I have made a conventional mill tot a CNC mill.
There was a backlash item in all axis.
The X and Y axis are solved since I measured the differences and studied the solutions in this forum (thamks)
However teh backlash compensation in the Z axis direction is not going to work.
Why, because the compensation must only work in one direction (only in upwards direction) because due to the gravity the spindel lowered too much).

Is there any solution for this becuse now if I lhave to lower the spindel 10 times , every time it lowers the drill holes becomes deeper due to backlash.

Ultimately I have to change the G coding with a backlash value every time it goes upwards again what is very anoying!

Please can anybody help??

Eliminate the backlash then you won't have a problem and you will be a much happier camper.
How is the Z movement done on the mill?

I solved my problem by raising and lowering the quill using a ball screw. Now hold  a 1/2 thou.
RICH

If gravity is eliminating backlash then I would say  it follows you dont have any backlash
Hood

Rich,

The movement is done via a wormwheel and a toothrack.
It is not easy to eliminate the backlash

I would disable for backlash in the Z and check to see that your steps per unit are accurate.
Hood

The typical drill mill spindle and rack combination is not very good for controlling the Z travel since there is play untill the gear makes contact.
I think mine was around 0.025" or so and that was using the sensitive feed gearing which helped.
Sometimes repeatability was excellant and other times poor.
Yes you could use BC, and maybe find a sweet spot, but it will never be right. Will just plain suck should you decide to do 3d machining.

A friend has a geared head mill which he applied a counter balance to the complete head with pulleys and now raises and lowers it very accurately with a small stepper and screw.  The head is probably 400 lbs.

RICH

could a much heavier return spring can do the job?

If you can eliminate the backlash by making sure the pressure (spring load)  bearing on the axis is enough to keep one side of the tooth of the rack engaged with one side of the pinion when other forces are applied (plunging for example) then yes it may help. Problem then may be that your motor has to be much larger.
I think it would be a lot easier to design and make up some ballscrew mounts and use that rather than the rack and pinion and also a lot more accurate.
Hood

off coarse you are right about that.
but the problem is that this is a major upgrade for which i need the same machines as I am dismantling.

I was also curious to know whether there are small (hobby) milling machines and lathes which have already ballscrew shafts implemented originally.
At this moment I am preparing a router with drive axis AND guiding existing of ball bearings shafts because i was able to buy a number of these axis (new) for a relative low price.

I was curious to know whether this leads tot a very precise router.

I suspect you are translating to english so I will attempt to explain this using words that will translate correctly:

I don't think your problem is backlash.

Backlash could be making it worse.

I think your problem is that your Z axis is dropping steps or losing position when it raises against gravity.

This will make successive holes deeper.  The Z axis moves less than the amount your G code specifies due to dropped steps and then does the drilling at a now lower position.

I have found that turning on backlash compensation makes stepper motor based systems more susceptible to dropped steps because of the way the backlash is done at a percent of maximum instead of accelerated.  This might have changed in recent releases.

Backlash will not be additive in any direction.  If you tell the Z to raise after it was lowered by .005 backlash + 1.000, Mach will do the number of steps to go .005" at a percent of maximum that you have specified.  This might cause your motors to miss steps since the heavier Z axis is not being accelerated correctly as it would under control when backlash is turned off.  If in this example, you then told the Z to move back, it will move the same amount of steps as it did to raise up.  Therefor even though the backlash has been removed by gravity, it will be forced back upwards taking up the backlash when the drill bit hits the work.  So you don't gain steps because of backlash compensation.

Jerry