AC Drive troubles

[Jiri Toman]

ED01,

I looked at the web site and couldn't find instruction manuals
in pdf format or any other format for that matter.
Am I missing something?
Why would you have a web site that gives you just a generic
info? Do you enjoy mailing catalogs and manuals to U.S. by snail mail?
If you have something innovative let us read about it right away.

 

[Alan Case]

Over voltage faults can be caused by trying to decelerate too quick for the size of the drive. The energy has to go somewhere.
Solution if you want fast decel is to usually oversize the drive. I find a simple rule of thumb is if the drive can't accelerate the load at the speed you wish to decelerate it then it will never decelerate it.
Regards Alan Case

 

[DickDV]

OkiePC, your first post says that a 5hp AC drive is replacing a 3hp DC drive. If properly done, that should be plenty to cover the application.

That it apparently isn't covering leads me to start my analysis at the basics. First, forget horsepower. Do your sizing with torque and speed. That leads us to the motor. Are the DC and AC motors capable of the same torque output? You must consider short-term accel and decel torque in this analysis. Assuming the AC motor is a NEMA Design B, the peak torque is about 220% of continuous rated torque. No amount of drive oversizing is going to get more torque than that out of the motor. The DC motor could be about anything for short-term overload up to about 300% but it is unlikely the DC drive would be able to drive the motor that far up in current. If you still have the DC drive output current rating and the DC motor nameplate amps, we could estimate how much overload was available for accel and decel.

How about giving us the DC motor nameplate rpm and the AC motor nameplate rpm and NEMA Design.

Once we have that info, I'll take it to the next step for you.

 

[OkiePC]

More details

The application is a cutter carriage travelling on linear bearings back and forth a distance of 11 to 14 feet. Adjustable proximity switches control the start and stop positions. I think the carriage weighs about 125 pounds. The stock being cut adds very little load. The carriage engages the material, and accelerates to the opposite end where the prox switch triggers a rapid reversal in direction. The old DC system made this "turnaround" better than the new AC drive system. We are controlling the new AB160 drive (V/Hz)with an analog command, a forward input, and a reverse input. I have considered going with a run input and a bipolar speed reference to see if that would help it change directions faster. The cycle time (down and back) is about 6 seconds. The drive spends about 10-12% of that time in decel. We have dynamic braking resistors rated for 5% duty cycle. When we first started it up after the contorls upgrade, it ran okay, then we occasionally got high bus faults on the drive. The technician reduced the top speed to correct this. I later increased the decel time from .2sec to .3. This morning, I changed the settings back to the original, and checked the fault queue (which only stores the last three faults) and the high bus fault was not present. The motor is cold to the touch, and seemed to be running fine. Another intermittent problem is that when the carriage comes to rest at the start position after completing a cycle, the drive seems to "let go" of it a little early and let it coast a little. Once in a while, it will slide into the end travel limit. This is rare, but worse at higher speeds. My gut feeling tells me that we've got plenty of motor for the application, but that our drive is the weak link. I think a sensorless vector drive will make the difference. I just need re-assurance! I have always been told that having a larger kw drive will improve decel ability...I don't know if that's true or if it's drive dependent. That's the reason we are considering oversizing the drive to a 7.5hp.

Original drive:
"SECO 6150 QUAD R"
0-180VDC ARM, 15A (via jumper setting...drive can put out 25A)
Original Motor FLA 14.5A

Current drive:
"ALLEN-BRADLEY CAT 160-BA10NSF1P1"
10.5 amp 460 v
Current Motor FLA 6.1A (design B)

Thanks
Paul C.

 

[DickDV]

OkiePC, I'm especially concerned that the DC and AC motor are of the same base speed. That way I know that the motors will develop the same torque levels. I am going to assume that this is true to make this go faster.

A 3hp 1750rpm motor develops 9 ft-lbs torque continuously and, with a 25amp drive will output at least 9 x 25/15 = 15 ft-lbs peak (that's assuming the 25amp drive rating is peak). The AC motor FLA seems too low for a 5hp but, assuming its ok, and using the 10,5amp AC drive rating as peak, the continuous output torque would be 15 ft-lbs and the peak would be 15 x 10.5/6.1 = 25.8 ft-lbs. Clearly, the motor is not the limiting factor in the new system.

Next, we look at the drive. Due to the V/Hz design and especially AB's tendency to use incredibly slow processors, the drive is not able to rapidly detect end of cycle and get the motor braking torque developed quickly. As a result, when the processor finally gets around to realizing that braking is needed, the braking level must be very high. That is where the High DC Bus faults are coming from even with the snubber brake package in place. I would expect that the decel distance is a little random as well for the same reason.

The 5% duty cycle rating on the snubber brake is simply saying that the resistor will reach its max temp limit if the brake is used at full capacity 5% of the time. Since your brake is not getting hot, I would not blame the duty cycle limit. It is important that the brake resistor be the lowest resistance permitted for the drive. Probably should check that.

My judgement here is that the drive is the culprit. I would change to a good fast sensorless vector drive and keep the same motor. I don't like saying this but, for your sake, AB is NOT one of those drives. Their sensorless vector products are at the low end of the scale performance-wise. If you must use AB, you will need to chose their flux vector drive and change the motor to a model with encoder provisions on the back side. Then add a 1024ppr encoder, an encoder option card in the drive (do you hear the cash register ringing?!!!), and add wiring from the drive to the motor for the encoder leads. You will have an excellent system with this package.

Or, pick a premium, fast, sensorless vector drive and that will do the job with the existing motor and no encoder. It's your choice.

Since I have been accused of brand favoritism here before, I will decline to offer any brand suggestions. If you want to talk about it privately, just contact me by email and I will oblige.

Having said all this, if my assumption about your motor choices is wrong, all bets are off and we will need to do this again!

 

[OkiePC]

DickDV...Thanks for the response.

Due to the V/Hz design and especially AB's tendency to use incredibly slow processors, the drive is not able to rapidly detect end of cycle and get the motor braking torque developed quickly. As a result, when the processor finally gets around to realizing that braking is needed, the braking level must be very high. That is where the High DC Bus faults are coming from even with the snubber brake package in place. I would expect that the decel distance is a little random as well for the same reason.!

This paragraph rings a bell with me. The part about slow processors is something that I always suspected just because of the way some of these drives have had random response times to a start input in my past experience. I was planning on using a PowerFlex40 drive which is sensorless vector capable and relativley cheap. We could go with a Reliance GV3000 also if I alter the panel layout some. It can operate in all three modes. I believe our motor already has mounting pad for an encoder. Getting an encoder and cabling to the drive should not be a big problem for us but, hopefully, we can do the job without it. Can we expect a great deal of improvement going from sensorless to encoder vector control?

Thanks Again
Paul C.

 

[DickDV]

The three slowest processor drives I have encountered are AB, Reliance, and Square D. The GV3000 is not a good choice unless, as I pointed out before, you go to flux vector control with an encoder.

In an AB or Reliance drive, you will get major improvements in performance between sensorless and flux vector. That is because the sensorless performance is not up to par. In your application, if you were to choose a fast sensorless drive, there would be no measureable performance difference between sensorless vector and flux vector.

I think its safe to say that flux vector control (encoder on motor) is going to be similar high performance across pretty much all brands. It's the sensorless vector performance that is highly variable.

 

[monkeyhead]

The AB 160 is pretty low performance. I wouldn't use it for a whole lot more than spinning fans or running a conveyor at constant speed.

Good call. Ever since i discovered the joy of scaling speeds with the analog signal follower mode on these puppies, seems like we burn 'em up pretty fast unless i give the ramp up/down times a good chunk of time.

 

[OkiePC]

Good call. Ever since i discovered the joy of scaling speeds with the analog signal follower mode on these puppies, seems like we burn 'em up pretty fast unless i give the ramp up/down times a good chunk of time.

We had 72 of these drives installed without shielded motor cabling, and they would fry themselves...I think the logic/CPUs were our main problem. We started adding shielded motor cabling and they began to live a lot longer. Another problem, was the original design had an INPUT safety contactor...big no no im my book...I moved the contactors to the output side of the drive and this helped a lot too.

I have used dozens of these things like a "poor man's servo". We already had analog position feedback, so I closed the loops in the PLC5s with an analog output going to the drives. It worked great, improving position accuracy (read product quality) and cycle rates tremendously.

OkiePC

 

[Alan Case]

I wish to amend my previous post to read a larger motor, not drive. Dont know what I waas thinking when I first replied but i have seen a lot of applications where a motor is sized correctly to accelerate a load but then the motor and drive are expected to decelerate the load in a fraction of the accel time/distance. Regards Alan

 

[OkiePC]

Follow Up

I replaced the allen bradley 160 series drive with a Powerflex 40 running in sensorless vector mode. I wanted to try another brand of drive, but our company has agreements with Rockwell to try to use their stuff exclusively. The performance improvement was remarkable. I only had two hours to remove the old drive and brake resistors and install the new drive (p/n 22B-D012N104) and resistor (p/n AK-R2-120P1K2) because I was interrupted twice. I never even had a chance to autotune it. Nonetheless, it stops and changes direction at least twice as quickly as the 160 series did. The overall machine cycle rate improved 8%. We have only had one fault (high DC bus) since installation about four weeks ago. It appears to have been a freak occurrence. The DC bus measurements look good under normal operating conditions. I am very pleased with the drive and plan to use it for other problem applications soon.

Paul C.