DC motor PWM flyback diode

[Moonwalker031]

I have a simple circuit for controlling the speed of a 12V 10A dc motor using 1 mosfet on the low side. For cost reduction I opted for no heatsink and chose a mosfet with the lowest RDSon. This worked well with no much heating.
However I am getting a lot of heat in the flyback diode. I tested different types HER602, MUR820, MUR1620 but all of them heated up after a few minutes. Since all diedes have a forward drop of about 1V, I cant think of a way to solve this without adding a heatsink.

Any help please?

 

[Skogsgurra]

Your swiching frequency has a lot to do with diode losses. Modern PWM drives are now using switching frequencies in the 1.2 - 2 kHz range. Try reducing switching speed and see what that does to temperature.

The speed of the diode is also an important parameter. The built-in diode in the MOSFET is fast, compared to most other diodes.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Moonwalker031]

I am using 20kHz since from what I read a lower frequency would produce an audiable sound.
The diodes I tested have 35ns recovery time which are fast enough. In fact I tried a mosfet instead of the diode and that heated much faster.

I will try a lower frequency and see the results. Thanks.

 

[Skogsgurra]

Remember, you run without any heat sink. That means you have already left trodden roads and may need to deviate also from standard switching frequency to make the thing work properly.

Yes, low frequencies can produce audible noise. If you cannot tolerate that, you will probably have to surrender and use a heat sink.



Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Moonwalker031]

I did some testing at 4kHz and 1kHz with no noticable improvement.
I had to go down to 250Hz to get it working with no much heating at no load. However when loading motor the diode starts to heat again. Will have to test it at normal working load and see how it goes.

 

[OperaHouse]

If effort is to avoid heat sink try two or three diodes in parallel. Forward voltage increases with current, look at the graphs. Can't escape the heating effects of forward voltage and current.

 

[Moonwalker031]

The diodes I am using are not cheap and I think I need 3 or 4 to minimize the heat. I may go for a heatsink with same price and better current handling.

How much current passes through the flyback diode? If the motor takes 10A how much will the diode take back - continuous and pulsed? If I can work with 2A diodes then I may consider paralleling otherwise I have to opt for heatsink.

Another question - I dont have any current monitoring at the moment - just a fuse. The mosfets can handle 100A and the fuse is 15A so they should be safe enough right? But if I monitor the mosfet/diode temperature, will that have the same effect as monitoring the current - temp/current should be proportional right?

 

[benta]

Try with a Schottky diode instead, it's faster and has lower forward drop.

Benta.

 

[Moonwalker031]

Nice idea benta. With schottky diode the forward drop goes below 0.5V while with the the other diode it was almost 1V. And viewing the waveforms there was no difference on the switching edges.
So the question is - if schottky diode is better than the Ultra fast diode why do the latter exist? Or when should one be used or the other?
Still need to know how big diode should be for a 10A motor please.

 

[OperaHouse]

Look at maximum voltage and leakage as they get hot. Schottky diodes are far from ideal. Everything has its place.

 

[benta]

Yeah, no free lunch.

Upsides to Schottkys:
Low forward voltage
Nice reverse recovery characteristics (capacitive, no stored charge)

Downsides:
Leakage current (not a problem in a 10 A application
Reverse voltage normally below 100 V, unless you start using exotic materials like SiC.

Cheers,

Benta.

 

[Moonwalker031]

Thanks again. Infact schottky is rated at 30V 15A and uf diode is rated at 200V 16A and the price is comparable. But for 12V operation high current the schottky wins big time.

Back to the motor flyback current. With the scope connected across the motor at 20kHz I have almost perfect square wave with a few ns ringing at the edges. On the OFF period I have a negative 0.4V which means that the diode is conducting. How can I determine the current passing through the diode please?
At 250Hz the wave form becomes a little complicated to explain.

 

[OperaHouse]

I have a little current transformer I pulled out of some power supply that has a hole in it to pass a wire through. Works great as long as switching speed is over 2 KHZ. These can even be found in some old switching power supplies with a single piece of wire built in. Calibrated mine with a power amplifier and series resistor. Just keep loading coil down till it is some even value. Its no Pierson but better than paying $200 for a used one on ebay.

 

[Skogsgurra]

You could also use a low ohms resistor between diode and gnd and measure the voltage across it. Use something like 10 milliohms if you have DC currents around 10 A. That will give you 100 mV at 10 A. Plenty of signal for any scope.

If you cannot find a resistor, you can make one from a piece of wire. Make sure the wire is heavy gauge so it doesn't heat and change resistance. Wire inductivity is not a problem at the switching frequencies you are dealing with. If you want to reduce inductivity you can fold the wire and twist it. That reduces inductivity at least ten times. For better results, use a coaxial shunt or a high speed DC clamp.

Gunnar Englund

http://www.gke.org

--------------------------------------
100 % recycled posting: Electrons, ideas, finger-tips have been used over and over again...

 

[Moonwalker031]

Lol that was elementary
Will get a resistor and have a try.

 

[OperaHouse]

Certainly a resistor is the easiest method to measure current. Many meters have trouble measuring pulses so a scope is the best option. Care must be taken not to introduce added inductance and resistance which can change performance. I've been scarfing up discarded Li Ion battery packs to replace nicads in equipment. They all have current monitoring electronics and a cheap source of low ohm SMT resistors. These are ideal for current sense because they can take a couple amps and size reduces the possibility of lead pickup. The picture shows a .05 ohm SMT soldered directly to a BNC connector. Just three inches of exposed scope lead could pick up a lot of noise. This method gives a clean signal and the ESL of the SMT resistor is typically only about 0.5nH.