Motor overvoltage and PSU sizing

If anyone is interested in using this PCB: You can order them very cheaply on Make sure though, to choose 2 oz copper weight, because the PCB won’t be able to handle 100 amps otherwise. There is a BOM list on the project page, where you can find links to the required parts.

1 Like

Honestly guys, I would say you are close to setting your garage (hopefully not your house) on fire like this.
Using nylon screws and washers to bodge a 100A PSU into running in series configuration when it wasn’t designed to do so? :scream: :jack_o_lantern: :japanese_goblin: :fire:

BTW, to drive 100A through the motor, at less than top speed you don’t actually pull 100A from the PSU.
The current is recirculated through the diodes during the off-cycle of the PWM.

If the motor is stalled (ie not moving) but producing full torque, then the PSU current draw is very low, because the only power used is that needed to overcome the electrical resistance of the windings, heating up the motor. There is no mechanical power output, because velocity is 0.
If you compute I^2 * R using your motor’s winding resistance (plus the MOSFET on-resistance), this gives you the power in watts that you need from the PSU when the motor is stalled at 100A. You will find that it is MUCH less than 24*100, because the odrive is effectively lowering the voltage during the PWM cycle (it is acting as a switch-mode buck converter, with your motor as the inductor)

Generally, you don’t need a large power supply, even to drive a very high current motor. Just so long as you aren’t still driving that torque when you have reached a high speed.

What is your application?

My applikation is a high precision laser machine with a heavy 4mm steel plate on one axis. It is used to produce PCBs. I am aiming at very high accelerations at a speed of 3000rpm. I know that dangers of this, but it is usually very safe if you isolate DC ground. Isolating PE is a horrible idea. I use ballscrews and linear rails, both with high pretention and need high torque for the acceleration even at the speed of 3000rpm.
Also I have a budged and ~50€ left for my PSUs.

@Noothless - please don’t use unpolarised bullet connectors for the DC Bus input. Well, unless you are happy to suffer a flash-over when you connect them backwards, and get blue-hot molten copper, aluminum and shrapnel in your face. :slight_smile:
XT90 connectors can handle well in excess of 90A for short periods, and TBH there is a lot more copper on those than there is on the ODrive PCB itself!
If you’re worried about your connector, then add a 90A slow-blow fuse. It will probably never blow on a 100A current-limited power supply, but it is guaranteed to protect a 90A rated connector.

And you should note that the ODrive can’t really handle 100A continuous, unless you can cool the PCB tracks themselves, as well as the MOSFETS, and add extra DC bus capacitance.


Do you know what voltage correlates to 3000 rpm? (what’s your motor back-EMF constant kV in rpm/V?)
Divide 3000 by kV to get the back_EMF voltage at 3000 rpm.
Multiply that voltage by 100A to find the maximum acceleration power (mechanical power). Add the electrical stall power (I^2 R) that I mentioned in my previous post.
Divide this total power by 24 to get the PSU current requirement.
When not accelerating, your power draw will be MUCH less, so you don’t even need a PSU which is continuously-rated for this current.

If you’re on a 50€ budget for your PSU, I’d start off with whatever (low current) one you might have lying around e.g. a bench power supply, and first seeing what performance you get out of that.

At 24V there is no danger of sparking, so what could potentially happen to the nylon screws? And even if something went wrong, server PSUs are very well protected. (Short circuit, over current etc.)

I am building a 6DOF robot arm, with a maximum length of probably about 1 meter. I use the BLDCs in combination with planetary gearboxes, to be able to generate the torque I need. I am aiming to drive the motors with as much current as possible because I want to achieve really high acceleration. Also, I still want to be able to lift about 2 kg.

The reason I bought these server PSUs, is mainly the price. I know, a smaller PSU would probably be sufficient, but you can’t compete with 30$ per 2400 watt.

The danger is not hv sparking, but that any small connection that shorts the main supply will weld together and possibly melt the cables.
Btw. I use 270kv motors on a machine that will have both axis running at the same time
I use xt90.
I bought the genuine ones from amass. These are the orignals and are sold from many different RC companies under the name xt90. It’s important to solder these. Creality once crimped the connetors on their ender 3 (xt60) and some failed horribly.

1 Like

@noothless the danger WRT fire is that if you haven’t accounted for everything, them something could very easily short, but would quickly melt and go bang when subjected to 100A, and yes you can still get sparks at 24V due to parasitic inductance (my MIG welder has only 15V open circuit voltage. for instance)
But you are right that server PSUs would probably detect this and shut down immediately (perhaps permanently).
However, given that the PSUs are NOT designed for series configuration, you may well blow them up or get unstable voltage regulation if you do this.

Agreed @Leo_TheCrafter definitely don’t crimp your XT90s :rofl:

1 Like

Yes, alright agreed. But as long as there is no metal debree inside the PSU or you don’t stick your screwdriver into it, there won’t be a short circuit. My PSUs will sit in it’s own fire proof cabinet, so that is a risk I can live with. :slight_smile:


No, they are not going to melt anything, because the PSU shuts down immediately. :slight_smile:

1 Like

Just make sure anyone else following your advice does the same.


Unfortunately I can hardly check, but if they read through this thread, they’ll hopefully know.