How to determine the caps parallel to the DC-supply.

Electric Motors and Controllers
Post Reply
arnespiessens   1 mW

1 mW
Posts: 19
Joined: Oct 26 2016 2:04pm

How to determine the caps parallel to the DC-supply.

Post by arnespiessens » Dec 11 2016 11:15am

Hello,

Attached is the scheme of Jeremy Harris his BLDC motor controller.

Regarding to the caps in parallel with the power supply I have the following 2 questions. (He puts 4 polarized (220µF) and 1 ceramic cap in each fase.)

1. Why are the caps there? Is it to stabilize the voltage and reducing the voltage spikes, or are the caps delivering the current to the motor phases?

My motor has to draw 15 amps peak current, do the caps have to be able to deliver this current?
For example one 1000µF cap (50V) from Rubycon has a ripple current of 3 amps. If I put 4 of these parallel to each other in each phase, is my motor gonna draw the 10 amps from these caps without blowing them?

2. Why the ceramic cap?

Thanks.
Attachments
Simple BLDC Controller V1.0.pdf
(11.72 KiB) Downloaded 49 times

User avatar
Schlafmutze   10 W

10 W
Posts: 71
Joined: Apr 23 2014 4:53pm

Re: How to determine the caps parallel to the DC-supply.

Post by Schlafmutze » Dec 11 2016 11:52am

The caps you're talking about are connected to the battery + input, they're put there to help smooth out the current spikes that would otherwise be loaded onto the battery.
The small ceramic capacitor is put there as well to help reduce EMI induced from the high frequency switching that the mosfets do, seeing as the larger electrolytic capacitors aren't as good at very high frequencies.

arnespiessens   1 mW

1 mW
Posts: 19
Joined: Oct 26 2016 2:04pm

Re: How to determine the caps parallel to the DC-supply.

Post by arnespiessens » Dec 11 2016 12:24pm

Ok, but why does he put 4 caps of 220 µF, why not putting one of 1000 µF in each phase?

If it's just to smooth the current spikes then this should also work, right?

User avatar
Schlafmutze   10 W

10 W
Posts: 71
Joined: Apr 23 2014 4:53pm

Re: How to determine the caps parallel to the DC-supply.

Post by Schlafmutze » Dec 11 2016 5:42pm

arnespiessens wrote:Ok, but why does he put 4 caps of 220 µF, why not putting one of 1000 µF in each phase?

If it's just to smooth the current spikes then this should also work, right?
Yes it should work to use 1 1000uf capacitors, but the reason why you often see lots of smaller capacitors used in parallel instead of 1 big one is either one or several reasons like.

1, cost, sometimes more smaller caps is cheaper.
2, size, sometimes you can't fit 1 big capacitor due to height restrictions.
3, ESR, usually many smaller capacitors that equal the value of a larger one have lower ESR (Equivalent series resistance). Also smaller capacitors have larger surface area which helps with heat dissipation and lifespan.

If you can get 1 1000uf capacitor that is good and has low ESR go ahead, the exact capacitor value isn't critical in such an application, and electrolytic capacitors have large tolerances anyhow.

You can also use capacitors with higher voltage values.

Njay   10 kW

10 kW
Posts: 836
Joined: Sep 09 2010 10:43am
Location: Portugal

Re: How to determine the caps parallel to the DC-supply.

Post by Njay » Dec 11 2016 9:41pm

See section "Main Capacitor" here:
http://www.4qdtec.com/pwm-01.html

User avatar
fechter   100 GW

100 GW
Posts: 14728
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: How to determine the caps parallel to the DC-supply.

Post by fechter » Dec 11 2016 10:30pm

A single large cap can't dissipate heat as well as 4 smaller ones. Less surface area. I've seen several controller caps that have exploded from heating due to excessive ripple current. You don't want the caps to even get warm really.

Ceramic capacitors and film capacitors can handle huge ripple currents without heating much.
"One test is worth a thousand opinions"

arnespiessens   1 mW

1 mW
Posts: 19
Joined: Oct 26 2016 2:04pm

Re: How to determine the caps parallel to the DC-supply.

Post by arnespiessens » Dec 12 2016 7:54am

Thanks for the reply.

I still have trouble with the ripple current parameter of the caps.

Let's say I have a motor of 250W and 36V that draws 10 amps. The caps are rated to a ripple current of 0.5 amps.
How will 4 of them in parallel deliver the needed current to the motor?

Thanks.

User avatar
Schlafmutze   10 W

10 W
Posts: 71
Joined: Apr 23 2014 4:53pm

Re: How to determine the caps parallel to the DC-supply.

Post by Schlafmutze » Dec 12 2016 8:15am

Few things to note is that battery current and motor current are different, the latter usually always being higher.
As far as ripple current on capacitors mean RMS, so a capacitor rated for 10a ripple can do 100a peaks without issues, 10a is just the average value.

With such low currents you're going to use you can probably get away with using most off shelf electrolytic capacitors in the right capacity value.

User avatar
fechter   100 GW

100 GW
Posts: 14728
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: How to determine the caps parallel to the DC-supply.

Post by fechter » Dec 12 2016 10:20am

The controller can work without any main caps. I've seen it. There was no noticeable change in performance. Until it blew up.
The problem is during PWM switching, the peak voltage across the FETs will be much higher and cause those to blow if the battery voltage is close to the FET rating. The caps are really there to protect the FETs from voltage transients. If you have a battery with very low ESR and heavy battery wires, the main cap isn't going to be doing much. There will always be some inductance in the battery wires, so the caps have to filter out the ringing caused by it.

If the battery wiring is long and has significant resistance and inductance, the main cap has to work harder to average out the current and will see a higher ripple current. Too much ripple current and they get hot and fail. Predicting the ripple current is not easy, as it depends a lot on the battery and battery wiring.

A 10A, 250W setup is pretty modest so won't require huge expensive capacitors.
"One test is worth a thousand opinions"

arnespiessens   1 mW

1 mW
Posts: 19
Joined: Oct 26 2016 2:04pm

Re: How to determine the caps parallel to the DC-supply.

Post by arnespiessens » Dec 12 2016 12:06pm

Thanks guys, helped me a lot!

Njay   10 kW

10 kW
Posts: 836
Joined: Sep 09 2010 10:43am
Location: Portugal

Re: How to determine the caps parallel to the DC-supply.

Post by Njay » Dec 12 2016 5:34pm

I use 2 x 330uF "low ESR" 105ºC electrolytics (can't remember the brand and the controllers are not with me) in a pair of 24V 250W controllers without any issues, and they don't warmup noticeably. Just for some reference.

arnespiessens   1 mW

1 mW
Posts: 19
Joined: Oct 26 2016 2:04pm

Re: How to determine the caps parallel to the DC-supply.

Post by arnespiessens » Dec 13 2016 2:58pm

Do you put 2 of them in each phase?

User avatar
fechter   100 GW

100 GW
Posts: 14728
Joined: Dec 31 2006 3:23pm
Location: California Bay Area, USA

Re: How to determine the caps parallel to the DC-supply.

Post by fechter » Dec 13 2016 3:15pm

They go across the battery connections leading to the FETs. Physically closer to the FETs is better, but for 10A it won't be so critical.

The Phaserunner controller uses a pair of 390uF caps. It's closer to 2500W.
"One test is worth a thousand opinions"

Post Reply