H Bridges for 2 directoin torque control

[MedievalMan]

For my application, I desire continuous dc motor torque control - the command torque can go from -'ve to +'ve quickly.

I made a previous post here regarding bidirectional motor control - with some good suggestions- basically, I'm having a tough time measuring the load current in the h-bridge, during +'ve to -'ve transitions (though I can have it run in either direction easily).

Besides providing electrical isolation, I don't see how the LEM or similar current sensors would resolve this problem (though I do have a few on order )

I've found h-bridges that have integrated current sensors using a few of the power transistor cells (like the LMD18200 by National SemiConductor).

http://www.national.com/ds/LM/LMD18200.pdf

However, when I read the application notes

http://web.mit.edu/rec/datasheets/AN-694.pdf

more carefully, I realize that they can only sense the current in the "forward" direction (the direction of motor operation), and if used with the "locked antiphase PWM control" strategy, the sense current is also discontinuous (only when PWM is ON).


I see SGS Thompson has a patent for a Bidirectional load current sense circuit for an H-bridge (

http://www.freepatentsonline.com/5644484.html),

but I can't find a product of theirs that uses it: I'm not sure if it is similar to the one I mentioned above.

Any suggestions on how I might solve this problem would be much appreciated

-Matt

 

[itsmoked]

Suggestions.
Have you looked at Maxim's forest of current sense amplifiers?

I have used them many times with good results.
They have bi-directional and unidirectional units.

http://para.maxim-ic.com/cache/en/results/30116.html

Are you running a single supply and using the H-Bridge to turn things around?

If you are then just use one of the amps in the single supply for current magnitude and you already know the direction. If you are using two supplies put one in each read the one of interest.

Put a filter on their outputs to ditch your PWM frequency and read the average current. I can't see why you would want instantaneous current but if you do you need to link the reading with your PWM to read while it's on.

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[MikeHalloran]

I think you could use a Hall Effect current sensor on one of the motor wires. I think you get an offset analog signal, e.g. 2.5V + [ k * I (signed multiply) ]. For a little extra, they come calibrated (k known).





Mike Halloran
Pembroke Pines, FL, USA

 

[blacksea]

Becide current feedback, you need to implemnt close current loop with PI-filter. See such structure of current mode amp from AMC (

http://www.a-m-c.com)

or Elmo (

http://www.elmomc.com).

 

[MedievalMan]

Thanks for the suggestions guys.

Keith, I am using a single supply and the h-bridge to turn the direction.

"If you are then just use one of the amps in the single supply for current magnitude and you already know the direction. If you are using two supplies put one in each read the one of interest."

This was my first idea (seemed the easiest way)... however, I found it didn't work when the command signal went from +'ve to -'ve (and then assuming the motor current does the same).. perhaps I should add a delay during this time to allow the motor current to switch direction in accordance with the applied voltage (b/c of motor inductance)?

Thanks again

-Matt

 

[itsmoked]

Can you be specific by why, 'it didn't work when signal went from +ve to -ve'?

You mean in one direction you get no reading at all?
Or immediately after either reversal the reading is gone briefly? OR only immediately after only a -ve reversal?

What, what? Inquiring minds want to know!

Keith Cress
Flamin Systems, Inc.-

http://www.flaminsystems.com

 

[sreid]

Knowing where the curent is flowing in a PWM H-Bridge is straight forward but can be tricky. A LEM Hall Effect Current sensor in one of the motor leads guarantees that you know what the motor current is at all times. To be specific, if one wants 4 quadrant torque control, you can have forward driving current with the motor turning backwards.

Blacksea is correct about the need for a current control loop. Send me an email at sreid at dac-intl dot com and I'll send you a paper on an analog implementation.

 

[cswilson]

I'll throw in my two cents here, basically agreeing with most of the above comments, but with some refinements thrown in.

If you are operating at a bus voltage low enough to fly under the UL and CE-low-voltage radar (~60VDC or less), I think you can use resistive sensing without isolation. For a simple H-bridge, you can use a resistor at the base of each side, making sure each one passes any current flowing "upward" through the lower flyback diode on that side. The difference between the voltages across the two resistors is proportional to motor current, and it is a signed quantity.

You can also put the sense resistor in series with the motor between the two half bridges -- I am doing this now with a 3-phase design. This is a little trickier because neither end of the resistor is at a fixed voltage reference.

For higher bus voltages you really want an isolated current-sense circuit. The nice thing about the magnetic sensors such as LEMs is they provide this isolation automatically. These are automatically bidirectional as well. Of course, you can use these sensors at low bus voltages as well, generally at somewhat higher cost than resistive sensing. Unlike resistors, they add virtually no losses to the power circuit.

In either case, with PWM control, you will have ripple in your sensed current waveform. A little filtering in the processing circuit takes this out to where it manageable for closing a current loop. Since torque is proportional to current, if you want decent torque control, you will want a current loop.

Curt Wilson
Delta Tau Data Systems