HobbyKing.com New Products Flash Sale
Reply
Thread Tools
Old Jan 28, 2004, 05:58 PM
Registered User
mmormota's Avatar
Budapest
Joined Jul 2003
2,567 Posts
LTSpice/SwitcherCad III is free and simply great.

http://www.linear.com/software/
mmormota is offline Find More Posts by mmormota
Reply With Quote
Sign up now
to remove ads between posts
Old Jan 28, 2004, 06:31 PM
Impossible? Hah!
KreAture's Avatar
Oslo Fornebu, Norway
Joined Jun 2003
3,074 Posts
getting it now.
KreAture is offline Find More Posts by KreAture
Reply With Quote
Old Jan 28, 2004, 06:33 PM
The blade numbers go up to 11
stumax's Avatar
Sydney, Australia
Joined Aug 2002
4,379 Posts
Julian, I see from the circuit you are pwm-ing the low side - doesn't it work better on the high side as it keeps the charge pumps up? Also, could you explain the function of D1,3&5? I assume you have 3 pwm channels on you microcontroller, and just use 3 output pins to switch the high side in sequence? BTW, what size bootstrap capacitor do you use, and what pwm frequency? Your commutation method is interesting. I have been thinking of different methods as well. One idea I have is to sample the back emf of all phases during the pwm off time via a 3 channel A/D. Two of the phases' voltages would then be phase shifted 120deg and 240 deg respectively, and a Fourier transform performed on all of the data, using as the frequency of the transform the motor speed (in Hz) derived from the previous commutation. This would result in the phase angle of the sampled waveform, and it would be very accurate due to the number of samples used. I use this method to read a codestrip on my homemade optical linear encoders on my cnc mill and I can resolve down to 1 micron with a 0.5mm period on the codestrip, which equates to about 9bit resolution which in angular terms is 0.7 degrees. I can do the transform using 128 samples in about 60us (15kHz) which is the same as the PWM freq, so I could know the rotor position very accurately at every pwm period.

Mr DIY (BTW, what's your name? ), you're correct about the IR2130. IIRC (haven't done this for about 12 mths) I get a rise time of about 120ns using small-ish fets, if I went to higher current and voltage ones I would be using 3 x IR2110 (which is what I've used on our 120A electric power steering controller at work) as they will need about 2A to get the gate up quick enough.

Stu Maxwell
StumaxAircraft
stumax@hotmail.com
stumax is offline Find More Posts by stumax
Reply With Quote
Old Jan 28, 2004, 06:51 PM
Registered User
mmormota's Avatar
Budapest
Joined Jul 2003
2,567 Posts
Quote:
Originally posted by stumax
One idea I have is to sample the back emf of all phases during the pwm off time via a 3 channel A/D. Two of the phases' voltages would then be phase shifted 120deg and 240 deg respectively, and a Fourier transform performed on all of the data, using as the frequency of the transform the motor speed (in Hz) derived from the previous commutation. This would result in the phase angle of the sampled waveform, and it would be very accurate due to the number of samples used.
Stu,

Another idea:
- no hw pwm, but sw pwm emulation, the on/of/on/off sequence is strictly keeps synch with the phase of the rotor
- because of it the a/d samples are representing similar phases, filtering is easier in the time domain
- there is no hw mix for the pwm - less parts, smaller board

Jozsef
mmormota is offline Find More Posts by mmormota
Reply With Quote
Old Jan 28, 2004, 07:02 PM
Impossible? Hah!
KreAture's Avatar
Oslo Fornebu, Norway
Joined Jun 2003
3,074 Posts
I will rather filter out the PWM and use the EMF directly. That eliminates the pesky adc with it's quirks...
KreAture is offline Find More Posts by KreAture
Reply With Quote
Old Jan 28, 2004, 07:26 PM
Registered User
mmormota's Avatar
Budapest
Joined Jul 2003
2,567 Posts
The drive voltage and the emf is mixed, at high throttle the drive voltage dominates. You need sampling to separate the emf.
mmormota is offline Find More Posts by mmormota
Reply With Quote
Old Jan 28, 2004, 07:38 PM
The blade numbers go up to 11
stumax's Avatar
Sydney, Australia
Joined Aug 2002
4,379 Posts
Jozsef, if you can do software pwm at up to 25kHz then go for it. Trouble is it will tie up the micro so much it has a hard time doing the housekeeping. My concept of using a Fourier transform is to determine the rotor position more accurately than zero crossing or sign changing due to the amount of data that's used to determine the result. For the average motor it's probably not necessary but for ultra efficiency and super smooth throttling and low speed operation it would certainly give better results.

Kreature, smarter minds than yours have long ago worked out that filtering the pwm out is not the way to go. Any filter is going to induce a phase lag which will corrupt you result. Sure, you could measure this phase lag at different speeds and compensate for it, but it will be a pain. Perhaps you could try an analogue multiplier and multiply the emf by the pwm signal and run the result thru a comparator, that way when the pwm is high and the emf is noisy the signal out of the multipleir will be zero (therefore it won't false trip the comparator).

Stu Maxwell
StumaxAircraft
stumax@hotmail.com
stumax is offline Find More Posts by stumax
Reply With Quote
Old Jan 28, 2004, 08:14 PM
Registered User
mmormota's Avatar
Budapest
Joined Jul 2003
2,567 Posts
Stu,
What is the target rpm? Probably 100 000 (magnetic field) rpm is enough (100k for a 2 pole, 50k for 4 pole etc).

That is 600us for a whole turn.
My estimation: 1 MIPS if I am lucky, 3 MIPS reasonable, 10MIPS - well, if _that_ is not enough, better if I choose another job...

25 kHz pwm: I think it is not necessary if the "pwm" keeps the sych. I suppose high freq is necessary only in asynchron to avoid interference with the mechanical rpm freq.

Fourier: if the speed is rapidly changing, the Fourier is not working well

Jozsef
mmormota is offline Find More Posts by mmormota
Last edited by mmormota; Jan 28, 2004 at 08:20 PM.
Reply With Quote
Old Jan 28, 2004, 08:44 PM
The blade numbers go up to 11
stumax's Avatar
Sydney, Australia
Joined Aug 2002
4,379 Posts
Jozsef, I think the Mega8 running at 16 MHz would do just fine (and they're cheap and small in SMD).

The pwm freq needs to be high enough for us not to hear the motor squeel (I learnt that making the motor controllers for my cnc mill - couldn't work out why it was so noisy until I rasied the pwm freq!)

How fast do you think the rpm can change during on revolution? I'de be surprised if it was more than a few%, in which case the Fourier transform should work OK. Not sure if it's necessary, and it would definitely require it's own micro!

Stu Maxwell
StumaxAircraft
stumax@hotmail.com
stumax is offline Find More Posts by stumax
Last edited by stumax; Jan 28, 2004 at 09:49 PM.
Reply With Quote
Old Jan 28, 2004, 09:52 PM
Registered User
Toronto, ON, CANADA
Joined Jul 2003
41 Posts
Toshiba TB6537F in stock;

http://www.glyn.de - they have 55pcs in stock.

Good luck.

Ian
IanS is offline Find More Posts by IanS
Reply With Quote
Old Jan 28, 2004, 09:52 PM
Registered User
mmormota's Avatar
Budapest
Joined Jul 2003
2,567 Posts
Stu,

I have a oscillogram of a Jeti controller. (I bourned it with reverse connection, and made some pictures during repairing it).
Unfortunately the range is missing. A CDrom motor was connected, probably the mechanical rpm is about 3000, the electrical is 18k, 300Hz. Seems that the switching freq is only 12 times higher, 3600hz. It is in the sensitive band of the ear - even if I underestimate the rpm by 3. The motor was silent.

The only motor that gives a high freq noise is my PJS550E (it's smaller brother, the 300 is silent), but I am not sure if it is because of the controller (CC Phoenix 25) or something pure mechanical.

Fourier: ones (upon a time...) I worked on a signal processing project including FF. It was very sensitive to slight shifts in the signal phase - we had to use autocorrelation function first.
mmormota is offline Find More Posts by mmormota
Reply With Quote
Old Jan 28, 2004, 09:57 PM
Registered User
mmormota's Avatar
Budapest
Joined Jul 2003
2,567 Posts
Forgot the picture...
mmormota is offline Find More Posts by mmormota
Reply With Quote
Old Jan 28, 2004, 10:15 PM
The blade numbers go up to 11
stumax's Avatar
Sydney, Australia
Joined Aug 2002
4,379 Posts
Quote:
It was very sensitive to slight shifts in the signal phase
That's the whole idea! Basically you're working out the phase angle of the signal with respect to a fixed sine wave of the same frequency.

I've found that some motors are noisy and some aren't. It also depends on the prop - or in my case the ducted fan which is a whole lot more succeptable to accoustic ringing due to low freq pwm. In serious motor controllers the DSP's use a slightly varying pwm freq to avoid any chance of audible noise.

This is a great thread, BTW!

Stu Maxwell
StumaxAircraft
stumax@hotmail.com
stumax is offline Find More Posts by stumax
Reply With Quote
Old Jan 29, 2004, 02:58 AM
Registered User
Mr DIY's Avatar
South Africa
Joined Aug 2003
587 Posts
I see some talk still going on about the feedback sensing. Whats wrong with Takao Shimizu scheme?

http://www.rcgroups.com/forums/attac...&postid=575019

The addition of caps filter out all the noise.

My sensing circuitry used – bummer cant add images like we used too. Too bad.

I have used it and it works really well. Once a phase is driven and the corresponding feedback signal THEN monitored, I find no jitter or PWM interference coming through at all. The other feedback lines can generate false information during this time with pleasure, as I am not interested in them. I suspect that other people like to monitor all three feeback signals together and drive the output phases according to the all 3 feedback signals. Hence, bad feedback information at any time will probably cause problems. From what I can understand, this is why there is so called ‘off times’ to measure/sense. Please correct me if I am wrong here.

In my case, the three op-amp outputs are digitally combined to provide a serial stream that is synchronized with the driven phases. Easy as pie. That way, all jitter, false triggers, ect are eliminated. The micro code becomes a lot simpler as it only has to look at one input pin as well. Under start conditions, the steam coming back will look like a mess until such time the load starts to turn properly. You then get short duration pulses that correspond to how much you are overdriving the motor. At that stage, you switch from ‘manual’ commutation to automatic so to speak.

Stumax ...Mr DIY (BTW, what's your name? ), it's Brian.
Mr DIY is offline Find More Posts by Mr DIY
Reply With Quote
Old Jan 29, 2004, 04:47 AM
Impossible? Hah!
KreAture's Avatar
Oslo Fornebu, Norway
Joined Jun 2003
3,074 Posts
stumax
Calculating the phase offset won't be much of a pain. You can calculate it based on time between the last commutation and the time you detect the emf transition. If the time between commutation and detection is half the time between the two last commutations, and it stays that way for some time, then it is in phase. If not, then finding out how much it is out of phase is simple math. A small phase increase since last time would indicate both that the motor is speeding up and that the phase distortion is increasing a tiny bit. (If I don't have this backwards.)

Also, I don't understand what the drive voltage has to do with anything. Only thing that happens as you increase voltage and drive rpm increase, is that back EMF increase too. The offset of the EMF wave will ofcource be shifted along with the center point of the drive vcoltage, but the amp/comparator is supposed to be balanced on the drive voltage and should handle this.
KreAture is offline Find More Posts by KreAture
Reply With Quote
Reply


Thread Tools

Similar Threads
Category Thread Thread Starter Forum Replies Last Post
Brushless speed control problem TomL Sport Planes 8 Jun 08, 2003 01:23 AM
TMM-18-3ph Brushless Speed Control nelsonjo1 Power Systems 5 Apr 04, 2003 12:33 AM
FS: Model Motors 40 amp Brushless Speed Control. Jim McPherson Aircraft - Electric - Airplanes (FS/W) 3 Dec 10, 2002 05:02 PM
Dragon 35 brushless speed control Steve H. Aircraft - Electric - Airplanes (FS/W) 0 Aug 17, 2002 12:29 AM
Hooking up a Jeti Brushless Speed Control to a Jeti Pashor 15/4 motor. HELP!!! Rob R Power Systems 3 Jan 10, 2002 06:37 PM