Machsupport Forum
Mach Discussion => Mach4 General Discussion => Topic started by: cncnc on April 08, 2018, 04:28:55 PM
-
Hi guys,
I have a problem with a jerky axis movement on my home built machine (with PoKeys 57CNC controller) and don't quite know where it comes from.
The only affected axis is the Y. It is a dual axis (two ball screws, two 8Nm closed loop steppers). Both bearings on the Y axis are quite preloaded but this shouldn't be a problem for the stepper motors.
During the milling cycle the Y axis starts to bang when it accelerates. At first I expected the worst, that both axis motors didn't move synchronously and the gantry is put under heavy stress. But turned out not to be the case, I measured the flatness of my X axis and it was just fine. If both Y motors would have moved different from one another, the gantry would bent. This shouldn't even happen though, because both stepper motors do have encoders and the same input signal. If they wouldn't move in sync, an alarm would be triggered.
So I watched both motors during a job and noticed that both seem to jerk simultaneously.
I don't think this comes from overload of the motors, they have plenty of power and in general they aren't equally strained during a job so a simultaneous jerking movement is quite unlikely.
Additionally these jerky movements don't appear when I jog the Y axis. This leads me to the assumption that there must be something wrong with Mach4 or the way I set it up.
I already thought this may come from too many steps per unit, but according to my calculations it should be ok and Mach4 (PoKeys plugin) didn't complain.
But I don't know how these step frequency calculations apply to a dual axis or when more than one axis is moving.
In order to give you an impression of what I mean I uploaded a short mp3 file where I face off my slot table. From 2,5 - 4 seconds you can clearly hear the banging of both motors in the background. AudioYaxismotors.m4a (https://www.file-upload.net/en/download-13072590/AudioYaxismotors.m4a.html)
I hope you have some ideas and can help me out!
Many thanks and best regards,
cncnc
-
Hi,
what steppers and drivers are you using?
What microstepping are you using?
What are your native units?
What is your current 'steps per unit'?
What pitch are your ballscrews?
Do you have any gear reduction/belt reduction between the steppers and the ballscrews?
I suspect that you will need to use microstepping, it results in smooth movement which is what you want.
If you don't like calculating steps per unit....bad luck...its part of successful CNC.
Post the relevant data and we'll do the calculation.
Craig
-
Hi Craig,
thanks for your help.
I'm using Leadshine Easy Servo ES-M23480 with the appropriate drivers ES-D1008.
They are configured to a 12800 steps per revolution. In combination with 10 mm pitch ball screws that's 1280 counts/steps per 1mm.
There is no gear reduction or similar involved, both ball screws are directly driven from the motors.
I don't think the problem is to less micro stepping. I initially thought it may come from too many step commands that Mach4 or the controller can't handle.
But I did all the calculations on it. And according to that there shouldn't be a problem. The PoKeys maximum step frequency is 125kHz, with 1280 steps per mm that leads to a maximum velocity of 5859 mm/s. I don't exceed this limit.
Did you listen to the audio file? What does is sound like to you? Does is it sound rather like a software or a hardware problem?
I already thought about loose couplings, but I tightened them as hard as I could and both couplings are quite oversized, they shouldn't have a problem to transmit the torque. And if they would slip, the axes wouldn't be square anymore, but they are.
In the end, I have no idea what's causing these problems and I'm glad for every help.
Best regards,
Yannik
-
Hi,
yes I listened to the audio and clearly the stepper is not happy. As to why sound alone doesn't tell me.
It looks to me that your 'steps per unit' is correct and given that you understand the signaling limit of the PoKeys board that should be OK.
How have you hooked your controller to the drivers? I thought the PoKeys boards are 3.3V output types and that would be insufficient for your drivers.
Craig
-
The PoKeys57CNC runs with 5V logic level, I don‘t see a problem there.
I think that this behavior is caused by the toolpath itself, not any communication errors or something like that since those jumpy movements exclusively occur at 2D adaptive toolpaths of relatively similar geometries.
Thanks for your time and help!
-
Hi,
reading the manual for your stepper/driver and its fitted with a 1000 line encoder which results in 4000 counts per rev. I'm intrigued as to how you get to a microstep regeime of 12800?
Craig
-
Hi,
found the page with the DIP switch settings and 12800 is certainly on the list.
Given that your encoder is 4000 counts per revolution is there any thing to be gained by even greater number of micro steps? Looking at the list there is an option for 3200 steps per rev.
Each step would be achieved because the encoder has slightly finer resolution whereas your existing 12800 setting could only be guaranteed at those locations where the encoder positions are,
ie 4000 positions in a revolution.
Even with the microstepping dropped back to 3200 per rev the resolution would be 3.125um or just a little over 1/10 th of a thousandth of an inch, plenty good!
It might be worth a try, after all you only have to flick a couple of switches and redo the 'steps per unit'
Craig
-
Hi Craig,
you‘re absolutely right. In terms of increasing the accuracy that high number of steps/rev is total nonsense, since you loose quite a lot of step torque and probably will end up being less accurate. But this isn‘t my goal. I chose this specific motor because it has a lot of holding torque and I can increase the steps/rev in order to smooth the movements and reduce the induced vibrations but without sacrificing performance (by not falling below the required application torque with the step torque).
Nevertheless I‘ll give it a try and see if that‘s the problem‘s solution (I‘ve already cut it down by half). If so, I might dampen the vibrations of each motor with a rubber pad between motor and machine.
As a side note: I programmed all drivers with the dedicated software. There are some options that aren‘t accessible from the DIP switches. For example you‘re able to set even higher steps/rev than with switches. In fact I could set steps/rev exactly to 4000 if I wanted to.
BR
-
Hi,
as you know most of us who use microstepping with regular steppers do it for smooth running. Its nice to think that we get increased resolution but due to the reducing
differential torque we get resolution of 400 steps per rev irrespective of the number of microsteps.
Getting smooth running without undue number of steps per rev is maximized with microstepping of 8 or maybe 16 microsteps per full step or about 1600 to 3200 microsteps per rev.
You have set your microsteps to 12800 steps per rev. Are you of the opinion that you are actually getting smoother motion as a result?
The reason I ask is because I wonder if your machine is attempting to make a huge number of very small steps at certain times, you mentioned 2D adaptive toolpath, not sure what that is, but
maybe it introduces a whole bunch of steps. I wonder that while your PoKeys controller can produce pulse streams of 100kHz it may be that at the top rate the pulses are getting attenuated by
the cabling and the drives are not getting a clean signal?
Craig
-
Hi,
at the moment microstepping is at 6400 per revolution. I‘ll reduce it to 3200 tomorrow for test purposes.
Initially I planned with 3200 microsteps because I‘m aware of step torque loss. But I found that there‘s still a little noise/vibration improvement with 6400 steps over 3200 steps.
The Pokeys should easily be able to handle the number of steps at the driven speed, it‘s not even in the top rate area. But I will reduce the microstepping furthermore, then we‘ll see, maybe there‘s indeed another transfer problem on the line
-
Hi,
I agree the PoKeys should handle this no trouble. I'm wondering if the signal is being transmitted faithfully to the driver.
As you know when using RF you have to use coaxial cables or other transmission lines to convey RF energy from one spot to another. Well 100kHz would be
VLF radio but there are radio signals that low and I'm thinking that you may be required to give some thought to the cabling between your controller and the stepper drives.
It may or may not be an issue and might be simple enough to experiment to prove one way or another.
Craig
-
There are no radio signals or similar in my basement.
I exclusively used shielded cables, both for power and signal.
Reducing the step per revolution didn’t change anything except increasing the noise level and induction of vibrations.
I stay with it: for the most part it‘‘s the toolpath generation. The installed motors are quite big, this leads to a increased noise generation. If I wouldn‘t have heard it, I would not have noticed this at all. Probably I‘ll dampen the motors‘ screw joints by adding some rubber layers between, that should minimize the noise to level where I don‘t notice it at anymore.
-
Hi,
I wasn't talking about radio signals interfering but rather that your 100kHz signal is a VLF radio signal. LORAN-C is 100kHz for example.
My suggestion is that you need to be careful that such a high frequency signal is cabled in such a manner that it is transmitted faithfully.
Craig
-
Yes, I understood that and agree with you that at these high frequencies usually coaxial cables should be used. I only use shielded cables, though. But I think it works just fine.
I thought you talked about external radio signals interfering with my control signals.
All cables in my electric cabinet are arranged in a way that signal and power cables are spatially divided from one another. I did the best I could for avoiding interferences and I think it's sufficient. I mean, I don't have any real disturbances or unpredictable movements. They always happen at the same code lines and all motors behave like that, the dual axis even synchronously. If they would be out of phase, the portal would change its orientation on the linear rails by little. But I let the Gcode run several times now and measured all angles between my axes afterwards. They didn't change a bit.
It get's a lot better though by increasing the speed and playing with different accelerations.
I don't there's a fast solution now, it's probably something that has to be worked out step by step. It could be everything. Next step is that I'll update my CAD/CAM solution the next days. I see a good chance that this could vanish this weird movement as Autodesk did some severe changes to their software.
I'll take note of every change I make on my machine in the future and check if anything has changed regarding this. If so, I'll let you know.
I want to thank you, Craig, for your time and your willingness to help. I really appreciate it, many thanks!
-
Hi,
glad you are making some progress.
My concern about cabling and the high frequency of pulses is not about interference, either interference leaking in to the step circuit nor interference leaking out of it
but rather that the signal may be attenuated.
Consider most coaxial cables have input capacitance of 60-90 pf per m. Thus two meters of cable between the PoKeys board and the drivers puts a load of 150pf on
the output of the PoKeys. The output is a low power TTL gate, excess capacitance could easily attenuate its output, further it would be worse at high frequencies than low.
A lot of very small moves as is often encountered in toolpaths in the vicinity of corners and the increase in frequency of the pulses might result in lower than expected
signal level at the drivers.
I'm not saying that it is the case but it could be. You will no doubt have noted that servo drives often specify an upper limit of 100kHz- 200kHz single ended signal but the same
drive will accept up to 500kHz when differentially signaled. Thus signaling a drive at high frequencies is not trivial.
Craig
-
Hi,
frankly, I didn't think about the "cable load" and also didn't pay much attention to frequency depending signal disturbance. I know that HF signals usually should be transmitted via coax cables, and for avoiding signal reflexions the wave impedance of the cable should match with the connected device. But I didn't use them because there aren't coax ribbon cables available (I took the best I could get: shielded (ribbon) cables) and I never heard that not using coax cables in applications like this caused problems. If so, most people (and even many companies) would be in real trouble. It sounds little overkill to me.
I only tried to prevent any interferences from happening.
As far as I can tell though, I don't have problems with the signals itself, otherwise I would probably see some really bad and more random behavior and it would occur at other tool paths (also very fast and small little corners), too, but it doesn't.
Nevertheless I'm gonna hook up an oscilloscope (as soon as I have my own) to the drive inputs in order to see if the signal level drops when the PoKeys is firing with commands :-D
Thanks for this interesting input!
-
Hi,
I have a servo as a spindle motor and was signaling single ended at 150kHz and I hooked up a my scope and was surprised to see how rounded the signal envelope had become.
I had about 4m of cable, shielded twisted pair and the output of the BoB is regular LS TTL. It did signal the drive ok but I was suprised at the signal degradadtion.
I rearranged the electronic gearing to reduce the signal rate to just under 50kHz and shortened the cable to 1.5m and now the signals look good.
Craig