Help with rotating head

I would say it's just a servo control defect.
On a single point, the CMM cannot go to the exact target, but gets the right readings.
For the sphere, even if you measure it with 4 hits, you shouldn't see any difference because the feature is calculated from actual values.
For a single point, there's no compensation available, so it gives the actual directly.
IMO

maybe "lobing error"
​

Hello all!

I am facing this issue and I am not sure if this is normal (it shouldn't be imo)

Setup: TIGO SF TP20 module
PC-Dmis 2017 R2

To put it simple, I am taking a vector point at -1,0,0 on a plane with a probe at A0B0 workplane Z+
I am repeating this exact same point but going from the side with A90B-90 (whatever workplane, gives the same result)... and I am having a 15 to 30 microns deviation between the 2 measurements. This applies to +1,0,0 also and to more than 1 probe.

I have 2 other TIGOS with the same setup and I have not noticed such a strange behaviour. Is it normal to have 30 micros deviation between scenario A hitting perpendicular to the probes axis and scenario B hitting parallel to it?

I find it very weird.

Finally, I am correctly calibrating, setting up parameters as move speed/ touch speed/ prehit and retract. When locating the calibration sphere in a seperate verification program, I see no deviation btw... Thanks in advance for your replies!

Are these manual measurements or are these DCC measurements?

Touching straight into a TP20 has the most pressure that the module will feel, as you are lifting all three lobes up at once.
In that picture of the triangle, hitting on the point of one, so that it teeters on the other two lobes is the least pressure.
everything else is in between.

That said, this seems like more error than I'd expect, unless you have a really long build.

So I tested more yesterday and got really weird results that I cannot explain.

To clarify a few things first! I used a cube hitting all four vertical sides. I did a fast exercise in DCC where I probe specific points in A0B0 and then rotate to A90B90 in 3 similar setups / TIGOS. the 2 of them had max 5 microns of deviation while the 3rd machine gave me 20 microns :O

Later on I tested Manually on the same cube same exercise but with 3 different probes / tips / modules. Only in X axis and when the A axis was at 90 degs I was getting 20-30 microns deviation. The weird thing is that I was probing at -X the left side parallel to the vector of the probe and I was getting 30 microns and when I was at -Y probing the "Southern" side (again parallel to the probes vector) I was ok. From the "Southern" position, while hitting the sides (left and right, perpandicular motion to the vector of the probe) though I was having he 20-30 microns deviation.

TLDR no matter the vector of the motion, when I was at 90 degs on Axis A, on all probes / modules etc, only when I was probing at X Axis I was getting the shift! Also the shift was in one direction, meaning that the width of the cube would be correct (both sides shift to -X) but its position in relation to a A0B0 probing (or all other angles for that matter) is off.

We will have a hexagon tech investigate the issue. They recently changed the whole head/controller of the machine so my best guess is that it was not a job well done either it being faulty hardware or the installation failed.

You say they changed the head ? Did it change the weight of the head ?
On a cantilever structure, changing the weight on the arm should impose a new compensation matrix, IMO.

The techs came for an annual calibration and after that the head was not registering some of the hits. We changed head using one from an unused CMM (same model + everything) and the head had issues rotating. Apparently we ended up changing the head and controller (when I say we I mean the hexagon techs). Not sure how they managed compensation matrixes etc but my hint is that something went wrong there. Most probably the hardware but who know

The whole thread was opened actually because a colleague of mine suggested that when the probe is hitting on the probe's vector is not the intended way of probing and that he had problems with it in the past. I have also read in the past that this is indeed something to note and is worth avoiding doing but for me 20 microns are hinting another issue with the machine. After testing I see that the issue is deeper but it is a good learning opportunity to hear from you guys if you had a similar experience with parallel or perpendicular probing.