Paul & Sabrina's cheap DIY 144v motor controller

[jyanof]

Originally Posted by princeton

They are actually 200V zener diodes that act as flybacks when drain voltage goes less than source (ground) but also act as voltage limiters to protect my 300V mosfets.

No there are no input capacitors. I will probably place some input capacitors across the battery terminals. I didn't see a need for them taking up space in my controller. The only real purpose I can see they serve is to protect the battery from rapid discharge current and prolong its life. They are not absolutely necessary in all systems, in my opinion.

Yes, the mosfets are holding up the heavy gauge copper wire but I intend to put some insulated bracing to hold the copper wire to help prevent the leads from being broken off.

There have been several responses about the diodes. I particularly like Paul's! Also note that diodes don't share current well unless they're thermally coupled - that is, one diode isn't allowed to get hotter than another (usually by placing them all near each other on the same heatsink). If one gets hotter than the rest, it hogs more current, gets even hotter, hogs even more current... thermal runaway ensues, and poof.

As for capacitors, they serve a couple of purposes. One, is that they help clamp the inductive voltage spikes from the motor. When the fet turns off, the voltage on the drain will rise due the motor inductance. When it rises above B+, the diode begins to conduct. At this point, the drain pin is essentially connected to B+ and the cap can help clamp the voltage spike.

Secondly, it smooths the supply current from the battery and the voltage spikes associated with it. Take the circuit as it is now without input caps. Let's say the controller is operating at 500A and maybe 50% duty cycle. When the mosfets turn on, 500A is drawn straight from the battery through the battery leads. When it's off, the current from the battery is zero (though the motor current keeps going due to the motor inductance). So, there will be this drastically changing current in the battery and it's wiring. (this is roughly what is called 'ripple current', btw).

The problem is that batteries don't like high current pulses - they'll likely heat internally and die early. Also, all wiring has parasitic inductance and that 500A won't stop instantly (see oprah) and'll cause (potentially destructive) ringing on the mosfets, and/or cause a ridiculous amount of EM noise (that long battery cable is like a big antenna!)

Adding capacitors will distribute the ripple current between the caps and the batteries, mitigate the parasitic inductance of the wiring, and bring a current source close to the switching section in order to minimize the parasitic inductance.

That's not to say the controller won't work as is, but it raises the question of how well it will work. If you find it doesn't meet the design goals, these are some possible options to look at when revising the design.

[princeton]

I'm not sure if I've convinced anyone but my point was that the capacitors across the battery terminals do very little to protect the mosfets. They do reduce battery fatigue, so I guess they are useful in that respect. Please correct me if anyone said anything to contradict this.

I do see a need for capacitors to protect the mosfets, just not across the battery terminals. I have a different configuration in mind.

I hope I'm not misunderstanding something.

[princeton]

I would like to correct my post that the Zener diodes in my circuit would act as flybacks. As I have configured them, they act only to limit drain voltage to 200V or less. For a diode to act as a traditional flyback, it would need to be connected from drain to positive supply.

[thecarfarmer]

Originally Posted by MPaulHolmes

Princeton, say 500 amps is going through the mosfets. Now, let's say they INSTANTANEOUSLY turn off. 500 amps INSTANTANEOUSLY stops flowing through the mosfets and has to go somewhere. Current has momentum.
INSTANTANEOUSLY (well, almost instantaneously) starts flowing through the diodes. 500 amps through the diodes.

It's like trying to stop Oprah from running to a chocolate cake. She can ONLY be redirected. NOT stopped. So, she gets redirected through the diode instead. A "cookie" if you will. That keeps her happy. Then when the "mosfets" turn on again, meaning the refrigerator opens again, she starts racing toward her ultimate goal of the cake again. The cookies could only hold her off temporarily. In the mean time, the cookies have to take the full brunt of her snarfing.

Now, if there could be some way of adding a diarrhetic to the cookie, it could help Oprah keep off some of the weight during the times where there is a very low duty cycle (and thus excessive cookie snarfing).

Uh, are you related to Dr. Phil?

And do you think Oprah will 'weigh' in on the Ezell's Chicken conundrum? (Seattle topical humor re: a chain of chicken establishments loved by Ms. W)

-Bill

[princeton]

I wanted to show some waveforms of my circuit driving a 1 ohm semi-inductive load (12 volt car battery) @ 12 amps. I'm using a 1 ohm, 3000W resistive coil. I apologize for the blurry pictures - I'll try to edit the post with better pics later. O-scope set to 20v/div.

The first picture shows drain voltage without a flyback diode. Note the initial voltage spike (when the mosfet opens) reaching about 180 volts and significant ringing afterwards due to inductive feedback.

Second picture shows drain voltage WITH flyback diode in circuit. Note a reduction of peak voltage to 80 volts and reduction of ringing. When I attach the diode to the positive terminal, it sparks considerably, so it's drawing lots of current.

Third picture shows drain voltage with a flyback diode and 20 ohm resistor in series. Note peak voltage hitting 110V but less ringing after resistor added.

Fourth picture shows drain voltage using NO flyback diode but instead using 2200uF cap and 50ohm series resistor from drain to ground (B-). Note peak voltage hitting about 110V (similar to flyback diode/resistor combination) but with less ringing.

Last picture shows combination of flyback diode/resistor attached to positive battery terminal AND ALSO capacitor/resistor hooked to ground terminal. Note peak feedback voltage is now back to about 80V (similar to flyback diode alone) but ringing is almost completely gone.

What I have learned is that a flyback diode alone can be very ineffective at dissipating inductive feedback voltage. A flyback diode is best used in series with a resistor to reduce ringing, otherwise the power in the inductive field has no way of dissipating effectively. Also, a combination of capacitor/resistor and flyback diode/resistor seems most effective at dissipating inductive feedback and thus protecting the mosfets.

I don't intend any offense to those that have contributed extensively to the current circuit design. I have been impressed at the work and degree of sophistication that has been put into the design but I also believe it can be improved upon and made more cost-effective.

[MPaulHolmes]

a resistor in series with the freewheel diode? That's making oprah slog through a swamp in order to get to the cookie...

I think the reason the resistor is reducing the spikes is that it is forcing the current to change slower. But if you use really really good layout, the stray inductance will be small anyway, so the spikes will be small.

P.S.: Do NOT encourage that woman to sing, if you know what I mean... 2012 will come early.

[jackbauer]

The battery cables act like springs. They get stretched during on time then they "twang" back during flyback. The purpose of the dc link capacitors is to remove the peaks and troughs caused by this inductive action. This is why the connections from the link caps to the b+ and b- busbars must be as short and stout as possible. The controller then sees a much stronger voltage source.

[princeton]

I am having a difficult time having an educated discussion with thoughts of Oprah and singing women in my head.

[jyanof]

Originally Posted by princeton

What I have learned is that a flyback diode alone can be very ineffective at dissipating inductive feedback voltage.

The pictures provided show massive voltage spikes at a low current of 12A. The voltage spikes will get larger with increasing current (they're proportional). Even still, the smallest spike shown is 80V.

With Paul's design, we measured a voltage spike of 20V or so... at hundreds of amps!

You're correct that a flyback diode alone can be ineffective at clamping these voltage spikes. It must be implemented correctly with a proper layout that minimizes parasitic inductance.

Also, note that a coil resistor load will have significantly higher resistance and lower inductance than a real motor. So much less inductance that those spikes may be caused by the lack of input caps moreso than the inductance in the load. An air core inductor would be easy to make and put in series with the load, if you have some extra wire. The larger resistance will dampen the ringing more. Maybe use an old starter motor or something?

[princeton]

Originally Posted by jyanof

The pictures provided show massive voltage spikes at a low current of 12A. The voltage spikes will get larger with increasing current (they're proportional). Even still, the smallest spike shown is 80V.

With Paul's design, we measured a voltage spike of 20V or so... at hundreds of amps!

You're correct that a flyback diode alone can be ineffective at clamping these voltage spikes. It must be implemented correctly with a proper layout that minimizes parasitic inductance.

Also, note that a coil resistor load will have significantly higher resistance and lower inductance than a real motor. So much less inductance that those spikes may be caused by the lack of input caps moreso than the inductance in the load. An air core inductor would be easy to make and put in series with the load, if you have some extra wire. The larger resistance will dampen the ringing more. Maybe use an old starter motor or something?

I have tried my circuit both with and without capacitors across the battery terminals in different locations but I cannot find that they do anything to reduce inductive feedback or oscillations on the drain of the mosfets. As far as my testing goes, these capacitors do absolutely nothing to reduce inductive feedback on the drain. Is there somewhere on the internet that I can find before and after oscilloscope readings on the drain for a controller with and without these capacitors in place? At this point, I have to rely on my own testing and will call this myth busted.