What's new
What's new

One vs two touch probe cycles

GiroDyno

Cast Iron
Joined
Apr 19, 2021
Location
PNW
We have a new Brother CNC and have been trying to automate more of the process for the machine to take advantage of its capabilities. Probe everything, adding batch number macros, templatizing (is that a word?) our CAM, etc... Its working great and now I'm noticing some opportunities to improve processes elsewhere.

We'd like to add some probing to other machines but have been reluctant to do so because of the added cycle time. We have two Haas VFSS machines using the Renishaw probes that ideally we could add a bore verification to. The stock Haas cycle is about 30 seconds, while the Brother does what looks like the same check in 10. Im sure some of that decrease is due to the single touch on the Brother vs. double on Haas.

The Haas manual says there are one-touch cycles available but it requires you to replace the two touch cycles. Is there a work around to allow us to pick and choose which type/what accuracy we want to use depending on the application?
 
There's a "fast break detect" macro available, which rapids the tool close to the setter to save time. The second touch only adds a couple seconds.

I'm talking about part probing BTW.

I saw there is a macro to optimize probe speeds but it sounds like it modifies the settings for every probe cycle. I'm fishing for an option to have a wide open tolerance/fast motion/single touch probe setting and another tight tolerance/double touch.

I know its only a few seconds, but over hundreds of parts it really adds up. We probe/set 6 different WCS for two different parts every cycle in the brother in less time than the Haas checks to see if a part is sitting flat on a mag chuck. Maybe were just spoiled now...
 
It is definitely doable. It would be writing different "9000" probe macro programs and editing some of the existing probe macro programs to be able to handle the different variables.

You would also need to add another letter variable in your line of probe code to be able to choose which probing you wanted to do.
 
This thread is a good reminder on the ease of configuration on a Heidenhain. They have a fast probe mode that the user can turn on in the program. And their cycles can set approach distance on top of a machine parameter that is also simple to set. And they only need to hit once.
 
Really down to pass/rejects at the end of the week more than cycle time for us. Or the difference between ending up with a scrap expensive multi-features part, or not.

This is where CAM shines. All of our features probing post-machining are programmed via CAM thus validated in-machining on the fly (tolerances, auto cutter offsetting, etc etc). This ensure e.g. a part with 100+ features gets 100 features to spec despite fixtures potentially moving, thermal, cutter wear... Shall I finish?

What I'm trying to say is; whatever extra cycle time it takes, as long as the part comes out as per requirements and greater time & cost isn't wasted when the week closes.
 
2-touch method need not necessarily be slower than 1-touch method.
In the 1-touch method, the feedrate cannot be made too fast, as this will affect accuracy.
However, in the 2-touch method, the first touch can have a high feedrate.
 
Comparing the two, the Brother seems just as accurate and is considerably faster. Some of that is obviously the fast machine, but some of that is removing half the cycle.
Stands to reason I could duplicate it on the Haas and see some improvements even if the machine itself is slower.

I'll start by increasing the first touch feedrate, see if that gains a few seconds without having to remove existing files.
 
Comparing the two, the Brother seems just as accurate and is considerably faster. Some of that is obviously the fast machine, but some of that is removing half the cycle.
Stands to reason I could duplicate it on the Haas and see some improvements even if the machine itself is slower.

I'll start by increasing the first touch feedrate, see if that gains a few seconds without having to remove existing files.

1. One touch probing is normally used on controls having High Speed skip input, which has shorter cycle time and as such the span between the actual trigger point and computer response is narrowed. I suggest to run trial of simple touching the surface in for example X:
G91
#100=#5021
#101=0
#102=20 (NUMBER OF APPROACHES)
#103=40 (FEEDRATE)
G103P1
#3004=2
N10
G31X-1.F#103
#[110+#101]=#5061
#101=#101+1
G0X[#100-#5021]
IF[#101LT#102]GOTO10
#3004=0
M0
M30
This will perform 20 touches for any of feedrates set in #103, the SKIP coordinate of each of them will be stored in variables 110=129. Now they can be analysed for scatter.

2.Increasing the first (approach) touch feedrate in 2 touch mode can impose enlarging of the back off amount to assure that the probe is reseating properly. If this distance is too short, it is possible that the final touch is done while the machine is still accelerating, not very healthy condition. In any case the calibration should be performed after any change of these parameters.

Stefan
Cogito Ergo Sum
 
I think you could skip the first touch if you're verifying a machined surface, so you already know where it will be within a couple thou. There's a risk though, that you'll crash your probe if something went wrong in your process (broken tool that hasn't been break checked) and that surface didn't get machined.
 
Double hit probing, at production speed, in a Brother Speedio:

Login • Instagram

1- Turn the Probe On just after the G100 line, so it has about 1 second for the Renishaw system to confirm communications are working while the tool is approaching the probe point.

2- Call your probe with the argument variables and the M3 command that doesn't switch on or off.

3- Turn the probe off just before the retract motion.

Using the Yamazen/Blum macros, setting the probe touch speed to below 60mm/min (macro parameter #127 in P8700) automatically turns them into double touch. A few experiments we ran found the double-touch to be slightly more robust with longer periods between re-calibration and slightly more reliable measurements.
 
I'll tell you my experience in writing custom probe cycles.
One of my large mills is a dual-column mill with 330" X travel.
Being a Japanese machine, it was exquisitely integrated with the Renishaw system 25 years ago and never updated since. (Machine is 8 yrs old) It uses the 2 touch method. First touch at 40ipm, second at 2ipm. Modern controls with the high speed skip signal input can do much better.

Trouble is, with cutting the large pieces I have, 50+ points is pretty typical. Average time on the 2 touch method was 11 seconds. Handling this much data and processing it using the conventional probing language really sucked, and the language was very clunky, so my first step was creating a much simpler language that also ensured all relevant comps were on at probing and integrated the probing into the 3d environment. (This is a portal mill) The language has many convenience features, including a custom data registry system and simple math/comparison cycles.

I then wrote a custom probing sub-routine that performs single touch probing at 120ipm. It is activated by the simple command, [#FASTPRB]=1.
After issuing that command, any probe cycle using my language will probe at 120 ipm until the variable is set to zero or the probe is put away. (Tool change macro resets it, and shuts the probe off too) My average time per point is now 4 seconds.

While setting this up, I discovered a few things.
1. With this much mass, I needed to allow enough pre-touch stroke for the machine to accelerate. Also need enough remaining stroke at touch to ensure no deceleration. A bit of logic buried in the routine checks that and triggers alarms if not met. Probing accuracy is ensured by having the same feed rate and triggering time as when calibrated.
2. Changing probe speed changes trigger error. This one is obvious, or maybe not. There is a delay from the probe trigger to the skip signal. As long as the feedrate of probing is the same and the delay is the same your accuracy is great. Even at 120 ipm I am within 0.0002" of the 2 touch method in all directions, as checked by averaging 10 touches on each direction with each method using a ground master block and mechanical verification.
3. You'll need a custom calibration cycle with extra common variables storing the trigger error values for this. Mine was simple as I incorporated it into the standard calibration cycle and based the trigger error on The standard slow method. The variables are set by the cal cycle and used in the probe result value. The trigger error difference between 2 ipm and 120 ipm is about .150".
4. Mass matters. When probing X with a loaded table I notice an increased deceleration time. My cycle accounts for this. If it were more consistent I could be probing at 160 ipm but that's getting heavy on probe deflection.

In all, this has been an excellent improvement to the probing system. My operators love it. We've done several direct comparisons using the same parts. Probe it the slow way and the fast way. The offsets are always within 0.001" on every axis
 
2. Changing probe speed changes trigger error. This one is obvious, or maybe not. There is a delay from the probe trigger to the skip signal. As long as the feedrate of probing is the same and the delay is the same your accuracy is great. Even at 120 ipm I am within 0.0002" of the 2 touch method in all directions, as checked by averaging 10 touches on each direction with each method using a ground master block and mechanical verification.

The problem is, that the “delay” is not “same” or constant. The scan time of the PLC (in case of standard skip input) or CNC CPU (in case of High Speed skip input) is constant, but the appearance of the probe trigger signal during this period of time is completely random. Therefore the amount of travel during time span from the moment of trigger to the finish of the scan is uncertain. The amount of this uncertainity depends directly on federate. The higher the feedrate the bigger uncertainity. Simple equation using the given data (0.0002” uncertainity at 120IPM federate) : 0.0002*60/120=0.0001 shows us the scan speed of 0.0001 sec, rather typical for high speed input.
If this really is the case, I would suggest simpler solution to the presented problem:
1. Use 1 touch as default method.
2. Setting measuring speed to 60 IPM will narrow the scan induced uncertainity to 0.00005”.
3. The probing program (which as I understand is used over and over again ) should be revised in such way, that the probe is positioned in RAPID very close to the touch surface (let’s say 0.25”), and then the measure routine should be executed. If the bore mentioned in the has large dimension, do not use the bore routine. 4 single surface touches and small macro calculation will give the same , but much quicker result. This, in my personal opinion, is more efficient way then dealing with original macro setups.

Stefan
Cogito Ergo Sum
 
The problem is, that the “delay” is not “same” or constant. The scan time of the PLC (in case of standard skip input) or CNC CPU (in case of High Speed skip input) is constant, but the appearance of the probe trigger signal during this period of time is completely random. Therefore the amount of travel during time span from the moment of trigger to the finish of the scan is uncertain. The amount of this uncertainity depends directly on federate. The higher the feedrate the bigger uncertainity. Simple equation using the given data (0.0002” /
You are correct. I was giving my results simply to demonstrate what is possible.

I don't think you quite understand what my problem is though. I'm not probing bores, if I were I would use 4 separate commands and process the math to save time.
I'm cutting large weldments, with lots of geometric and positional error. One family of parts has a milled surface about 220"x140." We take 38 Z touches on that alone to determine error and material allowance, then minimize air cutting time with the 8" high feed face mill.
Also cutting seats for linear rails and gear racks on these. Probe 5 points in X and 5 in Z on each rail before cutting, again to ensure alignment and verify material allowance.
Positioning the probe within .25" of welded pieces is fatalistic. My min approach distance for accel at 120ipm is .300. Most of my strokes are 2" and that allows the forgiveness needed.
I would prefer to use the single touch method for everything, but I'm fighting the old guard in the shop too. When I have more time I'll probably make my slow probe method single touch at 30ipm.

I appreciate your insight PROBE, it's hard to find people that can have an in-depth conversation about this technology.
 
We have a new Brother CNC and have been trying to automate more of the process for the machine to take advantage of its capabilities. Probe everything, adding batch number macros, templatizing (is that a word?) our CAM, etc... Its working great and now I'm noticing some opportunities to improve processes elsewhere.

We'd like to add some probing to other machines but have been reluctant to do so because of the added cycle time. We have two Haas VFSS machines using the Renishaw probes that ideally we could add a bore verification to. The stock Haas cycle is about 30 seconds, while the Brother does what looks like the same check in 10. Im sure some of that decrease is due to the single touch on the Brother vs. double on Haas.

The Haas manual says there are one-touch cycles available but it requires you to replace the two touch cycles. Is there a work around to allow us to pick and choose which type/what accuracy we want to use depending on the application?
does the brother have probing cycles like Haas? like the work probing cycles? WIPS?
 
A two touch is going to be more accurate but slower. If you really want to you could change the speed of the first touch and test a few probe cycles and see how accurate it is by probing a ring gage of known size.
 
The original question was about single touch on a Brother. Assuming Blum hardware then you can get single touch repeating to the single micron at 1,000mm/min single touch speeds

That looks roughly like this:

At 2k speed (on a C00 control) I notice that there can be a 1 micron wobble in probe measurements. This gives us some data on the time accuracy of the high speed skip in the C00 control. In theory the D00 control is faster. Perhaps you can run at higher speeds? Blum claim you can do 0.3 microns at 2k, but my machine at least can't achieve that

So I don't see any reason for an increase in accuracy with a 1 or 2 touch probe, at 1,000mm/min or less. However if you had huge unknown in the part location and were doing huge scans, you could imagine a double touch being faster: do the first scan at 2k+ to close the distance and then a slower second touch.

You setup the Blum using a calibration ring. So you take say a 25 micron ring and then it zings around at full speed to figure out it's error. Goes without saying that you can then measure it's diameter and come up with the same value. Bonus marks would then be to do an outside measurement of a gauge block and check that's bang on as well. Ideally use the same bit of ballscrew if you are trying to prove this stuff to the micron.
 
You setup the Blum using a calibration ring. So you take say a 25 micron ring and then it zings around at full speed to figure out it's error. Goes without saying that you can then measure it's diameter and come up with the same value. Bonus marks would then be to do an outside measurement of a gauge block and check that's bang on as well. Ideally use the same bit of ballscrew if you are trying to prove this stuff to the micron.
Too small of a ring. With Blum and Renishaw I always get more accurate calibration with a minimum of a 3.0 inch ring.
 
Assuming Blum hardware then you can get single touch repeating to the single micron at 1,000 mm/min single touch speeds
That is impressive. I did not know this. Thank you for information.
This must be using high-speed skip signal.
 








 
Back
Top