[RGlendinning]

Hey,

I am looking at getting some CL57y drivers and some P Series closed-loop steppers from Stepper online for my DIY CNC. The encoder on the motors has a resolution of 1000PPR, the driver can go to 40,000 Pulses per Rev. I am looking for 4,000 Pulses per Rev (or slightly higher) to try and achieve accuracy within a micron with 4mm pitch ball screws. So, my question is, will the resolution of the encoder be affcted by the pulses coming out of the drive(s) and vice versa? Can they be different? Will the low(er) res of the encoders affect/alter the surface finish/accuracy. Can this setup maintain sub-micron accuacy?

Thanks in advance for any in-site.

[joeavaerage]

Hi,
a closed loop stepper can, and reasonably does, interpolate microstepps up to the encoder count of the encoder fitted.

If your encoder has 1000 lines then it has an encoder count of 4000 counts/rev, which is what you require. Ideally the resolution of the stepper would be better than the required resolution,
but you can't have it all ways.

You might ask then whats the point of having microstepping setting of up to 40,000 per rev?. The short answer is there is no point. Just because you set microstepping to be 40,000 steps per rev
does not mean that you can actually get it. Its just to make it sound good to people whom don't know any better to get them to buy them.

With an open loop stepper, no matter how fine you set the microsteppeing the practical and achievable resolution you can get is with 1/2 stepping, and maybe at a push 1/4 stepping, i.e. 400 steps/rev and 800 steps/rev respectively.
With an encoder a closed loop stepper can interpolate better than that.......one of the only and few advantages that closed loop steppers have.

All steppers lose torque the faster they go. Closed loop steppers suffer the same way. Just being closed loop does not make them any faster, more powerful or anything else. Sure, if a closed loop stepper misses a step, and
that only occurs when its marginally overloaded, then the driver will insert an extra step to catch up, but guess what, the extra step is just as likely to be missed as a regular step. Thus a closed loop stepper starts to miss steps
in identical fashion to an open loop stepper when its gets marginally overloaded. Being closed loop alters that not one jot.

The only thing that a closed loop stepper can do is if it detects that its lost too many steps it will fault out and signal the machine that it is in error. That will usually will stop the machine rather than have it carry on and make a misshapen
part. The other advantage its that a closed loop stepper can interpolate between steps very much more than an open loop stepper can.

I still regard these slim advantages not worth the premium you pay for closed loop steppers.

If you want genuine closed loop performance, including very fine resolution and very substantial overload ( typically three to four times rated torque) capability get servos and be done with it.

I use 750W Delta B2 servos, and they eat any stepper ever made.

https://www.fasttobuy.com/flange-80m...er_p28084.html

Craig

[RGlendinning]

Thanks for the info!

The reason I'm not looking at a true servo is two-fold. 1) the size of the motors. Switching to a motor bigger than Nema 17 is going to be a TON of work. The machine will need to be redesigned. The big one is the enclosure will need to be reworked. And I still don't know if the enclosure will fit after modification(s) in my hobby room (Storage room). 2) Price, but if I can find what I need......

Does anyone know of a smaller servo, around a Nema 17 in size. The length of the motor is the major issue on one of my axis.

[peteeng]

Hi RG - The system you are describing will not maintain anywhere near sub micron "accuracy". Commercial mills that have 1um accuracy are 100's of thousands of dollars and 1M up. If the local temperature changes 1deg then you have moved 1um. So I think you maybe talking about possible resolution not accuracy. Here's a Kern machine that gives 1um accuracy its quite a big mortgage. Peter

https://cdn.prod.website-files.com/6...omprimiert.pdf

accuracy means if you tell the machine to go to (x,y,z) then its at (x,y,z) within its tolerance. Without feedback scales you actually don't know this.... for instance the HAAS mills quoted accuracy is 0.010mm or 10um so there's a lot of $$$ and engineering required to get down to 1um. Then there's repeatability and that maybe what your after. Consider a std NEMA stepper has a 5% tolerance on its rotation position I think that blows out the 1um.

[joeavaerage]

Hi,
accuracy is a rather complex subject....but resolution is not.

I have C5 ground ballscrews on my mill of 5mm pitch. Despite the servos having a 160,000 count /rev encoder I use electronic gearing (another big plus of servos) to have an effective encoder count of 5000/rev,
which results in a linear resolution of 1um.

I do not claim 1um accuracy as there are many things that go into that, resolution being one of them. I do reckon I've got about 8um accuracy over any 100mm cube, although how much that determination is affected by
temperature I've never bothered to try to find out. What I do know is that I can make parts with around +- 0.02mm tolerance, which is enough for me to make a living with this machine. Would 1um accuracy like a Kern be nice,
well sure it would, but if I had enough money to buy a Kern I'd retire and put myself 'out to stud' and be done with it!

The problem with highly accurate and rigid machines is the cost goes up and through the roof.....then you have to charge $300 an hour just to pay for the machine, let alone any business expenses or your own wages.
My machine has cost me around $35000NZD to date, and thus if I can charge it out at $30NZD /hour, and then some more for my own wages and expenses then I can make money with it......while I'd be losing money hand over fist with a Kern.

To OP; don't get too hung up about resolution, more often than not there are other factors like axis-to-axis perpindicularity, axis straightness, machine flexure under load all of which contribute to the degradation of accuracy and
very much more than axis resolution.

Craig

[joeavaerage]

Hi,
as an example lets say you have the X axis atop the Y axis, if they are not absolutely square to each other, that counts as inaccuracy.

Again lets assume you want to achieve 1um accuracy, and lets also assume a small target machine area, say 100mm x 100mm.
If the Y axis is out of square with respect to the X axis by just 1um over that same 100mm then

Angle (in radians) = 1um /100,000um or 0.000001 radians
Angle in degrees= 0.000001 x 57 = 0.00057 degrees or 2.05 arc seconds!!!!!

Hows your ability to read a protractor!

The upshot is that you have to keep your axes square to each other between 90.00057 degrees and 89.99943 degrees....and that is no easy feat!

If you want to achieve 1um over 10000mm the whole thing has just become ten times more demanding again.

I have a granite square I imported from the US, cost a bloody fortune despite being only 6"x6"x4" and it is 'AA' grade or 50 millionths of an inch over 6 inches
or 1.71 arc seconds. Thus my granite square is only just good enough to square your axes to the level of accuracy that you want to achieve.

I think you need to be more realistic about want you want or its going to cost a fortune.

Craig

[peteeng]

Hi RG - What are you trying to actually do? Here's Heidenhaim linear encoders. This is the sort of thing that you need FYI. In a deterministic system if you need 1um accuracy you then need all the systems to achieve at least 0.1um so that statistically you get 1um. The best encoders here run at +/-2um linear measurement and repeatability ... Peter

[v8kid]

Some really interesting posts here but they all assume the application is a 3 axis (or more) application. What about a 2 axis application I.e a lathe?
Ignoring the accuracy red herring for the moment is it reasonable to get resolution in the few micron range with a 1000 line encoded stepper?
It is not unreasonable for a skilled lathe operator using mid last century machinery to achieve close to 1/10 thou. I.e. a few micron. So can readily available Cnc techniques match this?
Note the op was specifically referring to stepper motors not servos as it turns out due to space and, importantly affordability.
Cheers!
David

[peteeng]

Hi V8 - 1/10 thou is 0.0001" = 2.5um. A skilled operator on a manual lathe achieves high accuracy by cutting then measuring then correcting, sneaking up on a measurement. This can be done on a CNC as well. We are talking about getting the required dimension off the machine with no creeping up manually. But this discussion also includes repeatability. Once the CNC is dialed in and the environment is stable then it should repeat its dimensional output properly. Peter

[joeavaerage]

Hi,

Ignoring the accuracy red herring for the moment is it reasonable to get resolution in the few micron range with a 1000 line encoded stepper?

Yes.

A 1000 line encoder or 4000 count per rev direct coupled to a 5mm pitch ballscrew will have a resolution of 1.25um. (1/4000 x 5= 0.00125)

Craig