Faulty KP3 Probe ?? <Solved, Tighten stylus>

[suntravel]

(4) 03-12-2025 17:07:31 8040 WARNING: PROBE TRIPPED
(4) 03-12-2025 17:07:32 9042 Probe Cleared
(4) 03-12-2025 17:07:32 8040 WARNING: PROBE TRIPPED
(4) 03-12-2025 17:07:32 9042 Probe Cleared
(4) 03-12-2025 17:07:33 8040 WARNING: PROBE TRIPPED
(4) 03-12-2025 17:07:33 9042 Probe Cleared
(4) 03-12-2025 17:07:33 8040 WARNING: PROBE TRIPPED
(4) 03-12-2025 17:07:34 9042 Probe Cleared
(4) 03-12-2025 17:07:35 8040 WARNING: PROBE TRIPPED
(4) 03-12-2025 17:07:35 9042 Probe Cleared
(4) 03-12-2025 17:07:35 8040 WARNING: PROBE TRIPPED
(4) 03-12-2025 17:07:36 9042 Probe Cleared

Sure there is not a loose wire on the probe inputs?

Besides from this the setup is ok IMHO. The probe is a "simple" switch, it should be repeat as precise as the machine could repeat a position used with the same speed.

Uwe

[RJS100]

Seems to me that if there was a loose wire on the probe input, the led would not stay illuminated.

[centroid467]

After doing some poking around, I think the numbers you are getting make sense for the machine and the probe. After the repairs, our test machine showed a maximum of 0.00017" variation between Z hits.

I would expect a KP-3 on an ideal or theoretically perfect machine to repeat within 0.0002". With 0.0011 variation on your test, I would take a look at the machine itself.

With typical dovetail ways you are looking at accuracy of 0.0005" / 12". I know the probing moves are not a foot long but it could have a small contribution.

You have 0.001" of backlash compensation set on the Z axis. That is on the larger side for amount of compensation. It also looks like you have not done the fine adjustment of the overall turns ratios. I suggest taking a look into reducing the backlash, doing a finer adjustment of the overall turns ratios, and then trying the Z repeatability test again.


As far as the probe LED being red when powered on, I would normally think of it as a sign of a sticking or soon to stick probe. However, it was recently repaired and we have procedures in place to avoid the causes of that issue.

My next thought would be high contact resistance on the kinematic seat but all cases of that I have observed do not occur when the KP-3 is at rest. Only after very large numbers of repeated triggers does that resistance tend to increase. The current production KP-3 spec does not exhibit this issue to any notable degree.

[RJS100]

Thanks for your thoughts. I am a bit surprised that you find these number to make sense. The dial test indicator that I used in the video has a resolution of .0005. While it does not make accurate linear measurements, it is designed to measure relative movement. To me it is showing that the machine is repeating movement in the Z-axis with enough accuracy that the .0005 indicator cannot detect any visible variance. The other test using the dial indicator that does make accurate linear measurements also indicates that the backlash compensation is accurately tuned to the point that no backlash is visible over .100 inches. How then is the probe indicating .0011 variance over a few hundred thou of movement. As I am new to this hobby, I am not suggesting you are wrong, but the figures don't make any sense to me at all. What am I misunderstanding?

One interesting note is that I found the exact same variance of .0011 in the x and y axis, and these have slightly less backlash compensation. Lastly, if my backlash compensation figures were off by let's say for example .0003, then I would expect the variance to be off in one direction, but it is not.

With respect to the LED light. I am happy to keep an eye on it for a while. I can't think of anything I could do on my side to make the LED stay illuminated (as shown in the photo), while the probe is not triggered. To me, it looks like an intermittent problem.

Thanks... Richard 

[centroid467]

Hello Richard,

What I was trying to get at is that there are many sources that can contribute to the amount of variance you are seeing. We have small numbers but a lot of factors to consider that can add their small contributions to get what you are seeing.


I do want to comment on the dial indicators not showing any variation. Every measuring device has accuracy and precision limitations. The test reports that came with those indicators probably gives the specifications regarding accuracy and repeatability. They'll probably show that the indicators perform far better than their specifications but the manufacturer's only guarantee is that they are within those specs. The test reports will show the actual measured performance which can then be used to see how the uncertainty specifications could apply.

For instance, the Mitutoyo model 2416S is specified to +/- 0.001" accuracy over the first 2.5 revs. I don't have a repeatability or hysteresis figure for that model. You were measuring over a single rev. I'd apply that as an uncertainty to the readings obtained. So, it returning to zero is read as it returning to zero +/- another graduation.

The 513-402-10 is specified to read within +/- 0.0005", repeat within +/- 0.0002", and have 0.0002" of hysteresis. In this case, I think we can ignore the hysteresis spec since you are travelling much further than that. The repeatability specification tells us that it will show within +/- 0.0002" of the previous reading at the same position. Then again, looking at the video shows that the cosine error is probably also large so any deviation we can see is probably magnified.

That was a lot of words to say that the measuring tools can sometimes mask errors.

[cncsnw]

If either dial indicator had a lack of repeatability large enough to mask a 0.0011" error in machine movement, what are the chances that the indicator error would exactly cancel out the machine error on each move?

[suntravel]

This is an example how I am testing...



Better dial indicator, more visible deviation

To test for the probe I would use the same travel and speed like the probing cycle.

Even with full closed loop now, I have to finetune the setting of the gibs ect. very often to get the best results with the cheap Chinese mill.

Uwe

[RJS100]

Thanks for your time & explanation. I agree with your thoughts regarding hysteresis and instrument specifications; however, I don't think they support a .0011" error. I agree with Marc (from CNCSNW) "If either dial indicator had a lack of repeatability large enough to mask a 0.0011" error in machine movement, what are the chances that the indicator error would exactly cancel out the machine error on each move". Keep in mind that that both dial test indicator and test indictor showed zero visible variation.

As Uwe pointed out, these tests were performed at the same speed as the probing speed of 5 IPM'

Richard

[suntravel]

Hi Richard,

I have no clue what is making this deviation. Your videos with dial indicator are looking ok.

I would get a cheap probe like this to compare

https://de.aliexpress.com/item/10050034 ... pt=glo2deu

I have one also, but the KP3 is outperformig it on my machine in any way.

Uwe

[RJS100]

Well Uwe, if you have no idea... you can be sure I have absolutely no clue

One thing I have not tried is to make sure the stylus on the probe is not loose. I will check this next time.

I made a friend on this forum while helping him convert the same mill that I converted. He ended up using the same JMC servo for the spindle you helped me with. I was bored hanging around the house after I had brain surgery, so I machine the mounting brackets for him while I was recoverying. We have become close friends, and I talked him into buying the KP3 probe. When he visits me, I will ask that he brings the KP3 and we can compare the results.

Thanks as always for your thoughts.

Richard