Protecting a VFD from another VFD malfunction

Hi Everyone,

Yesterday we encountered an issue where I had a 480V 3Ph 2HP Powerflex 525 blow up. Upon further inspection, we noticed that 10 other VFD's had also been taken out in the process. There were no internal power surges in the plant (we have facilities in place at the switch gear to protect over/undervoltage, loss of phase situations, etc), but I believe that one drive in particular blew its caps up and that surged the panel, taking out other VFD's. Upon troubleshooting, no phase-to-phase, dead short other ground fault issues were found in any part of the panel or any of the field devices.

The panel has 28 identical 525's that are zero-stacked (aka literally side-by-side mounted across the DIN rails), are individually protected by Eaton 4-6.3A adjustable MCCB's and all are fed from a 200 amp disconnect

My question is, what can be done to prevent one VFD explosion from taking out others? I've heard that using J-class fuse blocks instead of MCCB's will help as them blowing may prevent a cap-exploding VFD panel surge from passing onto neighboring drives. I don't want to create nuisance blown fuses by just replacing the MCCB's with fuses but I need to do something different. I was thinking about putting fuses upstream of the MCCB's, but rating them at some arbitrary value above the MCCB limits. Does this sound reasonable?

One caveat: I am making an assumption here that the one VFD failing was the cause of the others, so I'd also like hear any other ideas that could have caused it.

Thanks!

Do you have line reactors ahead of the drives?

how far are the motors from the vfd's?
there is a distance limitation before load reactors are required.

james

This panel nor the drives themselves were not initially equipped with either line or load reactors. I had thought about adding them a couple years ago due to some power quality issues we were having, but we've worked that out. Space is also at a minimum in this panel so I don't know where I would mount the reactors even if I needed to. I would likely have to use another. However, if that would protect from these issues then I would be willing to consider it.

Regarding motor-from-drive distance: The furthest motor is about 250 feet. We do have shielded VFD cabling running off all of these drives. What is the maximum distance beyond which a load reactor would be required?

Did you use SCR rated fuses on VFD power? MCCB is not fast enough.

Check these links about cable distance from VFD
PWM AC Drives: Maximum Motor Cable Length Between Drive and Motor
https://rockwellautomation.custhelp.com/app/answers/detail/a_id/52538

PowerFlex 523/525 Drive: Motor Lead Lengths, reflective wave chart
https://rockwellautomation.custhelp.com/app/answers/detail/a_id/544025

NetNathan said:

Did you use SCR rated fuses on VFD power? MCCB is not fast enough.

Click to expand...

No I only have MCCB's currently, there are no fuses. I was thinking about using SCR rated fuses as a supplement(upstream) to the MCCB's, but with a higher trip amperage ratings than the MCCB's. That way in 99% of cases the MCCB's would protect everything under normal running conditions, and the SCR fuses would protect the VFD's in the condition that I ran into yesterday.

tarik1978 said:

Check these links about cable distance from VFD
PWM AC Drives: Maximum Motor Cable Length Between Drive and Motor
https://rockwellautomation.custhelp.com/app/answers/detail/a_id/52538

PowerFlex 523/525 Drive: Motor Lead Lengths, reflective wave chart
https://rockwellautomation.custhelp.com/app/answers/detail/a_id/544025

Click to expand...

Thanks for the link! It looks like I'm in the clear as far as reactors are concerned on the load side. My max motor length is 250' and I have shielded VFD cabling everywhere.

With zero stacking, is there any way one drive blowing up could be sending conductive carbon soot, etc out through it's vents into the adjacent drives vents and so on?

I have been told to start considering load reactors starting at 50 ft.
check the manuals to be sure.


james

Cow said:

With zero stacking, is there any way one drive blowing up could be sending conductive carbon soot, etc out through it's vents into the adjacent drives vents and so on?

Click to expand...

Yes that actually happened on one of them. The one that I believe started this whole mess blew its caps out the side of the drive and into the one adjacent to it, destroying that one in the process. However, the other ones that blew were all over the panel with no apparent pattern or proximity to the others. In the end, 13 drives were killed.

After checking every inch of the power, ground and motor leads, no smoking gun was found so I have to assume that the initial incident surged the panel, taking out the others.

James Mcquade said:

I have been told to start considering load reactors starting at 50 ft.
check the manuals to be sure.


james

Click to expand...

Yes I checked the tech note that Tarik sent and it gives motor-to-drive distances and what you'd need as far as reactors go. My max distance is 250 feet and the max allowable distance w/o a load reactor is 275 feet.

I actually came into this situation mid-way. According to my maintenance team, one drive blew, they replaced it and the one next to it and then as soon as they turned on the disconnect, they heard a loud pop. They decided to leave the machine down until I was on site the next day. Upon inspection, I found the remaining 11 drives to be bad. It's possible that the first explosion destroyed multiple drives and that the maintenance didn't catch them. When they turned on the disconnect the already shorted diodes in the other drives (I confirmed 6 drives where all forward biased diodes had been shorted) caused another surge, taking out the rest of them.

If this is the case, would line reactors have caught this massive current surge in time to let the MCCB's catch it?

Trying to decide between line reactors and SCR fuses for my next step to protect these VFD's

IMO, the Fuses/CBs/MCPs are in the circuit to minimize the fire when something shorts.

IMO, input line reactors are very cheap insurance against line disturbances.

I suggest the line reactors, even if you have to move the MCPs and Line Reactors to another cabinet, or use enclosed Line Reactors and hang them on the top, back, sides, etc of the existing enclosure...

Gene Bond said:

IMO, the Fuses/CBs/MCPs are in the circuit to minimize the fire when something shorts.

IMO, input line reactors are very cheap insurance against line disturbances.

I suggest the line reactors, even if you have to move the MCPs and Line Reactors to another cabinet, or use enclosed Line Reactors and hang them on the top, back, sides, etc of the existing enclosure...

Click to expand...

Yea after some more research I tend to agree with your assessment. The problem for me is going to be space. It would be nice to use the bolt on reactors from AB, but there isn't enough space between the tops and bottoms of the drives and the panduit. However I might be able to get away with some DIN rail stand off spacing to get it to work.

Worst case scenario I'll use another panel or mount them on top.

Gene Bond said:

IMO, the Fuses/CBs/MCPs are in the circuit to minimize the fire when something shorts.

IMO, input line reactors are very cheap insurance against line disturbances.

I suggest the line reactors, even if you have to move the MCPs and Line Reactors to another cabinet, or use enclosed Line Reactors and hang them on the top, back, sides, etc of the existing enclosure...

Click to expand...

I'm have been in that camp too for years. If you think about a failure mode that could cause a high current to flow on the circuit FEEDING the VFD, it basically will be the failure of the VFD that causes it. Semiconductor fuses ahead of VFDs then only react AFTER the VFD has already failed, so they are good for stopping a fire, but they can't put the magic smoke back in.

But now I know that there IS an issue that can happen in a VFD, ESPECIALLY one without a line reactor or even a DC bus inductor, that CAN cause high current to flow through the diode rectifier PRIOR to a failure, which MIGHT be stopped by a semiconductor fuse. It's complex, so hang on.

Background:
Diodes don't conduct continuously, they have a minimum voltage (somewhere between zero and peak) at which they become "forward biased" and begin to conduct, called the Forward Conduction Threshold" (FCV).

The transistors in the inverter section pull DC power off of the capacitors and send it to the motor, then the capacitors pull current through the rectifier to recharge themselves.

Capacitors on a DC bus charge virtually instantly and discharge virtually instantly. That is why all VFDs have some sort of "pre-charge" circuit that reduces the charging current when you FIRST energize a VFD, usually by putting a current limiting resistor in series with the caps, then shorting that resistor out with a relay or contactor after a second. But once that pre-charge is done, it STAYS out of the circuit until the VFD is shut down. This means that once it is out of play, capacitors will be available to pull current at the available fault current in the circuit. That normally doesn't happen because as the DC bus caps are depleted by the inverter, a rectifier is providing energy immediately, and with 6 diodes feeding the DC bus, the caps are never far off from full charge.

But here is how that can go awry. IF there is a large ringing transient on the incoming AC line, and the "bottom" of the ring drops BELOW the FCV of one or more diodes right at the moment that it/they were supposed to conduct into the DC bus, it doesn't happen. If at that moment the VFD is sitting idle, nobody notices, no harm, no foul. But IF the drive inverter is modulating, meaning sending power to the motor, it instantly depletes the DC bus, but the DC bus is NOT being replenished by the diodes, so the DC bus voltage drops very very rapidly. Then BEFORE the VFD has time to react and trip on UV, the ringing transient is over with and the diodes CAN conduct again, so they do. But, the capacitors will recharge themselves by pulling current AT THE AVAILABLE FAULT CURRENT RATE! Remember, the pre-charge circuit is also out of play here because the VFD is running, so the pre-charge resistor is still shorted out. So if for example your AFC at the drive terminal is 40kA, the rate of current flow into those capacitors will be at a level approaching that value (limited only by the line impedance / resistance). That rate of rise (dI/dt) is VERY steep and can in fact be too steep for the capacitors and or diodes to handle, causing either capacitor failure or diode shorting, or both.

This is why reactors are so important, especially on small drives that typically do NOT have DC bus inductors. Inductance is the key to slowing down the rate of rise of a transient to the point of being within the ability of the VFD components to handle it. I actually believe in using line reactors even for VFDs that do have DC bus chokes, because the DC bus choke is AFTER the diode bridge. But the effect is dampened by the DC bus choke by virtue of the capacitor charging current being dampened.

In THIS CASE, the high speed semiconductor fuses MIGHT have been able to avoid the initial damage that caused your drive to fail in the first place (IF that was the cause) by clearing fast enough to avoid the dI/dt that caused the failure. The BETTER solution however is the line reactor.

A "rule of thumb" that I used on avoiding these damaging transient current effects is that if your VFD is one without a DC bus inductor and your SOURCE kVA is more than 10x the VFD kVA, you absolutely need a line reactor. As far as I know, all small "component class" drives 10HP and under, like the PowerFlex 520, don't have a DC bus inductor. Line reactors should be used in front of every one of them. Why don't the VFD mfrs just build the VFDs with line reactors then? Because there are LOTS of applications where that 10x rule is not violated. I once worked for a VFD supplier that did in fact put a line reactor inside of every VFD, their drives were hell-for-stout; they went belly up, because their drives were 3% more expensive than their nearest competitor. We (the market for VFDs) have created this issue.

Now that said, using the 10x rule, you can get a sufficient protective effect by grouping multiple VFDs behind one line reactor, so long as that reactor size stays within the 10x rule for the smallest drive behind it.