Also don't know if I mentioned but this problem isn't related to speed. If I turn the speed down to 100mm/min I get the exact same cuts as running 900mm/min. My acceleration is quite low also, 10mm/s/s.

Sebastian

Sorry earing should be gearing
Sounds like you may not be using it but I will give a wee explanation anyway
Ok when using servos with highish resolution encoders you will be limited to the max velocity you can get when using the parallel port. This is because the max pulserate of Mach is 100KHz but for most 60KHz is a likely max as it all depends on your particular computer hardware.
Ok so steps per unit.... say you have  4000 line encoders that is 16,000 pulses per rev as far as Mach is concerned. Assuming no mechanical gearing between the motor and screw and the screw being 5mm pitch your steps per unit would be 16,000/5 = 3200 steps per mm. Now if you had the kernel in Mach set to 25KHz that would mean the max Velocity you could achieve would be 25,000/3,200 x 60 = 468mm/min
 So say you find you can reliably pulse at 60KHz your max rapid would be 60,000/3,200 x 60 = 1,125mm/min
Now for most that is not a very fast rapid but that will depend on the machine for example 8 to 10m/min is sort of normal for a knee mill or lathe running servos. So as you can see that is not possible when using the parallel port, there are two things you can do.
1. Use electronic gearing if your servo drives support it, what that means is you tell your servo drive that every time it sees a pulse from Mach it treats it as say 5 or 6 or 10 etc. That way you can reduce your steps per unit in Mach by dividing by the electronic gearing ratio So say previously you were maxed out at 1,125mm/min as in the above example, if you had 5:1 electronic gearing you would now manage 5,625mm/min

2. The other option is to use an external pulsing device such as the smoothstepper, it can pulse up to 4MHz so you are unlikely to hit a velocity limit.

Now lastly, reason I asked about the electronic gearing was,  some of the lower end drives do not do it very well and can add a few or loose a few steps every now and then.

Hood

Ahh makes sense, I didn't know that such a thing even existed.

Can someone tell me more about tool compensation and where it's located? I've never used this nor set it up but I just want to make sure that it isn't set to something.

Sebastian

Your tool comp would be with a G41 or G42 in your code.
 I dont use the wizards much bit I dont think they use comp at all so it is unlikely to be your problem.
Hood

Can you post the G code for one of your test circles. Earlier you said it cuts perfect on outside cuts but OFF on inside cuts is that still true??

What version of Mach3 are you running?

(;-) TP

A little update. My dad said to cut a circle with the Mach3 wizard with the quill fully extended (around 6 inches) and barely extended, both times the circles had perfect finish and were 0.2mm too small. Following this I uninstalled Mach3 the latest most stable version (3.042?) and installed 1.84 which is the version that I have on my benchtop mill. I setup the travels again with my dial indicator and this time the circles were around 7.2mm too big. I triple checked the wizards settings, bit was 12.7mm but no luck every circle was a disaster. I then uninstalled and installed the latest development version. Reset the steps etc... Now the circle wizard was cutting within 0.08mm too small which I thought was great because the tool when brand new could be -0.05mm and it's cut a lot of aluminum so you could remove a couple more hundreths of a mm. I then proceeded to cut a test piece from solidcam and complete disaster, it cut the outside 0.2mm too big and the inside 0.2mm too small.

Everytime that I set the steps, I didn't copy the previous value I instead got the dial indicator out and set it with the configuration step option. Each version had a slightly different value, but between the most stable and dev version this difference was very small, while it was larger going back to the 1.84 version. I'll post gcode tonight, don't have it with me.

I'm running out of ideas.....

Sebastian

You have accurate balscrews, you know the count of your encoders and you I would imagine know the gearing between motors and screws if you have any. That is all you need to calculate the steps per unit. If you tell me the numbers I will calc it for you. Calculating is the most accurate way if your hardware is accurate.
Steps per unit do not change from version to version of Mach so there is something wrong if you are needing different values each time.

Also what kernel frequency do you have set?
Hood

Let me clarify, when going from the most stable -> the latest dev (3.0 something) the value changed by 0.1-0.3 something very small. When going from 3.0something to 1.84 it changed by around 10.

I'm sorry but I don't know that the ratio on the pulleys is as I purchased this mill with a damaged Anilam controller so the servos already the housings and belt conversion done, I just put in new ballscrews. Isn't that the point of the set steps funtion to make this calculation for you? It's not like I'm using an analogue dial indicator or a cheap brand with no reputation, it's digital and very accurate so that intrigues me as well as to why the values change. When I set the steps, I rerun the table to different distances using the step set function (but I don't type in the actual range travelled, just press esc to not change the set steps) ranging from 1-12mm and they are always withing 0.002mm.

It's the standard 25000, as I haven't changed it.

Sebastian

I recently had this problem and figured out the cause.  First - alert - I'm a noob... but I know that this caused my issue.   I was having issues with my part cutting fine, but then a second part of the g-code causing the finishing stage of my part to cut approx 1" to the right of where it should be (mill ended up going out into the stock where it shouldn't).  I was recommended to try to slow down the feedrates etc..  What I did was play with the Motor Tuning section of Mach.  However, I did it blindly and it caused my issue.  My unit is setup for 4000 steps per inch (157.48/mm).  I left this value, but cut the other two (velocity & accell) in half.  This caused my part to be cut at approx half the size it should have been.  When I changed it back, it worked fine.  Not sure if this is the same issue you have, but worth checking to ensure you have your Motor Tuning setup correctly.

What are the steps per unit you have?
What line count are your encoders, you mentioned 4000 but not sure if that is the line count or the quadrature output?
What pitch are the ballscrews? I assume they will either be 5mm itch or 0.2inch pitch but will need to know.
With the above info I can calculate out as I will know roughly your gear ratio from your steps per unit and it is most likely that te gear ratio will be a round number. You could of course count the teeth on each piulley and get the ratio that way but if you let me know the above it should be straightforward. Measuring distance is ok but it will only be as good as your measuring . Even having  good measuring  gear will not be as accurate as calculating when you have good screws etc

Hood
Hood