Powerflex 525 driving a flywheel

[Joe E. 259]

---

We have a handful of nearly identical flywheel-driven machines. About a year ago, we installed a 15HP Powerflex 525 drive in simple V/Hz mode with DB resistors on one of the machines. It's been running great for that whole time.

We have it set up so the STOP input (terminal 1) is wired through the DB resistor thermostats, input 2 is Run FWD ,and input 3 is Run REV. Since it's replacing an across-the-line reversing starter that's in the midst of a bunch of hard-wired relay logic, the drive's output Relay 1 is Ready/fault and will cause the relays that control drive inputs 2 and 3 to drop if the drive faults. This arrangement has been running fine on machine #1 for over a year.

We tried to duplicate that installation on another machine that has an almost identical motor (np is 1765 RPM instead of 1775, but otherwise the same including NEMA Design) but we're getting DC Bus overvoltage faults after the machine cycles for around 10 minutes. When trending in CCW, I see that the output frequency drops to 0 from one sample to the next while the STOP and Run FWD inputs are still ON and Relay 1 still indicates OK. The commanded frequency doesn't change. On the next sample, Relay 1 drops out, which also drops out input 2 (Run FWD). Then, about 400 ms later, the DC bus climbs past 810V and the drive faults on bus overvoltage. That's the only fault that's shown on the drive. The only parameter differences between the 2 machines are the network addresses and the v7 drive has flying start enabled since it doesn't have a flywheel brake that applies when the motor is stopped.

This screenshot  the values of the trended parameters as the output frequency drops:

 

Here's the next trend sample, at which point Relay 1 has dropped, taking Run FWD with it. At this point, the DC bus (the visible graph) is still well within its normal range. It trips at 810VDC, which happens about 400ms after the output frequency drops:


Per a suggestion from Rockwell, switching the drive to SVC (with a static autotune) accelerated the issue and I actually saw the STOP input open when the resistors got hot. The fault happened a lot sooner (in a few minutes of cycling instead of about 10 minutes). They also suggested changing p36 to 1800 to remove slip compensation and then to use P531 to disable the bus regulator, but neither improved the situation.

In case we managed to get a bad drive out of the box (or broke something during installation), we swapped in another brand new drive and got the same result. The only difference between the drive that's working fine and the ones that are faulting is the firmware version. The older one is v6.001 while newer ones are v7.001. The release notes don't highlight any issues. I would rather not flash the firmware down, but that's about my last option that I see here.

Any ideas?

[pcmccartney1 141]

Why is the DB so much hotter than the Machine #1?

Also disconcerted about the DC Bus voltage difference.

What is the result is you change the wiring to omit the DB thermostats signal?

[pturmel 160]

Can you swap the v6 and v7 units to see if it is a physical difference?

[Joe E. 259]

Thanks for taking a look, guys!

15 hours ago, pcmccartney1 said:

Why is the DB so much hotter than the Machine #1?

Also disconcerted about the DC Bus voltage difference.

What is the result is you change the wiring to omit the DB thermostats signal?

I was wondering about that as well. The DB resistors didn't get as hot in V/Hz as SVC (as expected). It's like it was trying to very aggressively clamp the speed, which we really don't need. I logged some brief data from both machines while they were cycling and the DC Bus fluctuation is very similar between the machines, but the problem child is going a little higher. Still nowhere near the trip point of 810VDC. I took the separately captured data from both machines and overlaid them on the same plot. The lower magnitude fluctuation at the beginning of Press 2 and the end of Press 4 is when the flywheels were running but the presses were stopped. The higher magnitude is while the presses were cycling. Both were sampled at a 50ms rate (per CCW...).

 

I haven't omitted the brake resistor thermostats while running, but when I disconnected them, and the STOP was asserted, relay output 1 stayed ON, as I expected. The press was not cycling at the time. I would expect, if the thermostats were blinking, that the input signals to the drive would change differently, but input 1 is generally staying on throughout the process. I only saw it open once while logging data. It stayed on every other time.

13 hours ago, pturmel said:

Can you swap the v6 and v7 units to see if it is a physical difference?

I thought about that, and about flashing the firmware back to v6 (if it will let me). If I have time, I may be able to swap the drives on 2nd shift when they're not running production but we already have machine #2 down for this upgrade so they're not going to be happy if we take #1 down at the same time.

[BobLfoot 302]

Couple of Question / Observations to mention.

1.   What is your Deceleration Parameter Setting on each drive?

2.    So the Unit with the Flywheel Brake is not tripping and the one without a brake is.  I know you said the Flywheel Brake acts as "parking brake" being applied after the motor stops, but is there any chance it is engaging before rotation is fully stopped and helping the non-tripping drive.

3.    Are you setup for Ramp Stop or Coast Stop?

[Joe E. 259]

Bob, thanks for taking a look.

1) P42 Decel time is set to 20s

2) The drive that trips is tripping while running, not while stopping.

3) P45 stop mode is DC Brake/CF

I ran some more trends at a higher rate after talking to a distributor rep. He suggested that the DB Resistor protection circuit may be forcing the output frequency to 0Hz if the DB duty cycle is too high, which it looks like it might be. The problem I have with that theory is that the ready/fault output is dropping long before the bus overvoltage happens. I can see the sequence being off by a sample or 3 (in fact, trending in CCW shows a slightly different sequence than trending in Logix Designer), but I wouldn't expect the ready/fault to drop until after the DC bus actually goes high, which is about 400ms after the ready/fault goes low.

Flashing the firmware back to v6.001 to match the working machine had no effect. I'll be back into this in the morning.

[chelton 66]

Are you certain the nameplate V/Hz match what is set on the drive? The V/Hz ratio makes a big difference. ( motor rewound maybe?)

What is the speed reference? Network/Analogue/Fixed - Have you confirmed it's not being commanded to change rapidly/noise etc.

 

[Joe E. 259]

I'm fairly certain the V/Hz on the nameplate is correct since it's a new motor on the problem machine. The speed reference is via Ethernet/IP, which is configured to "hold last" on comm loss since the start and stop are hard wired. While trending, the commanded frequency never deviates from 45Hz while the output frequency drops. When we had similar issues on another machine that had a high-slip NEMA D motor, the speed reference was the drive pot and was limited rather tightly by min/max frequency parameters, yet was acting very similarly.

I had to dig into a different problem yesterday, so I didn't get a chance to get into the data logs I captured. I'm going to work on that a little today (if I can...).