High voltage brushed motor control problem - back EMF

[George]

I'm designing a drive to control a Brushed DC treadmill motor. The drive is going to be current controlled PWM via comparitor in PIC.

I have a 600V 70A IGBT that I plan using to switch the motor (PWM), however I have some issue figuring out how I can protect it from back EMF.

The supply is going to be just over 300Vdc and I'll need to supply pulses of upto 15A to the motor, so back EMF will probably be several thousand volts.

I would ordinarily just put a diode across the motor, however in this application the motor can be driven backwards, and the torque the motor is providing needs to remain constant, A diode across it will not allow this to happen.

So far my thoughts are to use a gas discharge arrester across the motor (not sure will be fast enough), switch both sides of the motor and use diodes to throw the EMF back to the supply, or actively clamp the emf with a seperate FET or IGBT that triggers when the other turns off (I suppose this could be automatically done with a zener diode onto the gate to the negative side of the motor)

I'd rather stay away from switching the high side if I can,

any thoughts ideas or even bashing my ideas are welcomed as I'm a little stuck and only half know what I'm doing.

Thanks

[mackrackit]

Look at the pdf on page #3. Replace the DPDT with a relay for direction. Replace the SPST with your IGBT for speed.


http://www.picbasic.co.uk/forum/atta...4&d=1228220843

[George]

Thanks, but I'm not sure that's going to solve my problem. Electrically the motor is going to be only driven in one direction, however it will occasionally be forced backwards mechanically, and I dont want the freewheeling diode to get in the way and provide increased torque when motor is being 'overrun'.

So how to cope with the back EMF without a freewheeling diode is my question - at the moment I'm thinking of using 2 x 200V zener diodes onto the gate of a FET that's wired across the motor, when the voltage gets reverse above 400V, the FET switches on and quenches it out, the potential then changes and the charge on the FET gate should drain back through the zeners turning the FET off quickly.

The IGBT is good for 600V so the 400V FET trigger should be enough margin?

I'm hoping at the moment the tail on the turnoff of the IGBT will limit the back EMF to managable levels.

[mackrackit]

Sorry I misunderstood your first post. I was thinking the motor would be under power when running in reverse.

I do not know of any good way to not use a Fly-back diode, I never noticed much resistance using one either.

[amgen]

Hi,
is there any driving current to the motor when it is driven backwords by the mechanism ?
If not, the motor will develop voltage in reverse of the drive, but unless there is some load resistance , I don't think the motor will present any torque.. ?

don
amgen

[HenrikOlsson]

Don,
The problem, as I understand is it that if he places a flyback diode across the motor (or across the switch) IT will be the load when the motor is backdriven - in one direction but not in the other. As soon as the motor is backdriven fast enough to produce an EMF larger than Vf of the diode it will get forward biased and start conducting current.

George,
Crude solution but anyway: If the motor is only backdriven when power is off can't you put the diode in series with a relay that is activated when the system is powered. That way the diode is "in circuit" when the power is on and disconnected when the power is off.

/Henrik.

[George]

Exactly Henrik, There is going to be power supplied to the motor when it's driven backwards (overrun) so unfortunately the relay idea wont work. It's also going to be run it at high freq PWM so would guess the IGBT would see a few thousand plus volt spikes before the relay made contact on it's first switching.

My idea of using a Fet to quench it out is not looking that hopeful looking at some datasheets, the Vgs is only around +-30V on some high voltage FETs. I suppose tho I could use some more complex circuitry to activate the FET

Gas discharge arrestors look like they might be too slow - their spark over voltage dramatically increases with dv/dt speed.

Maybe an LC quench between IGBT and Motor?

Thanks for your input so far guys

[George]

BTW my dv/dt rise time for my motor back emf is around 1500V/us - also the motor is geared so when mechanically overrun is generating some 200V in reverse direction so that's well over the Vf of the diode and causes significant drag - pretty much locks up!

[mtripoli]

This is interesting... I worked at Powerex years ago doing "smart" power module design. Normally we'd use a bridge to drive motors so I can't think of a way out of this one... One thing comes to mind; clutch the gear train so it can't drive backwards...(sorry, had to throw that out there).

[Sneaky-geek]

Hi,
Just my 2C. Would not TVS work? I have used them in series to increase trigger voltage.
Some models can switch in just a few nanoseconds. Nice thing is they can also dissipate a few KW. Cheap TOO. Check out Digikey and the other usual suspects.

Cheers,

Terry

[PMController]

I think Speed controller is sense rotor's position based on back-EMF to manage acceleration, control speed.

[Acetronics2]

Hi, George

For relays, I sometimes use a RC series ladder to kill the back EMF ... but never tried it with such voltage and power levels.

also seen use of 2 Power Avalanche controlled diodes ( connected in series by anodes or cathodes ) in the high power welding equipment ...

Alain

[George]

Thanks for the comments, I'm not sure a TVS would do the job - unless it was huge. It would be dissipating a lot of power near enough to continuously? Arent they only made for transient voltages?

What do you guys think of an SCR triggered by a Zener?

I'm unfamiliar with them although I have set a pic to trigger triacs before for light dimming. I guess a Triac is jst 2 SCRs? So rather than dissipating 350-400V like a TVS would be doing it would only dissipate around 2V?

My understanding of them is when they see a voltage, they latch. They then unlatch when the voltage across them returns through zero.

So the main IGBT will turn on, charge up the motor to set flux (full flux is 10ms) then turn off for a set period. The SCR gate would be set with a zener - probably a standard diode also to prevent main power going through it. Then EMF spike rises, latches SCR when it reaches Zener voltage, this bypasses the voltage back upto the top rail. Voltage then drops and SCR unlatches?

Not sure - confirmation would be nice.

Thanks

[aratti]

Exactly Henrik, There is going to be power supplied to the motor when it's driven backwards (overrun) so unfortunately the relay idea wont work. It's also going to be run it at high freq PWM so would guess the IGBT would see a few thousand plus volt spikes before the relay made contact on it's first switching.

George, I think Herink's suggestion is still the best one!

You can activate the relay via a dedicated pic port (hope you have a spare one) so you can control the timing. Activate the relay, wait few millisecs and then fire your IGBT via pwm.

Reverse the concept when you will stop the motor: set pwm duty cycle to zero, wait a while till the motor stops turning, then open the relay.

All very easy to implement using PBP.

Al.

[George]

I've tried lots of things so far, to a point I have it working. But it's not working as smoothly or as nice as I would like, so I dont think there is anyway out of driving the high side also. - but how???

I'm planning on using the same NPN IGBT and driver on the top side as I do on the low side, this should keep switching time consistant. (Also I have a tonne of 600V 70A IGBTs)

The question is, what's the best way to generate 12 or so volts above my 300V rail, also - what's the best way to trigger the drivers with such a huge potential across them.

It all has to be done quite quickly and presisely, for eg at the moment I'm using an opamp for current limiting, however the slew rate is far too lesuirly. It's taking 5us to respond, from what I've observed that's going to have to be cut right down a hundred fold to 50ns or so. So an optocoupler is just not an option (25uS typical fall time)

Any ideas on large potential high side driving?

[HenrikOlsson]

Have a look at the IRS21884, logic-level input, adjustable deadtime to avoid shoot-thru, shutdown input for cycle by cycle current limit or whatever as well as the bootstrap circuit for the high-side driver.

If you're not intending to use PWM (which is needed for the bootstrap circuit to operate properly) you'll have to provide an isolated 10-15V supply for the high-side, conneted between Vs and Vb. But IIRC you are using PWM, correct?

/Henrik.

[George]

thanks Henrik I couldn't find an IRS21884 but I found an IR2181 which appears to do a similar job. However I cant get it working! I've only used low side FET drivers before and had no trouble.

The max Vin current according to datasheet is 60uA, I have 12V coming in through a 1k resistor (I have a 5.1v zener there to prevent over voltage), the IR2181 is pulling the input down to 2 volts (have checked it's not the zener) - so that means about a 10mA draw, the output to the FET is also floating around a couple of volts.

The FETs, the IR2181 and my logic circuit all share common ground.

The FETs currently have a 24V supply and the logic circuit and IR2181 have a 12V supply.

I've tried a second IR2181 incase the first was faulty but no luck.

Any Ideas? The circuit I've used is the same as documented in datasheet. I'm using a .47uF polyester bootstrap capacitor (I initially thought it had to be high voltage but now realise it only has to be 20V or so). Also got an ultra high speed diode charging it..

Has anyone here used these before and had similar trouble, or first time success?

Thanks

Thanks

[George]

Sorry - I'm an idiot trying to be an electronics design engineer!!

I looked carefully at the datasheet and have the pinouts all wrong I followed the circuit on the top of the page assuming it was layed out in pinout form.

Will let you all know how it goes.

Thanks

[HenrikOlsson]

Hi George,
Sorry about that, typo on behalf...The correct part number is IRS21844, here a link to the datasheet.

The 2181 should work as well but it differs from the 21844 in that it has separate drive inputs for the high- and low side so you need to generate the dead-time externally in order to prevent shoot thru. It has no cross conduction protection so IF you accidently get both inputs HIGH you WILL get into trouble. Another difference is that it lacks the SD-input which is handy to use for cycle-by-cycle current limit.

Let's see how you get along after sorting out the pin numbers. One thing I do wonder about though is that 1k resistor you mention. You really need to have a low impedance low inductance supply line to the chip as it drives the MOSFET gates with fairly high currents which has to be pulled from that +12V bus you say you have 1k resistor in....

12V supply straight into the chip with local low ESR bypass caps as close to the chip as possible.

Let us know how it goes and again, sorry about the typo on the partnumber.

/Henrik.

[George]

The 1K resistor was to the input of the fet driver, not the input of the FET, I used a 4.7 ohm resistor to input of FET, Apparently putting a resistor in line helps with ringing prevention.

To all those that are interested in the outcome - I dont know yet - however it's working well enough on low voltage, I havent had the guts or time to plug it into 240V yet.

The solution was to drive the motor top and bottom it was somthing I didn't want to do, however the FET drivers make things pretty easy these days driving high sides. The only issue I seem to have now is when the motor gets overdriven the voltage on the bus caps shoot up and will blow them if I let it, so I have to implement an active resistor system, or somehow try and feed it back into the mains.


I believe the newer Fisher and Paykel washing machines feed power back into the supply when slowing the drum down from spin cycle so shouldnt be too expensive to implement - maybe a little complex for me tho.

I also need to learn about quenching big spikes, when drive is under significant load transients are all over the place causing issues.

Thanks to all that took the time to respond