Lookin for low ESR capacitors for 600A brushed DC controller

[bearing]

I'm planning to build a 100V 600A controller for a DC motor with about 11 milli ohm winding resistance. I don't want the controller to increase that resistance by much because that will degrade motor performance. I'm aiming at <10% increase. This means about 12 mohm total resistance in the motor-FET-capacitor-wires-loop. Motor is 11 mohm and wires negligeable wich means FETs and capacitor has to be 1mohm total.

If the resistance is split even between the FETs and the Capacitors they can be 0.5 mohm each. Parallelling FETs to 0.5 mohm is not a problem. The problem seems to be the caps. I've been looking at low ESR aluminium electrolytics. Intense searching gets me to the conclusion I need several litres of capacitors, they will weigh several kilograms and cost hundreds of dollars.

I want the caps to weight under 1 kg and be less than 100 USD. Is it possible?

 

[ZapPat]

Bearing, look into using a combo of low ESR electrolytics for low frequency bypassing and something else for high frequency response. The high frequency caps also need to have low inductance and high ripple current capabilities. What I've decided on for my HF caps are MLCC's - multi layer ceramic capacitors which have some very good characteristics. You'll want to get a number of smaller valued, cheaper ones and parallel them to suit your needs. This way you will get a much better price than trying to buy the rare big-ass ones, specially as price goes down with quantity AND the big ones are more $/uF to begin with. Get X7R/X7S types if possible, but X5R/X5S would be OK too. BTW, you will want to place these small ones as close as possible to your FET's power connections on each phase, and anywhere else across the battery bus if you need to.

As for sources, I've found some 1uF, 100V, 1812 size, X7R MLCC's on ebay for about 0.20$ a piece (for 100), but they will be more expensive new unless you buy a lot of them. 50V rated ones are much less expensive per uF. BTW, these things are almost impossible to destroy since you can overvolt them a lot without damage. Beware though of their capacity loss as their working voltage rises, you actually can get half or less of their rated uF's at their max working voltage. Avoid "Y" dialectrics as they are really bad for this.

For the electrolytic caps, I use chemi-com's KZE caps as they have good ESR for the price, but I'm not sure what they have in the KZE's for 100V.

 

[bearing]

Wow, those KZE caps seems great. My mistake has been to look for the ones with the biggest capacity.

A bank of KZE-caps will to fit my goal of less than 1kg of caps.

Are you sure the MLCCs are needed? I'm not sure a few hundreds of uF will help at these current's, because the total energy they store are less than one single PWM pulse at these power levels.

 

[ZapPat]

Wow, those KZE caps seems great. My mistake has been to look for the ones with the biggest capacity.
A bank of KZE-caps will to fit my goal of less than 1kg of caps.
Are you sure the MLCCs are needed? I'm not sure a few hundreds of uF will help at these current's, because the total energy they store are less than one single PWM pulse at these power levels.

We tend to think of only the PWM frequency itself when deciding on what and how many power caps to use, but there are actually two different events to think about. The PWM frequency itself is fairly low, like 20kHz for example. This is where your bulk electrolytic caps come to play, to keep your input's ripple low specially if your battery feed is long or has high inductance. However, the FET switching transition itself is a MUCH higher frequency event, since your FETs might be switching in 100ns for example. This would give you a 10Mhz signal, but there's likely a bunch of even higher frequency harmonics on top of this. This is why you need good high frequency caps, and not only electrolytics, and that it is very important to place them really close to your FET's Vbus (as I mentioned above).

Even the slow-switching ebike controllers we have here have minimal high frequency caps close to each FET phase (three 0.1uF ones on my ecrazyman 6 FET). The quality and placement of these caps is more important than their value, although a minimum amount will be needed depending on your switching current.

 

[bearing]

Is it possible to use 50V MLCC's @ 88V nominal battery voltage?

I found a cheap reel of 10uF 50V X7R capacitors. 2000 pcs, 78USD.
http://www.sureelectronics.net/goods.php?id=929

Maybe It's possible to use only MLCC's...? that would be cool. Depends on how much capacitance they loose from the higher voltage. How much do they loose?

 

[ZapPat]

Is it possible to use 50V MLCC's @ 88V nominal battery voltage?

I found a cheap reel of 10uF 50V X7R capacitors. 2000 pcs, 78USD.
http://www.sureelectronics.net/goods.php?id=929

Maybe It's possible to use only MLCC's...? that would be cool. Depends on how much capacitance they loose from the higher voltage. How much do they loose?

:lol:

Well that's funny! Sure electronics is selling those capacitors only because I special ordered a whole reel from them about two weeks ago. They only had 0.1uF/50V ones for sale before... which would have required a LOT of them top get even the 10uF's of one of the bigger ones!!! And believe me, I've been there, stacking 50 or so of the little ones before I decided to shop around. The best deal at digikey for these comes to about seven times the price of what sure electronics is asking - totally crazy!

Bearing - using these 50V/X7R caps over their 50V rating will probably leave you with very little capacitance left, although I have never seen capacity tests done on them over their rated voltage. I do recall that they already might be down to 40% of their rated capacity at 50V, so using them at double voltage would likely leave you with next to nothing.

I am myself on the search for high capacity, X7S/X7R/X5S/X5R, 100V-rated MLCC's. I'll let you know if I find anything.

 

[bearing]

Hm, a bit strange coincidence yes.

Yes you are probably right about the capacitance getting too low when used at 100V. I found and application called Murata Chip Capacitor Characteristics Data Library. It did not have Capacitance vs. Voltage for their whole MLCC range, but enough to get the picture on how size, voltage rating and capacitance affects capacitance drop at Vr. It seems the bigger the capacitor and smaller capacitance the less it's capacitance drops from high voltage. The 1206 is a bit small for this I think. Muratas GRM32ER72A225KA35 is a 2.2 uF 100V 1210 X7R MLCC. It's capacitance drops to 20% of its specified capacitance at 100V. I assume capacitors with similar specs will drop a similar amount at their Vr.

Here is a 1uF 100V capacitor which seems resonably priced. I could not find any Capacitance vs. Voltage specs on it though.
http://uk.farnell.com/taiyo-yuden/hmk316b7105kl-t/ceramic-multilayer-capacitor/dp/1683539

 

[ZapPat]

Hm, a bit strange coincidence yes.

Yes you are probably right about the capacitance getting too low when used at 100V. I found and application called Murata Chip Capacitor Characteristics Data Library. It did not have Capacitance vs. Voltage for their whole MLCC range, but enough to get the picture on how size, voltage rating and capacitance affects capacitance drop at Vr. It seems the bigger the capacitor and smaller capacitance the less it's capacitance drops from high voltage. The 1206 is a bit small for this I think. Muratas GRM32ER72A225KA35 is a 2.2 uF 100V 1210 X7R MLCC. It's capacitance drops to 20% of its specified capacitance at 100V. I assume capacitors with similar specs will drop a similar amount at their Vr.

Here is a 1uF 100V capacitor which seems resonably priced. I could not find any Capacitance vs. Voltage specs on it though.
http://uk.farnell.com/taiyo-yuden/hmk316b7105kl-t/ceramic-multilayer-capacitor/dp/1683539

I'm not sure what 0.032 pounds sterling comes to in US - maybe about 0.05$US? If so then that's a pretty darn good price, on par with the quote I got this morning from sure electronics, but I have to get a whole reel of 2000pcs for this price. I'll probably get them anyways, since my controller use them for the HF bypassing. I usually prefer running ~50V max ebikes myself so far, but I can see why some would go for something towards 72V or so nominal. I like the 55V or so limit, since higher density low ESR caps are rated to 63V (rare and low uF's at 100V) and you can get 60V FETs with very good Rds ratings too.