controller capacitors melted

[SniperGaulois]

After overheating during a giant climb, my controller started switching off above 20A levels, so i looked at home, and 2 of the capacitors out of 3 had their wires melted off! i figure it will be fine if i solder new capacitors to it, they are jamicon 100v 220mf...

I can probably solder the old ones back on again, what do you think?

what are they for exactly, how come the controller seemed to have survived, and also where to buy new infineon? ebike ca is out of stock. i am in the EU by the way! thanks!

 

[heathyoung]

Yikes. I've seen that in very high power controllers, but rarely in ebikes.

Replace the caps with low ESR versions, search on farnell or rs or whatever your flavour - more capacitance - the better, lowest ESR you can get that still fits. If the ESR is not shown, then chances are the cap is garbage.

Infineons are available from Ecrazyman on ebay, the 48V one has 75N75 FETS (but 63V caps - replace if you are running higher voltage) and the 75V one has IRF4130 FETS and 100V caps + a switchmode power supply for the LV. Both are 12 FET, and cheap (but good)

 

[dnmun]

which of the caps had the leads desoldered? was it only the two input caps? i used 100V1000uF caps. 2 on the input and one on the S/D busses. you can test the caps to see if they function with the ohmeter and if it is shorted it will show a short but if it is open then the reading will change as the cap charges up with the low current for the ohmeter. reverse the leads and the apparent resistance drops, avoid reversing the cap. so put the red probe on the positive terminal of the cap first.

 

[SniperGaulois]

i have ordered some 330uf 100v phillips ones as replacements for the 220uf (it's ok?)

To solder them i have to unscrew 24 screws and 12 Fets on each side... to access the capacitor+ & battery+ rail on the reverse of the board.

So instead i am just going to put a new heavy gauge cable to extend the capacitor+ rail on top of the board, hoping that it doesnt modify the electrical properties, because the 4x 220uf capacitors were just joined in parrallel.... (what is the point in that???? in case one fails?) I figure i join the new ones in parallel it should be just fine


here are the pics, an Mk1 xlyte controller, 35a, can you ID if it is 48v or 72v? it has done 4000-8000k i lost count.
the ecrazyman 72v/1500w has an LVC of 61v, my battery is 50v/20ah. can i change the LVC?

the new ping battery is only giving me 16k but i figure that is totally unrelated, the controller cant damage the battery, i figure.

 

[scriewy]

ya you can change the LVC on ecrazyman, but when i frenzy searched few months ago on changing LVC on clyte and ecrazyman through hundreds of pages, i only found everybody were interested how to rais the LVC but didnt find how to lower. Most play with R12 resistor and maybe R10 R11

maybe you can even ask keywin to change it for you, he's a nice felow very cooperative about moding.

 

[ZOMGVTEK]

Yikes. That happened at 20A continuous? Those caps must be junk and have super high ESR, or your controller is poorly designed.

If you're happy with the controller, I would just replace the big caps with some new ones like it looks like you are going to. It will probably be OK.

 

[heathyoung]

I'd say very high ESR - junk - or dried out (which will increase ESR) - have a look at the lifetime rating - in a hot controller, 85 degree caps will have a limited lifespan.

Caps are put in parallel to increase capacitance (and also decrease ESR)

 

[SniperGaulois]

Okay thanks for the info

When they went, the controller was at 70'C/100'F, i just climbed 400 meters upwards in 15 minutes, 17A for 10 minutes and 35A for 5 minutes.

many times i put it through 35A for 5 minutes, but for 170 dollars, crystalite had rubbish capacitors in their MK1 72v controller.!!!
NEW ONES:

 

[Solcar]

After overheating during a giant climb, my controller started switching off above 20A levels, so i looked at home, and 2 of the capacitors out of 3 had their wires melted off! i figure it will be fine if i solder new capacitors to it, they are jamicon 100v 220mf...

I can probably solder the old ones back on again, what do you think?

what are they for exactly, how come the controller seemed to have survived, and also where to buy new infineon? ebike ca is out of stock. i am in the EU by the way! thanks!

Check to make sure there isn't too much resistance in the circuit between the controller capacitors and the battery. The result would be that the ripple at the capacitors would be higher than it should be because the batteries are not taking over as much as possible to shore up the voltage.

 

[amberwolf]

the controller was at 70'C/100'F

Just so you know, 70C is a lot higher temp than that--158F, rather than only 100F.

As for the caps failing by coming off the board--if they are like the ones in the older analog Crystalytes I have here, there is no support for the caps to keep them from vibrating, so they can break the leads off at the connection point very easily, just by the repeated bending of the leads at that point caused by vibrating around. Any component can do that, but big heavy caps (relative to the lead size/connection point) do it a lot easier.

So if they are not glued down securely, they will eventually break like that.

Once one goes, now there is A) less capacitance total, B) higher ESR total, and C) higher currents flowing thru the rest of the caps still attached.

The higher currents and higher ESR means they get hotter, and so do their leads. It's possible that if they put the minimum capacitance in there to keep ti working in the first place, then taking out caps (by broken leads) may well cause enough heat in the leads to desolder them, after the first cap or two comes off from breakage. So failure could be quite rapid once the first one happens from vibration, especially under adverse conditions like the sustained current flow and heat from the hill climb.

Having the caps mounted right there on the FET PCB rail area doesn't help any, because even if they're not glued down to it, they probably still touch it enough to directly conduct heat right into them from the FETs, and decrease their lifespan and cause other problems.


I expect most controllers only have just enough capacitance to take care of their expected typical usage, and heavy usage is really too much for most of them, and heavy usage when capacitance is significantly decreased for any reason is a death sentence waiting to happen.


When there is too little capacitance, the feedback spikes from the motor coil currents switching on and off could end up much higher than the FETs' max Vds, and easily blow them up (or blow one, and then the excess current flow diverted thru remaining ones in that phase cause heat that then directly or indirectly blows one or more of those up).