PWM motor controller for large motor, with low amps.

[Porch]

Schematic here.
https://dl.dropboxusercontent.com/u/24907255/CapOptions.png

I currently have the caps in "Option A". Where should I have them at?

I am not turning the motor at 1RPM. 1RPM is the final output speed. Possibly slower. The motor is belt and gear driven with worm drive gearbox. I don't have a way to measure the motor RPM, but I guess around 750RPM.

 Jeroen3: As in a FET driver, are you talking about a dedicated driver chip, or just driving the gate with a higher voltage? I can use another FET to drive the main FETs at a gate voltage of 12V.

[Phoenix]

Schematic here.
https://dl.dropboxusercontent.com/u/24907255/CapOptions.png

I currently have the caps in "Option A". Where should I have them at?

Option A will fry the capacitors, this is because you are applying a high frequency square wave voltage directly accross them. The place to put them is option C. This is not optional, it is the only place to put them.

The other problem is you are using electrolytic capacitors, even when located at option C they will have a large high frequency current drawn from them and thus heat up. You really need some non-polar capacitors capable of delivering high frequency current in parallel.

As for driving the MOSFETs, they look like they are OK with a 5V drive, but i doubt a single pin on the IC is properly capable of the current requirement for 5 in parallel. The use of a gate drive IC means you have a higher current drive (turning the FET on/off faster) and the option to raise the gate voltage (lowering the Vds during conduction). THe datasheet lists the absolute maximum gate voltage of Vgs=10V, maybe 8V is a good place to be.

[rob77]

5V drive for the gate is usually not sufficient... for most of the mosfets (even the "logic" ones) the lowest Rdson is defined at 10V gate voltage. of course 5V drive will work, but with higher Rdson and therefore higher loses and more heat.

[Phoenix]

5V drive for the gate is usually not sufficient... for most of the mosfets (even the "logic" ones) the lowest Rdson is defined at 10V gate voltage. of course 5V drive will work, but with higher Rdson and therefore higher loses and more heat.

Really got to look at the datasheet curves. I may have gotten the wrong mosfet in my previous reply.

At the start the OP says he's using a https://www.sparkfun.com/datasheets/Components/General/RFP30N06LE.pdf (refer to Figure 7)
Later on he changes to a http://www.irf.com/product-info/datasheets/data/irf1404.pdf (refer to Figures 1+2) which requires a higher gate voltage.

[Porch]

Thanks for the help.

I will move the caps to the correct location. If they get warm, I will switch them out to different caps.

The IRF1404 due seem to turn on fully around 10V, but can have a max Vgs of 20V.
The RFP30N06L do better at 5V Vgs, so I will use one of them to drive all the IRF1404.

Moving forward. !!! Thanks

[Phoenix]

I can't immediately see how you're going to drive an Nch MOSFET with an Nch MOSFET...

[Porch]

This won't work?

https://dl.dropboxusercontent.com/u/24907255/CapOptions2.png

[Yansi]

No it wont. Discharging 3 fets in parallel through 2k resistor is bad childish idea. Get yourself proper gatedriver and read please read some appnotes about paralleling mosfets, like this one: http://www.irf.com/technical-info/appnotes/an-941.pdf

Where can I find your schematic diagram of the whole power stage? I only saw some schematic on the first page, which was badly wrong.

MUR1620 is not suitable. Too high loss. For 24V operation, get 40 or  60V schottky. For 48V supply, get 100V schottky.

And also 20 to 30 amps is a work for one TO247 mosfet or two  pieces TO220.

The diode must be rated the same current as the mosfet is rated for - both are rated for nominal motor current - 30 Amps.

There should be absolutely NO capcitor in parallel to the diode. The cap will blow in ur face with almost hundred % probability.

The only capacitors needed here is decoupling on the diode-mosfet pair (2.2 to 4.7uF polypropylene) and properly rated (at least half the output current) electrolytics close to the power stage.

Not mentioning tha 30A switching stuff requires proper layout and circuit construction. The diodes should be directly near the mosfet, absolutely not on the motor itself. I talk about that almost in every PWM thread here. See picture below.

And I noted here a word "arduino"? I have suspiction, that arduino trash can't generate fast enough PWM signal to be used for such motor control. You need something like at least a few kilohertz, better 15k to 20k. That's work for hardware PWM timer, not that arduino software PWM libraries.

[peter.mitchell]

No it wont. Discharging 3 fets in parallel through 2k resistor is bad childish idea. Get yourself proper gatedriver and read please read some appnotes about paralleling mosfets, like this one: http://www.irf.com/technical-info/appnotes/an-941.pdf

Where can I find your schematic diagram of the whole power stage? I only saw some schematic on the first page, which was badly wrong.

Now now, no need to call it "childish", bad, yes, childish, no.

Using the nchannel fet you can turn the other fets off quickly, but they will still turn on slowly, ideally, you want to turn them both off and on quickly.

Unlike what some has already said, you don't need a gate driver chip, you can make a fairly simple push-pull driver from discrete components.
Do you have any small PNP + NPN transistors (eg 2N3904+2N3906), or small Pch + Nch FETs (2N7000 + BSS110)?

[Yansi]

Using complementary pair of mosfets requires cross conduction current problem to be solved. Use Bipolar complementary instead, like BD139/BD140 and level shifting stage in front of them, to shift the 5V logic signal to about 10-15V.

Very useful circuit which behaves well as gatedriver is this one. It also does level shift (needs only 5V input to turn on). Beware, that the output is inverted.
Don't forget to add base resistor. Ucc shoud be 12-15V. Diode should be 1A schottky (1N5819).

Or you can use small logic level N-MOS instead of the NPN on bottom. The upper transistor is required to be NPN in this circuit, can't be replaced by MOS.

Use suitable gate resistor for EACH mosfet. (about 47-100 ohms). The whole impulse current shoud not exceed peak ratings of components in the gatedriving circuitry.

[Porch]

The project is getting more complex then I hoped.

MUR1620 is not suitable. Too high loss. For 24V operation, get 40 or  60V schottky. For 48V supply, get 100V schottky.

Can you define too high of what loss?

And also 20 to 30 amps is a work for one TO247 mosfet or two  pieces TO220.

I am running 5  TO220.  The existing CanaKit circuit did well with 1 TO220 and a heatsink and small fan.

The diode must be rated the same current as the mosfet is rated for - both are rated for nominal motor current - 30 Amps.

I have 3 MUR1620 for a total of 48 amps.

Not mentioning tha 30A switching stuff requires proper layout and circuit construction. The diodes should be directly near the mosfet, absolutely not on the motor itself.

They are at the FETs. I also have the 8A8 at the motor itself as it really helped the CanaKit. So I left them.

Where can I find your schematic diagram of the whole power stage? I only saw some schematic on the first page, which was badly wrong.

They are all wrong. This is the last idea, but it's changing. https://dl.dropboxusercontent.com/u/24907255/CapOptions.png

And I noted here a word "arduino"? I have suspiction, that arduino trash can't generate fast enough PWM signal to be used for such motor control. You need something like at least a few kilohertz, better 15k to 20k. That's work for hardware PWM timer, not that arduino software PWM libraries.

I was thinking something along the same lines as for the frequency of the Arduino.

Can we solve most of this with out reinventing the wheel? How about I use an off the shelf motor controller and just boost the current capability with the MOSFET.
I have used this Polou controller before on smaller motors. https://www.pololu.com/product/1378
It will handle the volts I need, provide the fast switching of 22k, and give the control I need from the Arduino.
 
The gate of each FET will have about a 100 ohm resistor and connect to one output of the motor controller. Source to ground. Sure reverse won't work, but I don't need that.

Will this do the trick in providing a gate driver and the higher frequency needed?

Yansi and peter.mitchell have also been providing great help. But I am a bit worried about my skills in making a gate driver from scratch.

[peter.mitchell]

The gate of each FET will have about a 100 ohm resistor and connect to one output of the motor controller. Source to ground. Sure reverse won't work, but I don't need that.

Will this do the trick in providing a gate driver and the higher frequency needed?

Yansi and peter.mitchell have also been providing great help. But I am a bit worried about my skills in making a gate driver from scratch.

A gate driver would be no more complex to make than your motor pwm thing

[Porch]

Except I need to up the frequency. Can't do that with a gate driver.

[Yansi]

In my post from 01:02:00 PM, you gotta the almost easiest possible schematic for a gatedriver.

[Porch]

It's up and working.
I found a schematic for a 500AMP DYI motor controller. I copied the caps and diodes along with a cool FET driver chip. I didn't use quite as many FETs as they have listed as my amp draw is lower, and I use the IRF1404 I have.
http://ecomodder.com/wiki/index.php/Open_ReVolt/PCB_Schematics#Open_ReVolt_-_Cougar_500_Amp_DC_motor_controller_.28Rev2D.29

I missed with the timers of the Arduino and got 10khz out of pin 10 with a variable duty cycle.
http://www.oxgadgets.com/2011/04/creating-a-variable-frequency-pwm-output-on-arduino-uno.html

It's working so well. Only the diode's heat sink get very slightly warm. I can bearably tell. The FETs are cold to the touch.
I have the fan blowing across all of it just to be safe.

Duty cycle for normal speed is about 60%. The motor is barely warm to the touch. It really like this arrangement much better then the CanaKit. The motor is quiet and does not struggle nearly as much with the changing load. 

Thanks everyone for all the help.