To Servo or not to Servo .... that is the question

21

Yes, according to Clearpath you could just swap them out. I would bet they are the same exact moter with different firmware. They said the custom tuning in the motors that AVID puts in the standard motors could be copied over and would work just the same.

As Eric mentioned, there would be improvements in raw accuracy, such as the resolution of being able to make finer steps for backlash compensation or laser rastering, but for most things where you are just moving around following vectors there is a lot of dynamic stuff going on (i.e. the motors are nevers really exactly where they are supposed to be) and that would probably negate most if not all the improvements.
The other consideration is that this is going offroad a bit so AVID may not be able to help you much if you run into difficulties.

I have been planning on getting a set of those motors to test with laser rastering just as a fun project to do, but Iā€™m waiting on some other software improvements first, and i have been busy with another big project for the last 6 months. Maybe early next year.

1 Like
22

Thanks for the info guys. Sounds like its a good idea to hold off for now, but I would be very interested to see the real-life results if you try them out @jjneeb

23

Hey Greg - do you think moving to servos improved the off-the-machine quality of your work?

I still would like to get servosā€¦I just have a hard time paying the upgrade price when my current system is pretty darn good!

Corbin

24

Hey Corbin, I upgraded from Mach4 to the EX controller and switched from steppers to servos at the same time, so itā€™s hard to pinpoint exactly which improvements came from which change.

One big benefit has been crash detection, it gives me great peace of mind when Iā€™m away from the machine. The ability to run very high feedrates has also been a big time saver. Iā€™ve been roughing hardwoods at up to 350 IPM using a .5ā€³ bit, 0.35ā€³ stepdown, and 0.2ā€³ stepover with no issues at all. With your steel frame, I imagine you could push it even faster.

I am not sure if there was a huge difference in overall quality of parts coming off the machine. I think its slightly better with the servos but its hard to tell without something side by side. I wish I had done more before-and-after testing to compare.

Right now, Iā€™m following the steps in Jimā€™s axis calibration and backlash compensation videos to see if I can improve my machineā€™s accuracy any further.

25

When I get around to it, there will definitely be a video :slight_smile:

2 Likes
26

Yeah, makes sense! I definitely saw improvements when I went to the EX controller, but I still see some ā€œwavy-nessā€ in some specific operations that I wonder if they are caused by stepper movements being too harsh. It could be a number of other things too (smoothing, feed/speed, and the bit).

Iā€™ve been holding off buying a rotary simply because I think I might get servosā€¦and I donā€™t want to get the rotary motor twice. But maybe I should just stick with steppers and get it.

Most my use is hobby-based (and YouTube), so itā€™s hard to justify a big purchase that is just for fun when my basic system is working pretty good.

27

Todayā€™s task for me was to upgrade my machine from my old servos (salvage, and two of them died) to the clearpath servos. Remembering this thread, I put a test indicator on the spindle and did some measurements.

Note: my old servos had a 1000 line encoder, but I tracked edges, giving 4000 steps/rot, but it could only hold within a count or two, which makes it comparable to the 800 steps/rot clearpaths.

With the default settings for the servos, I saw it step about 14 times per 0.020ā€ but the step timing wasnā€™t smooth. I ran the speed slow enough to watch the steps every few seconds (iirc F0.1). The step count did seem to confirm that the servos are limited to 800 steps per rotation.

However, the default settings are 2000 steps/rot. Now, 2000 is not a multiple of 800, so it occurred to me that there would be some positions you couldnā€™t command accurately, so some of the actual movements might be timed differently.

So I changed my servos to 3200 steps/rot (in case the servos were tracking edges instead of lines of its encoder - I reprogrammed mine to run in quadrature mode as my controller supports that). Same test. I counted 8 actual movements per 0.010ā€ move but this time the steps were consistently timed.

Given the proā€™s drive is 1.018 rotations per inch, 7-8 steps per 0.010 is pretty close to 800 steps/rotation actual.

Backlash was on the order of 0.001ā€ or so but gently pushing the spindle gave a 0.005ā€ range of motion, which is Avidā€™s spec anyway.

Conclusions:

  1. The 800 step limit puts you at the limit of what the Avid PRO hardware can resolve anyway, so thereā€™s no real reason to go further.
  2. You should reconfigure your servos to some multiple of 800 steps/rot to make the most of the coarse resolution.

Given the enhanced option adds 15% or so to the costs of the motors (a couple hundred bucks per machine) , the 8x resolution might be a good call for some cases, but it would be tricky to justify for most PRO users. Maybe if you really needed smooth motion at fine detail, and the 0.0012ā€ steps are just too coarseā€¦

But what about steppers? The average stepper is 200 steps/rot actual, but microstepping can effectively increase that to 800 (4x) or more (I ran mine at 3200, or 16x). While that seems better, remember the microstepping is interpolated, and not mechanical, so YMMV. Accuracy not guaranteed, but again, higher microsteps might mean smoother motion at fine detail, even if the motion is not as accurate.

Reminder: steps per inch is precision; how close it gets to where you want is accuracy :wink:

Edit: the above measurements are for X and Y; I didnā€™t measure Z.

2 Likes