Hey,

Looking for some help to get started with a large sensored brushless motor(link at bottom)

Max voltage is 48v@135a=6.5kw(its apparent limit, guessing not continuous duty) though on average i dont think it will be going anywhere near that current. driving a cnc milling machine spindle, and possibly a bike!

After reading through this great document on atmel and an articl i found linked here on eetimes(links at bottom), pretty sure i understand the hall sensor, timing, backemf side of things but need guidance with a H bridge or whatever the term would be for a H brdige with an extra "leg" to drive a "3 phase" motor. I get the idea but with zero experiance im looking for guidance :)

Would it be best to use 6 single huge mosfets, or im thinking i could put a few in parallel to get the desired current? larger mosfets are slower but i think its still fast enough, would multiple smaller faster ones slow down been in parallel?

I think n channel would be more suitable? some people use p channel for the low side and n channel for the high side but i dont understand why?

Anyone with some experience suggest some common mosfets ill find other people using(in other projects even)?

I dont think ill be able to run the gates on these directly with the atmega, everything iv come across needs 10-20v and will need something to overcome the gate capacitive charge(i think i know what im talking about :) )?

Any other hurdles i might run into?

Powering with 44v 10A/hr lipo's

Motor: http://www.hobbyking.com/hobbyking/store/5142Turnigy_80_100_B_130Kv_Brushless_Outrunner_eq_70_55_.html

sensored brushless motor control: http://www.atmel.com/Images/doc2596.pdf brushless motor control: http://www.embedded.com/columns/technicalinsights/196701832

After some looking i decided on 12x 4110 mosfets setup in pairs http://www.irf.com/product-info/datasheets/data/irfb4110pbf.pdf

Now to work out mostfet drivers and flyback diodes.

Im so over my head haha....

I'm about the same as you are in this. I'm debating on the mosfet but I keep delaying buying anything because I'm paranoid it. Here are the mosfets I'm currently thinking of using http://search.digikey.com/scripts/DkSearch/dksus.dll?vendor=0&keywords=568-5756-ND

Befor i start im definetly no expert so take as a grain of salt :P

Compairing them to the 4110's they are alot slower and have a higher on state resistance. Means they will get hotter which i think is more of a limited factor to what they can output then anything else.

What were you looking at pushing through them, its generally not a great idea to run components right on the specs and expect any reliability. although the 4110's direct from IR(they set standards for these things) seem to be undervalued and run ok at specs apparently, there are alot of cheap chinese ones though that arnt nearly as good though so be carfull http://endless-sphere.com/forums/viewtopic.php?f=2&t=10655&start=105. Unless you are running really low at voltage like 25v you might also run into issues with back emf generated from the motor which can be higher then your battery ouput voltage.

Id get the 4110's( http://www.irf.com/product-info/datasheets/data/irfb4110pbf.pdf ), i cant see any benifit to your linked mosfets over 4110's. It leaves more room for the back emf voltage or upgrading system and its what 99% of ebike controller use.

I't will be inefficent if your running really low voltages, switching will generate more heat but i wouldnt be able to say if its enough to care about. There are tons more on irf site https://ec.irf.com/v6/en/US/adirect/ir?cmd=eneNavigation&N=0+4294841672+4294852589

I actually recently decided to go for these: http://search.digikey.com/scripts/DkSearch/dksus.dll?x=0&y=0&lang=en&site=us&KeyWords=BUK6E3R2

With this heatsink: http://search.digikey.com/us/en/products/576802B04000G/HS337-ND/1216344

And according to calculations with no airflow at 40C I should get 26.8A and that is DC so this would be true assuming full duty cycle. 26.9A would melt them most likely :P

I am going to be running 24-36V and I will be using a my1020 500W 24v motor. Not sure about its resistance so I'm having trouble getting myself a stall current to built the motor controller around.

Also is it a bad idea to not go pcb and do wire connections? I feel like it won't end well...

Lastly is it possible to calculate resistance based off 500w and 24V. P=IV gives you amps and then P=I^2R to give you resistance? Is it that simple or am I throwing the wrong equations at the wrong problem. My questions are for anyone to answer by the way so if you don't know don't worry about it :P

Ah a questions anyone who got a nerdkit up and running could answer easily (most likely). Does pwm output 0v on off or a low like 3.3v for off and 5v for on. If the latter is true then that sucks.

The new mosfet I found has half the rds(on) at 175C as the 4110 but I believe it may cost more coming in at $3.68 a piece on digikey.

Hi meltbox360,

The MCUs digital outs are 0V for low and Vcc for high. If you are powering the NerdKit off a 7805 then Vcc should be 5V.

About your earlier question. A 500w rating doesn't necessarily mean you can back the stall current of the motor. 500w is the amount of power you can safely put into the motor without it burning up, this will typically be less than the power dissipated in the motor at stall (this means its probably not safe to stall the motor for any particularly long amount of time). Your best bet for finding the resistance of the coils of a motor is to get your hands on a datasheet, or put a volt meter across it.

Humberto

Thanks Humberto! At this point I'm worried about the controller blowing up. I won't be driving the motor there for long but I would feel good knowing the controller could handle it if it happened. I wouldn't have the motor stall for more than a second at a time worst case.

I don't know what's safe. Assume 1200w @36v and I should be more than safe?

Ahhh one. Other question. Is gate source or gate body leakage the current that will be pulled from the microcontroller or is that something else. I have generally seen 1kohm resistors between gate and the microcontroller but my concern there is voltage drop to the point where the Mosfet will be on but with a high resistance.

Humberto! If you happen to read through this thread again I would like you to know that the 1kohm contrast resistor in the guide should be 10kohm or at least that is what worked for me. Do I have a wonky lcd? :O

I would also like to clarify my question. How many mosfets can be driven off a single pin. So how much current can a pin supply and what in a datasheet will tell me how much current I need.

I'd say either you misidentified the 1k resistor, have something wired differently, or have an odd LCD. Most people who use the 10k get a black display. The 1k definitly works for me and everyone else I know of.

I get a black display when I use the 1k resistor is brown black orange. Brown black red made it really dark. Really strange and I measures the old one was 1k and the new one is 10k ohms

Sorry in my last post I was saying brown black orange is the 10k and the brown black red is 1k ohm. I don't think I said that coherently.

I just realized now that I have a very very big problem. The mosfet I chose pulls around 2.2 amps for about a hundred nanoseconds. I can't pull that from a pin so whats the best way to go about doing this while keeping switching time as low as possible?

Perhaps I am wrong. Does anyone know how to calculate gate "turn on" current?

Hi meltbox360,

The fact that you get the a black display with a 1K in place makes me thing something is strange in your power connections. I would check that out before continuing.

It seems very unlikely that a mosfet will pull 2.2 amps into its gate. Large MOSFETs do have longer turn on times and will draw more current as they switch. If you absolutely positively need to use those fets and your MCU can't switch them fast enough, then you can use something called gate driver (its basically a power buffer stage). If you google MOSFET gate driver you will get plenty of examples and chips sold specifically for this.

Humberto