PowerFlex 523 reversing without command

[jraef]

The Stop Bit logic issue only related to network stop commands, this happens in hard wire too. But it brings up another point, there were other glitches in level 2 firmware, you should update that to at least 3.001, but better if you go to 4.001 (just released this month) if you plan on using ADC, because 3.001 caused an anomaly in Studio5000 not allowing ADC on that rev for some reason. If you don't know, you can flash the firmware yourself with CCW over the EtherNet port if the drive is powered up, or even with a built in web page interface that pops up when you connect via USB cable to the back of the Control Core (after removing it from the drive). If I were you, I would flash them to the latest firmware before spending much more time on this so that in case it was some combination of issue, your work around doesn't cause new ones later if you do flash it.

Still, that said, the fact that the problem gets worse as load on the drive increases is a glaring clue to the source being EMI / noise related. As current increases, so does EMI and common mode noise. So something is getting through to your I/O in my opinion.

 

[TripleJ]

I am sorry for not updating this thread more often. It's hundreds of miles to the PF523 installation and about a hundred to the PF525 installation. I visited the PF525 (EtherNet/IP) installation on 10/10/2014 and took the logs I posted earlier. I looked at how our techs wired the panel and how the electricians ran the wires to the drives. The electricians did use shielded VFD cable, but stripped them at the Panduit and left about 9" of unshielded wire to connect to the VFD. While I didn't like the amount of wire left unshielded, I do have to wonder how the VFD would react to noisy communications. I would imagine it would discard packets, not obey random commands. The terminals are turned off in the parameters, so a signal on a terminal, stray or not, should not do anything. The stop bit logic problem is more about if the stop is commanded while leaving the run command active, and then removing the stop command. I would post the program, but it's about 1.8Mb and too large for the posting limit. It is quite convoluted on how they got the commands to the drives, but it works. I did see that they (we hired a well known national company who I will not name as our shop was too busy at the time) maintained the run signal the entire time the drive was to run instead of a momentary command. The next time I run up there I may change the logic to issue a momentary start as outlined in the knowledgebase article regarding unintentional restarts.

The PF523 installation is harder to get to, but has no networking and is run completely off of the terminals. That installation does not use shielded vfd cable, but has the motor, power, and control wires run through separate conduits to a NEMA 1 adapter on the bottom of the VFD. What kills me on this installation is that the 1336's worked just fine, but not the 523's. The next time I run down there I will be disabling Flying Start and changing the decel to 0.05. When I turned off the Flying Start, the motor just stopped instead of physically reversing, but it took a while to go forward again. When I changed the decal down to nothing, it would just go forward, stop fast enough to blink, and then go forward again. I'm okay with the decel on the roller conveyors, as they are unloaded when they stop and start, but the PF523 job is loaded during startup and shutdown.
I will certainly continue flashing them to the latest firmware when I have the chance, but most of the time getting some downtime is a real pain.

 

[jraef]

... I looked at how our techs wired the panel and how the electricians ran the wires to the drives. The electricians did use shielded VFD cable, but stripped them at the Panduit and left about 9" of unshielded wire to connect to the VFD....

So they probably didn't ground the shield when they trimmed it at the panduit either, did they?

The reason why these sort of things are happening with new drives, but didn't with the older ones, is because of the changes in technology that have allowed the "cheaper faster smaller" drives. The 520 Series hare using the latest generation of IPMs (Intelligent Power Modules) and have 4th generation IGBTs in them now that switch on much much faster than they did in the past. The 1336's used bipolar darlingtons, which had a turn-on time in the neghborhood of .5-1.5mS, then the previous generation (as used in PF70, PF4 class etc.) used in IPMs were switching on at about .2mS. The latest versions, as used in the 520s and most other new drives, switch twice as fast, <.1mS. The effect is that, when combined with other circuit variables such as line capacitance, impedance etc., there is much less room for forgiveness in the installation techniques than there used to be.

Not grounding BOTH ends of the power cable shield eliminates the maximum benefit it has to offer: keeping the EMI / RFI inside. Shield grounding has always traditionally been taught as something you do on only ONE end of the shielded conductors. But this in because up until recently, the only place people used shielded cable was for SIGNAL wiring, not power. When you have POWER conductors, i.e. the power going from the VFD to the motor, those conductors are trying to become radio antennae. Think about it; FM Radio = Frequency Modulation, and what is a VFD doing? Then unlike signal cables that are dealing with milliamps of current, these are dealing with orders of magnitude in the thousands of times greater, so the effect is much stronger than people tend to think it is (which helps explain why your problem gets worse as load increases by the way). So the shielding on the VFD output cables is all about creating a FARADAY CAGE around the conductors, trapping the EMI and taking it to ground in as short a path as possible. When you ground only one end of the shield, it just becomes another transmitter, or at best it does nothing.

 

[ASF]

The way I was taught to do shielding was to think of it like a pipe full of water. If you want to keep the water IN the pipe, you have to connect both ends to something. If you want to get the water OUT, you have to leave one end of the pipe unconnected. Likewise, if you're trying to keep interference IN your VFD cable, and not let it out to screw with nearby wiring, you connect the screen at both ends. If you're trying to keep it OUT of your signal cables, connect the screen at one end only.

It's an analogy that you don't want to think too hard about, but it worked for me

 

[TripleJ]

Okay, problem resolved. I was wrong about the flying start, it WAS enabled on the PF525's. The PF520 series flying start will cause the motor to act funny when the motor is not actually running. Out vendor is talking with Rockwell to see what happened (I had their VFD guy out because I wanted to make sure I wasn't crazy). Also, when Flying start is enabled, the default Current for the flying start changes from 150% to 65%.

 

[TripleJ]

Apparently flying start messing with the startup of a vfd isn't limited to PF520's. It's too bad this didn't turn up in earlier searches. I've never seen flying start mess with PF70's or 750's.
http://cr4.globalspec.com/thread/89227/Powerflex-700-vsdhttp://cr4.globalspec.com/thread/89227/Powerflex-700-vsd

 

[danatomega]

I mentioned flying start in post #2

 

[DickDV]

I've seen a slight rollback of the motor (commanded by the VFD) on several brands of VFD's when operating in sensorless vector mode. This has generally been explained as a need for the motor magnetic poles to line up properly before the sensorless vector takes full control of the motor. In every case, it was cured by starting with a little DC premag.

I've never seen flying start react this way although it seems plausible that noise could "fool" it into thinking the motor was rolling backward.

There must be thousands of HVAC drives on air handler fans that have flying start enabled and are installed without shielded cable. While a little rollback on a fan is not a big deal, if it was at all common, I would think that it would be highly publicized.

That the OP is seeing this on V/Hz controlled VFD's makes the noise-sensitive flying start explanation a bit more likely, in my opinion.

 

[kamenges]

Originally posted by danatomega:

I mentioned flying start in post #2

YOUR post #2. That's really post #6 for those of you keeping score at home.

Keith

 

[jraef]

YOUR post #2. That's really post #6 for those of you keeping score at home.

Keith

Whew, yeah, I was. My score sheet was all messed up!

 

[OkiePC]

I thought this was expected behavior for flying start...I have used it with the old 13xx, PF70 and 700 and had the same documented behavior. I was going to mention it, but someone else did...

 

[danatomega]

I stand corrected - yes, post #6 ... my mistake!