• This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn more.
  • PLEASE: Read the FORUM RULES BEFORE registering!

Mach3: Unexpected Response for Steps per Inch

angelfj1

Active User
Active Member
#1
Hello:


Although I have completed the CNC conversion, I'm really having fits with the motor tuning and axis calibration. Something very strange is happening. I would like to take this real slow because I have limited experience with Mach3. I will list some basic facts regarding my cnc conversion.

My machine is a Grizzly G0704. Although this started out as a Hoss Phase II project, I changed my plans and purchased a pre-engineered kit supplied by David Clements, aka ArizonaVideo99. I had the saddle machining done by a local shop. They also did the oiling groves. Everything came together nicely.

Details
The ballscrews/nuts are metric - 5mm pitch
The steppers and other electronics are:
2 - KL23H2100-50-4B, 570 Oz-In (NEMA 23, for X & Y axes
1 - KL34H295-43-8A,960 Oz-In (NEMA34), for Z-axis
Drivers
3 - KL-5056D Digital Bipolar Stepper Motor Driver, set for 1/8 micro-stepping
BOB - C10, parallel port interface

I'm using Mach3, running on Win XP. I currently have the evaluation (free) version, but intend to purchase a license as soon as I iron out the strange symptoms I’ve experienced.

I have been able to get everything connected and responding to mach3. I can jog the axes.

My problem started when I tried to use motor tuning.

I set the X, Y and Z velocity and axxeleration very conservatively, 20 inches/min, and 3 in/sec/sec

I calculated the steps per inch first before trying the Mach3.

I did it like this:

My stepper motors - 200 steps/rev
Micro-stepping - 8 pulses/step
So, 200 x 8 = 1600 pulses/rev
no gear or pulley , e.g. N = 1
ball screw - 5 mm pitch or 5/25.4 in. pitch = 5.08 TPI
steps/inch = 200 x 8 x 5.08 = 8128.

So, I entered 8128 for steps per inch.

Then, I did an initial gib adjustment for the X-axis as suggester by Steve, Jumps4, ihis n YouTube video.
I made sure that the X-axis gibb wasn't too tight.

I set up a dial indicator to measure movement in the X-axis. For an input of 0.100, I was consistently getting .097. I did this several times and took an average. So, I went to the Mach3 settings page to use the steps/inch routine. I entered a value of 0.100 and the x-axis traveled .097. Then for "how far did you travel", I entered 0.097 and Mach3 returned a steps/inch value of 812.xxxxxxx, WHAT IS THIS? Since the X-xxis travel was only 0.003 short, I was expecting a number slightly more than 8128, maybe 8129 or 8130. But 812?????
OK, I know something is crazy, so I answered NO to the recommended steps per inch, closed the program, turned off my electronics, turned off the computer and restarted. I repeated the routine in Mach3 settings for steps per inch but this time I got a different number but still off by an order of magnitude.

Before I take any drastic steps, I wanted to reach out to others who have some experience with Mach3 issues.

Many thanks,

Frank
 

RJSakowski

H-M Supporter - Premium Content
H-M Supporter-Premium
#2
I ran Mach 3 for four years on my Tormach but never had to run the calibration routine as that had been done prior to shipment. I did have to run a similar routine on my lathe DRO though.

Rather than proceed the way you have, you could divide your expected travel by the actual travel and multiply that by the calculated 8128 steps/inch. You should get 8379 steps/inch. Enter that in place of the 8128 entered previously.

Run your expected vs. actual test again. To get a more accurate number, you should measure over a longer distance. I would use the longest distance that I could accurately measure. This will minimize any backlash/lost motion effects and round-off errors.

Actually, I wouldn't expect the pitch on the ball screw to be off by that amount. That is a 3% error.
 

Metal

Active Member
Active Member
#3
Oh you already found that screen

Dont move such a small amount, move an inch or more.

Any tiny amount of backlash and such is a larger percentage of the move the smaller distance you go, also make sure you are moving into the same direction

Ex move left, then calibrate left, do not move right, then calibrate left

if you go left-> right -> left without leading in you aren't taking backlash out of the picture, you set that up somewhere else.

It may take a few calibrations to get it "just right"
 

Bob La Londe

Active Member
Active Member
#5
You could have .003 backlash in cheap Chinese ball screws. AS somebody else said, if you do not advance a little to take up backlash and then advance further to measure you are not getting accurate measurement.

Yes your numbers are very conservative. You should easily get 10-15iss and 80 IPM with that relatively light machine. Maybe more. Where you might have some issue is with the weight of the head. You might have to add an air cylinder, gas spring, or a pulley with counter weight to lighten it up so it doesn't stall and lose steps when going up. If you just CNCed the quill and not the head never mind that last bit.
 

DAT510

H-M Supporter - Premium Content
H-M Supporter-Premium
#6
Have you gone through the Artsoft video tutorials? It been a while since I went through the setup of my cnc mill, but I found the tutorials really helpful.

I think I went through Paul Campbell's YouTube series that used video provided by artsoft. It went through all of calibration and other setup, along with the rational, which really helped me uderstand how Mach3 worked.

 

MontanaAardvark

H-M Supporter - Premium Content
H-M Supporter-Premium
#7
Frank,

Did you get this resolved yet? I'm running Mach 3 on G0704 with the parts Hoss recommends for phase 3. It's much like yours except I put a 570 in-oz NEMA23 motor on Z also.

What I did is like what RJ said. First, I went past where the start should be and then moved in one direction to a starting point. I did 0.100 at a time, starting out with the same 8128 you did (I think - it sounds familiar). I eventually started past the 0.000 line, went to 0.000 then commanded the machine go 1.000 " - full scale on my indicator. It was off (don't recall how much). Let's say I told it to go 1.000" and it went 0.992. So I divided 1.000 by 0.992, (1.00806) and multiplied 8128 by that to get 8193.5, which I rounded the decimal off. That got me to move 1.000 at a time. I wasn't using the screen you're referring to, just calibrating the motors from the pull down menu.


Bob
 

cs900

maker of chips
Active Member
#8
can i make a suggestion? Go buy one of the really cheap DROs (you know the kind with the little remote box readout) as long as the max travel on all your axes. Install it on each of your axies one at a time. Doing this you can:
- ensure your steps/unit is perfect
- map your ballscrew accurately (assuming you have home switches)
- accurately measure any backlash
- measure lost steps when motor tuning.

all in one shot. I know it takes a bit more money, and a bit more time to get set-up, but there's no question about the accuracy of your mill after that. At least within the tolerance of the dro scale.
 

Boswell

Hobby Machinist since 2010
H-M Supporter-Premium
#9
can i make a suggestion? Go buy one of the really cheap DROs (you know the kind with the little remote box readout) as long as the max travel on all your axes. Install it on each of your axies one at a time. Doing this you can:
- ensure your steps/unit is perfect
- map your ballscrew accurately (assuming you have home switches)
- accurately measure any backlash
- measure lost steps when motor tuning.

all in one shot. I know it takes a bit more money, and a bit more time to get set-up, but there's no question about the accuracy of your mill after that. At least within the tolerance of the dro scale.



this is what I did, but I only got one that was 12" long. I just clamped it to the table for the test and used a dowl pin in a chuck to move the head in the axis I was testing. Very easy to get accurate movement length and then convert that to steps.
 

angelfj1

Active User
Active Member
#10
Let me provide an update and I apoligize for this late response. After speaking to Dave Clements, aka ArizonaVideo99, I disassembled the table, ballscrews, x and y ballnuts, etc. It became obvious that the setscrews holding the ballscrew mounts in place had loosened. I added loctite and re-tightened. My backlash is now, X=.0015, Y=.0005, Z=.0025.

I'm happy.

Regards, Frank