[jviss]

A couple of things. First, I may have introduced a terminology collision here: "hybrid" can refer both to a stepper motor configuration, as opposed to permanent magnet or variable reluctance, as well as to a system that uses steppers for motive power and also incorporates encoders for position feedback. When I say hybrid I am referring to the later.

Someone suggested that if adding encoders to steppers made sense, why isn't it more popular? - saying it isn't more popular because it doesn't pay off.

I'd like to suggest that there may be applications where this make a lot of sense, if the benefits of encoders with steppers work out as I think they might. For example, if you have a G0704 with 200 step per revolution, directly coupled stepper motors, and a 0.200" lead ballscrew, the only way you'll get below 0.001" resolution is via microstepping.

In this application the use of steppers may make complete sense, with the exception of the resolution.

It is my understanding that a hybrid system with appropriately designed drivers can achieve higher positional precision than one without encoders.

In addition, it is my understanding that a hybrid system can recover from lost steps because it will detect the position difference between the encoder count and step count, and compensate.

Is this not so?

[SCzEngrgGroup]

First, on these machines, the stepper resolution is virtually NEVER the limiting factor on machine accuracy - it's not even in the top three. It's really picking the fly poop out of the pepper. Errors due to thermal expansion and machine flexing alone will completely swamp the stepper accuracy/resolution. Never mind the fact that no matter what level of micro-stepping you use, under heavy load the motor will be pushed off-position by a FAR greater amount than the theoretical resolution, as the motor has only max torque when at a full step position. Even the typical lead error of most ballscrews FAR exceeds any error introduced by the steppers. So, if extreme accuracy is your goal, you're throwing away good money to solve the wrong problem, and leaving much bigger problems unsolved.

Second, as has been stated several times here, a closed-loop stepper system **CANNOT** correct for position errors on-the-fly! It does not have the torque to do so, or it would not have lost position in the first place! It can only make you aware you that the error has occurred, and at that point your only recourse is the shut the system down, re-home, and pick up where you left off.

You need to forget all the greatly simplified, highly idealized, theoretical behaviors of stepper motors, and deal with the way the REALLY perform in the real world. Far better to figure out what accuracy you REALLY need (and RARELY is extreme accuracy required for the kinds of project I see people doing here), and set reasonable goals, or get ready to spend a great deal more money. And start with a machine of FAR better quality than those here, or you're just throwing good money after bad. It's not terribly difficult to get these machines to a reliable accuracy of +/- several thou **in the hands of a competent operator**! And that "competent operator" part takes MANY YEARS of hard work to achieve. Going beyond that, you're starting from the point of diminishing returns, and each small, incremental improvement will cost you almost exponentially more than the one that precedes it, as you will uncover more and more deficiencies in the basic machine and components from which you started. But NO attempt to achieve accuracy will succeed without first actually thoroughly characterizing the exact behavior of YOUR machine, through extensive measurement and testing, so you know where the problems are, and which are the most severe. You then must thoroughly understand the behaviors of all the components you'll be adding to the machine, their *real-world* behaviors, and what new problems you'll introduce. Otherwise, you're just playing darts with a blindfold on.

Regards,
Ray L.

[mactec54]

jviss
Quote
I don't think anyone has mentioned hybrid stepper systems. I saw one from Automation Technology, Inc.. They have a NEMA 23 kit with motor, digital driver and cables for $250; $440 for the NEMA 34 version.


You can get the Dmm tec Ac servo's for around the same money
There are a lot of Zone members using the Dmm Ac servo system for the same machine that you have

DMM Technology Corp.

[Fastest1]

Just for the sake of this argument, my G0704 is running 8000 steps per inch. This gives a theoretical accuracy of .000125". I don't have anything to measure something that small. I couldn't see it if I did. ;-)

[jviss]

Just for the sake of this argument, my G0704 is running 8000 steps per inch. This gives a theoretical accuracy of .000125". I don't have anything to measure something that small. I couldn't see it if I did. ;-)

Cool!

So, I assume you are direct-connected to the stepper, the stepper has 200 steps per rev, your lead screw is .2" per rev, and your are microstepping 8:1 to get to 8000 steps per inch?

How does it perform?

What is the torque rating for your motors?

Is it stable in between major steps? I mean, if the machine stops between steps, does it really stop there, or does it flop over to a major step?

Have you looked into the microstepping linearity?

Thanks,

jv

[jviss]

jviss
Quote
I don't think anyone has mentioned hybrid stepper systems. I saw one from Automation Technology, Inc.. They have a NEMA 23 kit with motor, digital driver and cables for $250; $440 for the NEMA 34 version.


You can get the Dmm tec Ac servo's for around the same money
There are a lot of Zone members using the Dmm Ac servo system for the same machine that you have

DMM Technology Corp.

Thanks, that's cool. I haven't come across any threads yet for guys here using servos. Are they direct-coupled, or driving through a reduction, i.e., belt or gear reducer?

[hoss2006]

Thanks, that's cool. I haven't come across any threads yet for guys here using servos. Are they direct-coupled, or driving through a reduction, i.e., belt or gear reducer?

Sdubfid's G0704 is one of several g0704's using servos, he's using DMM.

http://www.cnczone.com/forums/bencht...ids_g0704.html
Hoss

[SCzEngrgGroup]

Just for the sake of this argument, my G0704 is running 8000 steps per inch. This gives a theoretical accuracy of .000125". I don't have anything to measure something that small. I couldn't see it if I did. ;-)

First of all, knowing it runs 8000 steps/inch gives you nothing more than an *indication* of the theoretical resolution. It tells you nothing whatsoever about "accuracy". And if you DID have that measurement capability, you'd find the actual resolution is MUCH greater than that calculated value, and the accuracy, in terms of actual position vs theoretically "correct" position, is FAR worse. Accuracy and resolution are two VERY different things, and CANNOT be used inter-changeably. Nor is it meaningful to talk about one, without also talking about the other. And both are VERY dependent on a great many factors, and cannot be known from one simple, highly idealized, component specification. Otherwise, we'd all be running 1024X micro-stepping, and claiming 0.00000000001" resolution and accuracy.

Regards,
Ray L.

[Fastest1]

Ray I agree and I did mention it was a theoretical resolution. There are many relationships I dont understand in this field and leave it to the more experienced to explain the nuances. I believe my machine will go where asked very accurately so far within .0005 repeatably. It is directly coupled Nema 23/570/5056D and if the part doesnt look beautiful and or measure correctly, I would blame it on CAD/CAM/operator(otherwise ME) error. Most of these machines with a few modifications and attention to detail during assembly and set up will exceed most of our expectations and for sure our abilities, especially those of us who are hobbyists. Too much worry about a problem that might not be real.

[jviss]

Sdubfid's G0704 is one of several g0704's using servos, he's using DMM.

http://www.cnczone.com/forums/bencht...ids_g0704.html
Hoss

Thanks!