You are here

VESC on multirotors

33 posts / 0 new
Last post
jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30
VESC on multirotors

I just tested the VESC6 with a 36N42P motor and a 30 inch propeller. 
I looks nice, I have tested at 6S (22.2V) and it works very well, 


This graph is some seconds of hand throttle with a servo tester connected through PPM.

60A.JPG
60A.JPG (153.24 KiB) Viewed 22 times

I'm surprised that only at 6S it easily reaches 60 amps limit during accelerations

next test has been at 12S. At 12S it start ok, but when i try to give full throttle it makes a ugly sound like desynchronize

I have not experience with VESC so could you suggest how continue?

benjamin
Offline
Last seen: 6 days 20 hours ago
Joined: 2016-12-26 15:20
Posts: 357

Can you give some more details about the motor, such as kv and current rating? What are the current limits set to?

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

It is a Hengli Q9L 90kv Idle current 0.8A, max current 41A

Current limit is at 60A

Voltage is aprox 48V (12S)

 

This is the setting than VESC detect:

innicio.PNG

 

Ant it look the current when fail happens

 

problem.PNG

And this it the ERPM i can get at full throttle:

erpm.PNG

david_hickox
Offline
Last seen: 5 months 2 weeks ago
VESC Free
Joined: 2018-02-01 09:15
Posts: 3

It looks like you are hitting an erpm limit which if I'm not mistaken is around 60k to 80k on the drv chips. Also, may I ask why you are using a vesc for a propellor? Do you need reversing or breaking? You may want to look into an aero esc they are better for planes and cheaper, not to say the vesc isn't a great board it was just never designed for 400k erpm like a lot of the aero boards are. If this is in fact the issue I use cheap hobbywing skywalker boards (20 dollars for a 100a 6s 210k erpm from aliexpress) for all my aero stuff because nothing can beat their performance per dollar, granted they only go up to 6 s but they make hv 14s options as well but those get pricey.

brushlessdevil
Offline
Last seen: 1 week 4 days ago
VESC Free
Joined: 2017-10-10 01:48
Posts: 5

Hello JL cortex,

I'm not a vesc guru but I have 3 that I built from scratch myself.

 

With a 36N42P motor and 12S,you will burn it... period.

Too much poles,you are at the voltage limit, and the DRV won't handle it....(poles number x 3.7 x 12S) = BOOOOOM

 

Don't burn it...and I will doubt it will survive with even 6S

 

Regards

 

 

 

arvidb
Offline
Last seen: 5 months 2 weeks ago
Joined: 2018-01-16 23:09
Posts: 77

@david_hickox: VESC6 can handle at least 150000 erpm, according to Benjamin. The 60000 (or a bit more with FW3.30+?) erpm limit is for VESC4.

@brushlessdevil: Why would the motor get burnt because (poles number x voltage) is high? If it's a low Kv motor there shouldn't be a problem?

david_hickox
Offline
Last seen: 5 months 2 weeks ago
VESC Free
Joined: 2018-02-01 09:15
Posts: 3

@arvidb Ah my bad it seems like it is just not recommended a potential hazard to go above depending on the quality of the vesc but yes it is capable of hitting higher. Is there some sort of default erpm soft limit that could be doing this. My vesc has blown up 4 times on connection to 12 s batteries so I'm surprised his even worked for that long (it may be the lack of a large capacitor but I'm not really sure why mine blows up)

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

@david_hickox: This test is very interesting to me because we are driving big propellers. Above 30 inch propellers there are nothing than works correctly on a multicopter. of course there are 200A / 14S ESCs but they are for airplanes or helicopters, not for multicopters.

benjamin
Offline
Last seen: 6 days 20 hours ago
Joined: 2016-12-26 15:20
Posts: 357

The ERPM should be no problem, but It looks like the motor goes into saturation, making it difficult for the observer to track it. I had a look here

http://www.goodluckbuy.com/hengli-q9l-8314-90kv-110kv-brushless-motor-6-...

and they claim that the motor has 133 mOhm resistance. Running it at 41A is really pushing it, as it would generate 220w of heat. At 60A it would be almost 500w (square relation), and at that point the motor would be way into the saturated region. There are some parameters you can tweak to compensate for that, but I think it is pointless in your setup as there is no way the motor can deal with the heat losses without burning. If you want to get a bit more power out of the motor you can use a smaller propeller that creates less load at higher RPMs, but it will probably be less efficient.

If you want to try to push it anyway, you can experiment with the following parameters:

In FOC > Advanced: Set the speed tracker KI to 20000

In FOC > Sensorless: Set the saturation compensation to 5 or 10%

EDIT:

Just saw your detection result, which is the estimated phase to neutral resistance. Double that value should be the phase to phase resistance, which still is 110@40 and 240@60, which is a lot.

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

Benjamin, As far as I understand current is just to accelerate and decelerate. Not for continious working. It is a multicopter, and how fast you can aceelerate the propeller is key.

We use these motors between 6A and 12A in hover, the maximum value we measure is 32A in flight (full throttle but controller don't apply full full throttle because it need remaining 20% or so for control), ESC is rated at 80A

I am not measuring instantaneous current but the HW 80A FOC seems much faster accelerating the propeller than VESC so I understand HW ESC are applying more current than VESC.

So my target is to see about 80A during accelerations and about 40A at stable full thorttle. But maximum continious current at hover is about 12A.

benjamin
Offline
Last seen: 6 days 20 hours ago
Joined: 2016-12-26 15:20
Posts: 357

The observer will simply not track a motor well in saturation, and the extra acceleration you get by pushing is that hard is most likely not worth the efficiency loss. Again, you can experiment with the parameters I mentioned if you really have a need for that high acceleration, or you could go for the bldc mode instead of foc. I can't see in which situation you would need it though, unless you want to show off how fast it can accelerate straight up, or something like that. If your flight controller regularly pushes the motors into saturation during normal flight it is doing a bad job - there is absolutely no need for that, even in windy conditions. Also, in saturation the torque response is not linear on the motors, meaning that unless the flight controller compensates for that the control loops will not do too well if you analyze the details.

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

Just for understand how it works, I just tested the HW FOC ESC with the a current clamp and a oscilloscope, as I suspected it use a lot of current. i am measuring about 100A (at dc in) with fast throttle knob movements. I think it is the reason why this ESC is so fast and so violent. Could it have a phase current of about 130A?

100A.JPG

 

benjamin
Offline
Last seen: 6 days 20 hours ago
Joined: 2016-12-26 15:20
Posts: 357

Yes, it seems to push a lot of current, but I still don't see why you would want to do that with that motor.

Can you test the VESC with the tweaks I mentioned? You can also try to increase the switching frequency to 25 kHz in the FOC > Advanced tab. You can also test a current limit of 40A and see how much speed the motor can reach with slower acceleration to figure out if tracking works on 12S. As far as I can see the motor should work fine on 12S with the vesc unless you use too much current.

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

I continue doing tests, I am a bit scary to fry my VESC6. But this is my results:

KI to 20000 - I do not notice difference

Saturation compensation to 5 or 10% - tested at 0%, 5%, 7% and 10% - I do not notice difference

Motor Current Max to 40A - It works perfectly

Motor Current Max to 60A and Switching Frequency at 25000 KHz - The problem appears but it seems it works better

Motor Current Max to 45A and Switching Frequency at 25000 KHz - It works perfectly

Motor Current Max to 50A and Switching Frequency at 30000 KHz - It works perfectly, The logical question: What is the limit?

It seems now it is working, but there are appreciable differences, The HW ESC draw about 60A at full, not too much difference, but acceleration of HW ESC is much better.

HW ESC at full throttle (moving throttle slowly):

60A.JPG

Yes, it seems to push a lot of current, but I still don't see why you would want to do that with that motor.

benjamin, this HW 80A FOC ESC works really well, it is the best i have tested, we had stability problems and this ESC solve this problems, we don´t need to make any fine settings of PID, only change ESCs (and motors, it comes in a combo) and it fly solid as a rock, so i am trying to understand why. I suspect that this high current during accelerations has something to do

VESC is a great tool to understand how it works!!!

 

brushlessdevil
Offline
Last seen: 1 week 4 days ago
VESC Free
Joined: 2017-10-10 01:48
Posts: 5

Hello,

 

I'm not an electrical engineer,but a mechanical one😉!

I'm not the most qualified person to answer your questions,but speak with some experience.

 

42p is 21 poles motor.

 

In the worst case scenario with 12S fully charged you have about 4.19v per cell....

4.19 x 12 X 21 X 90 kv( which is a nominal and not real value..) you have aprox 91000 ,92000 ERpm.

 

From mine experience,is a little bit on the "high side"..

I continue to think,and I believe it's not only me, that the limitating factor is not the CPU neither the firmware,but alwaua the DRV, doesn't matter if it's the DRV 8301 or the 8302.

 

Other very important thing is the algorithm way the vesc works compared with a normal Brushless ESC.

 

But that is just mine opinion!

Best Regards

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

I continue researching about use VESC on a multicopter.

Right now VESC runs well at 50A (the HW-FOC gets 60A) but it is clearly heard that HW-FOC accelerates it much faster.

I am designing a simplified inertial dynamometer, I have experience with this dynamometers in racing application, i have a software to graph mechanical power and torque. it could be an interesting experience. 

flywheel copia.png

The flywheel is steel, it weight about 1.6kg and has 260mm diameter, at 4800rpm seem not dangerous. Right now, we are testing with 30" propeller. Right now we have a hole in the wall and the motor and the propeller is in a little room. This is not a small drone propeller, this is scary.

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

I just done very interesting experience. I have finished my inertial dynamometer, The dynamometer consist on a flywheel screewed directly to the motor, it has only a hall gear tooth speed sensor, it calculates the torque and power doing the math and knowing the inertia of the flywheel. I use the hardware and software of my racing motorcycles dyno. 

IMG_20180602_192544.jpg

I think it is very interesting experiment because I measured the DC current of VESC and HWFOC, and HWFOC draw a lot of more current than VESC (100A vs 40A) but I would like to see how this current becomes torque. And compare the mechanical torque directly and not the current. 

I have tested 3 ESCs:

HW BLDC 80A
HW FOC 80A
VESC6

I'm surprised that the ESC with the biggest torque is the BLDC.

TORQUE:

HW BLDC 80A - 10.29 N*m (skin color)
HW FOC 80A - 8.77 N*m (purple)
VESC6 - 4.57 N*m (blue)

The test shows the positive torque but also the negative (braking), I have done several tets to each configuration, but all tests for the same configuration is exactly the same, it don't happens with combustion engines!, here there a minor differences only in the start.

 

par.PNG

POWER:

Even though the torque is much higher on HWBLDC the power is a bit higher on HWFOC because it have the torque at higher RPM.

potencia.PNG

Now I see that this motor is at least capable of pulling 10 N*m

How to get it?
 

arvidb
Offline
Last seen: 5 months 2 weeks ago
Joined: 2018-01-16 23:09
Posts: 77

Nice measurements!

Did you try the VESC6 in BLDC mode?

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

Not tested, i can test but i am very interested in FOC control. 

HWBLDC ESC have more torque than HWFOC but HWFOC works much much better than HWBLDC, so I deduce than the FOC algorithm is better for multicopters because FOC have faster torque response, i mean than torque follows very fast the input signal. I would like to test this torque response speed.  

lizardmech
Offline
Last seen: 9 months 3 weeks ago
VESC Original
Joined: 2017-06-02 19:21
Posts: 80

Isn't that hobbywing foc controller matched to a specific motor that is different from the one you were using earlier? What was the current limit set to on the vesc with the hobbywing motor?

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

lizardmech, Yes, this time I'm using the XRotor 8120, it's the motor that comes with the hobbywing combo (esc + engine), it is almost identical to the other, this one is 100kv.

Motor Current Max - 50A / Switching Frequency - 30000 KHz. if i try more current it desynchronizes and makes a horrible sound. 

I want to improve my test bench and add DC current sensor, but last time with the propeller i was measuring about 100A on HWFOC (DC bus, and only during accelerations), that was my target. now i am not sure because I see the motor is capable of more and more torque! How can i get the saturation curve of this motor? or at least a aproximation?

Is it safe to test with more frequency, 35000khz or 40000khz?

 

 

TechAUmNu
Offline
Last seen: 21 hours 10 min ago
VESC Free
Joined: 2017-09-22 01:27
Posts: 346

For saturation I found that decreasing the resistance setting helps a bit and turning the stator saturation compensation on also helps. I don't think changing the frequency helped with saturation but you can safely go to 40khz without any problem. 

 

lizardmech
Offline
Last seen: 9 months 3 weeks ago
VESC Original
Joined: 2017-06-02 19:21
Posts: 80

I think that hobbywing motor is rated a bit higher, 66A or so. Have you opened up that hobbywing foc controller? It's a long shot but current sensing values, topology and MCU used might give some hints. I tried to find photos of the internals of that hobbywing FOC controller a while ago as FOC controllers for drones are uncommon but I couldn't find any.

TechAUmNu
Offline
Last seen: 21 hours 10 min ago
VESC Free
Joined: 2017-09-22 01:27
Posts: 346

It looks like it has 3 shunts and 6 fets. 

lizardmech
Offline
Last seen: 9 months 3 weeks ago
VESC Original
Joined: 2017-06-02 19:21
Posts: 80

The resistor value and fets would be the most useful info, if the MCU is one of those instaspin or other MCUs specialized for motor control it might offer some hints, I have yet to see them in any chinese products though. 

jlcortex
Offline
Last seen: 3 months 4 weeks ago
VESC FreeVESC Original
Joined: 2017-05-24 12:48
Posts: 30

hehehe, i just opened one, lizardmech, you are gonna like it! TMS320F28027 

this one?

http://www.ti.com/product/TMS320F28027F#

Captura.JPG

Captura2.JPG

Captura3.JPG

Captura4.JPG

Roger Wolff
Offline
Last seen: 6 months 2 weeks ago
VESC Original
Joined: 2017-05-24 12:27
Posts: 202

hehehe, i just opened one, lizardmech, you are gonna like it! TMS320F28027 

Ah! It has TI's proprietary FOC algorithm in ROM. 

lizardmech
Offline
Last seen: 9 months 3 weeks ago
VESC Original
Joined: 2017-06-02 19:21
Posts: 80

Oh, that was an unexpected surprise, first time I have seen instaspin in a product. Instaspin seems to have some way of enhancing sensorless tracking using the phase voltage sense, though I'm uncertain of the exact mechanics. Although the f28027 cpu is pretty weak I think the C2000 MCUs have many optimizations for real time control. It's been a long time since I read all the datasheets, I think the C2000s were faster at getting the ADC data and other things than the general purpose ARM MCUs, as to how much real world impact it has I don't know. You can also set them to run very high PWM frequencies with only a few lines of code, it's possible it runs at 60khz or more.

Roger Wolff
Offline
Last seen: 6 months 2 weeks ago
VESC Original
Joined: 2017-05-24 12:27
Posts: 202

I dropped that family of processors when I figured out that htey claim "32 bit", but it is actually a 16 bit processor with a few 24-bit and 32-bit enhancements. It breathes 16 bits all around, and then the address bus is upgraded to 16 bits, and some registers to 32bits. 

Next, to run things fast, you get to move your code from flash to RAM yourself. Ugly. 

Nah. C2000 is not my favorite.... 

lizardmech
Offline
Last seen: 9 months 3 weeks ago
VESC Original
Joined: 2017-06-02 19:21
Posts: 80

You don't use a C2000 for the CPU it's the peripherals that make them worthwhile, they have better PWM, ADC and specialized motor control hardware when compared to general purpose ARM MCUs. The latest ones have hardware accelerated trigonometry functions in the FPU, much of the functions which make up FOC and sensorless observers gets  done in hardware with barely any CPU use. On top of that they have a secondary stand alone FPU that has access to peripherals and can be running other control loops without touching the CPU. On one of the 200mhz versions I have it has an example project that does an entire FOC loop in 1 microsecond, I don't think even the STM32H7 with a 400mhz M7 could do that, you end up FPGA motor control performance in a $20 MCU. Of course the limitation is it only really does motor control and various power conversions well. 

benjamin
Offline
Last seen: 6 days 20 hours ago
Joined: 2016-12-26 15:20
Posts: 357

Very nice experiments, it seems it is possible to deal with saturation at higher currents with FOC. I will do some experiments myself and see how much I can push it. Maybe an alternate observer for high saturation applications could become a configuration option.

lizardmech
Offline
Last seen: 9 months 3 weeks ago
VESC Original
Joined: 2017-06-02 19:21
Posts: 80

Would they be using those odd phase voltage filters to aid the observer in difficult situations?

masterdev
Offline
Last seen: 2 weeks 5 days ago
VESC Free
Joined: 2019-09-29 06:09
Posts: 2

How VESC performs with multirotar drone? Or it is only built for RC Car skateboard users?