Page 1 of 5

MCU 20A 450VDC fuse investigation

Posted: Fri May 31, 2019 10:17 am
by kiev
If for some reason the Motor Control Unit fuse blows, then the On-Board Chargger (OBC) and DCDC converter (High Voltage-to-12VDC) will no longer function, since they have lost their connection to the Battery Pack.

If the fuse blows during operation of the OBC, then it is very likely that the two snubber capacitors in the HV output section will also be destroyed. This scenario has been reported dozens of times in the past 2 years in the OBC Troubleshooting and Repair thread.

So looking at this another way-- could the fuse have a limited lifetime due to age, wear and tear, etc?

Environmental lifetime factors for fuses are not listed in the datasheets, but i would suggest that automotive electronics has a significant vibration factor for which stationary applications are not usually exposed.

What if the combination of repeated thermal cycling during normal OBC and DCDC operation, plus the mechanical shock and vibration from normal driving on our typically smooth and pothole-free highways over several years, has caused the fuse filament element to develop metal fatigue cracks leading to breakdown?

Could this be the root cause of the numerous OBC failures?

edit: add photo of open MCU to show the fuse. It is behind the small access cover, so no need to remove the cover to replace it.


MCU 20A 450VDC fuse construction details

Posted: Fri May 31, 2019 10:19 am
by kiev
some photos of the fuse construction and of a failed fuse:


found ring of cracked solder on the endcaps--evidence of vibration

filament is inserted thru a slot and soldered to each end cap

slot and solder blob removed

when the copper filament melts, it glassifies the sand and forms a hollow sand tube

molten copper globules fused with the sand tube

Re: MCU 20A 450VDC fuse investigation

Posted: Sat Jun 01, 2019 7:39 am
by kiev
This is the datasheet for the 10.3mm diameter fuse from Pacific Engineering Corporation, PEC:


and the drawing showing the construction details:


Re: MCU 20A 450VDC fuse investigation

Posted: Sat Jun 01, 2019 8:14 am
by kiev
i measured the thickness of the copper filament at 0.101mm, and the width at 1.97mm. This cross sectional area comes to that of about a number 24 AWG copper wire, which has a fusing current of about 29 Amps. This can be seen on the datasheet for the 20A rated fuse.

But on the fuse drawing it shows that the filament is actually a perforated copper strip and not completely solid--there are punched out sections to thin the filament. So the resistance is given as 4.2mOhms, when i calculated it would be ~2mOhms for a solid strip.


i saw evidence of the perforations when i first dismantled the fuse--there were 2 parallel sand tubes at one of the break joints. But it fell apart before i could get a photo.

The soldered filament into the end caps can be viewed as a fixed-fixed beam clamped at both ends for mechanical vibration analysis.


This leads me to believe even more that vibration can result in fatigue cracks in the filament, especially in the thinned out sections.

So i think that the fuse mechanical lifetime in our cars is about 5-6 years or so, although a couple of folks reported really early failures which could be infant mortality, a manufacturing tolerance factor.

Maybe this should be a maintenance item--just replace it after 3 years to be safe.

i'm gonna buy some 40 Amp replacement fuses. If someone wants to swap out their fuse before it breaks, i would like to dissect a used older fuse to see if there are any fatigue cracks beginning to grow.

Re: MCU 20A 450VDC fuse investigation

Posted: Sat Jun 01, 2019 12:20 pm
by Don
Very interesting

Construction wise, that looks very much like a standard household fuse from the 1930's or '40's - I guess I would have expected something just a bit 'higher tech' from a 21st century electric car

Also seems a bit odd that the circuitry is designed so something else fails when (or because) the fuse fails - Kinda defeats the purpose of fusing a circuit, doesn't it?


Re: MCU 20A 450VDC fuse investigation

Posted: Sat Jun 01, 2019 1:53 pm
by kiev
Another clue that the MCU fuse is suspect of mechanical failure, is that the little HV output fuse in the OBC has never been reported to have failed, nor has the same fuse in the HV input section of the DCDC converter.

Here is the final HV output filter stage of the OBC showing the HV output fuse, it's 20A but in a much smaller package. The schematic is in the OBC troubleshooting thread.


Re: MCU 20A 450VDC fuse investigation

Posted: Sat Jun 01, 2019 2:24 pm
by JoeS
kiev, thanks for presenting this and further pursuing it. Like your detailed photos! High Voltage dc fuses are interesting from the standpoint of needing to be self-extinguishing, and hence the sand-surround. An intriguing thought: wouldn't the sand act as a vibration dampener? I always assumed the sand was packed in there, but I can't tell from their drawing.

Ever since you first identified this possibility, I've been meaning to look back at the circuitry to correlate the OBC/dc-dc component failures with suddenly seeing this HV circuit open up while it's running. This fuse has an interesting position, as it blowing out affects the output of the OBC and the input to the dc-dc - and you just pointed out that the fuses for each of those circuits have not blown!

Going to 40A is certainly a quick way of getting a physically-beefier fuse element in there.

I haven't done any literature searches to see if there is perhaps a MIL-SPEC equivalent, since such fuses usually have stringent vibrations requirements. I don't even know how to look nowadays...

Anyway, I'm game to replace the fuse in my two i-MiEVs with a 40A critter and be happy to chip in to buy a few and let you dissect my old one - send me a PM.

Re: MCU 20A 450VDC fuse investigation

Posted: Tue Jun 04, 2019 7:17 pm
by WReed82
Hi Kiev and Joe,

Are you comfortable replacing a 20A fuse with a 40A?
Would the 30A be a better choice?

If the mounting legs are vibrating and breaking, what about building up a 'mountain' of silicone around each leg?
This might also help lengthen the life of the fusible link.

Thanks and good health, Weogo

Re: MCU 20A 450VDC fuse investigation

Posted: Tue Jun 04, 2019 9:46 pm
by jray3
Excellent work KiEV, thank you! The rigid mounting of this fuse could be another factor, with thermal cycling moving the end caps back and forth at a different rate than the element. As in my EV conversions, I used a Littelfuse KLK20 as a temporary replacement, and when mounted in spring clips, it naturally will experience less mechanical stress on the mounting points. I found similar fuses also available with wire crimp terminals on the end caps, which could allow a very low strain mounting of the fuse.

Re: MCU 20A 450VDC fuse investigation

Posted: Wed Jun 05, 2019 2:15 am
by coulomb
jray3 wrote: when mounted in spring clips, it naturally will experience less mechanical stress on the mounting points.

I can almost imagine the metal fingers in the original fuse creating a "tuning fork" of sorts. Add impulses from normal car vibrations and jolts, and the fuse could be doing a merry dance, fatiguing the delicate fusible parts. A set of shorter, fatter, and stiffer clips might be enough to dampen the vibrations and keep the fuse alive longer.

Recall that the fuse blowing first is one of our several theories as to how the on-board charrgers are blowing up.

"The fuse protects the cable", so the fuse should ideally be suited to the cable. I agree that if we are forced to go to a higher current, 30 A or 25 A if available would be preferred over 40 A.

[ "fusible". ]