First off, I need to buy a driver board, anyone got a link? see requirements below.
Next, am I right to think a DIY solution is needed?
I had what I thought was an original idea, and then I see it has been done a few times. I’d make a driver board by placing some MOSFETs and a gate driver on a custom PCB then make a socket for an ESP32 Devkit board (or maybe a Raspberry Pico?) I see discussions of this and some design work in the forum but no one seems to have a driver board setup for production. Did any of these project go all the way to production?
If not I want to try again using DRV8305 and MOSFETs like DMT6004SPS (60V, 100A)
I want to get to work ASAP. Is there a driver I can buy off the shelf to runs a 5010 300kV industrial drone motor with specs up to this:
Motor KV: 300KV RPM/V
Motor Resistance (RM): 0.0946 Ω
Max Continuous Current: 36A
Max Continuous Power: 850W
This is a robotics application so the RPM will be very low. Same for average power but I can see 100 millisecond peaks over 50A.
What can I buy today to run a motor like this?
The longer term solution seems to be a custom PCB. That $0.90 MOSFET seems ideal and the DRV8305 seems the best gate driver I can find and then just some 2-cent passive parts are needed. Cost maters as I want about 20 boards made.
Hi, first take a look at the supported drivers for high performance BLDCs. As you can see there are a couple with somewhat high performance, although you might still need to cap the max current. A good price/performance driver is the B-G431B-ESC1 although you’ll need to supply it adequate cooling as I’ve seen some people talk about overheating.
Important considerations:
drone motors are rated for use with high amounts of airflow for cooling, using them in low RPM robotics applications you’ll have to derate it by a lot if you don’t want to burn it
doing a ESP32 diy board you’ll have to deal with it’s garbage ADC if you want current sensing/torque control
doing a diy board and slapping on 100A MOSFETs does not give you a 100A capable driver, check the STL180N6F7 MOSFETs used on the B-G431B-ESC1 driver, they are 120A rated and the driver is 40A rated
Probably the best immediately available driver is knockoff ODrive from Aliexpress. They run around $90 each and drive two motors.
DRV8305 is in high demand and short supply, so you probably won’t be able to get any boards made with it for a while. But if you don’t need the integrated current sense amplifiers you should be able to find a substitute. And then I think you could use those MOSFETs to make a ~$10 small size high power board by soldering on heavy conductors like I did here Lepton Deku mod
I intend to try it myself sooner or later. Although as discussed in posts 15-21 in that thread, I can’t figure out how to read gate driver datasheets, so I will need some advice on driver selection (limited to what is in stock at LCSC).
But first I need to figure out a good way to test those small boards at high current and see how effective the heavy conductors are. So far all I know is that running at ~3 Amps for a few minutes produces no discernible temperature increase.
All boards I designed went into production and I tested them. Problem is they are not cheap.
Could you please provide the following:
Current – 36A continuous, 50A peak
Voltage – min/max/nominal
Size/dimensions – width, height, thickness (make sure you account for heatsinks and fans if forced cooling)
Mounting holes – size, position, etc.
Comms / control – how are you going to control/telemetry the board? CANBUS, I2C, Serial?
Sensoring – the motor loop will close using what sensors? Current sensing?
Aux power – do you power the auxiliaries from the board or externally?
Cooling – passive or forced?
Housing – dimensions – the board must be housed in a box
Control – I very strongly suggest you go for bluepill or blackpill or G431 or Teensy. Avoid ESP32 or RPi Pico.
Budget – what is the expected cost per board.
Answer these first before you start choosing the board.
As I said, I reall have two problems I want to start prototyping a machanical system using real motors on a test stand and I also need to think about designing a custom PCB.
Cooling is so easy with drones. They use forced air cooling that is so powerful the entire assembly lifts off the ground from the forced air exhaust.
Ok in still air, I used a conservative model to see what size PCB trace is needed for 50 amps continuous at 10C temperature rise. It really can not be done even with 2 oz copper. However, if you cut a strip from 0.5 mm thick copper foil tape and solder it over a 2 oz trace, the trace only needs to be 5 mm wide. A roll of 0.5 mm copper foil tape is not expensive and it is easy to hand solder.
The other solution that works for a robot is to mount the PCB flush to the metal structure using an isolating thermal pad. Then place a heatsink extrusion over the top of the PCB and compress the driver board between two aluminum parts. I am not so worried about cooling the driver board.
Thanks for pointing out the poor quality ADC on the ESP32. I see that I can count on getting only about 9-bits of accuracy from it. The Raspberry Pico seems to be about the same.
However in a robotic application I’d use two rotation encoders. One on the motor shaft for FOC and one on the other end of the gear reduction to measure joint angle after backlask and belt stretch. So FOC’s torque target is adjusted from real-word feedback. I might not need perfect torque control in FOC.
Modeling this is really hard so I want to build a test jig hence the request for a working controller I can buy today. Looks like odrive is it. I only need one unit, I can buy the “real deal”