Control.com - Nerds in Control

Does anyone know what the specs are for an AB Powerflex series drive for how well it can control an AC Induction motor in closed-loop vector control as a percentage of max RPM, i.e. percent of 1800 RPM (which is generally an industry standard of measurement that I have seen)? No Load... just the motor.

Same question for an AB 1336E Impact Plus Drive.

Well, since the forum has had issues lately I found the answer to my question myself. AB states in the 1336E Impact manual that it can hold 0.02% of max speed for a 1000:1 speed range or .72 RPM on a 3600 PRM application... No Load.

Now my question is this, why can't I get it to control better than 9RPM No Load? I've run through the autotuning multiple times, using diffent bandwidths etc., I went through many iterations of manual tuning and the best I can get it is 9RPM... Any suggestions?

Drive manufacturers typically quote specs for long term speed regulation. If you averaged your speed over a period of time (minutes?) you may find that you fall within their spec. Short term speed regulation is largely dependent on the electro-mechanical system design and is out of the drive designers control.

Really, I have a hard time buying that. Never in any drive manual I have read (and I've read a lot of them) did it state it as long term regulation... especially for closed-loop vector control.

I do understand that somewhat the motor comes into play, but logic tells me if I buy a standard vector duty motor from a reputable manufacturer it should fall into the category of being able to be controlled by a vector drive... somewhat close to these spec’s, not 10 times over... no load.

In all the manuals I have read they state it as plus or minus ~0.01% to 0.02% of the speed range. And to do some sort of averaging... even over a second to meet a spec of 0.02% of the speed range would be extremely simple, you could have oscillations up to 50 RPM or more... and meet that spec... so if that is the case the spec is pretty much meaningless.


On January 18, 2008, William Sturm wrote:
> Drive manufacturers typically quote specs for long term speed regulation. If you averaged your speed over a period of time (minutes?) you may find that you fall within their spec. Short term speed regulation is largely dependent on the electro-mechanical system design and is out of the drive designers control. <

( Complete thread: http://www.control.com/thread/1026242822 )

It is kind of a meaningless spec, IMHO, if they don't specify the time frame in which the speed regulation is measured. I have seen it listed with a specific long term speed regulation and a system dependent (non-specified) short term speed regulation.

That said, if you can't regulate to +- 1 RPM with no load, then something is probably wrong.

I'm not sure about A-B, but I have used ABB ACS350's and ACS550's in closed loop applications. Under fairly steady loads, they held within 1 RPM even for short periods of time.

Well looking at the display it is less than 1 RPM, but adding a scope to the speed feedback shows 9 RPM, proving that the display's averaging/filtering routine is working very well and the motor isn't being controlled all that great.

Sounds typical, even well tuned servos can have problems with short term speed regulation. It can be difficult to achieve. Can you add a flywheel? Can you go closed loop vector and tune the velocity loop? It sounds like open loop "default" tuning is not going to work for you.

Since the VFD in a closed loop is effectively running a PID loop, it is fair to assume that it is constantly adjusting its output frequency to maintain speed. That being said, it would have to increase and decrease its frequency to the motor and compare this to the encoder feedback of actual speed to maintain a consistent flat level, and it does this constantly. When you use a scope to capture a rotating speed over a very, very short time, it can be extrapolated to give you the RPM, and this can be captured in well under 60 seconds. But in the end we are talking about maintaining revolutions per minute, not revolutions per second, or any other time constant other than minutes. When you do the math over a whole 60 seconds worth of scope waveforms, is the average of those adjustments up and down, within spec?