My new 18 FET TO-247 layout riding video page 10

[Arlo1]

I'm interested in the commutation frequency.

You mean the motor magnet VS stator frequency?
So its 0 at 0 rpm and its the magnet poll pairs times the mechanical rpm for any given rpm.
Or are you asking what the limit is?
I think its a variable limit based on design in the range from ~1000hz-2000hz.

 

[toolbag]

I have some high pole count inrunners that are on deck for load testing this and other controllers. They get up to ~1600 Hz when you're really pushing them, but normal operation is more like 1200 Hz. The two motors I usually use to test controllers are 20 and 22 poles, but there's a slower torquey 46 pole motor that'd be great for a whole lot of stuff if only there was a decent FOC controller out there!

The future of BLDC design is inescapably in higher pole count motors (as we're seeing in ebike motors more and more), so it's exciting to see what this particular inverter can do.

 

[HighHopes]

interesting toolbag, increased pole count is in my plan too. glad we feel the same about that.

 

[zombiess]

What is gained from a higher pole count?

 

[Arlo1]

From what I have read not a lot but I think a touch more torqe at lower rpm.

 

[HighHopes]

it totally depends on what you DO with those extra poles

 

[zombiess]

it totally depends on what you DO with those extra poles

Multi parallel/phase

 

[toolbag]

Currently spinning zombiess' controller with a load and getting some strangeness on the current waveform. We're erroring out in various ways at a variety of power levels that seem dependednt on the drive mode 3 coefficients (edit: amplitude control at 15x defaults let us do 10kW). The lower we make them the worse it gets. Above is a screenshot of the current waveform just before an error.

 

[zombiess]

Preliminary numbers are in. 9kW @ 110V sagging to 97V after 4mins on a 12uH 10 pole pair motor. MOSFET case temp max was 65C. 150A phase 88A batt.

Problem with tuning was the amplitude settings. I had to increase them almost 10x at 110v. Funky sine shape was caused by pwm freq. Increased it from 18kHz to 25kHz. Need to experiment more on this.

 

[toolbag]

The facts are in: this controller rocks!
https://youtu.be/gSKvJtzarUc

Lookin' goooood

 

[Arlo1]

Hey man. I don't know what happen to cut you off but I had to go anyways. The settings I used with colossus I have not used for ~2-3 years since v 1.xx I had the settings for 20 30 and 40 khz.
20khz I have written down as
Y= 1024
B = 31.326157
C = .106480
D = 1.33589211

Anyways good luck. The 3 phases look funny but PWM and F sample will do weird things as well I noticed a weird thing where pmw goes to a certain point and the switch never happens completely before the opposite command is made. IE the high side is on then commanded off for .5us but to turn off enough to get though the miller plateau and stop conducting it would need ~.8 - 1us minimum maybe your running into that.

As a matter of fact i remember lebowski saying something about a 3rd harmonic he overlay-ed.... That was also a long time ago

 

[HighHopes]

cool, well i think you're finally getting the hang of this power electronics stuff.

call me when u ready to do the full-up car :wink:

 

[zombiess]

We ended the day at 12.5kW running for 2mins 150A phase 120A batt. Tomorrow we will apply heat sink compound between the heat spreader bar and the case, right now its just metal on metal.

We might have to step up to a 2T motor as we pretty much maxxed out a Joby JM1-3TY with a 37x13 prop.

 

[Lebowski]

Hey, cool which version of the controller chip are ypu using there ?

Nice to see you're using Labview, I'm a big fan of Labview myself. I use it to real-time observe the internal variables of the controller ic (over rs232). With the 2.40 you can stick 8 temp sensors all over the motor and controller, and observe the temperatures over rs232...

 

[zombiess]

I'm using 2.21

So any thoughts on tuning? Does having to set the amplitude 10x higher to prevent the controller from conking out make sense?

 

[Lebowski]

I'm using 2.21

So any thoughts on tuning? Does having to set the amplitude 10x higher to prevent the controller from conking out make sense?

Can you post menu g, so the control loop coefficients ? The default settings are what works good on the few motors that I have, so i'm not really surprised you need much higher values for the amplitude loop in your high power setup. I'm more curious to see whether all the coefficients make sense to me.

 

[zombiess]

Here is a dump of my settings from where we stopped yesterday. We are going to try and get some "continuous" power handling numbers. I'm looking for thermal equilibrium and I don't really want anything to get over 65C for a continuous rating.

0x0AEF  0x0006  0x119B  0x0021  0x0002  0x0002  0x7FBC  0x01CC
0x0000  0x0074  0x8000  0x0180  0x0180  0x0446  0x0400  0x0400
0x0400  0xAAAA  0xAAAA  0xAAAA  0x0258  0x0267  0x001B  0x0301
0x000A  0xFF8C  0x0680  0x0199  0x0614  0x0851  0xFFFF  0xFFFF
0xFFFF  0x00BC  0x02A9  0xFFFF  0xFFFF  0x0C03  0x00E6  0x0902
0x0000  0x030C  0x0258  0x0196  0x0000  0x0000  0x0001  0x0002
0x0000  0x4CCD  0x000C  0x0000  0x00F0  0xFFFF  0xB333  0xFFF4
0x0000  0xFF10  0x0000  0x07AE  0x0018  0x0000  0x01E0  0xFFFF
0xF852  0xFFE8  0x0000  0xFE20  0x0003  0x0000  0x0078  0x0000
0x0000  0xFFFD  0x0000  0xFF88  0x0000  0x0000  0x04B0  0x003C
0x0000  0x0001  0x8000  0xFB50  0xFFC4  0x0000  0xFFFE  0x8000
0x00F0  0x000C  0x0000  0x0000  0x0000  0xFF10  0xFFF4  0x0000
0x0000  0x0000  0x1F4D  0x0321  0x011B  0x055B  0x0300  0x0038
0x00F0  0x02EB  0x014F  0x0008  0x00C6  0x0042  0x000F  0x0CCC
0x0000  0x03E8  0x00C8  0x4C31  0x03B6  0x6000  0x0399  0x0150
0xFFFF  0xC519  0x764B  0x5482  0x41B3  0x35C3  0x2D7A  0x276B
0x22C9  0x1F1E  0x1C28  0x19B5  0x17A6  0x15E6  0x1463  0x1312
0x11EB  0x10E4  0x0FFB  0x0F28  0x0E6B  0x0DC0  0x0D23  0x0C94
0x0C10  0x0B97  0x0B27  0x0ABF  0x0A5F  0x0A05  0x09B1  0x0962
*

 

[Lebowski]

0x0AEF	0x0006	0x119B	0x0021	0x0002	0x0002	0x7FBC	0x01CC
0x0000	0x0074	0x8000	0x0180	0x0180	0x0446	0x0400	0x0400
0x0400	0xAAAA	0xAAAA	0xAAAA	0x0258	0x0267	0x001B	0x0301
0x000A	0xFF8C	0x0680	0x0199	0x0614	0x0851	0xFFFF	0xFFFF
0xFFFF	0x00BC	0x02A9	0xFFFF	0xFFFF	0x0C03	0x00E6	0x0902
0x0000	0x030C	0x0258	0x0196	0x0000	0x0000	0x0001	0x0002
0x0000	0x7333	0x0012	0x0000	0x0168	0xFFFF	0x8CCD	0xFFEE
0x0000	0xFE98	0x0000	0x0B85	0x0012	0x0000	0x0168	0xFFFF
0xF47B	0xFFEE	0x0000	0xFE98	0x0003	0x0000	0x0078	0x0000
0x0000	0xFFFD	0x0000	0xFF88	0x0000	0x0000	0x0078	0x0006
0x0000	0x0000	0x0000	0xFF88	0xFFFA	0x0000	0x0000	0x0000
0x00F0	0x000C	0x0000	0x0000	0x0000	0xFF10	0xFFF4	0x0000
0x0000	0x0000	0x1F4D	0x0321	0x011B	0x055B	0x0300	0x0038
0x00F0	0x02EB	0x014F	0x0008	0x00C6	0x0042	0x000F	0x0CCC
0x0000	0x03E8	0x00C8	0x4C31	0x03B6	0x6000	0x0399	0x0150
0xFFFF	0xC519	0x764B	0x5482	0x41B3	0x35C3	0x2D7A	0x276B
0x22C9	0x1F1E	0x1C28	0x19B5	0x17A6	0x15E6	0x1463	0x1312
0x11EB	0x10E4	0x0FFB	0x0F28	0x0E6B	0x0DC0	0x0D23	0x0C94
0x0C10	0x0B97	0x0B27	0x0ABF	0x0A5F	0x0A05	0x09B1	0x0962
*

can you try the hex above ? I made the amplitude loop option j equal to 0, as there is a programming error there (fixed in 2.40) which
can be circumvented by using 0 for j. This may fix the amplitude problems, so I reduced the other two settings by 10. Further I made the
1st and 2nd order coeff in the phase loop equal for drive 2 and 3. To reduce conking out, try increasing the error filtering time (is now 5 msec) to maybe 10 or 15 msec.

Dont forget to re-do the motor inductance measurement when you switch to a different motor.

 

[zombiess]

Tried your settings lebiwski, but we ran into issues starting up, transitioning mode 2 to mode 3 and conking out while running over 10kW. Went back to my previous settings and increased the filter from 5ms to 10ms. We ran out of time to experiment more.

We did manage to get close to a "continuous" power handling with some airflow from the suction side of the prop. That was 7.5kW at 100v and the MOSFET case got to ~63C after 10mins and was leveling off at 0.75C rise/min. I feel comfortable calling this a 5kW continuous controller, but it should be better quantified.

I'll post up some pics and data later.

 

[izeman]

Nice to see you're using Labview, I'm a big fan of Labview myself. I use it to real-time observe the internal variables of the controller ic (over rs232)..

could you give more detail what i can see from the rs232 interface? as you may know, i plan to build a test stand for the motor to try different settings to make the MAC work over the autumn/winter period. it would be great to visualize some information the controller has. i guess it could be great in finding problems with the settings.

 

[Lebowski]

Tried your settings lebiwski, but we ran into issues starting up, transitioning mode 2 to mode 3 and conking out while running over 10kW. Went back to my previous settings and increased the filter from 5ms to 10ms. We ran out of time to experiment more.

We did manage to get close to a "continuous" power handling with some airflow from the suction side of the prop. That was 7.5kW at 100v and the MOSFET case got to ~63C after 10mins and was leveling off at 0.75C rise/min. I feel comfortable calling this a 5kW continuous controller, but it should be better quantified.

I'll post up some pics and data later.

Hm, strange. are you still using 1.5 for option j or did you make this one 0 ? (recommended !). The 60 should be fine but the 1200 is high (will introduce a lot of noise), I would try 300 instead...

The 2.40 has support for temperature sensors and can throttle down (limit the phase current) automatically if the sensors get too hot.

 

[Lebowski]

Nice to see you're using Labview, I'm a big fan of Labview myself. I use it to real-time observe the internal variables of the controller ic (over rs232)..

could you give more detail what i can see from the rs232 interface? as you may know, i plan to build a test stand for the motor to try different settings to make the MAC work over the autumn/winter period. it would be great to visualize some information the controller has. i guess it could be great in finding problems with the settings.

this is in the manual, page 20

 

[izeman]

Nice to see you're using Labview, I'm a big fan of Labview myself. I use it to real-time observe the internal variables of the controller ic (over rs232)..

could you give more detail what i can see from the rs232 interface? as you may know, i plan to build a test stand for the motor to try different settings to make the MAC work over the autumn/winter period. it would be great to visualize some information the controller has. i guess it could be great in finding problems with the settings.

this is in the manual, page 20

thanks. the section is so small and after the can setup section, so i overread it
do you think that the information collectable here and made visualized could help us/me/you to find out what's wrong with the MAC?
if so i would try to make it work.

 

[zombiess]

Hm, strange. are you still using 1.5 for option j or did you make this one 0 ? (recommended !). The 60 should be fine but the 1200 is high (will introduce a lot of noise), I would try 300 instead...

The 2.40 has support for temperature sensors and can throttle down (limit the phase current) automatically if the sensors get too hot.

I set option j to 0 when we reran the test. Unfortunately we ran out of testing time due to making too much noise on a Saturday. Spinning a big propeller is not a quiet experience

I'm looking forward to the features in the new version.

I'm not sure why I have to set the amplitude settings so high, but it does function. I need more test stand time to figure things out. The controller still conks out to Drive Mode 0 with very rapid throttle movement, but it can restart after the motor comes to a complete stop with the settings I posted. Unfortunately I had no time to test the flying restart option.

The settings you posted would cause a fault to occur on the gate drivers which most likely signifies an over current event of some sort. I latch the controller off when the gate driver signals a fault which requires a power cycle to restart.

I need more time on a test setup to improve the tuning. I'm really happy that I succeeded in finding a power rating for my controller. I believe it is also capable of short 20kW bursts @ 100V, but with a normal Cromotor I've only been able to get ~170A bursts because I don't have enough load / saturating the motor.

On the 12.5kW runs we saw sustained 150A phase currents which is 50A per MOSFET. The temps only made it into the low 60C range after 2 mins of run time and they started off in the 40C range due to previous testing. We were unable to perform some of the thermal / power testing I wanted to due to not having remote temperature read outs.

I did manage to get some good info on the internal case temp vs MOSFET case temp. Using a FLIR gun I was able to see that there was a delta of about 4C between the MOSFET case temp and the external enclosure temperature after server minutes of running.

I created this controller with the plan of having a high burst rating compared to continuous running conditions. I used a 0.25" thick aluminum heat spreader bar to add thermal mass before it interfaced to the case which has small heat sink fins on the top.

 

[Lebowski]

I'm not sure why I have to set the amplitude settings so high, but it does function. I need more test stand time to figure things out. The controller still conks out to Drive Mode 0 with very rapid throttle movement, but it can restart after the motor comes to a complete stop with the settings I posted. Unfortunately I had no time to test the flying restart option.

The settings you posted would cause a fault to occur on the gate drivers which most likely signifies an over current event of some sort. I latch the controller off when the gate driver signals a fault which requires a power cycle to restart.

That's strange... the 1200 and 60 (especially the 1200) makes the signals quite noisy, and I would imagine they make it harder for the output stage. And vica versa, 120 and 6 would be much more smooth. The controller IC is programmed such that it always observes the deadtime... what deadtime did you use again ? One difference with the 1200, this will cause it to clip the amplitude much earlier causing stretches of 100% or 0% dutycycle (in the peaks of the sine waves). Maybe this is what your output stages likes ? I can imagine 0% or 100% (so no switching of FETs) kind of resets your saturation detection ?