The best controller

[esoria]

Hallo boys, i have a question, my top controller for now is a Kelly KHB 136V 400A, but as i alredy said, i dislike poor throttle response and bad regen on this controller (KBL is more better), i want to know wich is the "Best controller" in the same range price? I can't spend 2000$ in controller
I read about sevcon?Kurtis?
Please help me to choose good controller in the range of 500-800$
Thanks


Ettore

 

[Lebowski]

my controller IC project is the best controller, it's only an IC though so it needs an output stage etc.
I'll start designing a complete PCB in a few months (integrating the controller IC, it's low voltage
power supply, FET gate drivers and the output stage).

My controller has torque based throttle with programmable throttle curve. Same goes for regen, it's
torque based variable regen with programmable response curve. It is vector controlled with sinewave
(or back-emf) out. Startup can be sensored or sensorless, once the motor runs it always runs
under sensorless control (as this is the only way to achieve perfect timing / perfect timing advance).

 

[esoria]

Good! Interesting
When it's ready? And how much for 136V 200A controller?

 

[nieles]

http://endless-sphere.com/forums/viewtopic.php?f=3&t=37377&hilit=sinewave

this controller is 100v max though

 

[Lebowski]

Good! Interesting
When it's ready? And how much for 136V 200A controller?

I'm starting with a PCB for a 6 FET version first which is
what, 150 V 50A when using 4115's ? 100V 100A with 4110's ?

 

[Ratking]

Good! Interesting
When it's ready? And how much for 136V 200A controller?

I'm starting with a PCB for a 6 FET version first which is
what, 150 V 50A when using 4115's ? 100V 100A with 4110's ?

Why do you want to use those small fets when there are much bigger better fets available?
Since you have to make a custom print board anyway?

 

[Lebowski]

Good! Interesting
When it's ready? And how much for 136V 200A controller?

I'm starting with a PCB for a 6 FET version first which is
what, 150 V 50A when using 4115's ? 100V 100A with 4110's ?

Why do you want to use those small fets when there are much bigger better fets available?
Since you have to make a custom print board anyway?

which FETs do you suggest ?

 

[esoria]

i need controller capable of 140Volts and 200A

 

[gensem]

which FETs do you suggest ?

TO-247

IRFP4468PbF for 100v
IRFP4568PBF for 150v
IRFP4668PBF for 200v but the rds is too hi, like the 4115.


http://www.irf.com/product-info/fact_sheet/fs10162b.pdf

 

[Lebowski]

can you fit 6 of those in a row less than 10 cm wide ? (I'm aiming for a 16x10 cm pcb)

 

[gensem]

Yes, they ll probably fit.

 

[nieles]

a 3 leg to-247 is roughly 16mm wide. so they will fit on the short side, although barely

 

[Ratking]

which FETs do you suggest ?

I don't know much more than what I have read at this forum, but to-247 package is superior to to-220 in shedding heat and low resistance.
Since I would have a low voltage controller I would use 6 of these: IRFP4368PBF

I think Rithee or other experts could give more detailed answers

 

[Teh Stork]

my controller IC project is the best controller, it's only an IC though so it needs an output stage etc.
I'll start designing a complete PCB in a few months (integrating the controller IC, it's low voltage
power supply, FET gate drivers and the output stage).

My controller has torque based throttle with programmable throttle curve. Same goes for regen, it's
torque based variable regen with programmable response curve. It is vector controlled with sinewave
(or back-emf) out. Startup can be sensored or sensorless, once the motor runs it always runs
under sensorless control (as this is the only way to achieve perfect timing / perfect timing advance).

I'm very interested in how your controller turn out. I've started out in the other end, by building the power stage. I've unconvered some rather huge efficiency gains (its up in the 97% efficiency area now) in the gate drivers and output fets. What is left for me is making foc, throttle adjust, 12v and 3,3v supplies, regen and field weakening work.

One of my limitations is my current sensors, witch cap out at 200A. So the maximum phase current the controller can handle is 200A, but I've been looking into transitioning to sensored if this cap is reached - for even harder startups. Then again, 200A phase should be enough for most of the buyers. My system is aimed at those running 2-3kW average, with bursts up to 10kW. The whole system is modularized. One module for decoupling caps, fets, gate drivers and current sensors. One module for dc/dc conversion and one module for the microcontroller with communication to android! Essentially it will replace the CA too with this feature enabled.

Trust me, modularize your design some. Mixing everything on one pcb is asking for trouble, makes shielding hard to do unless you use a 4-layer pcb or do your pcb layout extremely analytical. I tried the last one, and bit the sour apple several times. (Lucky for me, we have a in house cnc pcb router )

 

[Lebowski]

hm, I think initially a lot more people would be interested in a 1 PCB complete controller that can push 5000 W versus
a modular system. I think going modular is a logical step to take after the 1 PCB complete controller...

Personally I'm not interested in communication with android or anything like that. Maybe its a generation
thing (I'm 39 ) so all this iPad stuff doesn't interest me. A controller should make the vehicle
move when I push the throttle and that's it. I'm much more interested in using the best motor control
algorithm with automatic timing advance, variable strength regen etc etc but without the user noticing
anything of it. I want a 'twist and go' type thing. Single PCB, stick it in a box, best controller possible, done.

 

[gensem]

I tough you would do a 15kw version too.....

 

[Lebowski]

i`ll try to makt the 6 FET PCB for TO247 transistors, then the power rating depends on which FETs you use...

 

[gensem]

i`ll try to makt the 6 FET PCB for TO247 transistors, then the power rating depends on which FETs you use...

I know, thats why i did ask for TO-247 mosfets... they are better in every way but abit bigger.

I particulary like this one

fets v rds amps
IRFP4568PBF 150 5.9 171

 

[zombiess]

TO-263's are even better power handling wise. Check out the IXFK230N20T 200V 7.5mOhm, parallel 3 or 4 of them up and have some serious power on tap if you want to go high voltage.

I also like the IRFP4568's 150V 5mOhm FETs since they are pretty common and relatively low cost.

6 FETs is just so small power wise, should be a minimum of 12 TO-247/TO-263 FETs so you have the option to parallel them for more power

 

[gensem]

6 fet is really too small... but Lebowski is going to be pushed by a thousand of people as soon as he has like 10 small controllers working and doing 4-5kw...
Selling potential is huge

 

[Teh Stork]

hm, I think initially a lot more people would be interested in a 1 PCB complete controller that can push 5000 W versus
a modular system. I think going modular is a logical step to take after the 1 PCB complete controller...

Personally I'm not interested in communication with android or anything like that. Maybe its a generation
thing (I'm 39 ) so all this iPad stuff doesn't interest me. A controller should make the vehicle
move when I push the throttle and that's it. I'm much more interested in using the best motor control
algorithm with automatic timing advance, variable strength regen etc etc but without the user noticing
anything of it. I want a 'twist and go' type thing. Single PCB, stick it in a box, best controller possible, done.

Well, communication with Android or iPhone is certainly as easy as connecting it to a pc. Writing code for android is cake, compared to writing code for running a motor. You know, as well as me, that there will be some tuning needed before a "twist and go" controller is realized. Mine implements a special code to determine the speed-voltage constant of the motor. All the user needs to do is lift the wheel off off the ground, this value is then saved, multiplied with a sine table and saved in a lookup table witch is then written to eeprom. Doing this w/o help from a smartphone or computer would be very hard - having to implement all the code into the uC. As a tuning routine tho - easy.

Enabling the smartphone to show whr/km, estimated range, acceleration and speed (using inboard accelerometer or pole counting from controller) is certainly something one would like to have in the best controller. Couloub counting for bulk fast charging could also easily be implemented. Interfacing to a smartphone is a logical step. Believe me, writing code in java is not much different from c++.

Gensem, 15kW burst at what voltage? Amps and decoupling becomes a major concern, and while the fets might be up to it - the capacitors will not. I'm increasing pwm frequency to deal with this, and that is also how I'm doing 110A in a 6 fet controller. The package inductance on all the TO-fets are too high imo, there is a reason they're obsolete in modern switchmode supplies. RdsOn really is one of the lesser concerns of powerstage design.

 

[deVries]

i`ll try to makt the 6 FET PCB for TO247 transistors, then the power rating depends on which FETs you use...

I see you continue on in your inspirational work... can't wait for your success... so we can utilize your designs!

I'll finally have some money to spend on a new eBike in a few months or less...

Anyway, super progress Lebowski!