Digital Depth Good, Physical Depth Wrong

So I've built this machine and it's working quite well. I'm a couple weeks now into fusion 360 and am slowly getting a small grasp on it. I have made many small projects and always end up with a similar result. Z always ends up too low. Z never ends up too high, X and Y are both spot on all through the program. Z shows -6.2mm in Axis, but in reality it's 10 something.

I have a handful of dremmel router tools in the fusion library that I've used a very generous person's python script to convert into a linuxcnc tools table file. So all of my tools are in there. I'm using the right one. I'm setting my orientation right. I'm touching off.

What am I missing here?

Thank You.

**EDIT** Also in Stepconf I can put Z on a set test run like 20mm +/- and it does 40 mm over and over again without 'drooping'. not sure what else to call it. In test Z seems to always be where it should be.

Also. I'm not running very fast. I had set the speed in fusion at 400 or 500 mm/min. I was running axis at 120% feed override. I can't just sit doing nothing so I'm doing a test now. I found a G64 code but instead of 0.1 I used G64 P0.01. So far axis says z is at -2.25mm and it's tough to measure, but it looks very close. I also place a marker on the machine so I can measure as z goes -.


Please help if you can. I fear that if this doesn't work out I may lose my mind.

Thanks again,
John

**EDIT** The marker is doing well. I'm supposed to have a gap of 1.50 and can barely slide in a PCB that's 1.56 so I'm pretty happy right now. If this holds up, do I need to enter that G64 command every time I run?

The coupling could be slipping on Accel and decell.

Try g61, to exact stop also

Hey Leon. Thanks for that. I saw a tormach guy also mention the G61. You might think I'm crazy but isn't the whole idea with CNC supposed to be precision? So why do we have to enter special commands for CNC to be as accurate as possible?

Also, Will I have to enter in the G commands every time I run the machine?

Thank You,
John

PS. Right now supposed to be at 4.50mm and we're really at 4.65. It is a difference, but just. I'm not machining carbs, heads, or anything else really that .15mm would make or break the job. Closest would be circuit boards, but there I'm only doing one pass so this shouldn't be an issue there at all.

**EDIT** I see that the ini file is where I need to focus my attention next.

Hey Leon. Thanks for that. I saw a tormach guy also mention the G61. You might think I'm crazy but isn't the whole idea with CNC supposed to be precision? So why do we have to enter special commands for CNC to be as accurate as possible?

What's nice with linuxcnc is that it's user configurable. You can set it up to your liking. If you need exact stop mode you can run it that way or do path blending with the precision you prefer. A high speed router or plasma table needs different settings from say a slower heavy mill.

I ran default settings after upgrading a while back and was introduced to path blending that almost crashed a tool during a rapid not clearing the stock as planned on the lathe. On my mill, a drill will clip the side of a hole during retract as it's rapiding in an arc move to the next hole if the blend is too large and z retract height too low. Just got to be aware what settings do what in the g code.

I don't think the path blending is causing your Z axis issues. You would see more rounded corners.
How are you testing the travel on the Z?

Ok. so thanks for responding. I just shut down and was coming back to post some results. yeah... still at it.

I have attached a needle to the portion of z that travels. Marking it's points on a static part I'm tracking the movements. It seems that when there's a bit of chatter it will dig in. Or does it chatter because it's digging in?

For the most part everything is cool with z, but once in a while it just travels down more than called for. Also, sometimes unexpectedly. Like it's not even doing a level change and it's still driving down in tiny weird back and forth movements.

All was going smoothly no level changes, just eating through .5mm x .7mm of particle board and it suddenly dropped like it was supposed to be drilling a hole. I immediately paused linuxcnc and grabbed onto the leadscrew to pull the spindle up. It was still driving against me. There's a problem here and I'm beginning to suspect my driver board. Whether it's a bad component or one that's easily effected by some outside source.

I guess I really can't call it, but I'm leaning toward the driver board. I made it using just a few Tiny85's, LD293's, BC547's and of course resistors. It's a good clean board etched properly, but something's going on.

Once I get this resolved, I would like to know if there's a way to control the feed rate when the bit has a full bite. For example. The cutter is 3mm dia. I tell fusion to take 0.75 bites at 0.5 depth which works great. However, on initial plunge and on level change the cutter is actually encountering 3.0mm @ 0.5mm depth. It would be nice to say hey on the first pass take it at a quarter throttle ya know.

So anyways, it's a 3 axis cnc mill. 36" x 36" x 24", Steel frame, aluminum drive parts. 34mm NEMA17 bi-polars. Any good experience with a very affordable driver board that would do a good job for me?

Thanks again
John

These nema17's don't use a tone of current. Z was only supposed to move 0.75mm. It does it 100 times over two hours then all of the sudden it moves 2.6mm instead of 0.75. X and Y are running from the same driver and are doing an impressive good job. It's strange that after doing so well, sitting idle for 5 minutes then getting a signal to move just a few steps would cause this issue. Z is set (in fusion) to move at 114mm/min. Then in Linuxcnc I had it turned down to just 30% feed rate. I really doubt it's too much too fast for the l293s.

Is there a way to test signal on the parallel cable? Like to see if the computer is putting out funky bumps on the z pulse line...

I'm leaning toward strange computer/power supply interference, lab power supply.

The Z axis on a mill or router is almost always the 1st one to give problems loosing steps, because it is fighting the weight of the spindle. And for this reason it is also going to go down. Each upward move command, it will loose a few steps (or sometimes a lot), all this adds up until it is off a lot. Also the axis may perform fine while testing (only when moving one axis), but when working suddenly you are moving multiple axis simultaneously and if your power-supply is not quite up to snuff you will get a big voltage drop and the hardest working axis (usually the Z) is going to stall.

Things you can do to try to improve the situation:
-add a counter balance to offset the weight of the Z axis.
-increase the voltage to get more torque at speed.
-slow down the moves (step motors rapidly loos torque with rpm)
-decrease acceleration (lowers the torque required to move)
-make sure there isn't a mechanical issue restricting Z movement

Hey Todd,
I realized there was a little more weight on Z than I had originally figured would be once the build was done. Still the stepper had no trouble moving to and holding positions. Still though I did add a couple of pull springs. So now without the system running if I just touch the bottom of Z it will move up a bit on it's own. It's not a weight issue.

There is no mechanical restriction. The feed rate was 114mm/min in fusion then set to run at just 30% feed rate in axis.

Z wasn't supposed to be moving at all, but holding the flex coupler or leadscrew I was able to feel that it was trying to move down, but not with normal steps or power. Just kinda puking out muddled little twitchy moves. down down down.

X, Y and Z all stay perfectly on track where they should be. I've marked and measured them repeatedly. Just all of the sudden Z is being signaled to move way more than it's supposed to.