Fet current sharing? Noob alert..

[alban]

Hi
I've read around a lot but get confused by differing answers and general ignorance on my part... I just want to guestimate how many A's I can mod my tiny 6 Fet or a 9 fet controller shunt to before blowing a fet - as it can be an expensive experment - so looking for very rough 'rule of thumb' kind of answers rather than deep understanding. (Unlikely ever...)
Ok so it's 'phase current' I should worry about which may be, say, 3X battery (shunt) current.
6 fets, so 2 per phase, but each one of the pair takes the full phase current - right? So is the phase current each pair sees the whole current (approx 3X shunt say, modded to 25A =75A) in the motor or only 1/3 of that i.e. 25A?
Fets are STP75NF75 (see attached) rated 80A max with 11mOhms R at 40A. I worked out 46mOhms at 80A at 175C as dissipates 300W max in a perfect cooling environment. Rtheta(ja) is 62.5C/W. 2 thermal constants=about 1C/W. Plus anything else?
Any rules of thumb re heat dissipation away from fets with average alu block in the average Chinese mass-produced controller??
So can those fets likely take the 25A OK or is it just wild guesswork?
I have a mid-drive Tonaro so presumably the fact I can up rpm by changing down gears will help stop torque, and thus current, shooting up as motor Power=torque*rpm (I saw somewhere)? But how much of a protection is that grinding up hills slowly on full throttle?
Thanks

 

[liveforphysics]

You're doing very very well! Great job!

Add in 1C/W for the electrically isolating thermal pad, and model the inside face of the heatsink touching the thermal pad as ~70degC, work the numbers, and you're set.

 

[liveforphysics]

If you are looking for a compact controller with good performance, dump those FETs and replace with these:

http://www.digikey.com/product-detail/en/IPP023NE7N3%20G/IPP023NE7N3%20G-ND/2217697


They are perhaps the best 75v TO220 package FET around at the moment.
http://www.infineon.com/dgdl/IPP_I...b02d7&fileId=db3a304320d39d5901210a9da8c22702

 

[boostjuice]

Keep in mind, if you are looking to upgrade the FETs, you may be able to squeeze in some TO-247 or TO-262 packaged replacements if the linear array has some space between adjacent FETs (You may need to make up a new drilled and tapped heat bar to mount them though) . Some 6 FET controllers have the room for this, but basically all the common 12/18/24/36 FET controllers have adjacent FETS spaced as closely as possible which makes this option non feasable (unless not populating every third FET footprint).

One of the best 75V TO-247 FETS > http://octopart.com/partsearch#search/requestData&q=IRFP4368

A potential controller TO-247 candiate VVV

 

[alban]

Thanks for that encouragement and info!
Haha I think I may have given the false impression of more understanding than I actually have... it's some of the basic things I don't have a grasp of - on the level of 'that's not a cup holder, sir, it's for the DVDRom drive'. I was thinking that I would have to get new controllers if fets blew as swapping out fets would be a step too far for me... So can you just sub those new fets for old without altering anything else in the controller? (No,NO, I mustn't even think about that or I might just get sucked into the e-power addiction around here - doomed to eternal tinkering!).

So then does each fet in the 6 fet Contr. see only 1/3 of the total phase current at any one time - or all of it? My guess would be - all.
Cheers.

 

[boostjuice]

So then does each fet in the 6 fet Contr. see only 1/3 of the total phase current at any one time - or all of it? My guess would be - all.
Cheers.

For a standard 3 phase BLDC square-wave controller with a total number of FET's that are evenly divisible by 6 (ie. 6/12/18/24/36 FETS), simply divide the total number of FETS by 6 to calculate how many FETs are dedicated to each Hi Side/Lo Side arm of each H-bridge. Each of these arm 'banks' sees the full phase current, as phase current passes Hi Side > phase coils > Lo Side under differing commutation permutations which allows motor rotation.

So for example, a 6 FET controller with TO-220 FET's has a maximum theoretical potential of 75/120 peak phase amps (depending on the die wire bonding limitation of the chosen FET, and assuming the die current limit is higher than this), whereas the same controller with TO-247 packaged FETS, this rises to 160/195 peak phase amps (depending on the die wire bonding limitation of the chosen FET, and assuming the die current limit is higher than this)

Disregarding FET thermal limits (usually the first safe operating current limit threshold reached), equal FET bank current sharing and cabling/PCB trace capabilities, below is a list of the maximum theoretical peak phase current for different different 'FET number' models of standard consumer e-bike controllers that pretty much all use TO-220 packaged FETs. Also keep in mind though that as the number of FETs increases, the lower beneath the theoretical maximum phase current a controller tends to be due to current sharing issues. (covered elsewhere on ES)

6FET = 75/120 phase amps
12FET = 150/240 phase amps
18FET = 225/360 phase amps
24FET = 300/480 phase amps
36FET = 450/720 phase amps

When it comes to the 9 and 15 FET controllers, the extra FETs are allocated to bolstering the Hi-Side arms as these are more thermally stressed than the Lo-Side arms. This allows the FET banks thermal limit to be raised a little, allowing the controller to provide a little more power than without the 3 extra FETs.

So can you just sub those new fets for old without altering anything else in the controller?

Often yes you can. Although the FET gate drive resistor may not be optimally sized given differences in the replacement FETs gate charge thresholds, they usually remain within the 'tolerance band of successful operation'. (I have done this modification to 2 different controllers)

If you wanted to match it though, below is an example.

Total Gate Charge for the IRFB4368 is 380nC (typical), whearas for the STP75NF75 it is 117nC (typical). To compensate for swapping them in then, simply replace the gate resistors with others of ~1/3rd the resistance. Hopefully, the gate drive circuitry can handle the ~3times higher current to match the original turn on rise time.

 

[alban]

Cheers Boost. I've got bogged down in calcs for heat transfer and too many unknowns but that's all useful info. Maybe this thread ought to be included in the wiki for noobs?