Paul & Sabrina's cheap DIY 144v motor controller

[duncan]

Hi Paul

I have chickened out and decided to go with the standard bits for my new power board,

I have ordered the MOSFETs, Diodes, Caps and a new MOSFET driver from Digi-Key,

Where do I get some more of the isolating tape that goes under the MOSFETs and diodes??

[adamj12b]

Originally Posted by johnsiddle

Paul, was there a particular reason you used one mosfet driver instead of ten (apart from cost).
I have copied your power board using ten drivers and ten 130A 200V Mosfets and just had a spectacular blow up.
Also when your current sensor trips does it cut the motor drive and return as soon as the current drops or is there a delay or something? I think mine created a strong juddering effect just before blowing up.

More isn't always better. having multip MOSFET drivers increases the chance of them being out of sync from each other turning switches on off microseconds before or after each other. Paul used one MOSFET driver because that was all that was needed to drive all of the MOSFETs. The total gate charge of all ten MOSFETs is still less than limit on one driver.

the controller turns turn on off extremely quickly when current limiting. The motor actually makes like a groaning noise and switching on and off only a few hundred times a second. What you feel is probably individual MOSFET blowing up.

Adam

[johnsiddle]

Originally Posted by adamj12b

More isn't always better. having multip MOSFET drivers increases the chance of them being out of sync from each other turning switches on off microseconds before or after each other. Paul used one MOSFET driver because that was all that was needed to drive all of the MOSFETs. The total gate charge of all ten MOSFETs is still less than limit on one driver.

the controller turns turn on off extremely quickly when current limiting. The motor actually makes like a groaning noise and switching on and off only a few hundred times a second. What you feel is probably individual MOSFET blowing up.

Adam

Hi Adam.
Thanks for the reply.
Yes I was worried about multiple drivers switching at different times but thought maybe at the fairly slow switching speed it would not have too much of an effect. Someone had told me that each mosfet could require up to about 3.5amps to charge fully and with the 30ohm resistors they might not make it.

I will revert to the one, I had thought of connecting all the outputs of say three drivers together to combine the power capability... but maybe not.

I have decided to fit 75A 600V IGBT's instead of pure MOSFET's the info on the net seems to indicate that these are a better proposition.

On the current limiting front what do you think of the idea of instead of chopping the motor power zero and back up I over ride the throttle pot and drop it to say one third power with a slight delay on releasing. I feel it might be a little kinder on everything, it would still operate fast but a little smoother and I would still have some drive power.
I have already got a fail safe in that if the cct goes into over current for 1.6secs continuous (everything goes short and over current has no effect) then the main contactor is dropped out.

Regards John

[jyanof]

Originally Posted by johnsiddle

On the current limiting front what do you think of the idea of instead of chopping the motor power zero and back up I over ride the throttle pot and drop it to say one third power with a slight delay on releasing. I feel it might be a little kinder on everything, it would still operate fast but a little smoother and I would still have some drive power.
I have already got a fail safe in that if the cct goes into over current for 1.6secs continuous (everything goes short and over current has no effect) then the main contactor is dropped out.

It sounds like you're writing your own software... is that right? If so, is your control scheme 'motor current' based? As in, the throttle is interpreted as a commanded output current and control loop uses the current sensor feedback to adjust the PWM duty cycle to match this commanded current?

[mora]

Does anybody have precompiled software (.hex) for hall-effect throttle and higher switching frequency? (mosfet power board being used) I modded my controller for hall-effect throttle as per Jackbauer's instructions and Paul's custom software, but apparently that software is using 8kHz switching frequency as Jackbauer is using IGBTs instead of mosfets. Now my motor sounds like it is being driven by Cursit, err, Curtiss controller. Only difference being this controller is much better and whine doesn't go away as rpms build up. Luckily whining sound is not loud.

Maybe Paul could customize software for me too? Heheh. I'll email him just in case.

I got tired of jumpy starts. HE throttle made it all smooth again. Software update using RTD explorer was really smooth and easy too. About 20 000km done in two years using Revolt.

[johnsiddle]

Originally Posted by jyanof

It sounds like you're writing your own software... is that right? If so, is your control scheme 'motor current' based? As in, the throttle is interpreted as a commanded output current and control loop uses the current sensor feedback to adjust the PWM duty cycle to match this commanded current?

Hi Mate
You make it sound more complicated than it is.
It is not done in software but hard wired and totally autonomous.

Basically the PWM is generated by a pic chip that generate's PWM in response to a voltage from the throttle pot.

The current is sensed with a hall effect sensor as in the openrevo cougar.

When the current trips, the output from the sensor OP Amp is used to switch in a low value resistor across the output from the throttle pot so lowering the voltage fed to the pic chip. This will change the PWM duty cycle and so lower the current.
When the current sensor recovers the resistor across the throttle is removed and a smallish capacitor across the the input is allowed to charge up to the value set by the throttle. The current is thus ramped up fairly slowly to the value it was at the start unless the throttle has been eased off.

I thought this was kinder than chopping a very high current flow to zero and then immediately banging it back on possibly up to 100 or so times a sec.
I know this is exactly what the PWM generator is doing but very much faster.

I have a second timer cct that is allowed to fire if the output from the current sensor remains active for about 1.6secs. This trips my main contactor relay which remains tripped until I switch off and allow the timer to reset, again hardwired and autonomous. I figure if everything goes short circuit I wont get very far in 1.6secs even at full speed.

[jyanof]

Originally Posted by johnsiddle

Hi Mate
You make it sound more complicated than it is.
It is not done in software but hard wired and totally autonomous.

I don't think 'current control' is any more complicated than what you described. By the way, the open revolt software follows a 'current controlled' scheme.

If the software is tuned correctly, the open revolt will never trip the hardware overcurrent circuit. In fact, the hardware overcurrent circuit was put in mostly to prevent an errant software bug from blowing up the controller.

You could probably get a circuit like yours to work with enough time and spare silicon, but you might want to look at the open revolt software. Admittedly, there's a lot of extra stuff in there to handle things like various throttle sensitivities, overtemp protection, serial communication, EEprom settings storage, etc., but the throttle/output current control is pretty basic at its core. Feel free to ask questions...

Looks like Adam has an archive of the software on his website.

[johnsiddle]

Originally Posted by jyanof

I don't think 'current control' is any more complicated than what you described. By the way, the open revolt software follows a 'current controlled' scheme.

If the software is tuned correctly, the open revolt will never trip the hardware overcurrent circuit. In fact, the hardware overcurrent circuit was put in mostly to prevent an errant software bug from blowing up the controller.

You could probably get a circuit like yours to work with enough time and spare silicon, but you might want to look at the open revolt software. Admittedly, there's a lot of extra stuff in there to handle things like various throttle sensitivities, overtemp protection, serial communication, EEprom settings storage, etc., but the throttle/output current control is pretty basic at its core. Feel free to ask questions...

Looks like Adam has an archive of the software on his website.

Hi Mate.
Thanks for the info.
I wasn't saying current control was complicated I was saying that you made my system sound more complicated than it really is. Doing it in software is complicated to someone who struggles with software. I am a computer hardware engineer, software has always been a mystery to me. I like hardwired systems where a collection of bits have one dedicated job to do and are not dependent on a xtal clock running a big list. I know computers crash.

My system is about to get its last chance to work. If it blows up again I shall be getting an OpenRevolt system, so I hope it is listening...
cheers again john

[MPaulHolmes]

I just got done with a 1500 mile drive with my 1 year old in the back seat from Olympia to Maricopa. Boy that wasn't fun! We even ran over the big fat middle section of a dead deer. Rain and 60 in Washington, sun and 105 down here. haha.

You can definitely get multiple gate drivers to work. Ian Hooper (who is the reason I started working on controllers for cars in the first place) has made reliable controllers using up to 6 drivers and 12 mosfets. He didn't tie the outputs (pwm) of the drivers together though. He just had 2 mosfets per driver on a separate little driver board. Maybe the blowup was caused by something else? Those drivers have like a 15nS response time. Do you know at approximately what conditions (current of motor, current from batteries, pwm duty) the controller blew up?

mora, I'll try to find the .hex file. Man, I'm awful with knowing which version is which, and not keeping old versions. I tried SVN once, and wanted to stab myself in the eyeball with a fork afterwards.

[Nevyn]

Glad you're home safe and sound! I just finished re-reading this thread (yes, the WHOLE THING) and will hopefully be able to be somewhat helpful once again.

My version 1A of the 1kA IGBT board seems to be working. Other than backwards LED, that is. Man, I was soldering last night and forgot to fix that! Derp!

RTD connection works all good, except for the "restart" command - all it does is break communication! Such is life. Throttle input seems to work, current sensor seems to work (took me a while to read through the thread and find the pinouts for the 3-wire Melexis compared to the 4-pin header on my board!), so I think I'm ready to try and interface it with the old BG2A IGBT driver board!

Finally, Paul, one of your early testers is testing! I'm only what, 2 years behind Isaac_Alaska?