You are here

Experience with QS120 IPM motor?

13 posts / 0 new
Last post
marvinrau
Offline
Last seen: 7 months 2 weeks ago
Joined: 2020-07-24 12:10
Posts: 5
Experience with QS120 IPM motor?

Hello,

I am using a VESC 75/300 to control an QS120 motor to power an electric motorbike.

https://www.qsmotor.com/product/2000w-mid-drive-motor/

The rotor has integrated magnets in V-formation which makes the motor an IPM. I am controlling the vesc via CAN. The electric throttle is robust against electro magnetic interference. An STM32F103 reads in the signal, filters and maps the signal to write them on the can-bus. The motor is current controlled.

foc detection:

14 mOhm, 60uH, 35mWb

The motor is controlled in foc with hallsensors. At higher ERPM it runs without hallsensors. I had to lower the flux linkage from 35 mWb to about 14mWb to get the motor running. I was playing with the switching frequency between 20kHz and 60kHz (50kHz worked best). At this point I did not read out the Ld-Lq. Please take a look at the current in the attached diagram.

Motoreinstellungen_05_18.pdf

When switching to sensorless at 5400 erpm, the motorcurrent is way smoother. In the first diagram the switching point is marked with the vertical line. In the second diagram the switch is before the vertical line. Running foc completly sensorless is not an option. The motor only starts when applying a push. I have set the switching limit very low to 700 erpm but the motor sill has problems to handle phase currents over 150 A. The motor can handle currents up to 275 A says the supplier.

After reading out Ld-Lq and implementing the value of about 45 to 75 uH to motor runs worse than before. At higher rpm it sometimes stops. Driving the bike is not possible even below 150A. Please take a look at the video. Furthermore, the motor makes some electrical noise that is not common. It sounds like cracking. It is not mechanical.

I guess there is a problem with the hallsensors like electro magnetic interference. But the problem occurs even with small currents/no load as shown in the video.

Test.zip

My question: Does anybody have experience with this motor in combination with vesc?

I do not know what to do. Would be nice to get some help!

 

frank
Online
Last seen: 1 min 48 sec ago
VESC BronzeVESC FreeVESC GoldVESC OriginalVESC Silver
Joined: 2016-12-27 20:19
Posts: 553

Did you try out the latest BEAT VESC-Tool?

marvinrau
Offline
Last seen: 7 months 2 weeks ago
Joined: 2020-07-24 12:10
Posts: 5

Hi Frank,

thank you for the answer. No I haven´t tried it yet. Do you have a download link? I couldn´t find one by now.
 

Markula
Offline
Last seen: 1 month 1 week ago
VESC FreeVESC Original
Joined: 2019-03-22 12:33
Posts: 23
vadicus
Offline
Last seen: 1 week 4 days ago
VESC Free
Joined: 2018-08-17 07:26
Posts: 305

I've been using the same motor (at least it looks the same and also 2kw from QS) and have had my share of weirdness early on. In my case, most issues were self inflicted as I was in the process of developing the hardware. Today, after countless modifications of the hardware and running the 5.02 firmware, I can say it runs acceptable with or without load. 

One prominent issue was with the sensorless transition as described here: https://vesc-project.com/node/2588 

After some digging, I found this was a known issue with the firmware, at least, earlier versions. Running 5.02 today and with the deadtime set to 160nS, the problem is 'almost' gone. Under high loads, you can still hear the transition, but it doesn't cause it motor sputtering or at least not dramatic anymore. The sensorless control also improved quite a bit in the last few versions, so I can start the motor from standstill even without sensors. In fact, without sensors, it starts and runs smooth at any RPM. Another large motor I've run my controller with successfully is a qs205. I have it on my test bike running 109v 350A motor /100A battery without issues for a few months now.

Now, I don't know if my qs120 and yours are exactly the same but my detected values are somewhat different. It seems like your flux linkage may be a bit too high. If it's off, it would cause cogging and overcurrents. 

My hardware is very much different from trampa, so I wouldn't expect resistance or inductance to be easily transplantable to another controllers but flux linkage should be the same if the motors are the same.

But for the heck of it, give it a try:

2021-06-01_22-03-49.jpg

 

 

marvinrau
Offline
Last seen: 7 months 2 weeks ago
Joined: 2020-07-24 12:10
Posts: 5

Thank you for the link Markula!

marvinrau
Offline
Last seen: 7 months 2 weeks ago
Joined: 2020-07-24 12:10
Posts: 5

Hi vadicus,

thank you for sharing your experience! I will try updating the Firmware again and implement your measured flux linkage.
What Ld-Lq did you measure? Just to get a feeling for the value or are you not using this function?
Did you manage to run the qs120 2kW with 350A, too?

I was think of using the qs205, too. Are you running this motor in a motorbike?

I will try firmware 5.03, too. The improvements seem promising.

vadicus
Offline
Last seen: 1 week 4 days ago
VESC Free
Joined: 2018-08-17 07:26
Posts: 305

Regarding the Ld-Lq diff, my measurements will change depending on the duty cycle used. Below are some examples measured at 0.1, 0.2 ... 0.7 duty cycle. This is at 83v. As you can see, the inductance remains mostly consistent but the Ld-Lq will get smaller as the duty cycle increases. MTTPA works at values up to 10 or so as far as I remember. I haven't used it beyond just testing. There is a noticeable RPM bump above the based speed.

I've loaded the 120 up to 200A phase or so, that's what my bench setup allows to do. I don't see why it would not run at 350A phase. I am running the qs205 on my bike and 340A phase max and 100A battery. I can probably do more easily, just don't need that much.

 

Inductance: 66.17 uH, ld_lq_diff: 13.46 uH (9.89 A)

Inductance: 64.28 uH, ld_lq_diff: 13.48 uH (19.14 A)

Inductance: 67.04 uH, ld_lq_diff: 10.70 uH (36.29 A)

Inductance: 67.90 uH, ld_lq_diff: 7.69 uH (44.33 A)

Inductance: 67.73 uH, ld_lq_diff: 5.37 uH (53.49 A)

Inductance: 68.32 uH, ld_lq_diff: 5.09 uH (61.91 A)

marvinrau
Offline
Last seen: 7 months 2 weeks ago
Joined: 2020-07-24 12:10
Posts: 5

Hi Vadicus,
I could figure out one big problem. There were electromagnetic disturbances on the hall sensor cables. With smaller pullups and a protection for the sensorcables I could get the motor to run. The signal was stable until 250A phasecurrent.

 

vadicus
Offline
Last seen: 1 week 4 days ago
VESC Free
Joined: 2018-08-17 07:26
Posts: 305

Yep, it's pretty common for these motors to have the hall sensor wiring bundled wih the phase wires. I have not measured how much the high current in phases affects the hall signals. I use 2.2k pullups and isolate hall signals from the rest of the control circuit to prevent voltage spikes direct into the MCU.

Running halls at higher RPMs does make the motor run a bit rough. There seems to be also a gap/momentary power loss when switching from halls to the sensorless mode. I usually set the sensorless switching threshold pretty low, like 500-1000 eRPM. As soon as the observer picks up from sensors, the motor works fine at any load. I am running fw 5.02 which allows to basically start this motor from 0 without sensors at all. It doesn't work as well on qs205 that I run on my bike. Still need know the rotor position to get the motor going from a standstill.   

pe
Offline
Last seen: 1 month 1 week ago
VESC Free
Joined: 2021-10-22 13:03
Posts: 7

I am using the VESC 100/250 with a QS 6000W hubmotor and also find it very rough running on hall sensors. 12inch 5000w scooter motor for electric scooter, motorcycle, tricycle, etc. (qsmotor.com)

Do you suggest shielding the Hall sensor wires to avoid EMI ? I am going to use the beta FW 5.03 as was suggested by Frank. Will let you know the results. 

Thanks @vadicus and @marvinrau

benjamin
Offline
Last seen: 16 min 7 sec ago
VESC FreeVESC OriginalVESC Platinum
Joined: 2016-12-26 15:20
Posts: 475

I did some testing with a hub motor the other day and noticed that the hall sensor cables pick up noise easily. Increasing the hall sensor extra samples setting helped a lot in my case. You can try 5 or 6 extra samples. Also, I would recommend the 5.03 beta firmware and vesc tool: https://vesc-project.com/node/2859

vesc_tool_hall_samples.png

pe
Offline
Last seen: 1 month 1 week ago
VESC Free
Joined: 2021-10-22 13:03
Posts: 7

Tack så mycket Benjamin, this 5 samples did wonders!