Paul & Sabrina's cheap DIY 144v motor controller

[Ranger4Free]

Originally Posted by harlequin2

Not sure if it warrants a new thread! Maybe I'll give it a go and stick some photos up as well, but basically the space for motors in the BMW was long and thin so a single, fat motor wouldn't fit whereas two long skinny ones, end to end, just went in nicely. And give about the same total power and torque.

There does seem to be a lot of ignorance about controllers here, so let me see if I can shed some light:
The controller is merely a high speed switch with a variable on-to-off ratio. At zero speed, the controller is on for a very short time, applying full battery voltage and the current limiter switches it off as soon as the current reaches the limit, 500A in this case. So you get a series of short pulses of high current. This means the average voltage across the motor is quite low and it slowly increases as the motor speed comes up. At full speed, the controller is on 100% and the motor back emf limits the current. But the point is, the full battery voltage is applied to the motor every time the controller switches on. You can sort of think of volts being proportional to power and amps being proportional to torque.
A series wound dc motor will absorb whatever you force through it until it melts or flys to bits!
The faster a motor turns, the higher the "back emf" it produces and so the more volts you have to apply to force current through it and make it turn faster. This means that your top speed is limited by how many volts you have available and at full speed the controller is switched on all the time, connecting the motor directly to the battery. In my case, two motors in series across a 144V battery "see" 72V each and this gives me a top speed of 110 kph. When I had them in parallel, each saw the full battery voltage and I got 130 kph.
Next, capacitors. You don't need much capacitance to absorb the inductive spikes generated when the controller switches off, what you do need is the lowest effective series resistance ESR that you can get because the current - which at the instant of turn off is the same as the motor current - flows through this ESR and hence generates a voltage. If you have 500A flowing, you only need an ohm to generate 500 volts!
Also, the ripple current capability of the caps is important. If you have only 10 caps, then each must carry 50A which is asking quite a lot. So, to choose capacitors, don't worry so much about the capacitance but look for low ESR and high ripple current capability. Of course, high C values generally give lower ESR and higher ripple current capabilty.

Hope this was not too boring and maybe helpful!

Could you open a thread on linking motors in series and parallel,the reason being is you say you've done it on a BMW conversion and its not just that i want to link two 72 volt motors from one controller,but maybe two 48volt motors and just for kicks i can get about sixteen 24volt motors as well!!!!my thinking with them was to drive each wheel separately....bearing in mind i weld my own chassis etc,so i could maybe work out how to lay the power out to the wheels from a mechanical standpoint....but if it turns out to be too hard to engineer ill just drive the rear wheels....this is why im so curious about the possibilities of driving multiple motors of one controller!!I'll re-read your post as it's 6.39am here at the moment and i've had no sleep for a long time.But would you consider doing that please...new thread of the how's,why's,wont's,dont's and can'ts....all i've had before hitting this thread is bull crap...your the only person who has said it can be done and who has done.....(it).....*please*......(-;

[AllTerrain9]

I like it!

[isaac_alaska]

Originally Posted by harlequin2

Really, you shouldn't even consider IGBTs unless you are planning on driving an induction motor. Or perhaps a brushless dc motor.
Series wound dc motors such as are commonly used in EV conversions (Netgain, Advanced DC, etc) don't need voltages higher than 150 - 170 to achieve full speed and 200V FETs are quite happy at those levels. You do need the higher voltages for induction motors though as the higher inductances mean higher back emf and hence more volts are needed to push the current through the windings.
FETs are generally easier to drive than IGBTs and they don't mind being paralleled as much either. They also have lower losses - usually - because their on resistance is less. Switching times are much the same, so switching losses are similar.

as stated before, the reason for going with igbt's and higher voltage would be for efficiency purposes. right now, from what i've read, it looks like the mosfets in most motor controllers are dumping around 50 watts of heat each. multiply this by 12 or so, and that's 600 watts of energy that's being used to warm your motor controller instead of drive your wheels (or track.) From everything i've read, all of the switching devices (mosfets and igbt's) are more efficient at lower currents. because of this, the plan is to run 3 igbt's in parallel (which, from the data sheets i've read, is completely possible if you follow strict procedure to make sure they are all mounted to the same heatsink and are of the NPT type.)

This will keep current through each igbt down to around 50 amps during normal "cruising" operation and keep Vce down to around a volt. lower Vce, lower conducting losses.

Paul, do you know where i can find the igbt driver schematic or board on here? i did a quick search and didn't find it right off.

[MPaulHolmes]

Hi Isaac! check private message.

[mrbigh]

Originally Posted by isaac_alaska

Paul, do you know where i can find the igbt driver schematic or board on here? i did a quick search and didn't find it right off.

Please, refer to this specific thread

[DJBecker]

Originally Posted by isaac_alaska

as stated before, the reason for going with igbt's and higher voltage would be for efficiency purposes. right now, from what i've read, it looks like the mosfets in most motor controllers are dumping around 50 watts of heat each. .

A MOSFET controller has the potential for greater efficiency.

A IGBT will always drop a volt or two. Yes, voltage drop is reduced a bit under load, but not by much.

A MOSFET can be modeled as a resistor, and you can parallel many more of them. If the device has an on resistance of 6.5 milliohm, and you are putting a steady state 20 amps through each, that's only a 0.13V drop, about 1/10 the power loss of the IGBT design. Neglecting the switching losses (which will still be lower than for an IGBT), that's only 2.6 watts per device.

Yes, that's a unfair comparison. It omits many other efficiency effects. And the situation can reverse at peak loads. But in general a MOSFET design will be more efficient. It's just easier to throw in a few more device than buy a bigger heatsink.

[jackbauer]

I've been running an igbt based controller for over a year now without one problem. Not one. And its quite a "rough" build. I discovered this project and purchased a control board from Paul i think in sept 09. I opted for igbts in the power stage as they offer the ease of drive of mosfets with the ruggedness of a bipolar. Electrical issues aside they are also much more physically robust than pcb mountable plastic cased mosfets. This feature greatly assisted with the hardware build. I was able to use one device as the switch and one with its g-e shorted as the freewheel. This meant i had no parallel device issues to worry about. The 1200v rating also helped with the poor design as its much more tolerant of overshoot than a 200v mosfet. Its driven by a stock control board. No modifications beyond a miller clamp circuit comprising two resistors and a diode. The amount of naysing that went on on another forum when i was building the controller was unreal. One always sticks to my mind. "You won't drive 3 feet with that". Done 1500 miles actually and still counting. Recently upped voltage to 152v lifepo4.

Understand i'm not trying to prove anyone wrong here or say that anyone using mosfets is a moron. The point is that both parts have advantages and disadvantages.

I would gently disagree with the premise of old igbts via ebay beiing useless. The parts i have are fuji 1mbi800-u4l if my memory serves. Quite old 3rd gen devices of 2005 vintage. Purchased from a surplus dealer in Germany for 30 euros each second hand.
When i was having some problems driving the device sucessfully , Jeffrey Jenkins of soliton 1 fame rather criptically posted "There's a reason the solition 1 uses 8khz switching in performance mode". This set me off on the trail of lowering the switching frequency from 16khz that suited the mosfets to 8khz that the "old" igbts just loved Thanks to paul and the gang for doing the software mods to make it happen.

So in short i'm a happy camper with igbts. Others are likewise with mosfets. Now , the only way to settle this battle of the silicon is a drag race

[isaac_alaska]

Originally Posted by jackbauer

I've been running an igbt based controller for over a year now without one problem. Not one. And its quite a "rough" build. I discovered this project and purchased a control board from Paul i think in sept 09. I opted for igbts in the power stage as they offer the ease of drive of mosfets with the ruggedness of a bipolar. Electrical issues aside they are also much more physically robust than pcb mountable plastic cased mosfets. This feature greatly assisted with the hardware build. I was able to use one device as the switch and one with its g-e shorted as the freewheel. This meant i had no parallel device issues to worry about. The 1200v rating also helped with the poor design as its much more tolerant of overshoot than a 200v mosfet. Its driven by a stock control board. No modifications beyond a miller clamp circuit comprising two resistors and a diode. The amount of naysing that went on on another forum when i was building the controller was unreal. One always sticks to my mind. "You won't drive 3 feet with that". Done 1500 miles actually and still counting. Recently upped voltage to 152v lifepo4.

Understand i'm not trying to prove anyone wrong here or say that anyone using mosfets is a moron. The point is that both parts have advantages and disadvantages.

I would gently disagree with the premise of old igbts via ebay beiing useless. The parts i have are fuji 1mbi800-u4l if my memory serves. Quite old 3rd gen devices of 2005 vintage. Purchased from a surplus dealer in Germany for 30 euros each second hand.
When i was having some problems driving the device sucessfully , Jeffrey Jenkins of soliton 1 fame rather criptically posted "There's a reason the solition 1 uses 8khz switching in performance mode". This set me off on the trail of lowering the switching frequency from 16khz that suited the mosfets to 8khz that the "old" igbts just loved Thanks to paul and the gang for doing the software mods to make it happen.

So in short i'm a happy camper with igbts. Others are likewise with mosfets. Now , the only way to settle this battle of the silicon is a drag race

Thanks Jack. that makes me happy to hear. could you possibly pm me your power board schematic? i think i understand how it all has to go together, but i want to make sure i'm not missing anything. i'm pretty sure i could get away with a single igbt, but from what i can tell from the data sheet, the Vce increases exponentially (to a point) with higher Ic. My thought is that running three in parallel with drastically reduce the current through each one, and increase overall efficiency.

i've also got some giant SCR's....i wanted to use them for a controller but i don't know how to utilize them and i haven't done enough research on it to know for sure. maybe i can use one as the diode or something (:

[jackbauer]

Don't use scrs. They are and were used in forklift controllers with a capacitor discharge commutation circuit. real stone age stuff. Have a look at the videos on my website in my sig to see the controller build. I honestly never heard of vce rising at higher currents. If it did i don't think desat detection would work correctly. In any event i can get the hardware current limit to kick in without causing the devices any distress.

[DJBecker]

Vce can be modeled as a constant voltage drop in series with a small resistor. You may be looking at a semi-log graph if you see an exponential-looking curve.

A problem with paralleling three IGBT modules, as opposed to two, is simple geometry. It's harder to keep one from being hotter and getting most of the current.

I didn't want my post above to get too long, or I would have added a list of IGBT advantages. Foremost is that they are tougher than MOSFETs. Running a 300V device at 200V gives you plenty of room for inductive spikes, and they are fine running hotter than MOSFETs.

Second, they are usually packaged more conveniently for DIY builds. They have hefty, convenient screw terminals, a Kelvin gate connection pair, and an isolated base for easy cooling.

And they encourage using only one or two devices -- it's easy to forget how expensive it is to mount a dozen MOSFETs. Spring clips, screws, washers are easy to count. Board area, extra wiring, bus bars, drilling, tapping, isolation film, etc. can easily add up to more than the device cost.