I got Mach three from my mother-in-law and upon running a couple programs I am getting skipping in the x-axis, I think.

I have attached my xml file and in it you can see the last thing ran was a rectangular pocket that I selected through the wizards tab.
I am using a Gecko G540 which powers 3 380oz-in stepper motors powered by a 48volt power supply. I am running the software on a Dell with 2Gb ram and a Pentium D Core-2 Duo running windows 7 32bit. On the CNC I am using a 3.5hp Bosch router with a ¼” end mill and all I am doing is climb milling.

I have attached a picture of the last program ran where it skipped steps and plunged into the side. I also attached a picture of a circular vent I tried cutting a few days ago when the error first occurred. I am cutting fiber board as seen in the picture.

I am very experienced with 3-D printing and the small stepper motors used with that set-up. I figured the 380 oz-in stepper motors should have plenty of torque to take 0.020” depth of cut at 15 in/min at 800 rpm with ¼” end mill climb milling in fiber board.

Please, if you could be so kind, recommend options to help correct this problem. Thank you for your help.

First thing I would suggest is to reduce accel and Velocity and see if the problem persists, also might be worth increasing the Pulse Width on Motor tuning page to 10. I notice you are using a 45KHz kernel where 25KHz would be more than adequate so might also be worth trying that reduced, especially as some Dells are known to have issues with clean pulses.

Just curious how you have X and Y motors connected as your steps per unit are very low for an imperial system, only 4000 and that means it only takes 2 turns of the motor to move the axis 1 inch, do you have a very course pitch screw (0.500 pitch)?
Hood

Hood,
my velocity is 200 and acc is 8 is this very high? my only comparison is from my 3d printing experience where this is rather low. My step pulse is 2 and my dir pulse is 2 also  is the step pulse what you  mean by pulse width? My kernel was set to 25KHz but in a effort to figure the problem out myself i choose to bump it to 45 to see if it made a difference, I have since moved it back down to 25.
 If dells are known to have dirty pulses  could you recommend any aftermarket cards that work well with cnc's I have been looking on newegg and found a few but in the reviews i have seen some people where intending to use the cards for cncs and could not get them to perform.
 Lastly in regards to my x and y leadscrews i am using 1/2 by 10  lead screws  with 5 start  i have had alot of luck with these in a prototype 3d printer i am developing.

It may or may not be high, all depends on your setup and how heavy the axis are, any friction and how much cutting force.
Yes set the pulses to 10 and see, both of them.

Hood

Oh also regards the dell, some work some dont but if you are unlucky enough to have one that doesnt work then as far as I am aware swapping to a PCI parallel port wont help, only using an external pulsing device such as the SmoothStepper would help.
Having said that it may be nothing to do with it being a Dell as some work fine.
Hood

Hood,
Thank you for your help. I changed the values to 25000KHz and the pulse settings to 10 and 10  in both x  boxes and y boxes.  After changing these settings I decided to try cutting a 1" dia hole in hardboard ,  feed rate was 20, d.o.c was .050 with 20% step over I was using a sharp .25" endmill 8000 rpm. The hole came out to be .987 in the x  and .978 in the y.  When I cut my first circle on the machine when I got it running I thought I was a lot closer to 1 in both x and y, could changing pulses have affected this . Or do I have some kind of backlash, machine is brand new  fresh acme nuts and all. Or am I being too picky and is this what the community would call reasonable for a 1" dia hole

Best thing to do is measure as accurately as you can a move that you command from MDI then command back the way and see if its the same distance, if not then you have backlash.
Hood

Hood,
I did as you recommended the results are as follows
using a digital caliper that was fixed. moving +/- 1 inch in mdi
x axis
test 1
move to 1: 0 -> 1.0005
moved back 0: 0 -> -1.0055
test 2
move to 1: 0 -> 1.0055
moved back 0: 0 -> -.9925

y axis:
test 1
move to 1: 0 -> .9975
moved back 0: 0 -> -1.001
test 2
move to 1: 0 -> .999
moved back 0: 0 -> -1.000

i understand that just because a caliper goes to 4 decimals doesnt mean its accurate but I am wondering if my 5 start lead screws are to blame, or the fact that my velocity is 200 and acc is 12  could  i be over coming some friction that is making my machine be off? also this  test was conducted  when the machine has no load ( not milling)  so i would imagine that it would be even worse like i saw with my 1 " cut out of a circle.
when thinking about cutting out a detailed objects a couple of thou. here and there could really add up. Any suggestions?

If its backlash it will not add up.
Try this if you can, move one way, say X+ then set the calipers up and zero them and also zero the X DRO in Mach. Put Mach into step jog and have the step set to 0.1 and keep jogging the same direction  by pressing the jog key 10 times. The DRO in Mach should read 1 and your callipers should read 1
 Now step jog back ten times and Mach will read zero and if you have no backlash so will your callipers.
Hood

So I did as you recommended. I measured +X to be 1.002 and -X to be 1.000, I did it a second time and got +X to be 1.002 and -x to be 1.001. In the y direction I got +Y to be 1.002 and -Y to be 0.996, and a second time I got +Y to be 0.995 and -Y to be 1.000. After that I tried to mill another rectangular pocket using the wizard. It was suppose to be a 1" X 1" square pocket, its actual dimensions are 0.975" X 0.984".

Would that be backlash?

What steps should I take from here?