What is the use of a thread "Collection of Burned 100/250 Vesc´s"?
If it is about helping other to avoid burning VESCs or helping to improve the VESC tool, then more context is needed in what circumstances the damage occurred. Thus is would be good if the people who actually burned the VESC post this and explain what they did to help others to avoid that path.
@xenon321 You posted pictures of two burned VESCs that I took myself. I payed for the VESCs. Just because you got the pics, does not mean I'd like them to be published.
Now fot the context:
I already have two burned VESC 100 - 250 MKII (first and last picture above). That is testing at around 500€ / hour.
Both died on the test bench while working with a MAD M50C35 large brushless outrunner (54 poles, 34 KV, 140 cm propeller) and a big 20S, where the batteries were around 80 Volt. The target setup is a E-PPG trike / land plane aka powered paraglider.
The first one died on the stock firmware, the 2nd one died on the 5.03 beta 41 firmware (as of 2021-06-09). The 2nd one was still on and accessible after it stated to smell. This is the one and only error:
Fault : FAULT_CODE_ABS_OVER_CURRENT
Motor : 1
Current : 625.8
Current filtered : 407.8
Voltage : 76.73
Duty : 0.711
RPM : 70688.3
Tacho : 3562145
Cycles running : 1256087
TIM duty : 1990
TIM val samp : 2100
TIM current samp : 2100
TIM top : 2800
Comm step : 0
Temperature : 69.73
TBH I am in doubt if the VESC is the right controller for large brushless outrunners with lot's of poles, unless you are an expert of tuning in those motors. Using Auto-Detect won't save you from destroying the VESC. A clear disclaimer in that direction might prevent people from burning more VESCs.
If it is just a matter of doing it right then it would be great if someone can share a guild for the VESC on how to carefully tune large brushless outrunners correctly.
If you connect the VESC to a motor, test it in some way and it dies and then do the exact same thing again with a new unit you should expect the same outcome.
One problem with these larger motors is that I did not have many opportunities to test them and make the detection work well with them as they are expensive to buy and making test stands is also expensive and dangerous. So far I got a hold of a few and got all of them running well, but they were vastly different and required different setups.
Currently a T-motor U15 is on the way to me, which at least looks quite similar to this one but is smaller. I will try to get it running well, update the detection if needed/possible and show how I got it running. Then you can try the same approach with this motor and see how it goes. If you, for example, set the abs max current lower and stop the test once you start getting faults you can do it safely without burning the VESC. If the current gets much higher than the limits you know that the observer or current controller is not stable with the settings, and there is not point in continuing without adjusting something. Another note is that you should use a fuse with appropriate rating. The completely evaporated power stage comes from the fact that when they fail and short out, a huge current will flow until something in the systems burns through completely and stops the current flow. It is best to make that something a fuse rather than the VESC, cables or battery.
This is my experience also with tuning, for large motors to get them running well you normally need to do much more manual tuning. The best way is to set the abs max current to only 10-15% above the current limiit, then tune the detection parameters until you stop that from causing faults. Looking at the sampled current data and triggering on faults, you can easily see if the changes are helping or not. Once its tuned to the point it doesn't trigger faults then you can open up the limits a bit to stop random faults occurring.
I have seen a LOT of people having abs max current faults and their solution is to increase the limit! This is the completely wrong mentality and maybe needs addressed with something in the software to calculate what an appropriate limit should be. (If the abs max is set more than 1.2-1.5x the current limit something is probably wrong)
Changing the switching frequency during detection also seems to change measured inductance very significantly, especially on very low inductance motors.
Changing the deadtime compensation affects the measured resistance quite a lot as well.
Has anyone created a table of R, L, flux linkage, KI, KP, and observer gain values for different types of commonly used motors? That could help new users see if their detection values were sensible or diagnose detection errors.
> If you connect the VESC to a motor, test it in some way and it dies and then do the exact same thing again with a new unit you should expect the same outcome.
Sure, but that is not what we did.
We switched to the beta firmware and re-tested only after Frank suggested this would be the next step as you improved it quite a bit. Calculated values were different and the behavior was very different (better inital run up, but unstable at higher RPM, unable to reach the 10kW at all with 13kW on the old one) and we did not get ANY faults until the destruction (we got a few on the old firmware).
Therefore "... then tune the detection parameters until you stop that from causing faults" did not work, as the first fault was fatal. That is actually the main challenge for everyone trying a similar setup and I expect that folks shy away from using the VESC for those scenarios.
Benjamin, as requested to Frank a couple of times: I am happy to send you the motor for testing.
I value you time and a remote session is possible as well. You could get the remote control over the VESC tool via RDP or teamviewer and you can see the motor live on the test bench via zoom or whatever works best for you.
I am a bit concerned about the crash presented by xenon321, all the more so as soon I will be running the VESC 100/250, which will work with the freerchobby MP70141 15kw. Preliminary tests with the motor 2 kW passed without any problems, let me add that I used the battery pack 5s and so much less than the target voltage (20s).
How can I protect myself against VESC hardware damage, is a BMS with a current limit of 150A enough?
I would be grateful for any suggestions.
Vedder tested a T-Motor U 15 with the VESC100/250 recently. The design is similar, a high pole count outrunner. Problem: The stators go into saturation very early and in consequence standard parameters do not work after this point. Massive current spikes can occur under these conditions. With some manual tweaking Vedder could push the motor pretty far and reached very high output power in FOC. He will probably soon make a guidance for people using these kind of outrunners. These motors are a bit problematic, due to the fact that they saturate early and get instable in FOC operation quickly.
That matches what I have seen on these types of motor's. It's a problem for the smaller ones as well, such as MAD M9C12. Some way to map the saturation curve of the motor and account for it would go a long way I expect. Very interested to see what Vedder comes up with on this one!
saturation detection and mapping would be great
Hola xenon321 , he probado el Mad M50c35 en mi controladora personalizada asta los 11 kilowatios, https://www.youtube.com/watch?v=OP7dEkLCBeQ
To my understanding the max theoretical erpm on a VESC 75/300 e.g. is 150.000 erpm!
If the rule of thumb is correct that max shaft RPM is erpm/#polepairs, then with a 24 pole motor the limit would be 12.500 shaft RPMs ... before the VESC would "miss a lot" and eventually fry.
@dtydecks : RPM : 70688.3 / Current : 625.8 is reallly pushing it with a 24 pole motor on the VESC architecture ... I am not surprised ...
On another note:
Did anyone more qualified than me ever have the chance to look into a (Advanced Power Drives) APD HV_pro e.g.?
Those lill buggers are claimed to be good for up to 1.000.000 (1 milliion) erpms. (should take that 24 pole motor with ease!). What makes them that much faster/sturdier?
"the more I know, the more I realize how much I don't know"
It looks like those Advanced Power Drives are sensorless trapezoidal controllers judging by the product description. That method of control should be far less sensitive to sudden changes in motor parameters than foc.
@cmoalem the rpm shown in that post is erpm. These motors are normally very low kv and don't get close to max erpm. The problem is they saturate very early, not the rpm.
@TechAUmNu ... yes the difference to me is clear ...
thing is that FOC control loop in software on the current VESC hardware (32 bit, up to 180MHZ) obviously is burning some computing resources ... 150000 erpms (with tricks) are borderline for e.g. a 6 pole BLDC inrunner I am intending to use! (NOT a long board / eFoil application!!! Yes, they do exist too...)
Next iteration of VESC could consider using a small FOC in hardware!
They are available now! e.g. https://www.trinamic.com/products/integrated-circuits/details/tmc4671/
don't let "servo" and "encoder" confuse you ... they can very well be used for BLDCs (for now unfortunately only 2-phase) as well! see https://youtu.be/DjoWdWBKudw
just an outlook into the future ... hopefully 3 phase is not far away.
dug a bit deeper into the APDs ... and found in the documentation (yes the do publish a Fusion 360 3D schematic as well) for the HV_Pro 16s e.g. that they use 12 Mosfets per phase! (36 in total) ... have not even counted all those ceramic caps in the power stack stage.
No wonder they can stay "tiny" and not sweat it at all ...
The fancy Field Oriented Control implemented on the VESCs (unfortunately just in "software") is nice and most likely the future to come for all kinds of ESCs ... but the current hardware design is not cutting it for my application. Not the powerstage ... and not the compute power.
> @dtydecks : RPM : 70688.3 / Current : 625.8 is reallly pushing it with a 24 pole motor on the VESC architecture ... I am not surprised ...
@cmoalem 625A was the error message output. By no mean we meant to push 600 A, not even half of it. All settings were way below this value, which did not prevent the (partial) destruction. It is 2 month since the VESC has been pickup for a repair by Trampa in the UK. Still waiting for my first update...
> Did anyone more qualified than me ever have the chance to look into a (Advanced Power Drives) APD HV_pro e.g.?
Yes (as in we did - I not stating that we are more qualified than you ;-).
We tested ADP HV Pro 24S (https://shop.powerdrives.net/?product=hv-pro-200-24) + 100V ADP CapBank (https://shop.powerdrives.net/?product=capacitor-bank) with exactly the same massive 20S battery, motor (MAD M50 34KV) and prop (140 cm 3-Blade E-Prop). It worked out of the box without any issues what so ever. It is a rather small ESC and I expected it to have heat problems, but actually it did not even get warm. In our case it was mounted on a larger alu plate with a heat sink at the rear - same position as the VESC 100/250 before).
I'm hoping to drive a T-Motor U15L (which have 42 magnets) with a VESC 75/300 but after reading that topic I need to be very cautious before doing tests and using it on my paramotor
Is using trapezoidal mode instead of FOC could solve the motor's saturation issue ?
You can watch this video: https://cloud.vedder.se/index.php/s/jpm9Soiq68HBbdw
This is about tuning in a T-Motor U 15.
Thanks for the link, it is very instructive ! I now know more about safely tweaking the FOC parameters for this type of motor
I will test my T-Motor U15L with the latest firmware and report back on the forum as soon as I can
I just noticed in the pictures that the 100/250 is using 200v/120uF KXJ capacitors.
Isn't the 100/250 meant to be using 100v mosfets? So why are you using capacitors rated double that?
You can use 100v caps with 100v mosfets as the mosfets will go into avalanche and limit the voltage. Also if you are hitting 100v+ for any length of time the mosfets will explode long before the capacitors.
So it only has 720uF in total and those caps have awful ESR, do you not have really bad ripple on the input?
Why do you not use 100v KZN or KYB caps instead and get 4x the capacitance and lower ESR?
Also if its still using the 4.7u 1210 ceramic capacitors they will only have a total of ~10uF at 80v. With the large current spikes I would expect your bus voltage is spiking massively and shorting the mosfets.
A 100V cap will not have its full capacitance at full voltage. Interesting is the realistic capacitance at operating voltage.
It is wise to go higher in voltage rating and use overrated caps.
Most components on the VESC 100/250 are actually rated way above 100V. And in our experience the Mosfets don't brake down at rated voltage. We tested some of those Fets way above voltage rating and they didn't blow up, but still performed well! Not healthy for the Fets over time.... But they can cope with such beating astonishingly well.
"A 100V cap will not have its full capacitance at full voltage. Interesting is the realistic capacitance at operating voltage."
This is true for ceramic capacitors which is why I mentioned it for those. The 75/300 appears to use 4.7uF 1210 size 100v ceramic capacitors. I haven't seen any pictures or schematics of the 100/250 that show the ceramic so I am only guessing that you might have used the same part.
Electrolytic caps don't do this and Low ESR and high capacitance are much more important to reduce the bus voltage ripple. Especially when you allow people to have max bus voltage up to 97V!!!!
I would much rather if the product was specified with a proper safety margin. Say 15-20V instead of this constant push for bigger numbers with absolutely no margin for error. All it does is makes the hardware more likely to explode when something is not quite right with the motor detection. (Never mind the fact there is no protection on the logic connectors so the hall sensors can blow up the stm32)