Brushless DC Motors (BLDC)

I have been looking at BLDC motors as cheaper alternatives to steppers.

They are effectively three phase permanent magnet induction motors.

Given the way they are constructed they can function as low resolution steppers capable of high RPM. (30 or 60 degrees per step depending on how you drive them) so will probably need some form of mechanical reduction to increase resolution.

They are available with positional feedback (Hall effect) or positional sensing can be acheived using back EMF.

From reading the specs their power to weight ratio is pretty excellent. Efficiency is good also.

The drive looks to be complex but can be done from a micro controler and three half bridges reasonably effectively.

Last but not least due to the mass market presented by RC modeling etc there are some around for very reasonable prices. Particularly so when compared to Stepper motors with no positional feedback.

For use in driving extruders there are some very small ones also at reasonable prices that outperform brushed motors.

Has anyone else looked into these ?? what do you guys think ??

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

There's been a lot of talk about using them over the years, but I think you're the first to actually decide to do some hands-on.

You talk about low resolution, but have you uncovered some torque vs pulse rate charts yet ... if such things exist for this motor type, mind. Not all steppers have them.

Also, do you have some prices? The ones I've looked at have been a bit steep in price.

I'm really glad you're looking into this! If we can tap into the RC market with its large production runs of motors it could be wonderful.

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

Hi guys

It's good to see someone talking about brushless DC motors. They are a little trickier to drive than steppers but offer potentially excellent results.

I'm now going to stick my oar in in a big way :-)

The reason that BLDCs outperform normal (brushed) DC motors is that they have quite powerful magnets on the SHAFT, and the windings on the outside of the machine. This makes it easier to get the heat out, so the ratings go up. The downside is that they need drivers.

When it comes to driving BLDCs, you have three basic options:
1) get one with a driver built in. It will behave much like a normal DC motor. Easy, simple, probably fairly cheap but not the best result.
2) buy a separate driver. Varying types available, likely to be expensive.
3) make your own driver. A little tricky but potentially quite rewarding.

My personal preference is #3, but I'm doing a PhD in this sort of thing so my view is just a little bit skewed...

Overall, I would have to say that BLDC motors could offer better performance, but they will be more expensive (maybe a little, maybe a lot) and complicated.

Anyhow....

If you want to get into developing your own drivers, I have a couple of suggestions.
1) you'll almost certainly need an oscilloscope - there'll be switch mode stuff all over the place.
2) there are two types of BLDC. Some have trapezoidal innards while others have sinusoidal innards (technically a Brushless AC Machine - BLAC). Connect the motor up to the oscilloscope and give it a spin to find out what you've got. Trapezoidal ones are a bit easier to drive.
3) Atmel makes a range of AVRs (like the Arduino) which are intended for driving BLDCs. Have a look at the AT90PWM3 chip - it's fairly cheap and does a lot of the heavy lifting.

Forest:
The BLDC performance usually boils down to torque speed curves much like a normal DC machine [lancet.mit.edu] . Basically more voltage lets you go faster, provided your electronics can handle it! There is a trick called 'field weakening' which lets you get more torque at the top end but it's quite tricky to do and I wouldn't recommend it.

In terms of position accuracy, you could read the outputs from the hall effect sensors, but remember that they will probably be big cheerful square waves and only tell you the position to the nearest 60 degrees and the motor could be anywhere in that angle. If you need precision positioning you'll have to add some form of encoder such as the magnetic rotary encoder.

jbb

With an 8 mm leadscrew, 60 deg would give you 0.17 mm / step resolution. That's finer than my belt driven Darwin. You just need motors that have enough torque to drive a leadscrew directly, but based on the brushless DC plane I used to crash, that's probably not too difficult.

Wade

Or use the inbuilt hall effect devices and gear down the output to give you stepper level resolution with better torque characteristics.

Some of the sensorless controlers seem to be able to sense the position when it is'nt moving by applying pulses to the windings somehow. Not sure that I understand how they manage this. Given that it has to be moving to produce back emf.


jbb

Interesting input thanks for sharing that, much appreciated.

My bigest winges about steppers have been their open loop characteristics and poor torque at speed. Together with price of course.

brushed DC motors are cheap but arguably less reliable and you need to add positional feedback to use them at all.

BLDC motors on the other hand seem to have some of the better properties of both with ways to achieve positional feed back as a feature of the motor.

My initial interest was in looking for a very small but higher torque motor to run an extruder, (it needed to have a decent bearing in it at the drive shaft end to cope with the back pressure on a worm gear). and my curiosity was tweaked looking at the move in the RC area towards replacing petrol/glow fuel engines with BLDC's particularly for flying models and the Quadrocopter.

Still looking into this one, I have seen RC BLDC motors for around 5 to 6 UKP advertised on the internet.

I'll take the advice and have a look at the AT90PWM3 (I had been looking at the AT90 for USB on device, as a cost eliminator to do away with the FTDI cables/chips)

Cheers

aka45

PS this item suggests how to drive 3 phase bldc's at 30 degrees.

[bldc.wikidot.com]

jjb: what do you think to their sugestion ??

Edited 1 time(s). Last edit at 06/08/2009 11:40AM by Andy Kirby.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Wade Wrote:
-------------------------------------------------------
> With an 8 mm leadscrew, 60 deg would give you 0.17
> mm / step resolution. That's finer than my belt
> driven Darwin.
>
My Lord! What IS the resolution on your Darwin?

I'm getting 0.102 mm/step at full step with Tommelise 2.0 and have just shifted the controller to allow for half stepping. Mind, I'm using linear steppers with lead screws.

---

-------------------------------------------------------

Hell, there are no rules here - we're trying to accomplish something.

Opportunity is missed by most people because it is dressed in overalls and looks like work.

Thomas A. Edison

Right, on second thought actually that's a bit less than my Darwin. I get 8 steps per mm using half stepping (0.125 mm/step); I was thinking of full stepping, which would be 0.25 mm/step. Posting from my phone on the road, so I'm working from an imperfect memory.

Hi again

aka47: the article on getting more resolution out of a BLDC seems to be suggesting that you 'half step' it - much like a stepper motor. It would probably help with the resolution but I'm not sure how well the precision would fare.

Additionally, I recall people around here saying that gearboxes are undesirable due to backlash etc. At the very least I know that a gearbox will increase the inertia of the system and make it harder to get rapid response.

On the subject of sensorless control, yes it is definitely possible to do this. The catch is that most 'simple' schemes are based on detecting the back EMF of the motor and won't work at low / zero speed when there is little or no back EMF to be detected.

A more advanced approach is to inject a high frequency voltage into the motor to measure the inductance of each winding, which changes depending on the angle of the rotor in accordance with the 'saliency' of the machine. There are two catches: 1) you need to measure the phase currents and DC bus voltage quite accurately in order to accurately observe what's going on inside and 2) the controls will be a lot more complicated - probably beyond the capabilities of an AVR.

I hope this helps
jbb

On gear boxes I agree that backlash and stiffness can often be issues.

More gears usually means more backlash.

A worm drive though has a big reduction and greater stiffness. (Depending on the angle of contact it can be near impossible to back drive. Perhaps making the need us breaking to prevent things unwinding if you cut power (to save it).

Given Wade's contribution in conjunction with a work gear rather than a direct feed and your capability looks somewhere close to what is needed.

The essential question though is the positional feedback.

Given my doubts as to sensor less working at low speed/stationary coupled with jbb's confirmation it looks like with sensing is a must. (easier to control as well)

Unfortunately that seems to eliminate the RC motors as most that I can see rely on sensor less controlers.

There are plenty of BLDC's with sensors nut these look to be at the expensive end of the spectrum.

I guess this brings up back potentially to an opto wheel of some sort for positional sensing and accurate commutation.

hmmmmmm

maybe not as promising as first thought.

What do you guys think......

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

On waking up this morning I finaly realised what about BLDC's had been bugging me.

Following on from Forests experiments with stepper motors BLDC'sparticularly the outrunner style should be relatively easy to build/make with most of the dificult to shape parts Reprapable.

The outrunner patern of motor gives higher torque at lower speeds than an inrunner style.

If we are considering building RepRapping Outrunner BLDC's then we can design in photo interupt/reflective positional feedback for commutation and of course positional feedback.

An outer band (Outer or Rotor) of steel pipe/tube stops the magnets from flying off into space whilst making the magnetic circuit. An inner cage to position the magnets in the steel tube would be printed and would have in addition the shaft mounting and photo interuptor tabs as part of it's design.

An inner plastic core with (Inner or Stator) would have designed in voids to put the other half of the magnetic circuit into and formers to wind the coils onto. RC enthusiasts appear to rewind their own motors so this must be doable. It would also have mounting lugs at the bottom as well as bearings top and bottom to support the shaft.

All in all a vitamin count of 1 piece of steel tube, 2 bearings, magnet wire and something to make the stator magnetic circuit.

Sure the driving circuitry is more complex but can be completely digital. (CPLD or Microcontroled)

The essential questions being is it worth it for higher torque/speed + DIY Motor capability + Integral Positional Feedback.

Thoughts for what they are worth.

Cheers

aka47

PS never try to think and type ona keyboard with a defective space bar

Edited 1 time(s). Last edit at 06/09/2009 03:01AM by Andy Kirby.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

Aka47, I had the same thought! I think you won't even need positional feedback if you can get the output resolution down to ~0.2 or 0.1 mm. A 3 pole motor on a 1 mm pitch leadscrew should do open loop just fine.

Inrunner or outrunner, it should be doable!

Wade

The brushless motors used today for radio control don't typically have sensors. This is done by sensing the voltage of the one out of three windings not being driven.

Brushless motors can be easily designed for true multi pole operation, where as small electric motors (RC industry size), have only two poles at any particular time, but distributed over one or more winding hub.

Because the brushless motors can easily be designed for more poles, torque can be much higher, as the distances between poles diminish, and the inverse square law effects *Much stronger* attraction or repulsion.

Most brushless motors for RC use are capable of being run at very high current... some up to 100 amps or so. The motors that will run at the lower end, would typically consume more like 10 amps under load, and an amp free running. A brushless motor that is stalled has very little cooling, and a lot of heating, they are designed for performance operation, not heavy duty stalled operation.

STMicroelectronics have an application note on using their ARM cortex microcontrollers for power control.

If anyone is wanting to experiment, I recommend getting the lowest "KV" value motors you can find- more poles, less current, more torque. These motors can run at very high RPM with the right controller and on the high KV end.

Ive just realised that I actually have a boxful of these motors already, even better they are attached to a driver board!

The school that I work at lets me have any old kit for salvage. I can usually find bearings, opto sensors, power supplies and if Im lucky some decent steppers (the Epson C2600 colour laser has a NEMA 17 unipolar, a NEMA 23 uniploar and three other motors which I didnt recognise - until now!)

Anyway, the cheaper laser printers all have this strange motor that is attached to a circuit board. The actual case of the motor spins and the shaft sticks through the circuit board and has a helical kind of gear. The gear turns a similar but much larger gear inside the printer and there is absolutely no backlash at all - a real feat of engineering!

Anyway, I pulled one apart to see if it was useful and found a flexible magnet inside the case and a whole load of coils (probably about sixteen?)

Anyway, reading this thread, I will definitely keep them to one side because if they are useful, I now have about a dozen of them!

I wonder if anyone has thought about making a "bits I can salvage" site with a list of useful components available in old kit?

OK

Something I am currently working my way through and found to be worth recommending to any one interested is:-

[www.microchip.com]

AKA AN885 (application Note 885 from microchip the PIC people) It pretty much covers most of what is needed to be known.

Cheers

aka47

---

Necessity hopefully becomes the absentee parent of successfully invented children.

I found the book Brushless Permanent Magnet Motor by Duane C. Hanselman design quite helpful.

It is a bit pricey, but you may be able to find it at your local university library.

The authors website at [www.eece.maine.edu] has some sample pages available from the book.

Sir
I doing project in which i want to lift about 150 kg load using bldc motor then i want to know about how to select motor

sudhir

This isn't really the right place for this question - 150kg is way above anything RepRaps would handle.

Your cheapest alternative is to buy an automotive winch - except that it will be of low quality and use a brushed DC motor and a speed controller.

Alternatively, you will need a suitable BLDC and gearbox. Someone like Anaheim automation can sell you a nice big motor with gearbox, but it will be expensive. [www.anaheimautomation.com]

As a rough guide to how much power you need, lifting a 150kg load at 1m/sec (slow walking pace) will take ~1500W of power NOT including gearbox losses, so (I guess) you probably need ~2kW of motor power for 1m/s, ~4kW for 2m/s etc.

That's getting into serious power ratings, so I think that this is the wrong forum for you.

jbb

i did phd on BLDC Motor control
please help for this

I am doing phd work on BLDC motor control
any body working on this please help me

Check the major microcontroller chip manufacturer's application data sheets if you want examples.

I suspect you will need to do the work yourself though, if you wish to "earn" a PHD.

If you have the know how to design the control then the rest is just physics.

Work out how much energy to you need to lift your load at a given speed/rate.

Then back calculate this through a drive train to the motor to get the motor size/specs.

Once you have a motor size/spec if it is what you want great. If not change something until you have a model that works for you.

Don't forget to factor in generous losses due to efficiency at each stage.

Tweak variables until you get something you think you can live with.

Spreadsheets are a good tool to use for this, what if sort of modeling.

After this prototype it ie build one and see if it does as you predict.

If it doesn't repair your model until it matches reality. The tweak to get something you can live with.

Iterate in this sort of way until you have a final design for motor and drive train.

Then you can do thi thing you got your PhD for, ie design a control system for it.

Alternatively, buy something off the shelf that already does it.

As you can see it is not trivial and people are unlikely to put themselves to this much pain for you unless you are paying them.....

Hope this helps.

---

Necessity hopefully becomes the absentee parent of successfully invented children.

I work for a company that makes small DC motors that are used primarily for medical and military applications. I have access to obsolete motors, gearheads and drive electronics including both brushed and brushless motors. I would be willing to offer one of you guys some obsolete stuff in exchange for RP parts to build my own RepRap.

I am a design engineer and could offer much in the way of development, but I need to get my own machine built and running. Can you guys help me out?

bbook, are you in Florida? The local reprap user group might be up for 'loaning' you a set
of parts once you've got the rest of your setup together:

[forums.reprap.org]

Good luck,

---

-Sebastien, RepRap.org library gnome.

---

Remember, you're all RepRap developers (once you've joined the super-secret developer mailing list), and the wiki, RepRap.org, [reprap.org] is for everyone and everything!

Degree of step/commutation is influenced by how many poles you design into your motor. More poles gives a reduced step/comutation angle. (ie more commutations per revolution and greater positional accuracy, bit like a stepper motor).

---

Necessity hopefully becomes the absentee parent of successfully invented children.

I thought I should point out that I found a guy that already made a usable brushless DC motor using 3d printed (FDM) parts (plus neodymium magnets, a steel washer, bolts, magnet wire, and some sort of shaft). Looks relatively simple (not counting the controller).

Here's a video of it:
[www.youtube.com]

And forum thread:
[www.rcgroups.com]

If it can be modified to function effectively as a stepper driver, that'd be great. Getting the number of vitamins down is pretty great, plus more of the parts should be locally source-able and for cheaper. It should be possible to make an even simpler one by reducing the number of stator coils to 3 (and thus easier to put together) as well as substituting domestically-produced and cheaper ceramic magnets for neodymium (neodymium magnets are great, but seem to all be made in China). We're not going to be winning performance awards with this motor, but that's not really the point. IMHO, the more stuff you can offload to the microprocessor/microcontroller, the better. Those will remain a vitamin for quite a long time.