Double MOSFET in your controler. Kuteng 12FET to 24FET

Dariusz

100 mW
Joined
Oct 26, 2017
Messages
38
Want to make modification on my (kuteng) KT 12 MOSFET sine wave controller. My idea is to cut bottom part of controller (red lines on picture) and solder to the same pins of existing mosfet. In result from 12 mosfet will become 24 mosfet controller. New fets will be installed upside down and bend 90 degree. They will be attached to new aluminum sheet which will perform as heat sink. Thermal paste between aluminum plate and controller case. Of course i am going to add some extra capacitors. What do you people think about this idea? Let me know in comments.
controller mod.jpg

25129913_908810825950206_161049642_o.jpg

25075680_908806252617330_1664280326_o.jpg
 
Can't help you but I would love to see you run our OpenSource firmware on that upgraded controller :)

http://opensourceebikefirmware.bitbucket.io/
 
Dariusz said:
What do you people think about this idea? Let me know in comments.
Well, now that I thought more about it, seems a very good idea that will work.

You know, this controllers can look at the motor temperature sensor and reduce the power output as the motor temperature increases. And don't you apply a temperature sensor to the mosfets and our OpenSource firmware could be developed to read that value and reduce the power as the temperature increases?
 
You can buy 24fet for under 100$, but try it. Iam curious if it will work without modifiying mosfet drivers. I tried mosfet upgrade from 4410's to CSD19536KCS and its not even rotating well, noisy and phase cabels getting hot. I think you need to compensate for diferent reaction time, capacitance and these things when changimg to different type (so also if doubling) I hope someone can help us with this. BTW what setup do you want to run with this?
 
I am happy there is someone else who want to push big power from these controllers, but I think its better to get 18fet for 64$ (or way less if you buy whole kit) its 45A stock and you can modify traces, cables, connectors and shunt (or firmware+shunt) to 90A safe. If we get those CSD mosfets working then 150A safe and cooler than 90 with original fets.
 
Not sure about how many amps want to pull from it. Maybe 120A. I like this controller because its small and quiet. Not pure sine wave but has one current sensor up to 100A and 2 other phase are calculated little badly but its still quiet controller. I saw this info on casainho open source forum. Where do you find 18 FET controller for this price and is it sine wave?
 
casainho said:
You know, this controllers can look at the motor temperature sensor and reduce the power output as the motor temperature increases. And don't you apply a temperature sensor to the mosfets and our OpenSource firmware could be developed to read that value and reduce the power as the temperature increases?

I know it this controller have on display temperature of motor but i don't know where should i connect temperature sensor. I don't have cable going out from controller. Tried to find pin on circuit but without success. From my MXUS v3 it's white cable going together with hall sensors.

I think it can be enough to just test it on high hills for longer time and see how much temperature rise..
BTW i think my motor will get warmer than this controller. Big factor is actually if it's air cooled or sitting inside box. My will sit in this box
25198544_909288355902453_633393323_o.jpg
 
honya96 said:
I think you need to compensate for diferent reaction time, capacitance and these things when changimg to different type (so also if doubling)

It's good think to consider about. I am going to put the same as i have 4110 mosfets. Right now they are working in pair for high and pair for low line. So new mosfets should not be a problem... but... legs of new mosfet (bottom side) must have the same length as this on upper side..
higher resisatnce... delay in reaction.. This may causes some challenge. I have quite thick trace. I can spread and flatten

25188441_909292862568669_639835442_n.jpg
 
So it is working? :shock: 120A will be fine for the mosfets but you need like 6mm2 silicone cabels atleast, not the 2mm original + xt150 phase connectors and as150 battery will be best, if you dont have precharge paralelel with 2 phase c63 250vdc circuit breaker which I recommend. I think I will be getting hot really fast actually, you need bigger heat sink, search "dense heatsink" on aliexpress. https://m.ebay.com/itm/48V-1500W-Brushless-DC-Sine-Wave-Ebike-Controller-Regenerative-Function-/232088804925 they can add reverse for free and you can ask them about temp sensor (ntc 10k only) but I see you use 72v right? You will need to mod. Voltage regulator, LV cutoff resistor, capacitors and fets. Maybe not worth it? http://s.aliexpress.com/byyQjUjU
http://s.aliexpress.com/aeuaMZZZ (they have 72v in stock just not listed) you can get them to EU without tax if you know how..... :) so you use mxus 3k 5Turn? 20s 18650 pack?
 
honya96 said:
So it is working? :shock: 120A will be fine for the mosfets but you need like 6mm2 silicone cabels atleast, not the 2mm original + xt150 phase connectors and as150 battery will be best, if you dont have precharge paralelel with 2 phase c63 250vdc circuit breaker which I recommend. I think I will be getting hot really fast actually, you need bigger heat sink, search "dense heatsink" on aliexpress. https://m.ebay.com/itm/48V-1500W-Brushless-DC-Sine-Wave-Ebike-Controller-Regenerative-Function-/232088804925 they can add reverse for free and you can ask them about temp sensor (ntc 10k only) but I see you use 72v right? You will need to mod. Voltage regulator, LV cutoff resistor, capacitors and fets. Maybe not worth it? http://s.aliexpress.com/byyQjUjU
http://s.aliexpress.com/aeuaMZZZ (they have 72v in stock just not listed) you can get them to EU without tax if you know how..... :) so you use mxus 3k 5Turn? 20s 18650 pack?

A lot of good informtion. Thanks. I just didn't finish my mod. It's originally 72v with lvc and regen. I have 8gauge cables and i believe 6,5mm xt150 connectors. I have bought lots of new capacitors 1000F and 3500F. Will fit perfectly on right side in controller. I will add some covers for them for reduce interference. My plan was to install controller with some thermal conductive compound inside box in ebike (upper side). I use mxus 3k but not sure if its 5T. I have ordered 5T. My max speed on 80v was like 65km/h. Battery will depend how much i can fit. I ordered 340 pieces 18650PF so in best scenario 20s17p pack worst 20s14p. This will meter how much i can pull to my controller. Don't want exceed 7A a piece. Just for safety. Did you tried for ex. 90A and 120A. Is it big difference?
Everything (controller and battery) i need to install inside frame because its raining here every day.
 
Try to put capacitors paralel with input + - it goes the way where you want to put them :) motor seems to be 5T or even 6T so the controller will handle it. Fot that top speed your build it total overkill. At 7A/cell that battery will last forever, you can use 20s10p and draw 14A/cell because its not continous. 90 to 120 you will notice well but 90A is allready huge power if you have enough phase amp from standstill. With you battery and motor potential you will need the Opensource firmware which casainho is developing and I recommend to buy the 100$ 24fet in case something fails while moding the 12fet or testing the firmware, its twice the lenght but better cooling then.. also look at the mosfets i mentioned they have like 2.2mOhm vs 7.8mOhm 4110's so 1/4the heat, if we make it work.
 
They're usually marked inside on the stator, and the markings are typically correct.

But the no-load speed-at-voltage test shows for sure what it is. :)
 
So how is it going? Did you get it working? Seems that my controller got damaged mosfet driver. thats why my CSD mosfet modification does not work. Now its running 5 out of 6 steps, otherwise seems smooth so it should work ok also for doubling :) but rather than your design I would build an aditional double sided pcb with 12 mosfets on the other side of the original case/heatsink. Mainly because you cant beef up the original traces that much to hold 120A well.
 
Currently i am waiting for battery paper tabs and i can assemble my bike completely. Will make some test on 70A and test after modification slowly up to 120A. Probably i will record this. If you want mosfets on other side of controller need to consider that you need tick cables and its little space for that. I also dont know if you will not have some delays on new mosfets because they will be farther away - resistance. I can heat motor or just not run. You might also need few bigger capacitors. I am going to place them right there on this empty hole. Can you tell more about your modification? What is CSD?
About beefing up traces i am planing to go flat with copper wires or copper bar straight to the edge of pcb
25075680_908806252617330_1664280326_o.jpg
 
Resistance does not create delay (not noticeable) different mosfets do vary in delay betwen like 15-70 nano seconds - which also doesnt matter, so everything should be ok for your and my mod also :) maybe capacitance of the different mosfets is different but also not that much. By CSD I meant csd19536kcs mosfets which I mentioned before. My mod is basicaly just changing the mosfets to get twice the power or 1/4 the heat losses. Which means the smallest kunteng 6fet can be 3.5kw (72v) combining your and my mod it will probably be the most power dense controller out there :D. Doing 24 to 48fet mod means 400A from battery with a cool controller :D :D ...with the copper traces you cant cross the white line on pcb by much because of the case mount ;) you just cant fit enough to run 120A cool. Also it will not be much usable with that power until the Open source firmware gets to certain point.
 
I can't help but think if this works it will be by a happy accident!

Doubling up FETs, changing the physical layout, adding inductance to the high-current path is like undoing every design decision that went into the gate drive...
 
Punx0r said:
I can't help but think if this works it will be by a happy accident!

Doubling up FETs, changing the physical layout, adding inductance to the high-current path is like undoing every design decision that went into the gate drive...
I hope this info I bring is relevant.
I did measured the time between PWM signal change from STM8 up to power mosfet and is about 1us. On our OpenSource firmware I am using 1us of dead time (time that mosfets are disable on commutations to avoid short-circuits because of the circuit delays) -- and this time can be configured on the firmware, can be higher of lower than 1us, maybe this will be helpful??

Example of signal delay measurement -- all my measurements I did are documented here: https://opensourceebikefirmware.bitbucket.io/Motor_controllers--BMSBattery_S_series--BMSBattery_S06S--PWM_signals.html

33-8.png
 
1us makes plenty of room for changes. Anyway I dont think its all that sensitive, I have s06's with different types of mosfet, all having the same driver. Also 12fet and even 18fet running 3 parellel fets seems to have completly the same mosfet driver!
 
I am waiting for parts to come. I am going to use the same 4110 mosfets as original in this controller. Heatsink on bottom will be 9mm flat aluminum plate (i will look if i can find something ticker) and it will be in contact with ebike frame for better cooling. Maybe i will ad fan i will see. Will post some pictures when i receive my parts.
BTW. Do someone have idea how can i secure battery and controller from inside not drilling holes in frame. My 2 ideas is to drill holes and use spokes to pull parts to frame or separate with plastic or metal sheet and spray polyurethane foam like on picture below. Of course it will be smaller distance between batter and controller.
25198544_909288355902453_633393323_o.jpg
Just don't like heat with foam and i don't want to drill holes for spokes because in Bergen/Norway raining alomst every day. I will use silica gel boxes in frame to absorb moisture. Just microwave them every few months then reuse.
Some better ideas about how to secure. Maybe some threaded supports. Dont know yet how.
 
Punx0r said:
I can't help but think if this works it will be by a happy accident!

Doubling up FETs, changing the physical layout, adding inductance to the high-current path is like undoing every design decision that went into the gate drive...

This assumes engineering design decisions where made beyond what's the cheapest to function. It may barely be switching the FET's it does have, or it might be fine switching double.

I bet it's better than 50% odds of success if the person doing it is competent in the skills needed. Also might immediately auto-incinerate if it can't turn-off fast enough and experiences shoot-through, but I bet it has some timing margins.
 
Dariusz said:
BTW. Do someone have idea how can i secure battery and controller from inside not drilling holes in frame.

Self-adhesive Velcro? I bought some that is 50mm wide. Two or three strips full length would probably be enough.
 
When fitting the controller to the frame, atleast use some big silicone pad, if you screw it straight to frame just with paste sooner or later it will cause controller failure because of the vibration. Happend to me twice with these. :/ poor smd soldering job from factory.
 
honya96 said:
When fitting the controller to the frame, atleast use some big silicone pad, if you screw it straight to frame just with paste sooner or later it will cause controller failure because of the vibration. Happend to me twice with these. :/ poor smd soldering job from factory.

Thanks honya96!! Better dissipate heat and absorb vibration. Actually i have some big sheet used for GPU in laptops.
 
Back
Top