FYI & update on my attempt at a new spindle driver design.

(Don Kleinschnitz Jr.) #1

FYI & update on my attempt at a new spindle driver design.
Completed the design and physical layout using discrete parts from my bins.
The prototype will be hand fabricated [the “old” way] using pad drills and wire.

The key features of this design:
Optically isolated
Small footprint [2.5" x 2"], easy to mount on the gantry
Integrated heat sync
Single sided PCB
Fused output
Linear Gcode to rpm control via PWM
Optional mini fan
Power indicators

Really linear pwm control
Power Input:
positive 48VDC
2 wire PWM control:
pwm(-), [grnd true from Open Collector or Open Drain]**
Output: 48VDC @ 5-10A

** there will be an separate OC driver included to add at the TinyG side.

Next is to prototype and find out what I designed wrong! :slight_smile:

Schematic is here:

Layout is here:

Google Photos

(Rod Shampine) #2

Don, Thanks for your work on the old controllers!
On your low side drive section of the schematic, you left out the pull down resistor for Q5.
You should check the turn off speed on the Mosfet with this circuit. I would add a gate protection zener on the big Mosfet too.
I would lean toward having the PWM input be be isolated from the drive power; you’re close to that now, if you can drive the optocoupler from the PWM input directly, or add a gain stage to reduce the current requirements.

(Don Kleinschnitz Jr.) #3

@Rod_Shampine Thanks for taking a look at this design I appreciate more than just my eyes on a design like this.

Q5 Pull Down;
Turns out I did not need a pull down on this push pull config. :).

Gate speeds:
Do you mean that the turn off speed may be to fast or to slow? I ended up with the push pull to get fast turn on and off since the gate and miller capacitance of this MOSFET it pretty high and needs some beafey drive.

PWM Isolation:
In regard to the PWM being isolated from the drive power do you mean not to connect the Opto’s LED to the drive power and rather drive it totally from the PWM line driver? i.e. the open collector output is tied to one side of the Opto’s led and the other side is connected to a pull-up to tinyG’s voltage?

I thought about that but then convinced myself since the grnds were connected anyway having the driver power supply used as the pull-up voltage would not create a problem. I figured it was worse to run power from the tingyG to the Opto’s LED and it also meant I had to pull 12V power from the tinyG which I think is already marginal. It would also mean that the tingG voltage would be pulled off the board a long distance into noisy harnesses and I was concerned about injecting noise back into the controllers power supply. It would be easy to change it back but I have to insure that the tinyG supply will handle the load.

The gate zener:
…is that to protect it from over voltage if so what is the source? The voltage on the gate should not be exceeding its max with this setup but maybe there is a parasitic I don’t see. Interested in its use purpose and size?

I have prototyped this version and these are the results.

The gate drive is working as expected with sharp turn on and turn off current and voltage as measured with a scope. No noise very clean.

The mosfet turns on and off in sync with the gate drive when driving a resistive load. When driving an inductive load it turns on and stays on fine. The turn-off is still problematic. Its turns off and then depending on the particular motor turns back on or oscillates during the off portion of the cycle. If you observe the gate it is fully off during these conditions.
I am not sure why the mosfet is turning on when the gate signal is holding is clearly at gnd during the entire off cycle.
I assume that it must be a parasitic problem such as the “miller effect” occurring inside the MOSFET.

Last week I finally decided that since the protptype was flung all over the bench stray capacitance and inductance was killing me. I left my comfort zone and designed a PCB, it is out for fab.

I also did some research on the Mosfet I am using which is the same as the RioRand and decided that It was overkill. Its high voltage capability also carried with it unusually high internal capacitance’s. I have some lower voltage devices with 10x less capacitance ready to try.

I truly value your input on this thing. I seem to be overrunning most of mine and my engineering associates experience in regard to this thing turning on when its supposed to stay off, the same problem that I saw in the Rio Rand controller.

So I am interested in any of your ideas, criticisms of the design etc that helps get over this hump. We will see how the PCB performs.

In the end I plan to convert this PCB design to surface mount and I think that will allow CNC machines to enjoy a driver that is almost linear and very small in size so it can be easily mounted on the gantry.

Google Photos

(James Rivera) #4

@donkjr Just curious; did you ever finish this?

(Don Kleinschnitz Jr.) #5

@James_Rivera design and PCB done. Been waiting for test all summer. Why do you ask :)!