Can a 12FET controller handle 85A DC / 160A phase?

vex_zg

100 W
Joined
Mar 6, 2016
Messages
178
I have a KT40 kunteng controller, 12 x 4110 FETs

http://www.szktdz.com/en/news_show.php?article_id=502

I beefed up traces etc, and did a shunt mod. This made it go to some 85A battery current / 160A AC phase current measured with clamp current meter.

So in this 12FET controller of course two 4110 FETs are sharing this 160A phase current, and the FETs are rated 120A continous. I know they are not sharing the load evenly, so it might be one is taking much more than the other. Also an unknown is - what do transients look like and how much avalanche is happening.

They FETs are not heating much and have decent airflow around controller casing.

What do you think, can this last? What is the most you would push this controller, battery/phase current?

Thanks
 
It'll probably last a while, but for continuous reliable operation you'd be looking at more like 30-40A max per FET rather than >80A.
 
Punx0r said:
It'll probably last a while, but for continuous reliable operation you'd be looking at more like 30-40A max per FET rather than >80A.

Well there is no turning back now! :) I'll let you know how long it lasts. I like the smaller size of a 12FET controller. I have 18FET on the way as backup.

If my math is correct since FETs often fail in short circuit it might be quite a catastrophycal failure, probably taking out some traces and gate drivers.
 
I just don't get how those little legs support any more than 10amps, that's completely aside from the body. People say 'oh they're short so it doesn't matter'...I can't quite believe that. If I grabbed a piece of wire that thin and pumped 80amps through it, it would vaporise within a second so what's the deal with mosfet legs?
 
kdog said:
I just don't get how those little legs support any more than 10amps, that's completely aside from the body. People say 'oh they're short so it doesn't matter'...I can't quite believe that. If I grabbed a piece of wire that thin and pumped 80amps through it, it would vaporise within a second so what's the deal with mosfet legs?

since they are short their resistance is also quite low,and the heat spreads around, to PCB/traces and to MOSFET housing and the heatsink. I believe they copper plated with zinc or something. Copper is a good conductor. If it was just the leg in the air it would vaporize probably. I don't have a 80A variable source to test othewise I would surely :)
 
Alan B said:
The FET leads must carry the motor (phase) current. They rely on the circuit board traces to pull heat out of the leads. With high current they often unsolder themselves if insufficient heat is removed, or just vanish when there's a fault.

I didn't consider unsoldering, good point.

In my case it's 160A AC phase current divided by 2 MOSFETs (in a 12 FET controller), so hypothetically around 80A AC per that one thin TO-220 leg. I wonder how long would they be able to endure that 80A. Can't test it on the road, I don't like going faster than some 40-50km/h. Will test if they vaporize up a ski slope. Got one in my backyard.
 
Due to imbalance one of the FETs will carry considerably more than 80A. I had a 24 FET controller carrying those kind of currents that failed multiple times. So it really depends how well the balance is achieved. The more FETs the harder it is to have good balance.

You could carry a spare set for lower power and be prepared to swap it in the field.
 
Alan B said:
Due to imbalance one of the FETs will carry considerably more than 80A. I had a 24 FET controller carrying those kind of currents that failed multiple times. So it really depends how well the balance is achieved. The more FETs the harder it is to have good balance.

You could carry a spare set for lower power and be prepared to swap it in the field.

Yep, that's the plan, a spare one is on the way, until then not going more than few km away from HQ.

I'm a bit annoyed by these big 18/24 fet boxes so want to experiment with pushing 12FET to the limit, and really sorry phaserunner is not made in another stronger version.

I also ordered another set of a bit better MOSFETs to swap, I'll miller plateau match them before swapping.
http://eu.mouser.com/search/ProductDetail.aspx?r=595-CSD19536KCS

edit: Alan, what was the total damage bill? Just the FETs? Drivers too? more? was it repairable?
 
As I recall it was repaired with FETs several times, and one time required a new PCB. I sold it (cheap, with full disclosure) and the buyer had the last repair done. It ran fine after that, in his bike.

I think my problem was building my bike with very low impedance in the motor, wiring and battery pack (and my commute has a very steep hill to climb on a cold controller) all of which combined to make the peak currents very high. During the period of failures I commuted a lot of miles, then I would have a few failures back-to-back, then good miles, etc.

The 12 FET backup never failed, but ran quite hot and didn't have the same currents (acceleration). With the heat generated I would expect modding it for higher current would lead to unreliability. Eventually I bought a Sabvoton and tuned it to 80A battery / 120A motor current and it never misses a beat.

The PhaseRunner's bigger brother is available from the manufacturer.
 
Ah sh** now I realized that I was using the speed switch wrong and that those 85A were in mid setting and that the max setting is probably around 120A battery DC current so probably more than 200A AC phase current.

Well this can't last long. I will send pics when controller fries :)

Those new FETs are here, rain is just about to come here so I will be switching to them in a day or two, if the controller survives.

I wish I could use the scope to see the transients, those could probably be improved by snubbing or modifying switching times. I hate not knowing what the transients/ringing looks like.
 
My old analog Crystalyte controller did 80A for short bursts. Continuously it could probably do 50A. Those had a very hefty heat spreader and weighed considerably more than the typical 12 FET cans you see today.

If you have enough cooling to keep it from overheating, it might last for a while. The main caps can be a weak point too.
 
You didn't mention voltage, and if it is 14S or less, then swapping-in 3077's would run cooler.

Regardless of the voltage, beefing up the traces with copper (either wire or sheet) will not only lower resistance, the added copper mass will help to absorb heat spikes and dissipate that heat over time...also, I would add two fans, one pushing, and one pulling.
 
it's 16s, so 67.2 Fully charged. The caps are 100V, 3 of them. I could add more caps if needed. I think 3077s are too low rated with 75V Vds, since I wouldn't bet that transient spikes have been dealt with entirely.

The FET side of the casing doesn't get more than some 35-40*C. I'm in Sweden so it's 13-18*C here now, and anyway it's rarely above some 25*C, don't feel the heating needs improvement

I'll switch FETS with better ones (100V Vds, 2.3mOhm rds, lower gate charge, higher avalance rating) http://www.ti.com/lit/gpn/csd19536kcs)
 
fechter said:
My old analog Crystalyte controller did 80A for short bursts. Continuously it could probably do 50A. Those had a very hefty heat spreader and weighed considerably more than the typical 12 FET cans you see today.

If you have enough cooling to keep it from overheating, it might last for a while. The main caps can be a weak point too.

can you explain a bit more what about the main caps can be a weak point? Is there a benefit to adding more caps?
 
Your FET choice looks good. I haven't seen that one before.

If the cap internal resistance is high, the ripple current will heat it up and they can fail from overheating. More caps will help spread out the heat, or if you can find ones with a lower IR it will help. Usually there isn't much room for a larger can.
 
fechter said:
Your FET choice looks good. I haven't seen that one before.

If the cap internal resistance is high, the ripple current will heat it up and they can fail from overheating. More caps will help spread out the heat, or if you can find ones with a lower IR it will help. Usually there isn't much room for a larger can.

For FETs I used mouser search: lowest rds with TO-220 and Vds>=100V, those popped up.

Had no idea caps can fail from heat but makes sense if they are continuously being charged/discharged. These seem low esr since big sparks go off if I short them :) There is very limited room in the casing, have to see if I can add more.

So far I travelled some 100km with frequent 85A DC current throttle to the max up hills. I will not push the 120A DC setting until I change FETs and prep another backup controller.
 
Back
Top