OK, Wow, now we are really making progress! So glad this project is moving again!
I'll look through the datasheets as soon as possible and make a scetch, hoping you can verify it for me.
Thanks again!
/M
Hi DC_42 and all,
I was about to order a driver and MOSfets, when I stumbled upon this:
What do you think? It's all done already... designed for 24 v but seems to be capable of up to 60, so should be perfect for my 36v-need.
In any case, I just used a bunch of Amazon Vouchers I got from work and his book with the complete schematics and guide to building this ESC is not on it's way.
I think I just saved all of you a lot of work. 
And myself some time. Thanks all!
/Magnus
That solution uses P-channel mosfets for the high side switches. Unfortunately, P-channel mosfets generally have higher Rds(on) than N-channel mosfets. That's why mosfet h-bridge driver chips are always designed to work with N-channel mosfets throughout. The particular P-channel mosfets used in that design are specified as Rds(on) max = 0.026 ohms @ 23.5A. That's around 5 times the Rds(on) of the N-channel mosfets I suggested. With 3 of them sharing 50A, they will dissipate 7.2W each on average. So be sure to use good heatsinks.
The schematic image on that instructables page is so poor that I can't read the resistor values. However, it does not appear to provide any switching delay to avoid shoot-through, nor does it provide active pullup for the P-channel mosfet gate drive. From the parts list, I think the gate driver resistors are all 4K7. This is way too high to turn the P-channel mosfets off quickly. So you must expect the switching losses to be high when using PWM - another reason to use good heatsinks.
As you are using 36v instead of 24v, you should consider adjusting the resistor values to avoid exceeding the Vgs rating of the P-channel mosfets.
Hm! I see!
Ok! Well, the book is ordered. I will read up when it gets here in a week, If I can wait... Should have bought the e-version, dammit.
I do like the idea of designing something perfect, not just that works, meaning a design by your suggestion using a driver and only N-channels. My problem is ignorance, I understand the concepts but I could never select all the components around, I just don't know enough. I think I need a finished design if I'm ever going to get this done and working. Call it self awareness...
Question1: I note that all other setups have large capacitors. What do those do? Buffer or noise cancellation or what?
Q2: Now I just found this while googling: http://www.parallax.com/dl/docs/cols/nv/vol2/col/nv52.pdf
Another H-bridge design that looks to be more along your lines and good for for a microcontroller. Up to 900Watts, my motor is 800W, so perfect?
From the text:"Component selection for the HIP4081A is relatively simple. I selected the Harris HRF3205 MOSFET which has an Rdson of about eight milli-ohms (0.008 ohms) for the four H-bridge switches".
Question 2B: Would this be a better design? I would happily double up on mosfets to add margin.
Question 2C; The design states 15-30v, would this work with 36v (up to 44v when fully charged batteries?) Can it be easily adapted? Which are the components limiting the top voltage? The 78L12?
Still, why can't anyone post a complete design with part numbers, for noobs like me? That will be my legacy to the world if I ever get this working!
Again, thanks.
Also, I just want you to know that I read everything you write, several times, learning a lot! Thank you for the effort and time you put in! It may look like I skip along to my own next suggestion, but that is really just an iteration of you respones. As you write "this is good, this is bad", I try to focus on finding a new solution that fits in with what you say is good.
/Magnus
dc42:
That solution uses P-channel mosfets for the high side switches. Unfortunately, P-channel mosfets generally have higher Rds(on) than N-channel mosfets. That's why mosfet h-bridge driver chips are always designed to work with N-channel mosfets throughout. The particular P-channel mosfets used in that design are specified as Rds(on) max = 0.026 ohms @ 23.5A. That's around 5 times the Rds(on) of the N-channel mosfets I suggested. With 3 of them sharing 50A, they will dissipate 7.2W each on average. So be sure to use good heatsinks.The schematic image on that instructables page is so poor that I can't read the resistor values. However, it does not appear to provide any switching delay to avoid shoot-through, nor does it provide active pullup for the P-channel mosfet gate drive. From the parts list, I think the gate driver resistors are all 4K7. This is way too high to turn the P-channel mosfets off quickly. So you must expect the switching losses to be high when using PWM - another reason to use good heatsinks.
As you are using 36v instead of 24v, you should consider adjusting the resistor values to avoid exceeding the Vgs rating of the P-channel mosfets.
magnusvr:
Question1: I note that all other setups have large capacitors. What do those do? Buffer or noise cancellation or what?
A large capacitor across the 36v supply, positioned close to the mosfets, is needed when using PWM so that there is a low impedance path for the changes in current. This reduces the effect of the inductance of the wires to the power supply and reduces the radiated noise.
[/quote]
magnusvr:
Q2: Now I just found this while googling: http://www.parallax.com/dl/docs/cols/nv/vol2/col/nv52.pdf
Another H-bridge design that looks to be more along your lines and good for for a microcontroller. Up to 900Watts, my motor is 800W, so perfect?
From the text:"Component selection for the HIP4081A is relatively simple. I selected the Harris HRF3205 MOSFET which has an Rdson of about eight milli-ohms (0.008 ohms) for the four H-bridge switches".
Question 2B: Would this be a better design? I would happily double up on mosfets to add margin.
Question 2C; The design states 15-30v, would this work with 36v (up to 44v when fully charged batteries?) Can it be easily adapted? Which are the components limiting the top voltage? The 78L12?
That design uses the same mosfet driver chip that I suggested. The mosfets in that design have a higher Rds(on) than the ones I suggested, so they will run hotter. The design adds resistors and zener diodes to protect the mosfet gates, which slow down the switching and should not be necessary unless the circuit is poorly laid out. I suggest you use that design, but substitute the mosfets I suggested, and consider omitting the gate resistors and zener diodes. The input to the 78L12 is limited by the 27v zener diode, however at 36v input the power dissipated in the 100 ohm resistor and 27v zener diode will be too high. I would consider using a different regulator, one that is rated at 45v or more, and replace the 27v zener diode by a higher voltage one accordingly.