Thread Tools
Jan 27, 2004, 05:20 PM
Impossible? Hah!
KreAture's Avatar
Well, I was right
I added 3 large resistors to my phases and connected the other ends of em to read out a combined E-signal.

The signal loosk very promesing.
The signal has the PWM superimposed on it, but with low-pass filtering I should be able to remove that easily. After all, the PWM-frequency will be a lot higher than the commutation frequency. I think I will use 6 bit PWM control to get it up in the 100 kHz range. This should make it 20 times faster than the maximum commutation rate and thus very easy to filter out.

Then I won't have to bother with when I check for commutation.
Sign up now
to remove ads between posts
Jan 27, 2004, 05:56 PM
Impossible? Hah!
KreAture's Avatar
Here's a image:

Lower signal is 1 phase of my MGM controller running my MiniAC. The upper signal is my filter making a EMF signal from the sum of the phases and filtering out the PWM noise.

There is a phase-shift of maby 15 degrees now. For a PWM-frequency of over 100 khz, I can use a much smaller filter. Hmm, wouldn't the phase-shift be relative to the frequency ? It won't be same ammount all the way will it ?
Jan 27, 2004, 07:28 PM
The blade numbers go up to 11
stumax's Avatar
Guys, I just noticed this thread - cool! A few things to throw into the ring:

1. Why not use a separate micro to do the commutation with? That way there's one micro getting the rx signal, checking voltage for BEC (even running a switchmode BEC) and outputting the PWM, and another micro doing only commutation stuff. This would allow a multi engine controller to be built with one "housekeeping" board and multiple "commutation" boards.

2. See attached pic for a way to use hardware PWM and output pins to get PWM over 3 phases easily. Sure, it adds a logic IC but only a small one.

3. Kreature, the PWM frequency will be dependent on the type of motor you're running. I doubt that you'd want to go higher than 30kHz unless you start making ironless core motors. Besides, unless you have ball-busting gate drive circuitry the FET resistance during rise and fall times will heat the FETs up too much.

4. If you measure the voltage of the floating phase during the off PWM time you'll eliminate the need for filtering and have a more robust signal, allowing accurate detection at lower speeds.

Keep it up guys - this stuff is good fun, and very satisfying when you can make what people charge alot of money for!

Stu Maxwell
Jan 27, 2004, 08:00 PM
Impossible? Hah!
KreAture's Avatar
The diode puller solution is smaller than using a extra logic IC.

When it comes to measuring the commutation, it doesn't matter if you read from a undriven phase, or from all at once as they all have PWM-noise on em. (The noise on the EMF comes from the two driven phases.) You see it clearly on the bottom signal in the scope. Both rising and falling flank of the EMF have serious PWM-noise.

The aim is using the combined EMF signal after filtering and amplifying to TTL levels (using v/2 as threshold) to interrupt the PIC. It will then commutate based on some angle adjustment and possible time delay. The rest of the time, the PIC will not have anything to do and will easily be able to do PPM decoding and adjustments of it's hardware PWM generator. (Yes I use hardware PWM...)

As for the PWM frequency... I will have to use some small FET's to drive the larger FET's as they have a too large capacitance. (880pF) to be driven directly from PIC. (a 200 Ohm resistor should be used on the PIC to reduce current to 25 mA max and this would cause a max pwm of 5 kHz consisting of almost no ON time, just on/off slopes.

Since I have to use a driver or small FET array I might as well use the power I then get. I agree that switching time vs frequency must be thouroughly thought through.

My reason for the high switching frequency: I really want to be able to drive a CD-ROM motor up to 10k rpm and I want the filtering to do as little damage as possible to the actual EMF signal. The large the difference between the max commutation rate and the PWM signal, the better it is.

The MGM uses 10 kHz and has a max commutation frequency of 15000 commutations pr second. (150,000 rpm max at 2 pole = 900,000 comms pr min / 60 = 15000 comm/s) Something I don't get to add up unless it only is able to reach this rate under 100% duty.

I do however know it does not like to run 12-pole CD-ROM motors abowe 22,000 rpm wich is pretty close to it's theoretical max of 25k rpm for this type of motor.
Jan 27, 2004, 09:15 PM
The blade numbers go up to 11
stumax's Avatar
Kreature, I don't understand how 3 diodes can let you use 1 PWM signal to drive a 3 ph. H-bridge - can you draw a sketch?. My point about reading the undriven phase during the pwm off time is that there isn't any pwm noise there (you'll need to increase mag. on you scope window to see this), you don't need to filter it so you don't get a phase shift in the signal you're measuring (which means the detection is accurate regardless of rpm). You can use this signal thru a comparator and put the comparator output thru some logic (XOR the comparator output with the PWM signal) to eliminate false detection. I put the signal thru a flash A/D (Analog Devices AD7825 - 4 channel 2MS/s) and let the micro do the work. If using 100% duty cycle there's no choice other than reading the voltage during the PWM on time, however at 100% the motor is usually generating a very high back emf so the signal is good.

PWM switching frequency should be chosen according to the motor's inductance - that is what determines how fast the current can change, no point switching any faster than that. You'll find that for max efficiency you need to change the PWM freq according to rpm if you want those last couple of %'s efficiency.

I use a 3ph bridge driver from International Rectifier (IR2130) to drive the upper and lower FETS. If you choose high enough voltage FETS you can go up to as many cells as you like 'cause these FET drivers can handle a Vgs up to 600V. I also use opto's between the micro and the bridge driver to save on blowing up micros all the time! I've found that at 15kHz PWM freq a 10ohm resistor in series with the gate will help stop ringing (and associated EMR) yet still give fast gate rise times, and use a diode in parallel with the resistor to allow the gate to fall quickly, and a zener on the gate (Vgs + 12v or so) to protect the FET from spikes.

Hope this helps!

Stu Maxwell
Jan 28, 2004, 12:41 AM
Registered User
Mr DIY's Avatar
I see we are getting technical here …I like it.

Firstly, there is a simple easy way to get your feedback. No waiting for off time, no synchronization with PWM ect. What I did was use the LM324 amp made up like on mr mmomoto original circuit. The 3 outputs are fed directly into my logic chip and combined to produce a single stream output. The combining process notes the current drive sequence and applies the required logic to select witch output must be fed to the micro. What I get out is a perfect square wave at 50% duty cycle ( at Zero timing), without any interference from the PWM logic. To be honest, I did not expect it to work as well as it does. As I said before, there is no off time that must be allowed for in software. The logic device is programmed from the micro what the 10-bit PWM word is and then forgotten. Until such time as a change in speed is required, it will continue to use the programmed PWM rate. All I then basically do, it monitor my feedback with the Mega8 and send the new commutation code at the right time. Of course, when a new commutation word is sent, the PWM automatically resets for next FET pair driven.


You have hit on something that has puzzled me. The use of the IR FET drivers devices do not allow for fast switching. Just what kind of switching time are you getting that is deemed acceptable. The device you mentioned seems to only handle 420mA. Not sure if that’s sink or source, but other people have been using these devices as well and then drive FETs with 10nF of input capacitance … making for very slow on/off times. I have read about controllers running hotter at low/middle throttle settings than at full power .. and the lack of suitable drive current would explain this. I myself use the TC4426 devices that can drive 1.5Amp and found the controller running hottest at full power and not at intermediate power levels. Can you give hint on your switching times. Mine are under 1uS into 4 or 5nF

You'll find that for max efficiency you need to change the PWM freq according to rpm if you want those last couple of %'s efficiency.
Did not know that. Thanx .. Will experiment some more.
Jan 28, 2004, 01:33 AM
Impossible? Hah!
KreAture's Avatar
I wasn't after the last % of effichiency. I state in an earlier post, I want to make this as cheap and easy as possible.

But, I did use the wrong order of magnitude on the capacitance of the FET's

The FET's are 880pF and if I limit current to 25 mA with a 200 Ohm resistor I will get a rise-time of about 176 ns for a 70% swing. Since the FET's start to turn on after only 1 volt, the actual turn-on time should be no worse. If we assume off-transition takes equal time we have 352ns switching time.
At 20 kHz I'd have a ratio of 141:1 for on/off:switching. In real life, probably better. Only way to get some hard facts here is to wait for the FET's and then do some driving... My scope is fast enough to give me accurate measures as to how much power will be lost in switching.

You are right when you say the PWM-rate should be compared to the rpm or motor inductance. However the filter I will be using is too simple to work well with a wide range of frequencys.

Theese FET's have a incredibly small gate capacitance compared to what they can take of current, and comsidering I am only trying to make this drive cd-rom motors for now, 5 amps should be ok.
Jan 28, 2004, 02:05 AM
Registered User
Originally posted by dicker

Im one of the two guys behind

First of all I must excuse that I have few time at the moment. So our Project got Stuck a bit.
But we are much further than listed on the Page. The Redesign (Cyclone 6N) is running perfectly fine, driving my Little Torcman 280-5 at 14Amps quite Cool (for an 10A Fet :-))

A new design Im actually working on is back on the N/P driven by Transistors scheme. This is mainly for making it cheap.

The PWM technologie we are using is simple and cool (ofcourse
We are driving the Input Pins of our IR2102s with an 1K resistor as pre resistor. If the PWM Output (of the Hardware onboard PWM Generator) gos to Zero Level, the input Pins of the 2102s are taken down over 3 diodes, one to every Input Pin for the High Level Fets (HLF)

So the PWM is only cutting the HLF, which is afaik the same way as everyone does.

So, if thats not clear, I can make some Schematics from this.

so long,
Julian Wingert
Hi Julian,

I want to ask about Cyclone Mini. How is the performance? Will it work with CD Roms?

Jan 28, 2004, 02:31 AM
foamforce pilot
dicker's Avatar
Well, the Conrad motor with the original magnet ring is driven very well...

My Testing Motors are always unloaded CDroms.

Ive done no tests with normal CDRoms yet, but I'll do this when we are finished.
Jan 28, 2004, 02:49 AM
foamforce pilot
dicker's Avatar
Originally posted by KreAture
Here's a image:

What diff (Volt/Units) did you use?

because, this signal doesnt look small, the PWM interferes to few.....
You can test it by rotating the Motor by Hand without the esc put on. When you get the same signal then, it works. else youre ***** :-).

But believe me, the Back EMF sensing is not the hard part... Starting a Motor is much more complicated.

bye, Julian
Jan 28, 2004, 02:55 AM
Impossible? Hah!
KreAture's Avatar
what I have there on the scope is irrelevant. It will go through a self-centering amp first to make it a square wave. The threshold for the square wave will be half way between min and max applied voltage.

The lower signal is 5v/div and upper is 0.2v btw.
What matters is signal to noise ratio and I will get it better than you see there. Also, the fuzzyness of the traces is due to camera and screen glare. The traces are needle sharp in real life and about as thick as the white divider lines.

I am working on a schematic now that will show how to use 6 diodes and 6 resistors to do dual-sided PWM control of all 6 FETs.
Jan 28, 2004, 02:59 AM
Registered User
Thanks for the fast reply Julian,

I already ordered parts for SBL-Micro from Jo, But would like to build Cyclone Minin also. Is there a starting problem with mini?
Or it starts normally? Mr. Takao suggest that you flip the prop(like you do on an glow engine to start the motor)on his design.

Jan 28, 2004, 03:34 AM
Impossible? Hah!
KreAture's Avatar
Here's how the PWM-mixing can be done:

Basically, the P-side needs an extra set of FET's to handle the full swing from B+ or else the P-FETs would never turn off. (control voltage = 5v while B+ will usually be 2-3s LiPo.)
Since there is N-channel FET's on both sides they all use posetive logic. Thus, feeding them through a 200 Ohm or larger resistor will allow them to turn on, when the PWM signal is not pulling the gates down to GND+0.6v. (I think using transistors here would be better, or maby even fast logic, as there will be a lot of current wasted here during off-periods. Also, I am uncertain of 0.6v will turn the FET's 100% off.)
Jan 28, 2004, 03:37 AM
Impossible? Hah!
KreAture's Avatar
Smart idea! I'll use the flip-prop method first too, that way I don't have to do anything but look for commutation before starting to drive the motor. This would allow me to get one part of the system working well first, then I can concentrate on reliable starting.
Jan 28, 2004, 03:44 AM
Impossible? Hah!
KreAture's Avatar
I think using a AND- or NAND-gate instead of the 6 resistors and 6 diodes will give higher performance. I'm looking into it now. I think a gate array would lower the space needed for components, but increase the space needed for routing.

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