What encoder resolutions do you recommend for different applications?

[sunmix]

Well, we know CNC can be made for:

Profile
Mill
Lathe
EDM
Engrave
Route
and etc...

And we have such motors:
BLDC Servo
BDC Servo
AC Servo
Direct Drive Servo

and motors have many WATTs, and nMs too,

And there are many ranges of encoder resolution.

But how can you select a proper encoder resolution on certain application? I know the higher the best, but doesn't mean that I will use 20,000 encoder counts for each servo motor, on every application. There must be a reason why 2000, 1000, 500, 400, and 200 lines are made.

Could someone answer this question?

I hope that I can learn how to select a good servo motor, with enough torque, and enough resolution, yet affordable in its range too.

Thanks everyone again!

[TOTALLYRC]

Well, we know CNC can be made for:

Profile
Mill
Lathe
EDM
Engrave
Route
and etc...

And we have such motors:
BLDC Servo
BDC Servo
AC Servo
Direct Drive Servo

and motors have many WATTs, and nMs too,

And there are many ranges of encoder resolution.

But how can you select a proper encoder resolution on certain application? I know the higher the best, but doesn't mean that I will use 20,000 encoder counts for each servo motor, on every application. There must be a reason why 2000, 1000, 500, 400, and 200 lines are made.

Could someone answer this question?

I hope that I can learn how to select a good servo motor, with enough torque, and enough resolution, yet affordable in its range too.

Thanks everyone again!

Selecting an encoder resolution is like picking out the rest of the components.
What are you max speed requirments in ipm?
What is the max steps/counts that your controller can output per second?
What is the max the driver can handle per second?
What resolution do you need or can use?

An 80820 based router doesn't need and can't deliver tenths tolerances.
What is your lead screw pitch? higher pitches use lower counts and vise vera.
On my machine 500 count encoder in quadrature is 2000 per rev. on a 5 tpi screw it is 10,000 counts per inch times the 2.0 belt reduction is 20,000 per inch.
That is .00005 per count/step or 1/2 of one tenthousandth of an inch.Most machines can't effectivly use such a fine resolution. If you were to use 20,000 counts per inch on a machine like a plasma cutting machine which can cut near 400 ipm on thin material = 133,333 counts per sec. which is over the upper range of mach3 and many others even in 100khz mode, if my math is correct.
If you went with 2000 count encoders in quad at the same speed, your counts per second would be silly. I would suggest that you use the minimum or slightly above that will do what you want with what you can really use.

The reason that so many are made is for industrial applications and the differing needs of each machine.


That is really clear isn't it?

Mike

[Al_The_Man]

I hope that I can learn how to select a good servo motor, with enough torque, and enough resolution, yet affordable in its range too.

Here is a link to a graphic free sizing program Motor Sizing Program
You most probably will find out that the highest demand for torque is in the acceleration/deceleration rate you require.
i.e. the important motor/load inertia ratio factor, the industry standard is to design for under a ratio of 10:1.
Al.

[Mariss Freimanis]

Take a look at DigiKey (www.digikey.com) part number 102-1307-ND. It is a really nice encoder that has 16 dip-switch selectable resolutions (48, 96, 100, 125, 192, 200, 250, 256, 384, 400, 500, 512, 800, 1000, 1024 and 2048 lines). It draws 6mA @ 5VDC, is good to 30,000 RPM and costs $30. We have tested this encoder and it is excellent.

Mariss

[TOTALLYRC]

Take a look at DigiKey (www.digikey.com) part number 102-1307-ND. It is a really nice encoder that has 16 dip-switch selectable resolutions (48, 96, 100, 125, 192, 200, 250, 256, 384, 400, 500, 512, 800, 1000, 1024 and 2048 lines). It draws 6mA @ 5VDC, is good to 30,000 RPM and costs $30. We have tested this encoder and it is excellent.

Mariss

Hi Mariss,
I wish I had seen this about 4 days ago as I just bought encoders off of Ebay. I looked at the data sheet and they come with a bunch of different bushings to fit different shaft sizes and the selectable resolutions mean I can 4 or 5 on the shelf and not have to worry which machine they go on.
I bet that with your endorsement they sell a bunch to people on the zone.
Thanks for testing them for us.

Mike

[Mariss Freimanis]

Someone on this list turned me on to them (sorry, I don't remember who it was). We ordered 10 of them to try out to see if they are any good. Well, I'm impressed. Reasons:

1) Settable resolution. A lot of times I have had to switch encoders on a test motor because it didn't have the resolution I need for some breadboard circuit I'm working on. Goodbye to that irritation; just pop the cover and switch the resolution to what's needed.

2) Perfect quadrature. Optical encoders specify a 90 degree phase relationship between CH_A and CH_B. Problem is the tolerance is +/-60 degrees. Taken at face value it means 30 degrees, then 150 degrees between increments of motion for a "times 4" quadrature decoder. This means worst case the smaller increment of motion will be 1/5 the size of the larger one (30 / 150). I haven't seen an optical encoder this bad but it doesn't mean it can't happen yet still be within spec. The AMT102 measures 90 degrees +/-1 degree. Really remarkable.

3) 6mA @ 5VDC supply. This is 5 times less than a typical optical encoder. It works all the way down to 3.3VDC (spec'ed @ 3.6VDC). This means it's not real particular about what it's fed.:-)

4) Maximum feedback frequency (times-4 decoding) is 1MHz. This is 5 to 10 times better than what you can get from an optical encoder. I haven't tried to see what it's line-driving capability is, to see if it can feed a real-world 20' (6m) cable at this frequency.

5) The encoder is capacitive instead of optical. This should mean it will be less susceptible to contamination. All it takes is one fingerprint or a small bit of crud on the code wheel to put an optical encoder out of commission. The package is partially metallic instead of all plastic; that's good for interference suppression.

6) Easy to use. It comes with about 10 shaft diameter adapting inserts (from about 1/4" down to 'very small'), a centering tool and a gaping tool. Piece of cake to mount it accurately. Pin-out and cable connector is the same as USdigital or HP. Downside is you have to get the mating connector. :-(

7) No noise problems at all. We tried our 'motor from hell' with a G320 and it worked like a charm. The 'motor from hell' is a clapped-out NEMA-42 that makes the air stink of ozone because its brushes and commutator is so worn. It can drive most optical encoders insane without rigorous shielding. More thorough tests are pending to see if the AMT102 is as noise immune as it seems.

As a first impression, the AMT102 is a very nice, inexpensive and competent encoder.

Mariss

[H.O]

Hi,
I have one of those AMT's too. It works on my large Indramat servo-motor while my brand new USDigital E7P's does not. I've tried about every trick I can come up with but the USDigital encoder just refuses to work properly. Then, no matter what setup I have (almost) I slap that AMT encoder on the back of the motor and the problem is gone.

However, for some reason the motor is a lot more "nervous" with the AMT encoder than with any of the optical encoders I've tried. When grabbing the shaft and deflecting it from position the motor "vibrates" and "jerks" a lot more while resisting deflection than what it does with an optical encoder (no matter what resolution I set the AMT to). If this is due the encoder itself, my setup or something else I don't know.

Oh, there's also cable assemblies with line-drivers available from Digi-key that fits the encoder, very cheap too.

/Henrik.

[epineh]

I have been using these encoders and they have served my needs quite well, the biggest advantage for me is the selectable resolution and different shaft size options, both metric and imperial.

To add to your original question, at the moment I have threaded rod so I have the encoders set to 200 cpr, this gives me all the resolution I need, sooner or later I intend to change things to ballscrews, then as the leadscrew pitch will be a lot larger I will crank the resolution up a bit, probably 1024 and still get the same accuracy as before.

I wasn't aware of some of the points Mariss mentioned and must say I am now happier with my choice, the original reason I bought these was that they had the cheapest postage to Australia, a certain other manufacturer was quoting about $150.00 US for postage on three encoders, which was a little rich for my blood.

Cheers.

Russell.

[Xerxes]

However, for some reason the motor is a lot more "nervous" with the AMT encoder than with any of the optical encoders I've tried. When grabbing the shaft and deflecting it from position the motor "vibrates" and "jerks" a lot more while resisting deflection than what it does with an optical encoder (no matter what resolution I set the AMT to). If this is due the encoder itself, my setup or something else I don't know.

My best guess would be that this encoder does some resolution interpolation and low pass filtering on output signal. In other words (if I'm right) it adds some time lag to position information which leads to reduced motor stability. Optical encoders have instant response.

Anyway, very interesting encoder indeed! I'm going to get couple of those for testing on my next digikey order.

BTW, serveral companies also make magnetic encoder chips (rotary & linear) that are dirt cheap (around $5). See http://www.austriamicrosystems.com/

And about on-topic: the highest possible encoder resolution always gives the best performance and smoothest operation in all applications. The only cons are increased cost and increased signal bandwidth need (or limited motion speed if bandwidth is not high enough).

[Mariss Freimanis]

Nope. Doesn't do that.:-) I took a servomotor that has 1/4" shafts (6.35mm) front and back. I put a 500-line optical encoder on the front shaft, the AMT102 set to 500-lines on the back shaft. I put scope probes on both encoder CH_A outputs while servoing the motor with the AMT102. No lag, no delay; just a constant phase angle between both signals.

What's happening is the CH_A, CH_B drivers have less sink and source capability (this thing runs on only 6mA) so encoder to drive cabling is more strict. Use twisted pair cables (CH_A and GND on one pair, CH_B and 5VDC on another pair). That's what I use and I never have problems.

Mariss

[H.O]

Strange.... I don't drive the cable directly with the AMT102. I run the A and B channels to a DS9638 differential transmitter IC which drives the cable and reciever in the drive. Same cable, same transmitter IC yet it behaves differently when changing encoder.

[Mariss Freimanis]

H.O,

In that case go back to what you said about the optical encoders not working properly at all while the AMT102 work but are "twitchy". You have something else going on that the new encoders partially mask. Grabbing the shaft puts a torque load on the motor and causes current to increase through it. This increased current may cause more electrical noise (RFI) to be generated which then interferes with the encoder. Have you tried grounding the motor case back to the drive GND terminal to short this noise out?

Mariss