Paul & Sabrina's cheap DIY 144v motor controller

[princeton]

How is your explanation going to help me reduce inductive feedback voltages hitting the drain of my mosfets?

[princeton]

I don't guess I've explained my position very well. If you would be willing to allow me to explain. From your circuit diagram, it appears you are using diode STTH6002C for the flyback diode. Each is rated at 60 amps and you are using 16 of these total, for a current carrying capacity of 960 amps through the diodes. I may be stretching a bit but you may be thinking that using a capacitor in in parallel with the battery lowers the serial resistance of the battery and allows it to more quickly absorb the transient surges that go from drain, through the flyback diode and then to the positive of the battery. I agree, in theory that this sounds like a reasonable assumption to make. With my testing, I am unfortunately unable to reproduce this symbiotic relationship between flyback diode and capacitors across the battery. As you have stated, you are only receiving 20 volt peak spikes on your mosfet drains @ 500 amps by the use of your configuration. I am unable to reproduce such small voltage spikes on my mosfet drains by using a configuration similar to yours. The spikes that I am seeing are much larger and very transient in nature. I am concerned that, depending on your oscilloscope or other test equipment you are using, you may have missed these transient higher voltage spikes. With your configuration, spikes that rise above V+ should be shunted to V+ by the flyback diodes. Looking at the datasheet for your diodes, if I read it right, the forward recovery time is 350ns and reverse recovery is 40ns max. I am using 20khz clock speed, so each mosfet pulse lasts 1/20000=50 microseconds or 50000 ns. That means the diode can turn on and off a maximum of 125 times per cycle. The diode I used was slower than yours, so our results will vary. If you look at the o-scope traces I showed earlier, you will see that the initial spike is indeed about 1/100th of a cycle and the oscillations are close to this frequency also. This diode recovery time is one possible contributor for the transient spikes and also the oscillation/ringing that I was seeing. In order to dampen the oscillations, I have proposed a 3-pronged approach including voltage-limiting zener diodes, flyback diodes in series with resistors and also adding a capacitor with series resistor from drain to ground (v-). I acknowledge that I will need further testing to obtain the best possible combination of these and to determine what values of capacitors, resistors and diodes that will work best. It sounds like from your post that you are convinced that the simple flyback diode system you are using is the best overall. After some more testing, I should be able to determine which system is most cost-effective and most protective of the mosfets.

[MPaulHolmes]

Sorry, I was away getting pinewood derby supplies for my 3 year old. "He's" going to enter the competition on Nov. 20th. hahaha.

I'm using 10 stth6002c diodes.

If you use a diode and resistor in series, the mosfet drain voltage will pulse MUCH higher than if you don't have a resistor in series with the diode, assuming paths are already at a minimum length. I also simulated that, but I haven't posted it. With a resistor it was 300v at the mosfet, and without the resistor it was about 140v at the mosfet.

I think your layout is the reason you are observing different results. I think you should have each diode/capacitor/mosfet loop as small as possible, because that loop has the fastest changing current on it. Everything else is a slop fest (meaning the other stuff can be relatively long).

[wakinyantanka]

Originally Posted by MPaulHolmes

Sorry, I was away getting pinewood derby supplies for my 3 year old. "He's" going to enter the competition on Nov. 20th. hahaha.

I'm using 10 stth6002c diodes.

If you use a diode and resistor in series, the mosfet drain voltage will pulse MUCH higher than if you don't have a resistor in series with the diode, assuming paths are already at a minimum length. I also simulated that, but I haven't posted it. With a resistor it was 300v at the mosfet, and without the resistor it was about 140v at the mosfet.

I think your layout is the reason you are observing different results. I think you should have each diode/capacitor/mosfet loop as small as possible, because that loop has the fastest changing current on it. Everything else is a slop fest.

Hey Paul,
Your knowledge on this is simply astounding.
If may pick your brain a little, how does the SR controller handle these spikes and singing?
Thanks for all you do here.

[princeton]

Originally Posted by MPaulHolmes

Sorry, I was away getting pinewood derby supplies for my 3 year old. "He's" going to enter the competition on Nov. 20th. hahaha.

I'm using 10 stth6002c diodes.

If you use a diode and resistor in series, the mosfet drain voltage will pulse MUCH higher than if you don't have a resistor in series with the diode, assuming paths are already at a minimum length. I also simulated that, but I haven't posted it. With a resistor it was 300v at the mosfet, and without the resistor it was about 140v at the mosfet.

I think your layout is the reason you are observing different results. I think you should have each diode/capacitor/mosfet loop as small as possible, because that loop has the fastest changing current on it. Everything else is a slop fest (meaning the other stuff can be relatively long).

I realize that by using a resistor and diode in series, the voltages will be higher. How much higher is dependent upon the size of the resistor, of course. This is the reason I have been experimenting with the combination of zeners, capacitors and diodes. If I could duplicate your results and obtain waveforms with limited peak voltages and also control excessive ringing with the conventional flyback design, I would simply opt to use flyback diodes alone. I will try shortening some of the connections again but I can't commit to a design when I can't reproduce the results and feel safe my mosfets will be protected.

[MPaulHolmes]

Hello waktidlenayaandanadingdong! The usual goal of SR is to replace the diode with an "ideal diode". One that has a near 0v drop instead of like 1.2v drop of a standard diode. However, I don't care about that at all. I just wanted a TO-247 or TO-263 package "diode" that has the same 230amp rating as the mosfets. Such a diode does not exist. When life hands you lemons, you make lemonade! So, you can get rid of the diodes, and use 230amp mosfets. Hurray! You just have to make sure you don't have the freewheel mosfet and regular mosfet on at the same time, which is pretty easy if you are using a microcontroller with complementary pwm outputs. Heck, an AC controller has 6 outputs. 3 pairs of complementary outputs, so it can definitely be done safely. I don't know how the spikes will be. I think it will be pretty similar I'm thinking for similar current. That's one reason it might only be safe to run 1000amp at 120v. Spike of 20v at "hundreds" of amps, might turn into spike of 40 or 50 at "near 1000" amps, 50 + 144v = REAL DANG CLOSE TO 200V. hahaha.

[wakinyantanka]

Originally Posted by MPaulHolmes

Hello waktidlenayaandanadingdong! The usual goal of SR is to replace the diode with an "ideal diode". One that has a near 0v drop instead of like 1.2v drop of a standard diode. However, I don't care about that at all. I just wanted a TO-247 or TO-263 package "diode" that has the same 230amp rating as the mosfets. Such a diode does not exist. When life hands you lemons, you make lemonade! So, you can get rid of the diodes, and use 230amp mosfets. Hurray! You just have to make sure you don't have the freewheel mosfet and regular mosfet on at the same time, which is pretty easy if you are using a microcontroller with complementary pwm outputs. Heck, an AC controller has 6 outputs. 3 pairs of complementary outputs, so it can definitely be done safely. I don't know how the spikes will be. I think it will be pretty similar I'm thinking for similar current. That's one reason it might only be safe to run 1000amp at 120v. Spike of 20v at "hundreds" of amps, might turn into spike of 40 or 50 at "near 1000" amps, 50 + 144v = REAL DANG CLOSE TO 200V. hahaha.

So the SR microcontroller will have pairs of pwm outputs for driving multiple gate drivers? How many gate drivers could it handle and still be able to carry out the SR function? The 200v mosfets your using for this build, could they be upgraded to handle higher voltages? Let's say in the 160v range?
Thanks Again.

[MPaulHolmes]

Well, I'm editing the AC board right now (I started last night). The plan is to have perhaps 3 drivers for the "mosfets", and 3 drivers for the "freewheel mosfets". 3 drivers input with be from 1 optocoupler output. The driver inputs only use 10uA each. haha. Isn't that tiny?!

You could use higher voltage mosfets. Definitely. They will have higher RdsON, but whatever, it's all good. I just have 25 of the 200v ones right now.

[wakinyantanka]

Originally Posted by MPaulHolmes

Well, I'm editing the AC board right now (I started last night). The plan is to have perhaps 3 drivers for the "mosfets", and 3 drivers for the "freewheel mosfets". 3 drivers input with be from 1 optocoupler output. The driver inputs only use 10uA each. haha. Isn't that tiny?!

You could use higher voltage mosfets. Definitely. They will have higher RdsON, but whatever, it's all good. I just have 25 of the 200v ones right now.

So this current design could handle a 144v nominal ? Which is what I'm shooting for.

[MPaulHolmes]

Hopefully. Ain't nothing like actually trying it. What I think will happen is, the batteries sag right when the current goes up, so it probably won't get close to 200v. The spikes increase when the current increases, and the overall battery voltage drops, so everything is perfect!