Encoders for a Stepper System with a 7i85s

This is for a small stepper powered mill, I would like to run in a closed loop with linuxcnc.

Components I am working with:

NEMA23 dual shaft steppers, 1/4"(6.35mm) shaft dia. , .750"(19.03mm) hole spacing for encoder

Mesa 7i85s


The main constraint to work around seems to be the mounting dimensions this size of stepper requires.
Then the next factor is compatibility with the mesa inputs, but that looks to be a bit more flexible.

I have been looking through several US suppliers, and haven't found too many options with good
documentation yet.

At the time of this post, the most dimensionally compatible option I have found is a CUI AMT10 series
incremental encoder. I will attach some pictures of key electrical specs I found on this unit below.

This CUI encoder is marketed as using some form of "capacitive" tech with a "CMOS" output. It has a
5pin connector, 5v power spec, quadrature output, and an 8 pulse index. Looks like it may work with the
mesa board, but with single-ended outputs, although I am not entirely sure it is going to produce a
strong enough signal to work reliably. There does appear to be decent off-the-shelf shielded cables
available for this encoder, so that should help somewhat with EMI problems.


That is the general information I am working with, hoping to get some input on compatibility with the
7i85s, or hear other more functional options.

You are likely going to spend more on encoders & cables for those CUI/AMT encoders than a complete, turn-key closed-loop motor & drive bundle from one of the off-shore vendors.

I realize you want to close the loop back to LCNC, but I'm not sure there's going to be a significant performance gain vs. just using closed-loop drivers (in this case) and running LCNC open loop.

Besides, the closed loop drivers are likely better able to drive a similar size stepper at the upper RPM limit, as well as being quieter, than a 'dumb' drive.

If you want to 'close the loop' to LCNC I'd suggest looking in to adding linear scales to the axes.

The cost for one of these encoders with a 6ft shielded cable was only around $40 USD and that didn't seem too bad.

Serviceability has value for a machine like this as well, I could possibly get everything in a kit (from China?),
but knowing that individual components on the mill have available domestic stock and good documentation is more
important to me.

Also, I would rather not have encoders integrated within the steppers, that approach is somewhat less flexible.

Linear scales would be nice, but there will be a rotary table on the A axis that would still need an encoder,
and due to the compact size of this machine, it could be tricky to fit good scales. Integrating the rotary is part
of the interest in wiring everything back to LCNC as well.

As for performance, I am just interested in a more reliable control loop as opposed to upper limit feed rates.
The idea is to have more feedback on stepper loading to possibly identify missed steps sooner, an example case
may be working on a piece of exotic wood and encountering a bad knot for instance...

You did bring up several relevant points, I am just trying to work reliably with the available equipment and better
understand the available options. (Capacitive vs Magnetic encoders for instance?)

You're really not going to gain much closing the loop with LinuxCNC over a closed loop stepper combo. That CUI stuff is still made in Asia, if "domestic stock" is a big deal just buy some extra closed loop stepper motors and stick em on the shelf, you'll have enough money to do that saved over the whole setup anyway.

Besides that I don't understand wanting to stick an external encoder on a NEMA23 dual shaft stepper when you can just buy the separate closed loop stepper motor and just wire up it's encoder to your 7i85s. There's nothing special about the encoders on those motors, they aren't serial, they're incrimental differential encoders, generally 1000ppr which alleviates your other concern of the CUI encoders being single ended and sudceptible to noise. It's generally just a DB9 connector, break it out or cut it off. You can get closed loop steppers all day on StepperOnline.com

Adding a $40 encoder setup to every axis isn't too far off in price from just buying 400w AC servos by the time you factor in the stepper DC power supplies. Servo drives generally have emulated encoder outputs you can feed back to the 7i85s. Not sure what kind of control loop you wanna setup like that but you can certainly get the same visibillity from the encoder outputs on a servo drive that you'll get with discrete encoders.

While servos are a superior choice, the model of mill I am setting up was originally made with a closed loop stepper
system, so I choose to keep my component choices closer to the original, but without paying for a license for mach3
when LCNC could be more capable. For better or worse, I already have a 48v dc power supply and stepper drivers.

German or Japanese steppers are still my preference, but at the moment they have been hard to find here. Hence
part of the interest in a seperate encoder, I could just swap out whatever marginal stepper I had to settle for and bolt
the encoder to a higher quality stepper when they became available with less modification needed.

It is not my goal to be contrary here, and I am sure there may be options available from sources I have not been aware
of. So far, I was primarily working from what was available through Mouser and Digikey as they are well established
industrial sources.  For models in the 200 to 300 oz/in range, Mouser and Digikey didn't have anything good in stock,
and I ended up settling for Automation Technology the other week because I got tired of searching through a ton of
catalog pages for unavailable product.

The steppers that arrived from Automation Technology were horrible, (metal shavings in all the steppers, hard to turn)
and the customer service on the phone with them was rather poor as well. I got burned on those, so please try to
understand that I am a bit skeptical of a second go around with product marketed in that tier for even less cost...

Rodw had made a suggestion to look at moonsindustries as an option, and their stock looked reasonable, but even
their prices were still higher than stepperonline, suggesting a different target market. I may still end up buying something
from Moons as they appear to be the same supplier for the steppers that were originally equipped on my mill.



Again, I didn't really want to start a supplier debate as different people search for different grades of product for different
reasons. My rationale for starting this thread was mainly to discuss encoder technologies that could be functionally
implemented on a Stepper equipped Machine.

So far, the "all in one package" with the encoder being part of the stepper has been the suggestion, utilizing a differential
encoder of a magnetic tech variety.

Not much comment on the usability of a "capacitive" single-ended product with shielding has been made yet.

There is also "optical" models, but I haven't found any of those that would fit a valid stepper for my machine.
(*edit* usdigital does seem to have somewhat modular options available)

*EDIT*  (please disregard the "fullstep output" comment below, this has been coverd in later posts)
Another possibility could be to use the "fullstep" output from a driver, but I have not seen too much discussion
regarding the real differences there, that would seem like a less positive feedback however...

Your questions are all over the place here. You seem to have decided that you really like the CUI encoders but basically what everyone is telling you is that they are not more convenient, cheaper, or superior to anything you find  on a garden variety closed loop stepper.

If country of origin is your thing you're better off looking for a brand rather than a particular distributor like Digikey. AFAIK Oriental Motor is all produced in either Japan or US, so just find a distributor for Oriental Motor. No idea about other brands cuz if it's not made in the US it may as well be made in China AFAIC

StepperOnline is where I buy most of my steppers, they rebrand stepper products, they don't manufacture them but they're more than just a distributer. Kind of like (old) Craftsman  or Snap-On tools. neither of them made tools, they either directly rebranded or had an established manufacturer produce to their specification. You can tell by the different steppers they sell that many of them are manufactured by different companies. I've never had a quality issue with any of them though although I don't set high standards for stepper motors cuz honestly the way steppers are constructed it really doesn't matter that much. If they're made well they're all pretty much the same given the specs of course.

As for encoders I don't think anyone knows what on earth a "capacitive" encoder is. Really, don't get too caught up in magnetic or optical cuz if it's rated for a speed it will produce clean pulses throughout that speed. If it bothers you that much research the difference between photo diodes and hall effect sensors because that is the difference between optical and magnetic encoders. I guess in that same vein capacitive encoders are akin to capacitive proximity switches but I never use those so I have no idea.

Not much comment on the usability of a "capacitive" single-ended product with shielding has been made yet. 

What are you looking to hear? There's nothing special about any of that. If it generates pulses than yes, you can use the single ended output with the differential receivers on your 7i85s but why do you want to do that if you have a choice? I think you're trying to talk yourself into the awesomeness of "capacitive, shielded, single ended" cuz you really wanna like the CUI encoders but there's nothing special about any of that. A shielded encoder cable is the basic norm, differential of TTL. Single ended is inferior to differential in any respect to noise immunity, there's no question about that. There may be some question whether it matters but not what works better. Of all the signals in your PC, if they are very high speed or have to travel some distance or originate from a cable, they are differential, there's a reason for that.

Another possibility could be to use the "fullstep" output from a driver, but I have not seen too much discussion
regarding the real differences there, that would seem like a less positive feedback however... 

This one I don't understand at all. I think you're suggesting using a stepper driver output as an encoder output? It's not quadrature, it's probably not 5v, and it's not based on mechanical feedback so I don't really know where you're going with that.

As for encoders I don't think anyone knows what on earth a "capacitive" encoder is. Really, don't get too caught up in magnetic or optical cuz if it's rated for a speed it will produce clean pulses throughout that speed. If it bothers you that much research the difference between photo diodes and hall effect sensors because that is the difference between optical and magnetic encoders. I guess in that same vein capacitive encoders are akin to capacitive proximity switches but I never use those so I have no idea.

That was a more relevant response. I couldn't locate much information on what a "capacitive" encoder was and suspected it
was primarily just marketing hype, but there is always room for discussion so I don't like to just dismiss tech simply because
it is different. Honestly, I didn't want to use this type of encoder as it is not differential, but it does have good support for
different mounting requirements so it was a mechanically compatible option.

The more pertinent detail that still hasn't been covered is encoder compatibility with a 7i85s, more specifically, what is the
practical minimal signal current that the mesa input can reliably read. From what I saw on the CUI documentation in the first
post, it looks like that type of encoder may not have very strong signal output.

Now, I will apologize that I did not make it more clear in the first post that I was primarily focused on discussing encoders
from a compatibility standpoint, but that is what I am clarifying here, so please do not take this personal, and we can still
keep this discussion informative.

My comment regarding "fullstep output" was not meant as "encoder output" but rather as an alternative to a closed
feedback loop, ie. monitoring the "fullstep output" from the driver to throw a fault in the event of a stall. This was a thought 
that came to mind after reading discussions in several places that suggest closed loop stepper systems are not worth it.
Anyhow, please forget that I brought that up, the more useful topic is bebunking the usefulness of a capacitive encoder,
and outlining what the minimal signal current constraints for a 7i85s really are.

The 7I85S encoder inputs in single ended mode have a 2K ohm input
impedance (2K pullups to 5V) and a ~1.65V threshold so are compatible
with TTL, 5V CMOS and 3.3V CMOS output levels, as long as they can
sink the 2.5 mA pullup current.

That said if you want encoder feedback  to check for lost steps/stalls,
you are probably better off with closed loop steppers (and monitor
the fault lines from the drives)

Feedback from step/dir to encoders in LinuxCNC really does not have
any major advantages unless you are using linear encoders where you can
improve accuracy. (They can be used do check for missed steps but it's
probably more economical to just use close loop step drives.

As a point of reference, I'm using the CUI/AMT 31 (capacitive) commutation encoder on my lathe spindle servo.  Mine is a differential output, and the signal is going to a Copley XSL drive.  CUI shielded cable.

No (obvious) issues at 170khz (2048ppr @5kRPM)

EDIT
Note that I'm also using a fairly high res optical encoder directly on the spindle for positioning.  I wouldn't trust the CUI for really accurate positioning of a rotary axis unless there's a huge reduction... but it's working fine with the servo in velocity mode since LCNC manages the position via the 10k pp/r optical.

Utilitas wrote:

after reading discussions in several places that suggest closed loop stepper systems are not worth it.

With the prices of servos having come down substantially recently, closed loop steppers do not have the price:performance value that they once had.

They still have uses - in some applications it's nice to have the non-dithering (hunting) steady-state characteristics. And they are certainly easier to use than low-end servos from a tuning or novice installation standpoint.

And if your mechanical system simply cant take advantage of the higher speeds available with a servo, then opting for a closed loop stepper package might make some sense to save a few bucks.