How hard is it to drive a 3-ph brushless motor?
in [Embedded]
|
Prev: And for some Mechanical binary addition
Next: .NET micro framework ... end of embedded world???
From: Rich Grise on 9 Dec 2009 13:33 On Wed, 09 Dec 2009 13:47:00 +0000, Peter wrote: > > A different option is to use a stepper directly, 0-13rpm, no gearbox. This > might do but a failure of the electronics could make it go round very > fast, In what failure mode can this possibly happen? Shorted drivers will lock up the motor, open drivers will let the motor free-whee, kinda - steppers don't coast very well. If the micro locks up, no steps, no motion. I've worked with stepper motors before, and they have an inherent top speed - if you drive it too fast, it will just skip steps. I could spec an appropriate motor and possibly gearbox, and write a uP program to drive it at a speed proportional to your input control voltage. I could design this for you, for an appropriate conslting fee - if you're interested, make an email by taking richgrise at example period net (see headers), and replace the example period net with yahoo period com. Cheers! Rich
From: Rich Grise on 9 Dec 2009 13:36 On Wed, 09 Dec 2009 16:11:02 +0000, Peter wrote: > > Tim Wescott <tim(a)seemywebsite.com> wrote: > >>Yes there are single chip solutions, but I was not happy with the one >>that I was involved with -- it came on the recommendation of a vendor, >>would have worked very well for the application for which it was >>designed, worked like _crap_ for us, and we ended up rolling our own >>controller almost a year behind schedule. > > Well, an hour or two's of googling later :) I found what looks just the > thing > > http://www.datasheetdir.com/ECN30206+Brushless-DC-Motor-Drivers > > This seems to do the lot - except that the control voltage seems to > control the % pulse width 0-100%. There is no feedback control of motor > speed. Sure there is. Drive it with a variable constant current source or sink, and find what current runs it at 13 RPM. Then take the amount of current, make a fixed current source/sink, and the speed will be exactly proportional to the duty cycle. I've also done this before. :-) Good Luck! Rich
From: Paul Carpenter on 9 Dec 2009 15:03 In article <1spvh55fi473l7i80g14akf6ua7e8cdisj(a)4ax.com>, nospam(a)nospam9876.com says... > > Tim Wescott <tim(a)seemywebsite.com> wrote > > >On Wed, 09 Dec 2009 13:47:00 +0000, Peter wrote: > > > >> Hi All, > >> > >> I have been doing hardware/software (asm & C) development for 30 years > >> so not totally useless but have never done this, or stepper motor > >> control. > >> > >> I have a project which needs basically 0-13rpm coming out on a 1/4" > >> shaft. There is a whole pile of axial-style motors and planetary > >> gearboxes e.g. this > >> http://www.micromo.com/n49803/n.html > >> > >> Few do a 1/4" shaft (6mm is the nearest) but that is another problem :) ..... > >> But nobody does a suitable assembly which is short enough. To get the > >> max length I have to go to a bare brushless motor (no controller) and > >> mount the controller separately. ..... > >> The problem is lack of anywhere to put any off the shelf brushless > >> controller... so I have to build my own PCB. > >> > >> What I know is that the controller uses PWM to synthesise the 3 phase > >> waveform. Obviously the frequency will be the desired RPM and I assume > >> that the "rms" voltage will also be varied - the higher the rpm the more > >> volts. I guess they feed the raw PWM to the motor directly, in the > >> conventional way (I have designed loads of DC brush-servomotor PWM > >> controllers). Is there any standard approach to this, or preferably some > >> proved designs/chips? Quite common to get 3 phase pattern by PWM or Timing Pattern generator and gate this with PWM clock/DC to change the RMS level (part of torque) and adjust 3 phase pattern frequency for RPM. Watching your torque and RPM relationships for load and current speed to avoid stall. The 3 phase pattern is easy to do even for reversing in software or hardware. I am seeing a customer tomorrow who have an ASIC solution for doing this used for missiles so long life time product. Let me know if you want a datasheet or other details. > >> And I would quite like some kind of closed loop motor speed control. > >> Even though the torque is small, the application needs the speed to be > >> within about 10%. This can be done with back-EMF; should not need a > >> tacho. +/- 1.3 rpm at 13 rpm or higher rpm at the actual motor? 13rpm is not even 0.22 revs per SECOND, your issue will be measuring movement and speed accurately. Cannot remember how good the back-emf method is at that slow a speed, from memory not the best method, as it gets very difficult to detect changes as the back emf is VERY low. Even with hall effect/opto detection, knowing you are moving is an issue. At those sorts of speeds I tend to have driven things with no feedback, or if high precsion required, a very fine 'toothed' opto/hall detector, to get lots more pulses per second. Measuring motor speed does not guarantee measuring final drive speed or distance. > >> A different option is to use a stepper directly, 0-13rpm, no gearbox. > >> This might do but a failure of the electronics could make it go round > >> very fast, whereas a motor+gearbox cannot go beyond a certain point > >> which is perhaps 2x max rpm. Brush motors are inherently speed limited > >> but I can't use one... This application cannot deal with a runaway > >> motor. Even if you use a micro other cutouts for limits of travel, max speed and NO movement should be considered. > >> Can anyone suggest an easy way to do this? I don't fancy programming up > >> a microcontroller to do the 3-phase synthesis - a wheel which must have > >> been invented by so many people...... > >> > >> I found this > >> http://ww1.microchip.com/downloads/en/AppNotes/00857a.pdf Check its comments on sensorless motor control (page 9) "- The motor must be moving at a minimum rate to generate sufficient back EMF to be sensed - Abrupt changes to the motor load can cause the BEMF drive loop to go out of lock - The BEMF voltage can be measured only when the motor speed is within a limited range of the ideal commutation rate for the applied voltage - Commutation at rates faster than the ideal rate will result in a discontinuous motor response" I would not consider back emf measurements below 100-200rpm (depending on motor and number of magnetic poles). Remember you have a commutation cycle, which is NOT RPM but RPM is determined by pairs of magnetic poles. So how many commutation cycles are you likely to be going through for 1 RPM AT THE MOTOR? .... > I don't *need* a microcontroller for this application because the > input is analog anyway, -10V to +10V. So a micro would need an ADC, > and all the pulse timing, current limit sensing/latching etc, would > take a week or two to put together. You need something to loop through commutation cycles depending on pairs of poles and analog setting to determine RPM of motor. You need some form of RMS level/PWM control, to get torque right. All of which is highly logic orientated (even if 555's). For back emf at lower rpm you will probably have to sample with a/d or sample and hold all three phases of 3 phase windings in energised and open state. Then determine which to use. You need some form of level translator to drive your half bridges last time I used IR2131, for a small sensorless BLDC. Watch your back-emf and currents t higher speeds. > Looks like I will need a comparator sensing the control voltage > passing through 0V, and switching the direction signal to these chips. Or switch direction or starting point in table reading of pattern. Oh for small PAL to do that. As I said there are chips around, depends on sensored or unsensored motor, I can get details of some other devices if you wish. -- Paul Carpenter | paul(a)pcserviceselectronics.co.uk <http://www.pcserviceselectronics.co.uk/> PC Services <http://www.pcserviceselectronics.co.uk/fonts/> Timing Diagram Font <http://www.gnuh8.org.uk/> GNU H8 - compiler & Renesas H8/H8S/H8 Tiny <http://www.badweb.org.uk/> For those web sites you hate
From: Rich Grise on 9 Dec 2009 16:56 On Wed, 09 Dec 2009 20:03:34 +0000, Paul Carpenter wrote: > > I am seeing a customer tomorrow who have an ASIC solution for doing this > used for missiles so long life time product. Let me know if you want a > datasheet or other details. Missiles? They have a very short lifetime, indeed! (albeit, they might be on the shelf for awhile.) Now, _satellite_ should imply a very, very long life. Speaking of that, how long do geostationary comm satellites last? When one of them goes down, do they just scuttle it and launch a replacement? I've never heard of anyone going up to repair one! Thanks, Rich
From: Jan Panteltje on 9 Dec 2009 17:27
On a sunny day (Wed, 09 Dec 2009 13:56:31 -0800) it happened Rich Grise <richgrise(a)example.net> wrote in <pan.2009.12.09.21.56.29.114217(a)example.net>: >On Wed, 09 Dec 2009 20:03:34 +0000, Paul Carpenter wrote: >> >> I am seeing a customer tomorrow who have an ASIC solution for doing this >> used for missiles so long life time product. Let me know if you want a >> datasheet or other details. > >Missiles? They have a very short lifetime, indeed! (albeit, they might >be on the shelf for awhile.) Now, _satellite_ should imply a very, very >long life. > >Speaking of that, how long do geostationary comm satellites last? 15 years spec or so. >When >one of them goes down, do they just scuttle it and launch a replacement? They let it burn up before the fuel to control its position is used up. They Launch a new one every one in a while, for broadcast they may divert to a different transponder on a different sat. You need not re-align your dish, as those sats are in groups. For example Eutelsat hotbird has several close to 13 east IIRC. http://www.eutelsat.com/satellites/satellites.html >I've never heard of anyone going up to repair one! Neither have I. >Thanks, >Rich |