Comparison of Absolute Encoders

N

Thread Starter

Niraj

We are looking for reliable angular measurement solution & it seems rotary encoders are best suited. Our angular measurement span is just 35 degrees & we need 4-20 mA output. Is there a place where the encoders are rated on parameters like make-model wise => accuracy, repeatability, MTBF etc. If you can help us that would be great....
 
S
An encoder will not generally output 4-20ma. If you only need to measure 35 degrees of arc, I recommend a potentiometer with a signal conditioner to translate the pot output (variable voltage) to 4-20ma.

You will apply your DC voltage high and low to the CW and CCW terminals of the pot, and the wiper will give you a voltage output proportional to position. If your pot has more than 35 degrees of freedom of motion, you can still use the signal conditioner to give you 4-20ma over just the 35 degrees that you're interested in.

Novotechnic and some others have some nice industrially packaged pots. I've also used throttle position sensor pots for this application, but the ones I used don't seem to be available any longer.
 
J

James Ingraham

"Is there a place where the encoders are rated on parameters like make-model wise => accuracy, repeatability, MTBF etc."

No, there's not. We've talked about an industrial "Consumer Reports" several times on control.com, but apparently nobody has taken the idea and run with it.

Having said that, at this point encoders are pretty much a commodity. Yes, you can pay more and get more rugged versions. For the most part, you can trust the manufacturer's spec sheets.

As I have recommended many times before, choose based not on technical specs but on vendor relationships.

-James Ingraham
Sage Automation, Inc.
 
J

James Ingraham

"An encoder will not generally output 4-20ma."

Disagree. There are plenty of encoders with 4-20mA output. I've personally a used Kubler by Turck with 4-20mA output. The one I used had a full 360 degrees, so if you're only going a 1/10 of a rotation you're only getting 4-5.6mA. However, they sell one with a 45 degree measuring range, which fits the application much closer.

I didn't go looking too hard, but I also found BEI has 4-20mA output. Oddly, the two big names in encoders, Sick/Stegmann and Heidenhain, don't seem to offer a 4-20mA option.

-James Ingraham
Sage Automation, Inc.
 
S
>Disagree. There are plenty of encoders with 4-20mA output. I've personally a used Kubler by Turck with 4-20mA output.<

Odd. If you're not going to use it as an encoder, why buy an encoder?? Repeatability?
 
J

James Ingraham

"If you're not going to use it as an encoder, why buy an encoder?"

There are a handful of reasons I can think of, but let me start with the Big Picture. You are correct that "encoder" may not be the proper term. "Encoder" brings immediately to mind something that spits out pulses as it turns, even if it's a fancy Sin / Cos with an additional serial channel for multi-turn absolute. So in this case we need an "angular displacement sensor," rather than an encoder. However, "angular displacment sensor" is harder to say, the device we're looking for is made by encoder companies, and is in identical housing, with the same kind of shaft options. It LOOKS like an encoder, even if it has a different purpose in life.

More specifically as to why you would choose a single-turn absolute encoder with analog output for measuring angular displacement, the most obvious feature is comfort. 4-20mA signals are widely used and understood, reasonably easy to troubleshoot, and available on virtually any controller architecture. The points I mentioned above about the physical properties of the encoder are also nice; we all know how to couple a shaft, and it's a nice, small, rugged package. Finally, what else could we use for measuring angular displacement? There aren't a tremendous number of options that are easily understood and integrated. Sure, you could use an inclinometer, (especially in this case, limited to 35 degrees). But we're back to comfort zone. Having never used an inclinometer, I don't have lots of manufacturer's bookmarked, I don't know what kind of ouput they have, I don't know how they mount, I don't know their resolution and repeatability, etc., etc., etc. But an encoder with 12-bit 4-20mA? I know EXACTLY what that means.

-James Ingraham
Sage Automation, Inc.
 
S
<i>"So in this case we need an "angular displacement sensor," rather than an encoder. However, "angular displacment sensor" is harder to say, the device we're looking for is made by encoder companies, and is in identical housing, with the same kind of shaft options.</i>

I get the feeling you didn't read my original post all the way through, which I didn't realize till now. I said "pot" which is even shorter than "encoder".

<i>"More specifically as to why you would choose a single-turn absolute encoder with analog output for measuring angular displacement, the most obvious feature is comfort. 4-20mA signals are widely used and understood"</i>

Now here you're just responding to the OPPOSITE of what I said. My beef wasn't with the 4-20 part but that using an encoder to get there was the hard way to go about it, when there is a device specifically intended for measuring part-turn angular displacement.

Anyway, it sounds as if we might be sort-of agreeing while using terminology differing sufficiently to make ourselves think we aren't. I recommended industrial pots (some of which are packaged with 4-20ma conditioners on board), though my experience is that the companies offering those are generally not big in encoders. Just a variation in experience no doubt.
 
J

James Ingraham

<i>I get the feeling you didn't read my original post all the way through, which I didn't realize till now. I said "pot" which is even shorter than "encoder".</i>

I did, in fact, read the whole post. Yes, "pot" is shorter than "encoder." However, "pot" doesn't necessarily mean something you connect to equipment. When someone says "pot" I immediately assume we're talking about a small rotary dial for a person to manipulate. Naturally, that's only one possible meaning. "Encoder" on the other hand is almost always a machine-mounted sensor. Encoders that are manaully turned are generally refered to as jog wheels or hand wheels. Also, "encoder" implies some sort of connection mechanism (shaft, bearings, etc.) designed for continuous use, although again that's not strictly true. When I think of a pot, I think of something designed to be turned every once in a while, not something that's part of contiunous process. Yet again, not a definitive description of a pot, just the first that that springs to mind for me.


<i>Now here you're just responding to the OPPOSITE of what I said. My beef wasn't with the 4-20 part</i>

Fair enough. My thinking was that your statement <i>"if you aren't using it as an encoder"</i> implied why use an encoder if you're not getting pulses, which is they way most people think of encoders. Sorry for the misunderstanding.


<i>It sounds as if we might be sort-of agreeing while using terminology differing sufficiently to make ourselves think we aren't.</i>

Agree.


<i>Just a variation in experience no doubt.</i>

Probably so. Which is why I like these discussions. Out of curiosity, what are some manufacturer's with 4-20mA output industrial pots? I mentioned BEI and Kubler / Turck for encoders, but I really don't know where I'd go looking for what you're describing.

-James Ingraham
Sage Automation, Inc.
 
S
<i>I did, in fact, read the whole post. Yes, "pot" is shorter than "encoder." However, "pot" doesn't necessarily mean something you connect to equipment. When someone says "pot" I immediately assume we're talking about a small rotary dial for a person to manipulate. Naturally, that's only one possible meaning. "Encoder" on the other hand is almost always a machine-mounted sensor. Encoders that are manaully turned are generally refered to as jog wheels or hand wheels. Also, "encoder" implies some sort of connection mechanism (shaft, bearings, etc.) designed for continuous use, although again that's not strictly true. When I think of a pot, I think of something designed to be turned every once in a while, not something that's part of continuous process. Yet again, not a definitive description of a pot, just the first that that springs to mind for me.</i>

Actually, the way I define the terms, "pot[entiometer]" means a voltage dividing device operated by whatever means, and an "encoder" is a discrete linear or rotary device, whether single channel or quadrature, whether pulse output or binary ("absolute"). If the device gives analog sin and cos signals rather than square edge pulses, then it's a "resolver".

<i>Fair enough. My thinking was that your statement <i>"if you aren't using it as an encoder"</i> implied why use an encoder if you're not getting pulses, which is they way most people think of encoders. Sorry for the misunderstanding.</i>

Now you're really confusing me. That's exactly what I meant. What made you think otherwise? When I originally said "my beef wasn't with the 4-20ma" I meant I had no beef with the original poster WANTING a 4-20ma signal for his control system, which you seemed to be trying to defend. I WAS objecting to the use of an encoder as the easiest way to get to an analog signal to sense 35 degrees of rotary position. My question could be more explicitly phrased as "If you want an analog output, why start with a fundamentally discrete device and force it to fit, when people make perfectly good, and potentially cheaper, devices that are analog by their nature?" You and I seem to parse sentences VERY differently.

<i>Out of curiosity, what are some manufacturer's with 4-20mA output industrial pots?</i>

Novotechnik and Celesco (big in string pots) are the two that jump to mind. There's another European one whose name escapes me at the moment.

 
J

James Ingraham

Steve- Let me start out by saying that I'm aware that I should probably just keep my mouth shut at this point, but that's never been something I'm good at. As you stated earlier, we're pretty much both on the same page here, with just minor quibbles on semantics. Minor quibbles on semantics make for the best Internet flame wars.


<i>"pot[entiometer]" means a voltage dividing device operated by whatever means</i>

That would be hard to argue against. Nevertheless, I maintain that people often have a more specific idea in mind, based on their background. For me, pots have always been little manual adjustments. Also, I think of pots as being generally crude devices; they're affected by temperature, and it's difficult to set it precisely. Other people may have different things come to mind.

<i>an "encoder" is a discrete linear or rotary device, whether single channel or quadrature, whether pulse output or binary ("absolute").</i>

Here's a partial list of encoder output types:

Incremental encoder:
Pulses (single or quad, single with direction, with or without Z channel, RS-422, TTL, Line-driver, push-pull...)
Sin/Cos

Absolute encoder:
Parallel digital signals (Gray's Code or Binary)
4-20mA / 0-10V analog
SSI
BiSS
EnDat
Hiperface
Fieldbus (e.g. DeviceNet, Profibus, etc.)
Ethernet (e.g. EtherCAT, EtherNet/IP, ProfiNet, etc.)

It's not really the output that defines whether the device for measuring linear or rotary displacment is an encoder. It's the WAY that it measures displacment. Traditionally, encoders have been optical devices, although there are also magnetic versions. While the line may be blurred sometimes between an encoder and something else, there's still an "I know it when I see it" aspect to encoders. Internally, a pot works differently than an encoder, even if they are both outputting 4-20mA. This is also true in the linear world; a magnetostrictive sensor is NOT a linear encoder, even though they do the exact same thing from an external standpoint.

<i>If the device gives analog sin and cos signals rather than square edge pulses, then it's a "resolver".</i>

Here I disagree completely. Again, it's the internals, not the signal interface, that makes a device a resolver or an encoder.

Here's AMCI's definition of a resolver:
"A resolver is a rotary transformer where the magnitude of the energy through the resolver windings varies sinusoidally as the shaft rotates."
http://www.amci.com/tutorials/tutorials-what-is-resolver.asp

From the same page:
"Recent advances in technology have enabled the integration of a resolver and on-board electronics in one housing... available with the following output types; Absolute Parallel, Incremental Digital, Analog Current, Analog Voltage, and DeviceNet."

So a resolver can have an output other than sin/cos. It's the internals that matter.

On the flip side, here's a line from Sick-Stegmman on their SinCos encoders:
"The sine and cosine signals from the incremental track are output on separate analog channels, and can be used directly by the drive, or squared to provide a conventional A quad B digital signal. Thus, the SinCos® encoder can be used as an absolute, sine/cosine, or incremental feedback device."
http://www.stegmann.com/product/servo/datasheets/1044.pdf

Not to belabor the point, but here's a line from Emerson Control Techniques which clearly allows for the concept of an encoder with sin/cos output distinct from a resolver:
"When precision and fast machine startup are critical to achieve production goals, the Absolute Position Motor Feedback SIN/COS encoder is the best alternative over incremental encoders and resolvers."
http://www.emersonct.com/download_usa/literature/appFlyers/AbsolutePositioning.pdf

<i>I WAS objecting to the use of an encoder as the easiest way to get to an analog signal to sense 35 degrees of rotary position.</i>

All right, my bad again.

<i>My question could be more explicitly phrased as "If you want an analog output, why start with a fundamentally discrete device and force it to fit, when people make perfectly good, and potentially cheaper, devices that are analog by their nature?"</i>

Okay, now that I'm on the same page (I think), I'll answer that. (a) If I had to come up with the solution quickly, I know the Kubler / Turck line has it, so I wouldn't have had to go looking around. (b) Arguably, starting with a discrete device and switching it to analog can be more precise. e.g. Encoders with analog output are not as affected by changes in temperature as potentiometers. (c) Though the 4-20mA output probably renders comparison moot, an encoder can have vastly higher precision than an analog sensor like a pot or resolver. While theoretically the analog devices have infinite resolution, in reality encoders have higher usable resolution by orders of magnitude.

Having said all that, (depending on application specific factors) a pot or resolver or a dozen other technologies would all work. I guess I've been going on about this mainly because I don't understand your seeming animosity to encoders. I don't really have a strong opinion here. We used to use servos only with resolvers; lately we've switched to Hiperface (sin/cos with absolute data) because our vendors did. We've only used pots for small manual adjustments, as I mentioned before, but I would be perfectly happy using an industrial pot for angular measurement.

<i>You and I seem to parse sentences VERY differently.</i>

Again, sorry about that.

Novotechnik (I've used their magnetostricive linear sensor in the past) has both what I describe and what you describe when saying "pot." Take their little PC 90 series for example. Small, designed to be turned by hand, has a mechanical lifespan of 4000 movements. Useless as a machine-mounted sensor. Meanwhile, their IP6000 has an IP66 housing, a shaft clearly designed for coupling, has pretty impressive repeatability and linearity, a connectorized output, and a 100 MILLION movement life-span. It even physically LOOKS like an encoder. In fact, their IPE6000 "precision angle encoder integrates electronics into the potentiometer, eliminating the need for an external transformer." They actually call it an encoder! But we know it's not; it's a potentiometer with integrated electronics.

Okay, you'd think after a thousand words I would have, you know, a POINT. I'm better at rambling than summarizing. Still, it mainly comes down to:
An encoder with 4-20mA output is a perfectly valid solution to the problem of angular displacement measurement.

-James Ingraham
Sage Automation, Inc.
 
B

bob peterson

> "pot[entiometer]" means a voltage dividing device operated by whatever means<

That would be hard to argue against.  Nevertheless, I maintain that people often have a more specific idea in mind, based on their background.  For me, pots have always been little manual adjustments. Also, I think of pots as being generally crude devices; they're affected by temperature, and it's difficult to set it precisely.

Apparently you have never heard of a motorized pot. Or 10 turn
precision pots with locking dials. Neither are typical Radio Shack
items, but pretty common in industrial controls until recently.
 
S
<i>Steve - Let me start out by saying that I'm aware that I should probably just keep my mouth shut at this point, but that's never been something I'm good at.</i>

Me too, as I'm sure you've deduced! ;-)

<i>As you stated earlier, we're pretty much both on the same page here, with just minor quibbles on semantics. Minor quibbles on semantics make for the best Internet flame wars.</i>

After I read this post, I agree. I think we're finally understanding one another.

<i>Incremental encoder:
Pulses (single or quad, single with direction, with or without Z channel, RS-422, TTL, Line-driver, push-pull...)Sin/Cos

Absolute encoder:
Parallel digital signals (Gray's Code or Binary)
4-20mA / 0-10V analog
SSI
BiSS
EnDat
Hiperface
Fieldbus (e.g. DeviceNet, Profibus, etc.)
Ethernet (e.g. EtherCAT, EtherNet/IP, ProfiNet, etc.)

It's not really the output that defines whether the device for measuring linear or rotary displacement is an encoder. It's the WAY that it measures displacement. Traditionally, encoders have been optical devices, although there are also magnetic versions. While the line may be blurred sometimes between an encoder and something else, there's still an "I know it when I see it" aspect to encoders. Internally, a pot works differently than an encoder, even if they are both outputting 4-20mA. This is also true in the linear world; a magnetostrictive sensor is NOT a linear encoder, even though they do the exact same thing from an external standpoint.

Here I disagree completely. Again, it's the internals, not the signal interface, that makes a device a resolver or an encoder.</i>

I think you're right about this. I'll adjust my internal dictionary entries on "encoder" and "resolver"!

<i>Okay, now that I'm on the same page (I think), I'll answer that. (a) If I had to come up with the solution quickly, I know the Kubler / Turck line has it, so I wouldn't have had to go looking around. (b) Arguably, starting with a discrete device and switching it to analog can be more precise. e.g. Encoders with analog output are not as affected by changes in temperature as potentiometers. (c) Though the 4-20mA output probably renders comparison moot, an encoder can have vastly higher precision than an analog sensor like a pot or resolver. While theoretically the analog devices have infinite resolution, in reality encoders have higher usable resolution by orders of magnitude.</i>

Fair enough. This is the answer to my original question.

<i>I guess I've been going on about this mainly because I don't understand your seeming animosity to encoders.</i>

Not at all. I've probably used more encoders than pots for industrial position measurement. My original post was from the perspective that the guy might be running down the path to one measurement technology before gaining at least a passing familiarity with the alternatives, one of which sounded to me like it would be a good fit. Like you, I'm not super invested in the issue. All my seeming belaboring came from seeing post after post misapprehended and trying to restate my point in more precise language to clear up the misunderstanding, not from any distaste for encoders.

<i>It mainly comes down to: An encoder with 4-20mA output is a perfectly valid solution to the problem of angular displacement measurement.</i>

True. Depending on the app and the precision requirements, I might still go with the pot, but I'm more open to the encoder idea than I was at the beginning of the thread, so you've done what few have been able to do!
 
Top