham Lincoln
For the best use of Eng-Tips, please click here -> faq731-376

 

[itsmoked]

Might act as a high frequency choke.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[rockman7892]

Jraef

I should have explained my meter setup better, but I did not have the current probes hooked up both rather only the three voltage probes on the line side of the drive. I figured I was only looking to capture transients on the incoming voltage supply so I was just monitoring the voltage and was not concerned with the current.

My apologies for not explaining better.