Calculating fuse size for capacitor (ripple current?)

Thread Starter

MrSoftware

Joined Oct 29, 2013
2,186
I'm using a capacitor to smooth power on a motorcycle without a battery, and I'm going to add a fuse to protect the bike (prevent a fire) in case something happens shorting the capacitor or the user connects it backwards. So I'm trying to calculate the smallest fuse I can get away with. If I ignore the initial charge (it's too quick to pop the fuse) I believe all I need to calculate is the current due to the ripple voltage. Lets just use idle speed for now. The screen shot below shows start-up and idle. It's a 10,000uF cap, and the worst case ripple appears to be about 5v at about 100Hz. I don't know the ESR, so lets assume an ideal cap to get the worst possible case current. Q = CV . So worst case change in charge would be 5v * .01F = 0.05 columb per cycle. 0.05 columb/cycle * 100 cycles/sec gives me 5 columb/sec, or 5A. ... OK how badly did I butch this?



upload_2016-12-22_8-38-54.png
 

MrChips

Joined Oct 2, 2009
30,621
What you need to know is the load resistance or the load current.
Working backwards from

ΔV = I/2fC

I = 2fC x ripple voltage = 2 x 100 x 0.01 x 5 = 10A

You will need a 15-20A fuse.

This does not take into account the surge current when the capacitor is fully discharged. This will depend on the source resistance of your charging circuit.
 

Thread Starter

MrSoftware

Joined Oct 29, 2013
2,186
Thank you very much. I'm off by a factor of 2, I think my math neglected to account for the charge and discharge currents.

I'm not doubting the math, but intuitively this is much higher current than I was expecting. I think adding in ESR and noting that the ripple current average is probably closer to 2-3v, not the 5v worst case will bring it down, but still higher than I thought it would be.. Next I need to rev the thing up and see if the numbers get better or worse. ;)

Thanks for your help!

PS-> Here are the 2 caps I was considering using for this, it sounds like the 25v model is definitely out as the data sheet lists ripple current of only 3A.

25v - https://www.digikey.com/product-detail/en/nichicon/UVR1E103MRD/493-1071-ND/588812

40v - https://www.digikey.com/product-detail/en/epcos-tdk/B41456B7109M/495-4220-ND/2269429
 

Thread Starter

MrSoftware

Joined Oct 29, 2013
2,186
I don't know why I didn't think of this earlier, but is there any reason that a clamp-on AC current meter wouldn't give a reasonably accurate measurement of the ripple current?
 

RichardO

Joined May 4, 2013
2,270
A side comment... A failure mechanism of fuses is mechanical stress such as vibration. I think an automotive style fuse is probably more rugged than a glass fuse such as normally used in electronic equipment.

There are few things more aggravating than having a fuse fail from non-electrical reasons and not knowing why.
 

Thread Starter

MrSoftware

Joined Oct 29, 2013
2,186
@RichardO, indeed I'm going to use an inline ATC blade fuse, automotive style.

@MrChips - we weren't far off. I put the clamp AC current meter on it, and I got a max of about 2.4A at idle, and 3.4A if I revved it. Is there any reason I shouldn't trust the readings on this thing?

Idle:
upload_2016-12-22_19-58-8.png

Revving:
upload_2016-12-22_19-58-47.png
 

Thread Starter

MrSoftware

Joined Oct 29, 2013
2,186
Great question! The manual doesn't say explicitly, but it does say "50/60Hz True RMS", but it doesn't say whether other frequencies are supported or not.. So there is definitely some doubt there... hmm.. I've got a Fluke 87v that I can put in-series with the cap, though I'm not sure how much the meter will affect the behavior, and I'll have to check the manual on that one too. Thanks for the tip.
 

crutschow

Joined Mar 14, 2008
34,201
When I speak of load current I'm referring to the DC current the motorcycle draws when running.
I assume that current also goes through the fuse (and your clamp ammeter only measures AC).
 

Thread Starter

MrSoftware

Joined Oct 29, 2013
2,186
OK guys, results are in. The DC current component was zero, see image below.

I used a pair of 10w power resistors in parallel to give me a 0.7ohm load. They are wire wound, but I'll pretend the inductance is insignificant since I only need a ballpark answer. Firstly wow those things got hot, definitely some current flowing. The capacitor stayed stone cold though, so it's handling the job just fine.

Here's a screen shot from the scope, measured across the resistor with some revvs going. Vpp is 7v, so that would indicate a peak current of 10A, wow! I'm going to have to dig up my math book to figure out RMS current since that's nothing like a sine wave.

Revving, probably 4-5k rpm:

upload_2016-12-25_18-9-49.png

Idle, ~1800rpm:

upload_2016-12-25_18-12-21.png





upload_2016-12-25_18-8-33.png




upload_2016-12-25_18-7-42.png
 

Thread Starter

MrSoftware

Joined Oct 29, 2013
2,186
OK I'm not that bright.. right in front of me on that top oscilloscope screen shot was Vrms = 2.23V, which comes out to be about 3.2A RMS. So if I'm understanding correctly, I'm dealing with a ripple current of only about 3.2A RMS, but about 10A peak. Do I have this right?
 
Top