Control.com - Nerds in Control

The Freescale DSC I am looking at has a Reference Design using the encoder pn: ES 28-6-1024-05-D-R from
http://www.inducoder.de/dat_en/datsheet/es28.shtml

Just wondering if there are lower cost alternatives out there.

Encoder Products sells good encoders for pretty cheap. Or you could try AutomationDirect.com though their selection of styles is much smaller, but if they have the one you want, they're pretty good and the price is right.

Thanks! I am finding that a cheap encoder is 50 bucks!! I need to find the 2 dollar part, since this is very low cost motor drive using vector control.

Good luck....

For anywhere near that price point, you're going to have to make your own.

Look at the encoders used to replace pots at the electronics emporium of your choice. You may hit your price point. I would think anything worth controlling to that degree is worth quality components. That said, there _are_ inexpensive ways to encode rotary motion, like something similar to mouse guts. It might be cost effective to find a motor with the capability built in as that would seem to be cheaper than
separates components as there are less bearings, etc.

Regards

cww

Thanks Curt:

Since I am designing the motor, the encoder will be inside the housing, and so it will be possible to design my own solution. Thanks for the tip. The only problem I can see is finding the part that would hold up for millions of revolutions... most of the encoders for front panels are meant for occasional use and are specified in 10s of thousands of revs. Do you know of any high rev count parts ...that are inexpensive?

Thanks,
Tim

Cheapest would probably be an optical disk and a pair of optical interrupters, the small commodity U type that, e.g. DigiKey carries. The disk can be done photographically or even on mylar with a laser printer. I have even seen them done on glass epoxy board by a printed circuit fab. These won't get you 5000 line encoders, but they would be good enough for high motor to motion ratios and can be non-contact for zero wear.

Regards
cww

Thanks Curt, I am leaning in this direction. My only reservation is contamination inside the motor. There are only 4 openings, the 2 bearings and two side covers. Maybe I should look at water and dust proofing the motor, using good grease in the bearings, and O ring seals on the two end housing covers. Then I would not think about contamination from without.

Thanks again for your comments...
Tim

Try Renco Encoders or US Digital...

Tim:

If you want an inexpensive encoder for this application, you will probably have to look for one primarily designed for the "office automation" market, not "industrial automation" as the one that Freescale recommended. The encoders that go into inkjet printers and the like are now made for less than $1.

I'm presuming you are planning to go into reasonably high volume here, so it would be worth significant engineering time to drive the piecepart price down. The least expensive solutions involve "kit" encoders where you work out the mounting. One of your challenges will be figuring out how to make the installation physically robust enough for your application.

Check out Avago encoders (formerly Agilent, before that HP). Last I checked, they had 70% of the world market for "office automation" encoders.

Curt Wilson
Delta Tau Data Systems

Thanks Curt and all:

I am looking at the Avago parts, and yes this will be high volume. I am designing the motor also, so I have the freedom to design-in the encoder guts. I am not afraid to roll my own, but as you say, the challenge is to make the assembly robust for long life and reliability.

For environmental reasons, I am not too excited about optical solutions however. I am cosidering digital hall effect devices and also reed switches, but I have to find a fairly high freq capable reed switch. Also for a high life application, I would need 20 million operations or more.

Thanks again...
Tim

Mechanical considerations can be more difficult with hall effect encoders than with optical ones. If you are using a magnetic disk and a hall effect sensor, then you have to position the sensor within a fairly narrow distance window for reliable operation. The electronic drawings may look simple, but the mechanical assembly on a production line to the required tolerance may not be. I'm not sure what your motor looks like, but this is something to keep in mind.

I was thinking 180 deg opposite (sorry for the obvious pun). Since this is a BLDC exterior PM motor, the mag disk or optical wheel (whichever) would be on the rotor (8" dia), and if magnetic, the sensor needs to be close to the disk, on a PCB on the stator. But with optical, the wheel would have to be "inside" a U type transceiver, which is the only low cost optical solution.

This would take more manufacturing considerations, assuring that this mating will happen when the motor is buttoned-up...I think. The magnetic solution looks to be too expensive it appears, so the only low cost option is optical. The trick will be to either make it self cleaning or make the motor totally-enclosed.

Thanks for the thoughts...
Tim

You could also buy reflective type sensor assemblies from the same vendors that would work even with an opaque disk. But you would want adaptive electronics that could deal with changes in reflectivity and keep the threshold centered between the high and low levels. Not very difficult, but something you would want to have. The one sided approach would make the mechanicals and assembly easier and more tolerant of parts stack up. etc.

Regards
cww

If the max diameter where the sensor can be located is 7", then the circumference at that point is 22", and at 1024 pulses per rev, the reflector stripe (I see a alternating pattern of reflector and dark stripes) would be 0.021", which I believe would require a laser to detect don't you think? So for a normal optical tx and rx pair, maybe 0.1 resolution could be achieved which would require maybe 5 sensors, and a reflector stripe width of 0.1".... right?

I do like the one side assembly idea...that is the right way to go...

Tim

Speed might be another factor, some of the optical assemblies aren't optimized for high speed. 6000 rpm = 100Hz. You might check into the sensors used for optical mice. Volume makes the price reasonable and they may be adaptable for the purpose.

Regards
cww

Assuming 1024 is the fundamental resolution and not post quad, the solution is most likely going to have to be optical. Finding the parts for the optical bench (disk, sensor, light source)at a $2.00 price point is going to be difficult. Even if these are found cheaply, there are usually interface electronics needed to get to some sort of usable signal.

Using a hall sensor is not a bad idea, as it is inexpensive, but the magnetic "disk" for this many counts would likely be pretty expensive, if it could be done.

You can probably get by inexpensively if you are able to interpolate off of a sine wave, but this is really just passing the buck to the interpolating electronics. This method will probably also give you a low quality signal.

Jim Miller
Application Engineer
Quantum Devices Inc.
www.quantumdev.com

Yes that is the fundamental resolution needed.

I can find the optical parts for that price, assuming I design the wheel, and use off the shelf, led and receiver, but I am worried about moisture and dirt contamination inside the motor, it wont be waterproof.

You are right about the magnetic solution. I can find sub $.50 hall effect parts, but the magnetic wheel is in the 70-dollar range!!

I can work from a sinewave, or any waveshape, and just use a comparator to square things up to get a good signal, as long as it is symetrical around a reference voltage. I am feeding a micro, so it will directly measure the time between pulses.

Thanks for your thoughts!!

Tim

Now I see my mistake!! I have been thinking only of digital solutions...even about the encoder insides too. I now see there is an analog option, one that might fit this application exactly. The PM rotor (BLDC hub motor) has very strong neodymium magnets, 32 of them. Why not use them? There is a pcb on the stator, that can be stuffed with one or more analog hall sensors. Each magnet will produce I think a haversine waveform from the hall sensor. For 1024 steps, using two sensors, required is 16 steps per magnet, per sensor (2 hall X 32 mags X 16 steps = 1024) which would be pretty easy and cheap to do with 4 quad comparators and a bunch of resistors. Total cost under 2 bucks...but will it work?

On second thought... if I know the shape of the waveform and the amplitude of the sensor output, I can know what thresholds to set on the comparators, so the encoder output would be reasonably accurate....right??

Hmmmmmm
Tim

That is a great idea...

I have seen motors that used digital hall sensors to sense the magnets for trapezoidal commutation.  It was just two TO-92 packages on a PCB.  It worked fine.  I don't see why analog could work as well, with a little more signal conditioning/processing.

Let's back up a bit. What are you doing? If you want angular position to accurately control some external device (which is what everyone has been assuming), then no, it won't be accurate. The size, position, and strength of the magnets will vary and this will affect your analogue signal.

If you are doing this for commutation sensing, then this signal is what you really want anyway. This is a very conventional way of sensing magnet position for winding commutation. Many new digital servo motor / drive combinations use optical encoders (since they have an encoder anyway), but most (if not all) older analogue drives used hall effect sensors to detect magnet position.

Hi M:
The app note I referenced in the first post is a design for a motor control using an off the shelf encoder, that is more expensive than my whole ebike motor will be. So yes, the encoder is used for commutation feedback. The only place I have seen hall sensors used is for BLDC trapazoidal drive motors, using digital hall sensors for crude angular detection (uses 3 sensors for 3 phases). This design uses field oriented control with sinewave space vector modulation, which requires more precise location detection.

Thanks again...
Tim