mconcepts
try these settings for a starting base.play with a program that had notacable bad move.

set backlash speed to 45% -- set shuttle speed to 0.1 , this is a real good setting --- set look ahead to 5 lines.
play with this setup a little bit and see if you notice something.

bill

Rule of thumb i have recommended to start with is  50-50-50.
-BACKLASH SPEED = 50%
- SHUTTLE SPEED = .050  ( i use .005, but try the slower speed first )
- MOTOR VELOCITY SETTING = 50% of max velocity ( if motors skip at 100 ipm reduce to 50 ipm)

Now increase your velocity up to 70%  of max ( you are leaving 30% margin for cutting ) and try,
then start decreasing the shuttle speed down in increments to .005, trying each downward setting.
I would leave the backlash at 50% but if your velocity setting is at 70% of max, then 100 % may be just fine.

You need to play some as each machine is different.

RICH

OK. I'm playing around with the setting but it's really frustrating. Each time I think I have it set properly it screws up again. I can get it set to complete a Job correctly (after having to cut test pieces), then I move to a new Job and the X axis is messing up, test again, works fine with compensation turned off, change the settings and now it's working again, then 30 minutes cutting later the Z axis screws up, test again now I've fixed that problem... Now I have to rerun the entire job to make sure it isn't screwing up somewhere else. I'm testing 90% and cutting 10%. I can't seem to find a setting that will make it work correctly all the time. Each time I can change the settings to fix the current problem but then it screws up somewhere else. It's messing up on really easy code that I don't understand. Below is the latest example.

N59767 G02 X15.5063 Y9.8769 I-.5063 J-.8769
N59768 G01 X15.5688 Y9.9851
N59769 G02 X15.5688 Y9.9851 I-.5688 J-.9851
N59770 G00 Z.1
N59771 M05
N59772 G53 Z0.
N59773 G53 X0. Y0.
N59774 M01

All it has to do is finish the pocket, move to Z +.1 turn off the spindle and move to X0 and Y0. Instead it moves to Z +.1 then moves right back down to -.1875 where it finished pocketing and moves across the part (cutting where it shouldn't) and stops at some random 8. 9.? postion and stops. The funny thing is that the code is fine and the probem is repeatable. Only thing that makes it work correctly is turing off compensation. I Still don't understand what these settings actually do. What does Shuttle Accell have to do with Backlash compensation and what does the speed % have to do with why it would work correctly or not. It didn't work at all when set at 100%, worked but messed up occasionally at the default setting of 20% and now it is messing up at 50%. Do I just try 75% and see what happens? I know it can't be anything else like noise or grounding issues since when it does mess up It's always repeatable in exactly the same place in the G code and it always goes away when Backlash compensation is turned off. Very Strange. I'm also only running at 20-30ipm with a 1/4" 2 flute cutter so I'm not pushing alot of speed for wood routing.

 
First off.....you need to have the appropriate backlash set for each axis. Your feedrates are not that high.
What works for one axis may not be appropriate for another. So three axis need to be played with to get the combo right for you.

Shuttle Accel - It's  how fast the backlash will be implemented during non-motion. Try this so you can see what i just wrote. Set the shuttle accel to 5, jog at 10 ipm or so and then jog the other way. Backlash will comp will come in on a change of directiion, and if  you watch a handwheel on the motor  or the pulley, you will see it takes forever. Now change it to say .5 and see the difference. Try .005.

Let me remark here, should you have a lot of backlash and a fast shuttle acceleration ( i use .005 ), and use a small end mill, don't be suprised if you break them. The cutter can't take the forces. I use 1/64" ball ends and it will break in a heart beat and they are not cheap!

It sounds to me like you are missing steps. When you miss a step, the axis has a mind all of it's owne, and it's off to where ever.

There are folks in here that run routers and work with wood and provide some first hand experience.
RICH
RICH

I have different Backlash settings for each axis. I'm not using any for Z but I have tested and set X and Y correctly to get a perfect Circle. I tried the setting for the Shuttle Accel but got wierd results. With the setting at 5 and moving in one direction and then changing directions it changes directions quickly, slows down over about a 5 inch distance and eventually comes to a complete stop then slowly accelerates over the next 5 inches of travel until it's back up to full speed. I found this strange becasue I would have thougth it would have stopped at the point of switching directions then accelerated back to speed not change a full speed, slow down to a stop then accelerate back to speed. Does this seem right?

It's been some time since i fooled with it a lot to see just the affect of the different settings.
I just tried it in version .028 and it does behave differently than what i remember.
Dosen't seem right as you said.
RICH

Had the issue again on another Job last night..... I haven't gotten a chance to test this yet but I think I might have figured out the issue. For the last two times I have had the compensation issue it has been right before a tool change, after finishing the final cut. I think even the time before that it was the same situation but I didn't notice it util it started cutting after the tool change. I started having the backlash problem reappear after I upgraded to BobCAD 23 although the code looks fine and I hadn't cut enough after thinking the issue was originally fixed to assume it had anything to do with the BobCAD upgrade from V21 but BobCAD 23's CAM Tree and Mill/Router post automatically turns on and off the router for me in between tool changes (I really like V23 CAM alot, so no complaints here).  I think I'm getting the bad movements in the table when backlash compensation is turned on and the router is being turned off after the last cut. The Mach Mill/Router post was setup to turn the router off immediatly after finishing the last cut on the tool change and then move to home X0,Y0. This is where it misses steps, it never properly makes it to X0,Y0. I've now modified the Mach Post file to have it turn the Router off after it reaches X0,Y0 and not before so I don't get any electrical surges/interference when the router gets shut off and backlash compensation is occuring after the final cut on it's movement to X0,Y0. I think this will solve my issues even though turning off the router at that point without backlash compensation turned on has never caused me an issue once it just seems to be a combination of the two that is causing the issues but I'll test and let you know. I have several posts that I can recreate the issues with.