Three Phase Brushless Motor Amplifier

[dareon]

I am working on designing a simple linear amplifier (current feedback) used for driving three phase brushless dc motors (for work). Currently each phase is controlled via a current cmd signal from an Aerotech controller. The DC motor windings are connected in a wye configuration. The amplifier works great if the center is connected to ground as well. However if the phases are just connected with no reference to ground the voltage outputs just end up controlling straight to the rail and getting stuck there and confusing the amplifier.

In the future I will be able to make the ground connection if need be but for testing at this point it will be rather difficult (not impossible). It would also be nice to have the added flexibility if it doesn't degrade the performance of the amplifier. I have attached a simplified schematic.

http://img820.imageshack.us/img820/9145/singlephase.jpg

I know its off topic but any suggestions would be greatly appreciated.

Thanks

Russell

 

[richie00boy]

Are you using linear for simplicity, or for some other reason such as power factor or EMC?

 

[dareon]

EMC we work with precision measuring equipment and we can't tolerate the switching.

 

[bob91343]

Trace the current path. If you have no neutral connection, the current through phase A must return through phases B and C. So your driving circuitry has to allow for that.

 

[benb]

I'm thinking add a "servo" feedback loop. Either take the Y center connection, or perhaps(!) equivalently, three equal resistors from the three phases to a common point, and have that go to an integrator with a time constant of several seconds (or lower than the lowest frequency this thing uses, but then if it's real 3-phase AC, the average should always be zero and it shouldn't matter). This output would then be summed with the CurrentCMD signal of each of the three phases. An op-amp integrator inverts, so this will give negative feedback, keeping the Y center connection at zero volts.

And of course this solution should be robustly tested to insure it never goes into oscillation, etc.

 

[dareon]

benb,

That sounds like it could work. I will give it try this morning and let you know.

 

[dareon]

The suggestion didn't work. I have tried on the motor and on just resistor in a wye network with basically the same results.

http://img217.imageshack.us/img217/3739/singlephasewyefeedback.jpg

Basically when the integrator figures out that the center of the psudo resistor WYE network has gone from zero it puts some output in an attempt to correct. But since all three channels get this signal summed at their current command input they all try to change their outputs to attempt to put out this new current. However since they are all changing in the same direction/rate the amplifier just rails because it is never able to make them put out any current.

Perhaps the problem is where the integrator feedback is going. Instead of summing it with the current command I could use U4 as the current feedback amplifier and then sum its output with the integrator out using U2. How does this sound?

http://img683.imageshack.us/img683/9774/singlephasewyevoltagefe.jpg

This is the schematic for the updated feedback.

Thanks for the comments

 

[benb]

Now I see my mistake - I set the integrator to sum with the CURRENT control signal, where it actually needs to control the VOLTAGE. Your second schematic puts the integrater's feedback inside the current feedback loop, at the point where the integrator DOES control the voltage of the output. I think that ought to work.

 

[dareon]

I hooked it up like that with no real luck. It kept going randomly unstable and I could not figure out why. I'm gonna have to call it good enough for now and just open up the drives and ground the common connection.

 

[benb]

Hmm, sorry to hear it doesn't work. I'd think this should be doable, but I'm completely out of ideas.

 

[dareon]

I think the reason the integrator feedback didn't work is that the circuit is inherently unstable when driving only the inductive load(regardless of the integrator). I discovered this after connecting the wye on a three phase motor to ground and attemping to drive it with this amplifier. Basically all of the phase get a large DC offset combined with with a high frequency offset. I tried a couple different solutions to fix this including a snubber resistor network at the output as well as a lpf filter in the feedback. The only way I have got the circuit working was with paralleling power resistors with the motor phases as shown in the attached circuit.

Do you have any suggestions on improving the stability of the circuit.

http://img130.imageshack.us/img130/8942/stabilityissues.jpg

Thank You

 

[audiomik]

Dareon

you will find that using current feedback will cause substantial phase issues in your feedback loop, this may well be the cause of your problems.

Having built a number of similar Amplifiers for 3 phase motors as you are wanting, I always used as minimal as possible voltage feedback from the Amplifier output with a very low open loop gain for the Amplifiers themselves. Also return each of the 3 motor windings to ground such that for each amplifier they appear as single loads (star connection).

Monitoring the current into each phase - use a decent 1:1 600 ohm audio line transformer with the primary across your 'shunt' resistor - and compare the Amplifier output voltage with the current sensing transformer output on a dual trace 'scope.
You could try also your parallel resistance and Zobel networks to help reduce the phase shifts in the current.
One other point is you probably need to fit diodes to conduct anything in terms of 'back-EMF' over your output supply voltage on the output back into the PSU rails.

Some sort of similar work on inductive loads on Amplifiers at:
Amplifier Damping Factor - Dynamic Testing - Speakerplans.com Forums - Page 1

Hope this assists and makes sense - getting late here!
Mik