encoders/scales on X2?

[X2cnc]

I'd like to have table position information independent of stepper counts. Preferably, rotary encoders, since they would be simpler to mount and less likely to be damaged than glass scales.

Perhaps my google-fu is weak, but I'm not finding much on position feedback on the desktop mills. Is it a price thing, or are the stepper counts accurate enough that it's not a problem?

[Crevice Reamer]

Originally Posted by X2cnc Is it a price thing, or are the stepper counts accurate enough that it's not a problem?

Hi X2cnc. Welcome to the Zone!

It's both. Stepper systems cost less than servo systems and are less complex. They can be configured to be quite accurate without losing steps.

Stepper motors have their maximum torque at standstill. As the speed increases, torque decreases. Stepper motors lose steps when they stall. They stall from lack of power at rpm. Steppers wired Bipolar Parallel maintain their power into more rpms than other wiring methods. Properly sized and powered stepper motors do NOT lose steps until they reach fairly high rpm.

The trick is to use a large enough power supply and suitably rated drivers to provide them with their BEST PERFORMANCE voltage. This varies from motor to motor and is established by the formula: 32 times the square root of the motor phase inductance in Mh.

For example: A Keling N23 425 oz stepper has 6.8 Mh of inductance when wired (as is the best) bipolar parallel, and draws 2.8 Amps. The square root of 6.8 is 2.6076 and 32 times that = a best performance voltage of 83.4 volts. For THIS motor, you would want to find a PSU and drivers that will provide as close to that as possible.

The Gecko 203V will run 6A and up to 80 volts. Keling offers a 72V PSU that will get as close as possible.

This combination should let the motor run as fast as possible with the most power at speed--thereby minimizing the possibility of lost steps.

Now if you were going to run this motor at 24 volts with a Zylotex controller--You would get much less power at much lower speeds and would probably start losing steps at a fairly low rpm.

CR.

[sansbury]

Originally Posted by Crevice Reamer This combination should let the motor run as fast as possible with the most power at speed--thereby minimizing the possibility of lost steps. Now if you were going to run this motor at 24 volts with a Zylotex controller--You would get much less power at much lower speeds and would probably start losing steps at a fairly low rpm.

I would agree with everything you said but I think "fairly low" needs to be qualified a bit more with regards to the specific application, which I'm guessing to be an X2 based on the OP's name.

A typical cutting speed on these mills is anywhere between 8 and 20IPM. If you're using a direct-drive 5-pitch ball screw (a typical X2 setup), that equates to motor RPM of 40-100, which is pretty low.

Also, 425oz is not a marginal motor for any of the X2 axes. You need RPM up into the hundreds before the torque drops by more than half, and there are a lot of X2s running happily with 24V/<300oz motor setups from what I've read here. You aren't going to see 100IPM rapids with 1/4 microstepping but you don't need that. Unless you're already going all-out, the extra money the fancier drives would cost is probably better spent on something else.

OTOH if you were running a wood router, it can be a whole different story. 60IPM is a cutting speed here and when your axes are 3+ feet long, rapid speeds start to mean a lot more. If using a typical 10-pitch acme screw then you are going to need a lot of RPMs to get there.

Though it's worth noting that a lot of production machines this size (like the Shopbot) use low-ratio rack drives with higher-torque motors. Ultimately if you need torque and speed you end up in servo-land. Steppers are better for slow and strong until you get to fairly high torque levels where AC servos start to become cheaper.

[Crevice Reamer]

Hi sansbury! You are, of course correct. We are both talking about the same thing, but in different ways.

Not to argue, but the motor example was just one which I had the specs in my head for. I used it to demonstrate how to choose the best power/speed voltage. That motor would NOT be my first choice to run at 24 volts.

The original question implied to me, that X2cnc has heard a lot of myths about steppers losing steps. My discourse was just about a way of making SURE there are no steps lost.

CR.

[sansbury]

CR: Yes, agreed. There is a problem in that people get simple ideas stuck in their head which become popular not because they are correct but because they are easy to understand.

'Steppers are less accurate than servos because they can lose steps' seems to be one of the more widely-held semi-truths around here. My pet peeve is guys putting 300oz motors on Sherlines.

[LongRat]

I'm not sure this has much to do with the original question...? Please correct me if I'm wrong. But if you want absolute accuracy, nothing will beat linear scales mounted on the axes. The mechanism used to drive the axes is completely immaterial. You are not limited to glass scales by the way, check out the steel tape scales made by Renishaw, designed for harsh environments and equally accurate.
Fundamentally though, an X2 doesn't justify going to this level of expense or complexity. The component parts aren't of a good enough quality to support the data from linear scales. For a more accurate machine, I would spend the money on a higher quality base machine, with ball-track slides etc. Like the Wabeco machines. Also, finding control electronics to interface with linear scales is going to be difficult. I've never seen anything hobby-level to do this. There are plenty of hobby-level servo controllers out there though, however all of these work from feedback from motor-mounted rotary encoders. Your mechanical arrangement after that point will therefore contribute further error.

[X2cnc]

Originally Posted by LongRat I Also, finding control electronics to interface with linear scales is going to be difficult. I've never seen anything hobby-level to do this.

I haven't either, which is why I asked.

I have an X2, the Xylotex 425oz 4-axis kit, and the CNC Fusion bracketry and ballscrews, which haven't arrived yet. While I'm putting it together, I want to make sure I haven't missed something obvious.

[sansbury]

Originally Posted by X2cnc I haven't either, which is why I asked. I have an X2, the Xylotex 425oz 4-axis kit, and the CNC Fusion bracketry and ballscrews, which haven't arrived yet. While I'm putting it together, I want to make sure I haven't missed something obvious.

Sounds like you are doing well so far.

There is a thread going in the past few days talking about interfacing with cheap Chinese scales to do this kind of closed-loop feedback. The electronics are non-trivial and there is also an issue because the cheap scales only update 50x per second which limits speed. You can go to glass scales but by the time you're done it's cheaper to put a proper screw setup on it.

[tooManyHobbies]

Originally Posted by Crevice Reamer This varies from motor to motor and is established by the formula: 32 times the square root of the motor phase inductance in Mh.

[Anal mode ON]
Your explanations have been spot on for every post I've read of yours, so forgive my little nitpick here: it's mH for inductance. Henry is capitalized because it's named after a person (Joseph Henry). 'm' means milli (10^-3), while 'M' means mega (10^6).

Sorry. I'm an EE & I can't help myself sometimes.
[Anal mode OFF]

I've often thought of adding linear encoders to my X2, and did some costing to see how bad it would be. USDigital has some very nice mylar linear strips for $6.30/inch (500 CPI) for lengths greater than 7 inches. They also have the encoders to read them for for $35.70 (500 CPI). You could use the 250 cpi versions of both and use 4x resolution multiplier in the encoders to get up to 1000 cpi to reduce cost. Of course once all these items have been purchased, the next step would be to design a PCB controller for interfacing to all of them, and then writing firmware to provide a fast enough update rate to ensure no missed steps. Not sure how this data would get into something like Mach3, but I'm still intrigued by this design and would love to do it some day!

After reading many posts here and other places, I decided that it's not needed, and there's no system setup to properly take advantage of it (but future Gecko drives might change that?).

tom

[Crevice Reamer]

Originally Posted by tooManyHobbies Your explanations have been spot on for every post I've read of yours, so forgive my little nitpick here: it's mH for inductance. Henry is capitalized because it's named after a person (Joseph Henry). 'm' means milli (10^-3), while 'M' means mega (10^6). tom

Thanks, Tom! Nothing to forgive! In the interest of accuracy, I'll try to remember that.

CR.

[The Blight]

About that prefix. Once saw a package containing 500Mg of sugar. Thats a lot of sugar!

About positioning with stepping motors. I too am concerned with loosing steps, and so I have designed a simple encoder that will be mounted on the front of the motor. This has a resolution of 200 counts (50 holes). And so if the motor is off by 1 whole step (when you loose steps, you have actually lost 4 whole steps as far as I can figure out. So you have to be more then 2 whole steps off to actually loose a step), it sends a signal to the E-stop pin of the controller. You can also hook it up as a limit switch signal if you want to. When you first start up the machine it will calibrate itself by finding the center of one hole and working out from that. I have not yet tested this system, so I can't say if it will work as well as I hope, but in theory it should work. I'm going to make the encoder and housing myself. I have also put a resonance dampening disc on the encoder.

[sansbury]

Originally Posted by The Blight About positioning with stepping motors. I too am concerned with loosing steps

Just out of curiosity, why?