Paul & Sabrina's cheap DIY 144v motor controller

[Tesseract]

Originally Posted by bigh

Hi Tesseract, sorry It wasn't my intention to reverse engineer the Soliton1. I was intruiged as to "Cutting edge capacitor: Special purpose high reliability/low loss 600V film capacitor. No electrolytic capacitors are used in power stage. "

No problem. I'm not all bent out of shape or in anguish or anything, I just wanted to let you folks know that we can all get along. If nothing else we serve entirely different markets

Originally Posted by bigh

...I'm looking to learn what my options are and select what the best option for my purposes is. Ie high voltage long lasting, and easy to hook up. The fact that your controller used no electrolytic caps is very interesting, as I found the rated lifespan of the electrolytics to be very poor.

Yes, film capacitors are superior to electrolytics in all aspects EXCEPT one: capacitance per unit volume. If you need lots of capacitance you want to use an electrolytic; if you need lots of anything else you pretty much are better off with film. Specifically, polyester or polypropylene. Mixed films are good too.

You don't really need a high dV/dt rating for this application - i.e., pulse rated capacitors aren't so important - but you do want low ESL, ESR and Rth. The latter is very important because of I²R heating (the ripple current through the ESR).

You can use many PC board mounted capacitors in parallel or you can use screw mount or stud mount caps attached to a bus structure or bus bars. Lots of ways to go with this.

There are even film caps that are direct replacements for electrolytics, with higher ripple current vs. lower capacitance.

BTW - the rule of thumb for sizing electrolytics based on ripple current is 20mA/uF max. Secondly, give them at least a 50% voltage margin when used in hard-switched converter circuits. Thirdly, every 10-11C drop in internal temperature doubles the life of the capacitor.

[squiggles]

Welcome to the forum Tesseract.
It is good to see someone with some knowledge of the capacitor application here, it seems to be the last of the technical aspects to be discussed.

Given that you have stated that the guide is 20mA/uF is there a rule of thumb for estimating the ripple current of a particular set up?

Also is there any wisdom about the best location for the capacitors, it seems they are generally built within a controller but is this any better than say a separate capacitor box located closer to the batteries?

[Tesseract]

Originally Posted by squiggles

Given that you have stated that the guide is 20mA/uF is there a rule of thumb for estimating the ripple current of a particular set up?

I don't actually subscribe to the 20mA/uF rule of thumb; I only mentioned it because it is a common shortcut to get a first pass approximation. The input capacitor for a buck converter has a very hard life so it pays to calculate what specs are really needed.

As mentioned before, the worst case for ripple current occurs at 50% duty cycle and can be estimated by simply halving the output current. Thus, if your controller is delivering 500A at 50% duty cycle then the ripple current will be 250A (this is an approximation, too, btw because the ripple current waveform is not a simple square wave, rather, it is a "ramp on a step").

Originally Posted by squiggles

Also is there any wisdom about the best location for the capacitors, it seems they are generally built within a controller but is this any better than say a separate capacitor box located closer to the batteries?

The capacitor(s) absolutely must be placed as close as possible to the switch/freewheeling diode. This is to minimize the spike produced both during switch turn off and diode recovery. Another way of minimizing this spike is by simply slowing down the switching times for the switch (IGBT, MOSFET, whatever). But this results in higher switching losses. Always a balancing act between switching losses and voltage spikes.

The battery pack is, in some respects, already a capacitor, so putting more capacitance next to it doesn't really help. On a related note, though is that it is beneficial to twist the battery cables together as this reduces their inductance (and reduces radiated electromagnetic noise).

[Greg Fordyce]

I have started putting Paul's kit together. I have got the control board soldered. Is it safe to power it up and try connecting to it with RTD explorer before I build and connect the power section? I would like to make sure it is working before finishing the power section. Any thing else I should check when it is powered up?

Thanks,

Greg

P.S. Paul it must have taken you ages to bag up all the parts and label then. Much appreciated though.

[adamj12b]

Originally Posted by Greg Fordyce

I have started putting Paul's kit together. I have got the control board soldered. Is it safe to power it up and try connecting to it with RTD explorer before I build and connect the power section? I would like to make sure it is working before finishing the power section. Any thing else I should check when it is powered up?

You do not have to worry about powering the controller without the power stage attached. I have one just like this running on my bench for testing of RTD Explorer.

If you have all the chips in the correct orientation and there are not bridged solder connections, you should be all set.

-Adam

[Greg Fordyce]

Powered up the control board and got a green led and flashing yellow led. Connected to RTD Explorer o.k. Then decided to connect the throttle pot, RT1 and current sensor. This time the yellow and green leds stayed on. Connected to RTD explorer and could see throttle response as I worked the pot and temp increasing when I held RT1. Also saw PWM commands, so I would have to say it is all working.

Now to set my sights on the power section.

Greg

[MPaulHolmes]

Oh heck ya! I love it when a plan comes together! Please God, don't let anything blow up...

[harlequin2]

Yea! I finally reached the end of this thread. Only taken a month or two.
I have an E30 BMW powered by a couple of L91 motors and I also have a LogiSystems controller. That blew up. So I decided to build the Cougar and bought a couple of boards from Paul. While waiting for the rest of the parts to arrive, I repaired the LS lump (24 new IGBTs, 31 new caps, a whole bunch of comparators and op-amps!) but it still suffers from the jerky starts for which it is infamous. But now I have my Cougar controller nearly complete and am hoping for much better things! Look out World.
Anyone want a LS controller going cheap?

[MPaulHolmes]

That's awesome! Is it an IGBT module approach like Jack Bauer? Having the LEM current sensor be sort of the center of the design was one of the best decisions I ever made. It allows almost instant hardware overcurrent protection, and doing throttle proportional to current cures all the jerky starts, and Curtis whines, etc... It sounds like the Logisystems controller went the Analog route. I still don't know what the heck an op amp even is. I really need to read a tutorial on that. They sound important.

[harlequin2]

I'm told that the LS is derived from the old Curtis 123-something. I think it used MOSFETs originally, but this version has 24 x 600v x 78A IGBTs in parallel. Its the 1000A, 156V model.
The control board doesn't use a micro, just a large number of comparators and op-amps and the current sensing is just a diode off the IGBT collectors. That was disconnected in my one, by the way. It had a resistor missing from the control board that was needed to make it work. When I did connect it, it behaved very erratically going into current limit as soon as the throttle was opened. I guess they couldn't get it to work either! Also, the thermistor that is supposed to provide overheat limiting was replaced by a fixed resistor. So, no current limiting and no thermal control - no wonder so many of them have blown up.
I played around with the current sense and fitted a low pass filter to stop the spikes from causing it to shut off and it now works quite well. Also fitted a thermistor and that causes the PWM frequency to drop from about 13.5 kHz to 4 kHz when it gets hot. Hopefully, it is now more reliable than the original.
I am toying with fitting your controller board with the Hall sensor instead, but I'd have to redesign the layout to fit in the case.
How's the ac controller coming along?