Impact of Humidity on Metrology Measurements

[[Lu...]]

Hello, everyone,

Does anyone have experience with humidity in the metrology area?

We are using a Zeiss Prismo Ultra for measurements.

We performed the usual calibration process on Monday, which was successfully completed without any deviation in the sigma values. The first calibration was done at 7 AM (20°C, 70% humidity). The first measurement of the part was done at 1 PM (20°C, 60% humidity). The results showed total axial and radial runout out of specification. The second calibration was done at 2 PM (20°C, 60% humidity), and the second re-measurement of the part was done immediately after re-calibration. The results showed total axial and radial runout within specification. The part wasn't re-clamped and stayed in the same position without any movement.

Based on these results, my main suspicion is that the fluctuation in humidity could negatively affect the results. The humidity changed from 70% to 60% between these two measurements.

Zeiss Prismo Ultra requirements:

Requirements for operational readiness:
Relative humidity: 40% to 70% (without condensation)
Ambient temperature: 15°C - 35°C

Ambient requirements:
Measuring reference temperature: 20°C - 22°C
Per day: 1 K/d
Per hour: 0.5 K/h
Spatial: 0.5 K/m
Floor vibrations

I have found the following general information on the website: 
It's important to ensure that humidity and temperature levels are stable, not just within the recommended range, but also with minimal fluctuations. The normal operating humidity is 40-70%, with an allowable fluctuation of +/- 1%-5%.

Questions:
Can humidity fluctuation negatively affect the measurement results? Based on my research, I think it's possible.
Are there any recommendations from the Zeiss company?

Thank you for any help with this matter.

Best regards,
Lukas

 

Edited Tuesday at 05:37 AM

[[Ma...]]

From Gemini:
 

Please sign in to view this quote.

I am not sure how much is 10% difference, but temperature is not same in area, rather it fluctuate - same for humidity.
Next is in what condition was part at beginning ( temperature ) and after and how tight is your tolerance and form of that feature.

I would also monitor how much are probe data changed between calibrations - this will eliminate or prove the cause.

[[No...]]

My first question would be: What material is your part made of? Steel? Plastic?

Plastic parts for example are affected a lot by changing humidity. Many plastic materials can absorb somewhere around 1-2% humidity without much effort.

I'm not aware of any functional parts of a CMM that are sensitive to humidity, but of course fluctuating temperature distribution caused by changing humidity could play a certain role. It depends on the tolerances you have to deal with.

[[Lu...]]

The temperature changes met the specified environmental requirements in both measurement cases. Unfortunately, I do not have a temperature record from the room as it is equipped only with a thermometer without a data logger.

The temperature of the measured part is not recorded, but I believe such large differences in readings would not be caused by it. The part was tempered in the given room for 24 hours. The part is made of metal.

The measurement is performed by two sensor in position 3 and in position 5 on the one measurement assembly.

	First measurement	Second measurement
Total radial runout 0.004/C-D	0.0074	0.0026
Total axial runout 0.01/ C-D	0.0161	0.0020
Sensor calibration (sigma)	Dynamic tensor Sensor stiffness about 47 Measuring force 100mN Sensor in position 3-0.0002 Sensor in position 5- 0.0001	Dynamic tensor Sensor stiffness about 47 Measuring force 100mN Sensor in position 3 -0.0001 Sensor in position 5 – 0.0001

I am trying to understand where such a large change in the readings comes from. If it is not humidity, then according to the calibrations performed, it should not be the sensors either, as the sigmas were fine in both calibration cases. The temperature change of the part would not affect the results to such an extent (although it is not recorded). In both cases, the ambient temperature was between 20.0 – 20.5 degrees.

What should I focus on?

Thank you for any advice.

 

[[Ke...]]

Although the sigma values look good:

Did the recorded probe diameters & positions change at all?

Did you clean & inspect the probes and try different/new probes?

[[Ow...]]

Based off what you have described, I don't believe humidity or temperature are affecting the results. 

My guess is the problem is related to how the part is fixtured or the measurement strategy is not repeatable. 

We don't know the size or shape of the part or how it is fixtured on the CMM.

If you could attach a picture of the part set-up on the CMM, it would help.

 

 

[[Lu...]]

Please sign in to view this quote.

I didn't record anything except the sigma values. I have only tried one measurement assembly with probes in positions 3 and 5, nothing more.

If we calibrate all sensors, what kind of information is good to include in the protocol? We only record the sigma values of each sensor.

I have measured five parts, and all parts have the same deviation of about 0.15mm. So, I decided to remeasure the last part and then leave this part in the same position without any movement. Here are the steps then I followed:

1.I fixed the calibration holder with the ball (20mm) and measured it with the reference probe.
2.Calibrated the probe in position 3.
3.Re-fixed the calibration holder with the ball (20mm) and re-measured it with the reference probe.
4.Calibrated the probe in position 5.
5.Remeasured the part without any movement.
6.The part is within tolerances.

Edited Wednesday at 11:55 AM

[[Lu...]]

Please sign in to view this quote.

Thank you for your point of view.

I understand that many factors can affect the overall measurement. Is it possible that this uncertainty is coming from the manual alignment before the CNC starts? Or could it be an issue with the calibration process?

I have only re-calibrated the sensor in positions 3 and 5. The clamping and fixturing of the part were the same without any movement.

I have attached the pictures.

[[Ma...]]

Every probe which measures cylinder C-D, Plane A and cylinder 13,002 should be calibrated - others are not important for this calculation.

But if you use manual start alignment and then CNC base alignment, then it's fine.

Numbers are in INCH or MM?

Let's say - if your probe system is loose, then reporting or comparing it's XYZ values of a tip can tell you if your probes are tight or loose. Perhaps while scanning it's a little bit rotated ( main middle screw ).
What head do you have?

[[Lu...]]

Please sign in to view this quote.

I apologize, numbers are in mm.

The head is: VAST XT gold

 

[[Ow...]]

Please sign in to view this quote.

Knowing which unit you're using (MM) helps and I'll admit I was first thinking "inches".

Being such a small part does open the question of whether it moved or not during the first measurement.

Assuming the part did not move, I might question if the stepped cylinders recalled points were not fully or accurately captured on the first or 2nd measurement, especially if you are using lines in the strategy or less than 3 circle paths in each cylinder strategy. It's rare but I have seen where Calypso did not recall and evaluate the points equally.

Did you evaluate the form of the stepped cylinder between the two to see if they were different?

Have you measured the part un-moved more than two or three times to see if the inspection is repeatable?

If it is not repeatable, you might consider recalling the cylinder points into a 3D line and see if repeatability improves.

A lot of things to consider and I don't have much time, but some input is better than no input, right?

 

 

 

[[Ke...]]

Please sign in to view this quote.

I don't record any values; I just mean the values that are recorded by the CMM after a qualification. BUT: I may take a screenshot and compare XYZ values before & after calibration. a change in values may be a loose stylus or extension, etc.... but I also look at the probe diameter; a larger than expected diameter could mean a loose stylus, even with a passing sigma value.

Depending on the complexity, or where very tight tolerances are involved, of the program/evaluation: it is also good practice to check the repeatability of the program. It's nothing that is required where I work, but I will often 1) measure the part 5x without refixturing and 2) measure the part 5x with refixturing. After that, if I see any potential issues, I may: doublecheck filter/outlier settings, loop the base alignment or secondary alignment as needed, and/or modify data collection