Machsupport Forum
Mach Discussion => Mach4 General Discussion => Topic started by: jevs on October 27, 2019, 10:39:46 PM
-
I am using a Wildhorse Econoprobe
I adjust the tip with a dial indicator and get it pretty much dead on.
Then I put a new 1.0937" ring gage in the vice.
With a dial indicator in a collet, I center the ring gauge up and zero out x and Y.
Switch to the probe.
Go through the calibrate xy offset and radius.
After that I measure the bore of the ring gauge and it measures right on the money at 1.0937. A couple times I got 1.0936 after calibrating, but this is still excellent I think.
Next I probe the X edge and then load the edge finder and I was getting within .0001 of each other which is again really good.
Then I try to do the Y edge and keep getting about .0036 difference from the probe and the edge finder.
Not sure why this is happening, but I am not sure I can trust the probe or use it with it being this far off.
-
Not sure if this helps, but after many cals and tweaking, the Y axis always comes out with more offset.
Here are the current calibrated settings from the machine.ini (which have been about the same any time I have ran through the cal and re-adusted the centering of the probe tip etc.).
XOffset=-0.000213
YOffset=-0.001520
Radius=0.056210 (actual radius of tip is .0625)
I do wonder how/if backlash amount using compensation affect this?
-
If you rotate the probe 90 degrees, does that change your results?
-
No. Same measurements at 90. (Without recalibrating at 90)
-
I was thinking that the error could be due to the probe. My Wildhorse probe has a very stiff spring. It appears that the stiffness varies depending on the angle to the 3 axis of the probe. The stiffness appears to be strong enough to cause the probe shaft to deflect slightly. I never actually measured it though.
-
Hi,
I do wonder how/if backlash amount using compensation affect this?
Backlash comp is a realtime process and is handled by the motion controller not Mach4. If indeed backlash comp is screwing it up
you need to take it up with the manufacturer of the motion controller.
Craig
-
Right now I am not sure what it is. I am trying to narrow down and understand what Mach 4 is doing better.
It is strange that it reads a bore within .0001, but has trouble with edge finding accurately. If anyone has some suggestions on methodically narrowing it down to where the issue is that would be great.
At this point I am not sure if it is in the probe, the machine (the amount of real backlash), the motion controller (ESS), or something else.
I do have a second wild horse probe that is being used as a tool setter. I hate to mess with it because I have it all dialed in and the height set. If I don't figure something else out, I guess I could set that one up as a 3D probe and see how it differs in results.
Or, I could buy another probe that might be better than a Wildhorse and try that, but I am not sure what would be better and within a reasonable price frame?
-
I was thinking that the error could be due to the probe. My Wildhorse probe has a very stiff spring. It appears that the stiffness varies depending on the angle to the 3 axis of the probe. The stiffness appears to be strong enough to cause the probe shaft to deflect slightly. I never actually measured it though.
Wouldn't the calibration account for the flex (if you don't rotate it)? I have a mark on mine so I put it in the same spot each time. It is also in a dedicated collet tool holder. Since it calibrates in X and Y at that position, you would think they would be pretty much right on if always used in that position?
-
Yes, if you do it that way. I calibrated mine by rotating the probe against an indicator, so that the probe is centered in the collet. I should also then calibrate as you describe , to compensate for any variance in flex. That would compensate for the amount of deflection before the probe is triggered.
-
I also calibrated mine with an indicator before doing the Mach 4 calibrations. I am just not sure yet why I get the discrepancy when edge finding, yet it measures a bore right on (using ring gage for cal and checking).
-
I guess my next troubleshooting attempt will be to perform all the calibrations with the probe at a 90 to what I have been doing. Then I can see if the offset values change or stay on the same axis, and then I can do some comparisons to the edge finder to see what/if anything happens different. I need to narrow down if this is due to the probe itself or the machine/ESS/Mach.
-
Is the error reversed if you probe the opposite face?
Also is there an error in the other axis?
Have you another gauge ring you can measure, the other may be oval
-
The ring gage is brand new. I bought it for this purpose. It also checks pretty much right on with my calipers to the thousanth and it measures within +/- .00005 with my Mitutoyo inside mic., however I calibrated the mic with the ring also, but double checked it to my my outside mic and my calipers referencing a brand new gage block set.
So, I kinda have to trust the ring gage is okay. It is certainly more precision than the probe or even my machine I would say.
The error did not change when I rotate the probe 90 degrees and check (without doing a new calibration).
However, I also rotated it 90 and did do the calibration all over. It still had too much error.
I tried changing the Y offset manually in the machine.ini just to see how it reacts. I can dial it in to get the Y- correct, but then I did not check the others (Y+,X+,X-). I just wanted to verify it was working and kinda how it works.
After that, I set things all back up again and rotate the probe to a new position where the LED is facing the operator. This kinda falls where none of the 3 internal pins are in line with an axis. I wanted to see what it does. I centered up the X and Y to the ring gage with an indicator again and recalibrated the probe.
This is what I got this time:
XOffset=-0.001253
YOffset=0.002593
Radius=0.056330
I then did more measurements than other tests to have better data. I probed Y-,Y+,X-,X+ to get zero with the probe at each location followed with a check at each location with the edge finder.
This is what I got with that check.
.200 Dia Edge finder measurements:
Y- -.1014 (Target -.1000)
Y+ .0958 (Target .1000)
X- -.0979 (Target -.1000)
X+ .0996 (Target .1000)
So, I consider this still crappy.....
After every calibration I always measured the ring gage with the probe and it was almost always within +/- .0001. It did go out +/-.0002 a couple times. So I seem to be able to measure the bore of the calibration device very accurate, but edge finding sucks for some reason.
The backlash on my machine is a little worse than I would like at .0014 in Y and .0030 in X. I have backlash compensation on and if I go back and forth .015 it will usually land less than +/- .0005 from target or better (my dial indication is .0005 resolution and it usually lands on or within half way of that first mark off zero (so likely +/-.0003 at worst).
Open to ideas? This should at least be better data than I had before to work from.
When I first got this probe up and running, I did have to disassemble and clean all the contacts. The LED was kinda sensitive and flickery (indicating some resistance in the contacts). After I did this, the LED and signal seem fine.
Another thing of note is that if I probe the same thing over and over, the repeatability is always on or within +/-.0001.
I am wondering if I could just manually adjust the radius and XY offsets to dial this in? Is it possible the auto calibration is just not great, or do I have a probe or machine issue?
-
You could always have the machine laser aligned and mapped but the backlash is your main issue, to maintain the accuracy you think you need your machine mechanics need to be a power of 10 better, so you are not going to get it.
-
So you think the backlash is causing the probe accuracy issue with edge finding? Seems odd that it will measure a bore right on and the probe is repeatable, but it cannot edge find decent (compared to an edge finder) and the backlash compensation seems to work as it should.
-
place your setting ring at the farthest most point of the machine in X&Y from the home position, this on most machines is the least used place so is the lease worm part of the screw and slides. calibrate there and then edge find there.
If you find the same errors then its time for a new machine. End of problem.
-
I don't think the backlash is in the screws. I think it is in the end support bearings allowing end play. Long story not related to the probe, but it looks like there is a collar to adjust the slop in the bearings, but it may be all the way in at this point. I am just going to have to tear it apart and see.
New machine is a bit drastic don't you think :)? Other than this, the machine is in excellent condition with little wear and tear. Looks better than most few year old machines in a shop. Do people buy a new machine when it needs a couple bearings or a shim? Even a decent Chinese machine to replace this would probably be $25k. This one was 35k in 1978 when it was new (equates to about $137k in todays dollars). Yeah it is old, but low usage. I had it sitting around tinkering and upgrading it for probably 15 years or something and cut only about 6 parts.....before that it was just used at a college for training. It never was in production. I think I will go the route of fixing the slop instead of buying a new machine. That would kind of defeat the whole purpose of this whole mach conversion.....
Besides this, I am not even sure backlash is causing this. It may be, but It may also be the $100 probe. I don't really know what kind of results people typically get with these things. It just seems odd I can measure a bore to ten thousandths, but not edge find. I am not positive how accurate a Chinese edge finder is either. I found a couple references saying the trip point of an edge finder is pretty tiny, but it is a Chinese edge finder. However, even if the edge finder was off, the difference would be the same at each point and it is not.
-
The more I read about 3D Probes, 3D tasters, and regular edge finders, the more I wonder if my Chinese edge finder is even good or what I should be using as a reference or accurate...
How do I know that is not what is causing the discrepancy I see here. It seems odd I can measure with a known ring gage and the probe measures excellent, but only when I compare to the edge finder measuring edges is there a perceived issue.
Maybe I should be comparing the probe in a different way to make sure the edge finder itself is not the real problem?
It is a bit of money, but those 3D Tasters look nice to have for a mechanical way to check things. Just losing the .1000 math even though it is simple is almost worth it.