20Kw motor and 300A ESC (ultralight aircraft propulsion system)

nicobie · Aug 11, 2022

Impressive :warn:

Glad to see you've done your homework. Mad skills.

Mihai_F · Sep 1, 2022

I'm back with some updates, here is a continuation form the magnet coverage topic. https://endless-sphere.com/forums/viewtopic.php?f=30&t=117251

So, basically i added two 5x5x20mm N45 magnets end to end to form a 40mm long magnet that i inserted between the existing magnets, i positioned them so that the 40 rotor poles still go N-S-N-S-N-S with equal spacing.
After the magnet insertion i still had left 0.35mm between poles, i placed them to touch the big magnet all on the same direction.
This little spacing (0.35) was almost enough space to fill them with some lead foil in order to balance the rotor, but i was still 3 grams out.
The final result after this modification was very good, the before mod Kv was 36 (3300rpm no load at 91V) and after the mod is 34Kv (3070rpm no load at 91V), but now comes the good part, before the mod i had a 1000RPM drop from no load to full load (porp), now i have only 700RPM drop from no load to full load (porp), witch means it gained some torque as the Kv went down.
And the best part is that it runs supper smooth, no more jerkiness, no more wrong and weird hall signals, and the radio interference has dropped a bit and now the squeltch can clear it, mainly because it runs smooth and draws current more smother, all that jerkiness creater spikes of current and EMI.
The motor max temperature at 16kw went down by a few ~4 degC.
The weight pelanlity is only 300 grams added, but is way outweighed by the benefits.
I simulated at the ground a 30 minute flight, 40 sec 16Kw, 2 minutes 10Kw, 6 min 8Kw, and 21 min at 6 to 8 Kw.
All went well, except one thing, during this simulation i monitored the system parameters all looked ok, but at the end of the test in last 2 minutes i noticed an faster than normal decrease in motor RPM (throttle lever not moved), so i started to look for the cause, and i noticed that one of the batt cells (no.15) has dropped to 1,3V all other cells ware at 3,4V to 3,45V , i cut the power immediately, battey temp was only 40 degC, ambient was 25degC. A after the motor was stopped i left the system powered for about 5 minutes to monitor the battery, and cell no.15 went from 1,3V to 2,7V in about 3 minutes then stabilized there. So i said this is not good, why would that happen cell no.15 (and no.16) is only 2 years old (2 years ago i mod the pack from 20s to 22s) and all other 20 cessl are 7 years old, it makes no sense, then upon pack disassembly and investigation i noticed that cell no.15 witch is made from 12 cells (5000mAh) in parallel, had one tab of one cell unsoldered from the rest 11. Now one cell off out of 12 meas 55Ah (11p) instead of 60Ah (12p), but that 5Ah difference it might not be enough to cause what happened, or maybe being 5Ah short it got discharged faster than the other cells, but then the rest ware at 3,4v, a bit far from being discharged compleatly. I checked all other cells and all are fine and soldered well. So i have to charge cell no.15 back slowly and discharge it separately to check for current and capacity, but this is no good i broke the li-po rules, and i did not have the worning set if one of the cells goes below 3V, bummer.

So here are a few photos from motor mod, here is also my rotor puller.

Mihai_F · Nov 14, 2022

Hy guys, i come with some updates, i completed an endurance flight test, it flew 30min (27min in air), used capacity was 41Ah, and remained 10min for rezerve, plus a 5Ah down rating because of old battery. Average power usege was 5kw. Battery is fine, (no damage after last test when i over discharge the cell no15), all cels are +/-0.07v of each other, after this flight. Max batt temp was 46degC. Ambient temp was 16degC. Motor max temp was 71degC after 3min at 16Kw, controller max temp was 26degC. Altitude gained was 280m. Next plan is to try to go all the way to the max capacity to see maximum possible powered flight time.
One problem that i still did not solve is radio interference, i managed to reduce the received amount by moving the antenna further way from motor and controller, but there is still plenty, i can hear the comms very well no buzzing, i can transmit well, the ground control tells me that i have some buzzing, but when there is no comms my squelch at max does not clear my buzzing interference, all these only when motor runs, at 75% throttle buzzing is worst.
These EMI stuff is, preaty difficult to fix, i have no way to measure it to see what emits it... I tried using the FFT function of the scope, but the PC and others things powered with SMPS generates more EMI than my system, AM reception is verry susceptible to EMI. I talked to some technical guys from the airport, and they said i might not be able to eliminate it from being emited, and my best chanse is to shield my system and my radio and antenna to not pick it up... If you have any ideas or suggestions please feel free, i have reached the edge of my knowledge with this one....
Here is the link to the flight video : https://youtu.be/-KKLN59jOF8
Cheers!

mxlemming · Nov 16, 2022

Mihai_F said:
Hy guys, i come with some updates, i completed an endurance flight test, it flew 30min (27min in air),

One problem that i still did not solve is radio interference, i managed to reduce the received amount by moving the antenna further way from motor and controller, but there is still plenty, i can hear the comms very well no buzzing, i can transmit well, the ground control tells me that i have some buzzing, but when there is no comms my squelch at max does not clear my buzzing interference, all these only when motor runs, at 75% throttle buzzing is worst.
These EMI stuff is, preaty difficult to fix, i have no way to measure it to see what emits it... I tried using the FFT function of the scope, but the PC and others things powered with SMPS generates more EMI than my system, AM reception is verry susceptible to EMI. I talked to some technical guys from the airport, and they said i might not be able to eliminate it from being emited, and my best chanse is to shield my system and my radio and antenna to not pick it up... If you have any ideas or suggestions please feel free, i have reached the edge of my knowledge with this one....
Here is the link to the flight video : https://youtu.be/-KKLN59jOF8
Cheers!

Congratulations! Always great stuff.

Ideas... Just ideas for the emi and radio interference. I do a lot of this for work, and honestly, having a test center with 100kusd of equipment helps a lot... But you can still replicate the problem and try things so...

I'm presuming you're using AM on the 12xMHz bands? I recall you have ringing at 60 ish MHz which almost certainly has harmonics at 2x that frequency. That ringing is probably being modulated at the commutation frequency. You have a few angles of attack... Eliminate the carrier or eliminate the modulation or eliminate the coupling.
1) the carrier frequency can be attenuated potentially with ferrites, like the clamp ones you sometimes get on per supply cables. Try putting them around the input and output cables to the controller. Try common mode (one on all 3 cables for the motor and the battery pair) and one on each cable.
2) the modulation probably comes from the commutation rate... You could try various things, tunning your pids out changing the advance angle to avoid current spikes, or perhaps move to sinusoidal or even Field oriented control.
3) the wavelength at 120MHz is about 2.5 meters, so theoretically the aerial is fairly long (quarter wavelength is 67.5cm) so it's likely the cables transmitting, though it could be a dipole antenna from your PCB.
4) try grounding to your frame at many points or completely isolate it.
5) controller in a big metal box with ferrites on the exits
6) I presume you've got the radio powered from a separate power source? And that is isolated from the frame etc? Conductive coupling is usually far stronger.
7)Test to see if it's the battery cables by adding extra bus capacitance and seeing if the interference changes. You could also try adding extra FET output capacitance, perhaps you can change the ringing frequency?
8) Completely screening the cables with copper tape can help but you need to eliminate all gaps. Depending on the nature of the emission source the copper tape can just radiate equally.

Usually, most things will make barely any difference and one thing makes a huge difference. Good luck

GbR · Jun 15, 2023

Mihai_F said:
Hy guys, i will move here the discussion about my 20Kw motor and 300A esc. This is been a project for my past 6 years, it's been a huge learning curve. Before i started this project my experience with electronics was with power audio circuits, analog stuff, and nothing digital.
I will start first with the electronic speed controller because that took most of the time to learn and develop.
So i begun with a BLDC monolithic driver chip, the MC33035, and i made the circuit with wires, it took me a month to get it running and be able top spin a small BLDC sensored motor, it looked like this:
View attachment 305026
It was capable of about 12w ,24v at 0.5A, then i learned what MOSFET stands for: Magically Obliterated, Smoke and Fire Emitting Transistor, i was amazed how much smoke can make a TO220 packge , inductance was my friend (bundle of wires), but at that time i was blissfully unaware of it. Then i learned a few things and moved the project to PCB, i was so impressed about my first toner transfer technique PCB, but from electrical stand point was crap, again i did not know that, this is how it looked,
View attachment 305028
This was the first year and it was capable of about 48w ,24v at 2A, before it wood magically let the smoke out of the FET-s, at that level i learned about that thing called "ringing"... .Then i learned a lot more things, i added a few functionalities like safestart, rpm, voltage, current and tep readout, all made on 4 digit 7 segment display, it was a lot of work, and it looked like this, (the big motor was made at this point) ,
View attachment 305029
This was the second year it was capable of about 120w , 36v at 3.3A, o about that time i realized i need to move up a bit and enter the AVR microcontroller world, i knew nothing about them, and all i knew about C programming language was that there was some things called "if" and "else"... , then another year of huge learning curve and another variant of the controller came witch was way easier to customize and change things, hence programmable microcontroller, it looked like this
View attachment 305030
This was in the third year became V1.0 it is capable about 2.5kw 20S(84v) 30A (current sensor limited),IRFP4468 FET-s , it has BMS and a lot of functionalities including 128x64 GLCD with menu where a lot of parameters can be customized, then the learning curve was the most steep making the ESC to work up to 16kw, then i learned what inductance can really do, and what miller plateou, reverse transfer capacitance, stray inductance, and a lot of goodies that give a lot of headakes.... , then came next variantwitch looked like this
View attachment 305031
View attachment 305032
View attachment 305033
This was in the fifth year, v1.2 witch worked up to 16Kw at 20S (84V), IRFP4468 FET-s (6/plase), PCB-s made by me. It was the first and last time i tested the motor by holding the table just myself... :lol: :lol: , it was exiting seeing my work succeed, but kinda scary because i could barely hold the table, the prop was pulling 78kgf

This was how later tests were made, the motor was strapped by the car, no more worries about flying with the table
View attachment 305034
This was in the 6-th year the flight testing.
The motor was made by scaling up a smaller motor, (was still 20pp and 30 teeth but only 10kw), i did calculations for the winding, i drew it in 2D cad, went to a machine shop and told them what parts i need, the stator laminations were laser cut (unfortunatly 0,5mm thick), then a full day of winding. It was not even by far such complex to make as the ESC.
It's mass is 7.7kg, 216mm OD, 40mm stator hight, 40x10x5 N45 magnets, 6mohm and 10uH per phase, delta terminated, 3 halls and 1 temp sensors. It can do 16Kw continuous and 22Kw max (1 min) at 20 deg C ambient, being air cooled ambient temp affects max power.
And here are some photos of the system.
View attachment 305035
View attachment 305036
View attachment 305037
View attachment 305038
View attachment 305039
View attachment 305040
View attachment 305041
View attachment 305042
View attachment 305043
View attachment 305044
View attachment 305045
View attachment 305046
View attachment 305047
View attachment 305048
View attachment 305049
View attachment 305050
View attachment 305051
And in the present the 7-th year V1.3 (IRF150P220 FET-s) gives headakes as mention in my other post (300A ESC help...). The change (FET-s) was made in order to go with 22s battery.
It looks like this, with PCB-s made by specialized company, power board has 1mm thick water jet cut busbars.
View attachment 305052
View attachment 305053
View attachment 305054
This is just a short story, the hole experience has waaaaaay more detalis and waaaay more headakes.
There is a lots of room for improvement, this is very low tech, and to some may be even foolish, i know, up to this point it was the best i could do, but will see what future shows up.

Amazing ! On your first few picture i see you are using MC33035 and later on looks different chip ? I'm also working on sensored bldc controller. Do you recommend MC33035 or something else ?

Mihai_F · Jul 3, 2023

GbR said:
Amazing ! On your first few picture i see you are using MC33035 and later on looks different chip ? I'm also working on sensored bldc controller. Do you recommend MC33035 or something else ?

No, (mc33035 is an ASIC), go with microcontrollers, a 32bit one, you will have waaay more flexibility and options. For start, you can go with an 8bit AVR (328) to get things going, then upgrade to 32bit (STM32).

Mihai_F · Jul 3, 2023

mxlemming said:
Congratulations! Always great stuff.

Ideas... Just ideas for the emi and radio interference. I do a lot of this for work, and honestly, having a test center with 100kusd of equipment helps a lot... But you can still replicate the problem and try things so...

I'm presuming you're using AM on the 12xMHz bands? I recall you have ringing at 60 ish MHz which almost certainly has harmonics at 2x that frequency. That ringing is probably being modulated at the commutation frequency. You have a few angles of attack... Eliminate the carrier or eliminate the modulation or eliminate the coupling.
1) the carrier frequency can be attenuated potentially with ferrites, like the clamp ones you sometimes get on per supply cables. Try putting them around the input and output cables to the controller. Try common mode (one on all 3 cables for the motor and the battery pair) and one on each cable.
2) the modulation probably comes from the commutation rate... You could try various things, tunning your pids out changing the advance angle to avoid current spikes, or perhaps move to sinusoidal or even Field oriented control.
3) the wavelength at 120MHz is about 2.5 meters, so theoretically the aerial is fairly long (quarter wavelength is 67.5cm) so it's likely the cables transmitting, though it could be a dipole antenna from your PCB.
4) try grounding to your frame at many points or completely isolate it.
5) controller in a big metal box with ferrites on the exits
6) I presume you've got the radio powered from a separate power source? And that is isolated from the frame etc? Conductive coupling is usually far stronger.
7)Test to see if it's the battery cables by adding extra bus capacitance and seeing if the interference changes. You could also try adding extra FET output capacitance, perhaps you can change the ringing frequency?
8) Completely screening the cables with copper tape can help but you need to eliminate all gaps. Depending on the nature of the emission source the copper tape can just radiate equally.

Usually, most things will make barely any difference and one thing makes a huge difference. Good luck

Hy, thanks for you reply, it's been a lot of time since my last post.
A lot has happened, one of the main things is that this project got me a JOB in electronics industry, big career change for me.
Anyway, here it goes.
I di try shieling the phase cables, unnoticeable effect.
I did put ferrite beads on every cable that goes in and out of the controller, a bit of improvement ,enough for the interference not to open the squelch at max.
Antenna grounded or isolated no effect.
Radio is own battery powered.
Controller is in a metal box well grounded
Did change the PWM SW freq up to 24khz, little effect.
All in all, moving the antenna in the tail and ferrite beads on cables had improvement.
My next approach is to try to slow down my rise/fall time for turn on/off , and tune a bit the snubbers to eliminate the ringing.
Also on my list is to migrate from 8bit AVR's to STM32, and try different controlling modes, like FOC, because at this time it is clear that the current ripple that i have is mainly part of the cause of the interference.
At this time, i'm dealing with a phase damage, it got fried at very low load (when reducing power after a runup on the ground), basically the controller worked for about 2,5 hours in total time with no functionality problems. Upon investigation i found cracked and and shorted ceramic caps (the ones on the rails between fet-s) on the damaged phase, and one completely blown ceramic cap on a good phase, I have reasons to believe that they ware stressed by to much ripple, they ware flex terminated 4,7uF 100V mlcc, so mechanical stress is out of the question, this also proves that i have to much current ripple in the controller.
If you have any thoughts with regard to mlcc failing due to current/voltage ripple.

mxlemming · Jul 3, 2023

Mihai_F said:
Hy, thanks for you reply, it's been a lot of time since my last post.
A lot has happened, one of the main things is that this project got me a JOB in electronics industry, big career change for me.
Anyway, here it goes.
I di try shieling the phase cables, unnoticeable effect.
I did put ferrite beads on every cable that goes in and out of the controller, a bit of improvement ,enough for the interference not to open the squelch at max.
Antenna grounded or isolated no effect.
Radio is own battery powered.
Controller is in a metal box well grounded
Did change the PWM SW freq up to 24khz, little effect.
All in all, moving the antenna in the tail and ferrite beads on cables had improvement.
My next approach is to try to slow down my rise/fall time for turn on/off , and tune a bit the snubbers to eliminate the ringing.
Also on my list is to migrate from 8bit AVR's to STM32, and try different controlling modes, like FOC, because at this time it is clear that the current ripple that i have is mainly part of the cause of the interference.
At this time, i'm dealing with a phase damage, it got fried at very low load (when reducing power after a runup on the ground), basically the controller worked for about 2,5 hours in total time with no functionality problems. Upon investigation i found cracked and and shorted ceramic caps (the ones on the rails between fet-s) on the damaged phase, and one completely blown ceramic cap on a good phase, I have reasons to believe that they ware stressed by to much ripple, they ware flex terminated 4,7uF 100V mlcc, so mechanical stress is out of the question, this also proves that i have to much current ripple in the controller.
If you have any thoughts with regard to mlcc failing due to current/voltage ripple.

Congratulations on the job situation! Somehow I'm still doing mechanical engineering for robotics not electronics but it's all good work so hey!

I've not seen mlcc crack from ripple or even get hot yet but iirc you were running way way up at their max voltage and generating a huge amount of ripple. Try putting 2 in series. You'll find that due to the way they derate with voltage you don't lose much capacitance.

Then go to FOC as soon as you can. If it's an option, put an abi encoder on your motor. I wired one into my controller a week ago and it blew my mind how easy it was to interface and get perfect angle measurement from it. With an encoder working, FOC is barely a few evenings work to implement on stm32. Don't bother with the lower end simple ones, go straight to f4 with triple ADC or H7, and try to get one of the ones with higher clock rate. Ignore G4, they have a serious ADC errata.

If you stuck with BLDC, try counting the pwm periods per hall segment and either implement a phase advance or just roll back the duty on the last few pwm pulses to avoid the current spikes (they normally happen at the end of the commutation phase).

Mihai_F · Jul 3, 2023

mxlemming said:
Congratulations on the job situation! Somehow I'm still doing mechanical engineering for robotics not electronics but it's all good work so hey!

I've not seen mlcc crack from ripple or even get hot yet but iirc you were running way way up at their max voltage and generating a huge amount of ripple. Try putting 2 in series. You'll find that due to the way they derate with voltage you don't lose much capacitance.

Then go to FOC as soon as you can. If it's an option, put an abi encoder on your motor. I wired one into my controller a week ago and it blew my mind how easy it was to interface and get perfect angle measurement from it. With an encoder working, FOC is barely a few evenings work to implement on stm32. Don't bother with the lower end simple ones, go straight to f4 with triple ADC or H7, and try to get one of the ones with higher clock rate. Ignore G4, they have a serious ADC errata.

If you stuck with BLDC, try counting the pwm periods per hall segment and either implement a phase advance or just roll back the duty on the last few pwm pulses to avoid the current spikes (they normally happen at the end of the commutation phase).

I have a dev board with a stm32f4, i made a few tests with ADC , timers, some filtering with floating point math, and i'm blown about it's computational power compared to 8bit AVR's.
The controller has phase angle advance implemented, basically i measure the time between commutations in us resolution, then schedule next commutation ahead of it's time (to a predeterminated phase angle advance). One interrupt is from halls that records the time, anther interrupt is timer to schedule the commutation in advance. It has 2 settings, a time offset in us, that parameter is fixed in time but variable in phase angle, the more rpm the more advance, and another parameter in electrical degrees, that is variable in time but fixed in phase angle, same advance at any rpm. So i did in excel some calculations based on current, voltage, motor inductance, and RPM, and it gave a phase angle advance curve, witch i matched with my 2 parameter setting in the controller. Basically i could predict the advance angle because my load (motor drives propeller) is increasing with rpm, so at max rpm i can only have max load (i know that phase angle changes constantly with current, voltage and rpm). So i fiddled with that phase advance curve until i got maximum RPM and power for a certain amount of current.
Now, what you say to roll back the duty cycle at the end of each commutation, i thought of that recently more and more, i would like to try that at the beginning of the commutation and at the end of the commutation in order to better match the current with the trapezoidal BEMF. This is all SW, so from a time point it can be done sooner than the hole migration to the stm32 witch includes HW, so i'l give that a try.

Mihai_F · Jul 18, 2023

One thing that i read lately, is in automotive industry in order to reduce EMI in traction inverters, a common technique is to use modulated carrier frequency for the PWM. Basically if you have 20Khz PWM carrier freq, there would be a dominant pole at 20Khz (and its harmonics) , a modulation between 10 to 30 Khz, would spread the energy of that pole across a wider range thus reducing its intensity, hence lower EMI. Here is a document explaining that.

https://www.merl.com/publications/docs/TR2020-150.pdf

I plan to implement that freq modulation technique, mainly because as i observed, the highest EMI received by the radio is at duty cycles between 70-95%, when duty cycle is at 100%, i have only the commutation freq 5-5.5khz present, a lot less often switching pulses then at 20khz. So if i spread the switching pulses across a wider range, say 16-32khz in theory i would have less peak power in the EMI.
Reducing the di/dt of the switching can be done up to some degree, ferrite beads on cables helps a bit, carrier freq modulation they say it helps, so all those added might be enough to reduce the EMI enough to accepted level, it's a case a small improvements from many sources.

Mihai_F · Aug 10, 2023

I'm back with a bit of trouble,

i made some DPT in order to fine tune the desat overcurrent protection on each phase, i found this [FAQ] UCC21750: How can we adjust the DESAT detection threshold in UCC217xx & ISO5x5x? - Power management forum - Power management - TI E2E support forums
it helped me to better understand the adjustment.

Now the problem, i was testing desat on phase W low side and phase V high side with an inductor (one phase of the motor connected) between them, i was giving 100us pulse for 450A DPT. At some point due to my mistake the microcontroller commanded 4 long pulses (certainly longer than 100us) in a row, witch damaged the power stage, it fried phase U witch strangely enough, was not connected to any load (it's switching node was not connected to any thing) , phase W had desat set up, i suspect that saved it. But what could kill phase U, witch was not commanded to do anything and not connected to any load, could the spike generated by the DPT on other phases (W and V) transfer thru to phase U. I thought that the phases involved in the test would get fried, but not. This is baffling, also adds to confusion that phase W got fried last time when throttling down the motor from high load to light load (10kw to under 1Kw) at light load it got fried. Gate driver checks out fine for phase U, it behaves and measures as other phases (except desat).
So any thoughts on this might be helpful .
Thanks in advance.

mxlemming · Aug 13, 2023

Mihai_F said:
I'm back with a bit of trouble,

i made some DPT in order to fine tune the desat overcurrent protection on each phase, i found this [FAQ] UCC21750: How can we adjust the DESAT detection threshold in UCC217xx & ISO5x5x? - Power management forum - Power management - TI E2E support forums
it helped me to better understand the adjustment.

Now the problem, i was testing desat on phase W low side and phase V high side with an inductor (one phase of the motor connected) between them, i was giving 100us pulse for 450A DPT. At some point due to my mistake the microcontroller commanded 4 long pulses (certainly longer than 100us) in a row, witch damaged the power stage, it fried phase U witch strangely enough, was not connected to any load (it's switching node was not connected to any thing) , phase W had desat set up, i suspect that saved it. But what could kill phase U, witch was not commanded to do anything and not connected to any load, could the spike generated by the DPT on other phases (W and V) transfer thru to phase U. I thought that the phases involved in the test would get fried, but not. This is baffling, also adds to confusion that phase W got fried last time when throttling down the motor from high load to light load (10kw to under 1Kw) at light load it got fried. Gate driver checks out fine for phase U, it behaves and measures as other phases (except desat).
So any thoughts on this might be helpful .
Thanks in advance.

Normally when things like this happen, it just means you did not realise you dropped a screw on it or added some code that commanded a shoot through or whatever. It's pretty unlikely that you blow through a completely separate unconnected phase, unless it was not even connected to the gate drivers/MCU left floating/... something that would allow the gates to turn on.

Good luck...

Mihai_F · Aug 28, 2023

I'm back with some study on an older issue, if you remember i had still some ringing on the gates that i attributed to source bond wire inductance.....But back in my head i was never fine with that, and it might be a culprit for my sudden mos-fets failures lately.
Well, back then because i was fixated on miller clamp, i did not even considered to increase the individual gate resistance of the paralleled 3 gates, and i have 1 ohm for the 3 gates in parallel.
After some study i believe this resistance affects that ringing, if i increase it to 2.2 or 3.3 ohms, the formed "LCR", C-iss + L-bondwire + R-g, will have more damping and that might significantly reduce the oscillations.
What do you thing on individual Rg for paralleled gates?

https://www.ti.com/lit/an/slpa020/slpa020.pdf?ts=1693225279254&ref_url=https%253A%252F%252Fwww.google.com%252F

page 3

mxlemming · Aug 28, 2023

I've tried both with and without individual gate resistors and not personally seen an issue without but I do understand the reasoning and now fit the extra resistors.

Contrary to that appnote I have found that small gate capacitors per MOSFET can be effective at taming ringing. I built up a simulation model and noted that the extra capacitor in combination with the mos gate resistance acted as a very effective damper against ringing. You have to have a resistor per gate though otherwise the capacitors served to ring against each other. This proved effective in curing someone else's board I looked at recently even though the capacitance i added was only 25% of the most input capacitance.

I'm very sceptical that this caused your failure though, you double pulse tested to over 1kA and it failed when that leg wasn't even being used? This seems very unlikely.

Mihai_F · Aug 28, 2023

mxlemming said:
...... but I do understand the reasoning and now fit the extra resistors.
......
I'm very sceptical that this caused your failure though, you double pulse tested to over 1kA and it failed when that leg wasn't even being used? This seems very unlikely.

With regard to resistors, they are there from the beginning of this version, but what i want is to change they're value from 1r to 2.2/3.3/4.7r, and see what changes. Strange enough is that capacitors 4,7n or 10n in parallel with the gates made things worst.
I want to tackle that ringing problem also for EMI reduction.
As with regard to the cause of the failure, i agree with you, very unlikely for when i did DPT.
My no.1 suspect is a cross conduction event caused by the uC going berserk for unknown reason at this point...
Since last flights where it worked, the only difference in HW is i added max232 modules on the lines for comms between display and controller, but the max232 is NOT driven by same uC for motor control...
I'l investigate....

Mihai_F · Aug 29, 2023

Well i'm back with some testing, since the controller has a phase missing (last failure) and i'l have to order parts anyway, i decided to do some barbaric tests, and that means short circuit the phase leads to test desat protection.
So basically i shorted the phase leads, the calculated inductance of them was in the range of 0.6uH, so that asks for a lot of amps in a shot time. I commanded a DPT, with first pulse 100us, and wait for 5us and than a second pulse of 10us, basically if desat would not catch it would self destruct.
But guess what, i did this test 8 times, 16 shorts, and it survived, desat protects it, each time after pulse start in 5us it stopped the pulse, with 110ish us in between pulses just like i commanded it, but the first and second pulse stopped after 5us instead of being 100us and 10us long, that is so impressive.
Now the conclusion, a pulse at 90V in 5us thru 0.6uH it calculates about 700A, in normal overload operation desat activates at about 550A on 90us pulse, so my guess is that it can't detect any faster than 5us. At this point desat has a blanketing time of 3 to 4us to ignore false triggering, i can shorten that down to 1us, but that is not fast enough to catch a shoot thru (cross conduction) event because of low inductance of the layout of the power stage. So a protection against shoot thru made from logic gates before the gate driver is mandatory, so it will never ever allow it to happen.
Now, on the quest of making it "fail proof" at this point the only thing that it can destroy it is a shoot thru event, except from water, and excessive mechanical shock

), the point is to conclude what could cause last 2 failures.
I also exchanges the individual gate resistors from 1R to 4R7, that reduced the ringing amplitude and length.
Here are the screen shots of the shorted test.
DPT

second pulse zoomed

second pulse turn on

second pulse turn off

Mihai_F · Sep 24, 2023

I'm back with some good updates, i was never completely fine with IRF150P220 mosfets on the power stage, they had some strange turnoff ringing that could never be resolved, so after some talking with zombiess regarding "new generation fets" vs "old generation fets", and i decided to swap fets to IRFP4568. There are some nice projects with these fets on the forum and they give good results, plus all the "known standard practices" work on them, like added Cgs makes them "calm down the ringing".
So after swapping fets, put back the miller clamps in the gate drive, put the turn on turn off gate resistors 2R2 ohm, individual fet gate resistors 1 ohm, added 30nF Cgs (see schematic), exchaged 4,7uF 100V bus ceramics with 1uF 200V ones , 9 pcs per phase .
I conducted all the DPT tests, and set the desat at about 700A, also narrowed down to 2us blanketing time, and did a short circuit test, all fine, i was blown how good and clean are they switching, and after ensuring every thing is fine at its place and in specs, i went out with the table and did a high power test, i simulated a 30 min flight regim test, all went fine. One thing that i was verry impressed is that with clean switching all the radio interference has decreased by about half in intensity, now i could have the radio antenna at 1m from propulsion system and with squelch at 60% and stay clean, (before all fet mods the squelch was at 90%, antenna at 3m and still buzzing).
During the test day was sunny and 27 degC ambient, so before the test start after staying in the sun for about 20 min for the setup, motor and controller was at 30 degC, battery was at 25degC, all the min, max, avg, values are in the pitcure.
The other main thing is i moved my project to Kicad, yep, big move lats of time invested, i have done so far the schematics, for the command module is version 1,5 where i migrated to STM32 uC, soon to be produced, and made the pcbs layouts for power stage and command. Next big move will be migrating motor control from 8bt AVR to STM32, code, tool chain all that stuff....
Here are some shots wit the Vds and Vgs from 700A DPT:
low side Vds turn off

low side Vds turn on

low side Vds Ggs(green) turn off

low side Vds Ggs(green) turn on

low side Vds Ggs(green) turn off (low freq ringing bwtween ceramics and dc link caps)

700A DPT low side Vds Ggs(green)

700A ish short circuit low side Vds Vdesat(green) blanketing in 2us and current ramp up in 3us total detection 5us

min max avg values

power stage schematic

power stage kicad pcb layout

Command module kicad pcb layout

zombiess · Sep 24, 2023

What is the thought process of those 1uF ceramic caps on the DC bus? Notice any difference with them vs without them? I tried something similar, but with polypropylene caps as large ceramics are prone to cracking under thermal stress + suffer dielectric effect. I don't recall seeing any notable differences when I played with it.

Looks like your gate traces are running through you power pass section, this is generally considered a bad idea at higher power levels due to coupling. Twisted pair flying leads could be a work around, however with the vibration issues for your application it may not be a good idea unless you they are mechanically secure and vibration isolated.

Mihai_F · Sep 25, 2023

zombiess said:
What is the thought process of those 1uF ceramic caps on the DC bus? Notice any difference with them vs without them? I tried something similar, but with polypropylene caps as large ceramics are prone to cracking under thermal stress + suffer dielectric effect. I don't recall seeing any notable differences when I played with it.

Looks like your gate traces are running through you power pass section, this is generally considered a bad idea at higher power levels due to coupling. Twisted pair flying leads could be a work around, however with the vibration issues for your application it may not be a good idea unless you they are mechanically secure and vibration isolated.

The 1uF mlcc on DC bus, i see them as them as HF decoupling on DC bus, and good to have there, although when going from 9pcs 4,7uF to 9pcs 1uF i did not see any big difference on the waveforms, and probably they do not do much, since there are RC snubbers on each mosfet. They are flex terminated mlcc.
Gate traces are on bottom layer and inner lower is gnd return plain, 2 separate, 1 for each HS and LS, inner upper is gnd shield, top si DC bus +and-, and phase out is bottom in between gate traces. Although fets are directly under the gate traces, they have relative short legs and bond wires, so not much inducing from there. One idea is to have gate traces between 2 Gnd planes with via stitching on a 6 layer board, but that might be overkill. So far i did not see any induced glitches on the gates, since they have miller clamps, they are held down pretty decent.
inner lower gnd plain and bottom gate traces

inner upper gnd shield and down the rest

Don Pournelle · Dec 6, 2023

Mihai_F said:
Hy guys, i will move here the discussion about my 20Kw motor and 300A esc. This is been a project for my past 6 years, it's been a huge learning curve. Before i started this project my experience with electronics was with power audio circuits, analog stuff, and nothing digital.
I will start first with the electronic speed controller because that took most of the time to learn and develop.
So i begun with a BLDC monolithic driver chip, the MC33035, and i made the circuit with wires, it took me a month to get it running and be able top spin a small BLDC sensored motor, it looked like this:
View attachment 305026
It was capable of about 12w ,24v at 0.5A, then i learned what MOSFET stands for: Magically Obliterated, Smoke and Fire Emitting Transistor, i was amazed how much smoke can make a TO220 packge , inductance was my friend (bundle of wires), but at that time i was blissfully unaware of it. Then i learned a few things and moved the project to PCB, i was so impressed about my first toner transfer technique PCB, but from electrical stand point was crap, again i did not know that, this is how it looked,
View attachment 305028
This was the first year and it was capable of about 48w ,24v at 2A, before it wood magically let the smoke out of the FET-s, at that level i learned about that thing called "ringing"... .Then i learned a lot more things, i added a few functionalities like safestart, rpm, voltage, current and tep readout, all made on 4 digit 7 segment display, it was a lot of work, and it looked like this, (the big motor was made at this point) ,
View attachment 305029
This was the second year it was capable of about 120w , 36v at 3.3A, o about that time i realized i need to move up a bit and enter the AVR microcontroller world, i knew nothing about them, and all i knew about C programming language was that there was some things called "if" and "else"... , then another year of huge learning curve and another variant of the controller came witch was way easier to customize and change things, hence programmable microcontroller, it looked like this
View attachment 305030
This was in the third year became V1.0 it is capable about 2.5kw 20S(84v) 30A (current sensor limited),IRFP4468 FET-s , it has BMS and a lot of functionalities including 128x64 GLCD with menu where a lot of parameters can be customized, then the learning curve was the most steep making the ESC to work up to 16kw, then i learned what inductance can really do, and what miller plateou, reverse transfer capacitance, stray inductance, and a lot of goodies that give a lot of headakes.... , then came next variantwitch looked like this
View attachment 305031
View attachment 305032
View attachment 305033
This was in the fifth year, v1.2 witch worked up to 16Kw at 20S (84V), IRFP4468 FET-s (6/plase), PCB-s made by me. It was the first and last time i tested the motor by holding the table just myself... :lol: :lol: , it was exiting seeing my work succeed, but kinda scary because i could barely hold the table, the prop was pulling 78kgf

This was how later tests were made, the motor was strapped by the car, no more worries about flying with the table
View attachment 305034
This was in the 6-th year the flight testing.
The motor was made by scaling up a smaller motor, (was still 20pp and 30 teeth but only 10kw), i did calculations for the winding, i drew it in 2D cad, went to a machine shop and told them what parts i need, the stator laminations were laser cut (unfortunatly 0,5mm thick), then a full day of winding. It was not even by far such complex to make as the ESC.
It's mass is 7.7kg, 216mm OD, 40mm stator hight, 40x10x5 N45 magnets, 6mohm and 10uH per phase, delta terminated, 3 halls and 1 temp sensors. It can do 16Kw continuous and 22Kw max (1 min) at 20 deg C ambient, being air cooled ambient temp affects max power.
And here are some photos of the system.
View attachment 305035
View attachment 305036
View attachment 305037
View attachment 305038
View attachment 305039
View attachment 305040
View attachment 305041
View attachment 305042
View attachment 305043
View attachment 305044
View attachment 305045
View attachment 305046
View attachment 305047
View attachment 305048
View attachment 305049
View attachment 305050
View attachment 305051
And in the present the 7-th year V1.3 (IRF150P220 FET-s) gives headakes as mention in my other post (300A ESC help...). The change (FET-s) was made in order to go with 22s battery.
It looks like this, with PCB-s made by specialized company, power board has 1mm thick water jet cut busbars.
View attachment 305052
View attachment 305053
View attachment 305054
This is just a short story, the hole experience has waaaaaay more detalis and waaaay more headakes.
There is a lots of room for improvement, this is very low tech, and to some may be even foolish, i know, up to this point it was the best i could do, but will see what future shows up.

Wow. I might eventually need 20kw + motors for a full sized version of an experimental aircraft: Login • Instagram
I had assumed that would use off the shelf motors, but maybe you could make custom motors to match my 5-bladed variable-pitch rotors if it turns out that a good match is hard to find ?
Of course this is many years away, but it seems a long running project hasn’t fazed you!

Mihai_F · Dec 6, 2023

Don Pournelle said:
Wow. I might eventually need 20kw + motors for a full sized version of an experimental aircraft: Login • Instagram
I had assumed that would use off the shelf motors, but maybe you could make custom motors to match my 5-bladed variable-pitch rotors if it turns out that a good match is hard to find ?
Of course this is many years away, but it seems a long running project hasn’t fazed you!

Finding an off the shelf for these high power motors for a specific application is not very common, they might be a partial match, but that is not ideal. For specific application you need the motor to be made to match you requirements. Electric propulsion in aviation is not very common yet, so off the shelf electric motors for typical applications are not here yet, like ICE counter parts have rotax582 or 912, etc. . There are a few manufactures that make these high power motors and have a customizable rage of specs, although they are kind of pricy. I made an assessment in order to produce my motor (better updated version) to be sold as one of from time to time and it will be around 2800 euros, or 2 pieces at once is about 2200/piece , and so on. So one of is the worst case, that is why off the shelf is not very common, no one keeps this things "on the shelf" to sell them rarely because they cost a lot. China does it, but there variety is not very broad, quality and specs are ambiguous, generally specs are over rated by a factor of 2 or 3, plus they are not very cheap either. So on that note, it can be done, but a price must be payed.

Don Pournelle · Dec 11, 2023

Mihai_F said:
Finding an off the shelf for these high power motors for a specific application is not very common, they might be a partial match, but that is not ideal. For specific application you need the motor to be made to match you requirements. Electric propulsion in aviation is not very common yet, so off the shelf electric motors for typical applications are not here yet, like ICE counter parts have rotax582 or 912, etc. . There are a few manufactures that make these high power motors and have a customizable rage of specs, although they are kind of pricy. I made an assessment in order to produce my motor (better updated version) to be sold as one of from time to time and it will be around 2800 euros, or 2 pieces at once is about 2200/piece , and so on. So one of is the worst case, that is why off the shelf is not very common, no one keeps this things "on the shelf" to sell them rarely because they cost a lot. China does it, but there variety is not very broad, quality and specs are ambiguous, generally specs are over rated by a factor of 2 or 3, plus they are not very cheap either. So on that note, it can be done, but a price must be payed.

Thanks a lot for that advice!

SammyMc · Feb 15, 2024

zombiess said:
What is the thought process of those 1uF ceramic caps on the DC bus? Notice any difference with them vs without them? I tried something similar, but with polypropylene caps as large ceramics are prone to cracking under thermal stress + suffer dielectric effect. I don't recall seeing any notable differences when I played with it.

Looks like your gate traces are running through you power pass section, this is generally considered a bad idea at higher power levels due to coupling. Twisted pair flying leads could be a work around, however with the vibration issues for your application it may not be a good idea unless you they are mechanically secure and vibration isolated.

unrelated question ,. you always emphasize on the idea of traces running through power section and routing gate/return in differential pair or kelvin connection i donot remember .
How can you apply those principles when you have mosfets in parallel ?

Mihai_F · Mar 17, 2024

Hello Guys!

Due to some personal reasons i have to sell my project, all its contents.

Selling 22Kw motor

Hello Guys! Due to some personal reasons i have to sell my project, all its contents. This is my project: 20Kw motor and 300A ESC (ultralight aircraft propulsion system) Here is what i'm asking for my propulsion system, detailed description is in the upper link on the first pages. BLDC...

endless-sphere.com

Mihai_F · Apr 2, 2024

A bit of an update. Battery degradation after 9years, stored at 20degC and at 3,8V/cell most of its life, it had a total of 10 cycles on it. One year ago it started to die, first 2 cells at once went to 0v over a time of 1 month (at that time i was doing batt measurements in storage once a month) . This year another cell went to 1 volt over a time of 2 weeks this time i'm doing measurements of storage twice a month. The idea is there is nothing to do about this from happening, the cells are puffed up , because of age, but they are constrained inside the battery case so they can not expand, they have pressures inside, and from time to time some leak somehow due to their age, and that electrolyte corrodes the pouch from the exterior, once that happens to a single cell of 12 in parallel they all die because of one that died and that discharges other 11 that are in parallel with it. This happened 4 times so far, once i catched it in time and managed to save one banc of 11parraleled (12P 22S total). All in all its time has passed sadly, they will all die one by one.
A new battery is needed for this project to continue. From VW ID 3/4/ e-tron/else battery, 2 modules 12s 48v 6.85KWh 0Z1915 592 H , 32kg each, they are NCM712 made by LG chem, and are a good fit and match, but the are kinda pricy...
Anyway, sunk fallacy cost, maybe....
Here are some pictures of the dead cell from this year, and how they are stacked in the batt case.
The extracted cells are not puffed any more because i vented them in order to be able to extract them.

20Kw motor and 300A ESC (ultralight aircraft propulsion system)

Moderator

100 W

100 W

100 kW

1 µW

100 W

100 W

100 kW

100 W

100 W

100 W

100 kW

100 W

100 kW

100 W

100 W

100 W

10 MW

100 W

1 µW

100 W

1 µW

100 µW

100 W

100 W

Similar threads