Machsupport Forum
Mach Discussion => Mach Screens => CVI MachStdMill (MSM) => Topic started by: Fastest1 on August 14, 2010, 10:39:09 AM
-
Dave, just to help me understand. The "edge width" is used to calcualte the amount of angle you are varying from being perfectly square with your axis? What does the "Calibration Diameter" do? I am getting better at operating the probe but these variables dont seem to make a change in the operations that I can see. I am having much better luck at center finding a hole than I was. There might be a bit if noise on my machine. I had my debounce set to 500 (though I dont remember why) prior to your suggestion of setting to 100 for the set up. I do see it say face not found occasionally but I cant tell if noise or incorrect parameters triggered it. The way I understand the debounce, doesnt it require the signal be of a minimum time period to trigger so errant spikes wont trigger it? If the setting of 100 is too low during this testing is there a safe maximum debounce setting? If my debounce is increased to accomodate the noise should I slow my feedrate? Seems like the longer the trigger signal would also mean the axis had traveled farther prior to reacting. Is that close to correct?
-
I just read the whole guide and then went thru and printed pages 111-122 in the manual. There is more to it than I thought. The manual did answer my earlier question of probing a hole of a specific diameter to calculate the tool tip size. Are you or have you ever considered teaching professionally? You are very thorough in your explanations and I cant see you rolling your eyes at me! It is wonderful to have people like you share their knowledge and work so openly. Maybe all teachers start out like this and then some of us students wear them out? Great work.
-
Also I see you refer to "down & over" probing. It does even show the probe doing just that in the screen but the green oval select point wasnt obvious to me. In fact I cant wait to go out in the shop and try it. Eyes arent what they used to be. Dont laugh too hard when I couldnt figure out why it wasnt following that toolpath. By the time I finish asking you all of the questions I will need there wont be anything else to address for the rest of the users. LOL
-
Some MSM manual references:
Edge width: 7.1.3.3.5 & 7.1.5.11 & 7.1.6
Calibration diameter: 7.1.3.1.2 & 7.1.4.10
I just added the following info to the How to post:
[Info added 8-14-2010]
The basic mach probing sequence:
1) G31 starts, mach starts moving machine
2) Probe input triggers; mach records the position of the probe when it was triggered
3) Mach decelerates the probe and movement stops. Note that the stopping position is not the trigger position. The difference is “overrun”. A bit of Overrun is expected and good – as it keeps the probe in the triggered state.
4) Mach returns from internal G31 execution to program that issued the G31 (for the MSM probing operations, this will be a return to the code in the MSM probing library)
5) Script looks at probe signal –
a. If probe is not active, we ran out of distance w/o triggering the probe
b. If probe is active, we found something and were triggered.
6) Probing Lib gets the trigger position and does what it needs (For example sets axis zero to the face we found).
7) Library backs off by the overrun amount (this lets the probe become un-triggered)
08) Library back off the user defined clearance amount.
Noise on the probe input line will make this sequence not work correctly –
Example: A short noise pulse at step 2 can stop the G31 motion; but at step 5 if probe will not be in the triggered state when checked. MSM thinks nothing was found and so puts the probe back to the starting position.
Adding some Debounce may help:
Mach's debounce is really a software fix for a hardware problem.
The ideal solution is to remove the noise from the system and use a debounce of 0.
Many people either can't or don't want to expend the effort required to accomplish that.
The pragmatic approach is to lower the debug number until the problem appears, then raise it back up a little.
That leaves the issue of "How do I tell if my required debounce setting is too high"?
"Debouce" is the amount of time a signal must be continually "on" before mach will consider it as just having turned on.
Each number in the debounce DRO represents about 40usec of time (25kHz Mach: 1/25KHz = 40uSec)
A debounce of 40uSec* 4000 = 160mSec. That is a long time for a computer, it's even pretty noticeable to a human.
At 4000, the probe signal has to be on for a minimum of 160ms before mach will see it and stop the probe motion.
How fast are you probing? Let's just pick a number for illustration:
Say you are probing at 50ipm; during the debounce interval the probe will have moved (50ipm * 160msec ) / 60 sec/min = 0.133 inches....
So the probe will travel 0.133 in past where the edge is before mach will think the probe is triggered - and then it has to decelerate and stop.....
Consider - how much tip travel does the probe have? 0.133 inches over travel could be hard on the probe - depends on the probe design.
Same scenario with 200 debounce: 50ipm * (200*40usec) / 60 sec/min = 0.0067 inch over travel
Personally, I prefer to see a debounce of no more than a few hundred. but the real answer is whatever combination works that makes the machine run reliably.
Play with the numbers and decide what you are comfortable with.
[end added info 8-14-2010]
Dave
-
Excellent. That was the way I percieved the debounce to work. I didnt know the calculations in time versus distance. I was thinking of maybe going to a max of maybe 500 and of course I probe at much slower speeds as I am not in any production scenario, maybe 10 and down to 1 for the slow speed. Tune as I work thru the steps. All of my cabling is sheilded and the sheilds are connected at the controller only but I always am trying to clean up the operation and performance in preparations for understanding a future bigger machine 1 day.
-
be a bit wary of slow probe freed rates -
at low FR the probe is moving so slow that there is little speed or momentum - that can cause worse switch bounce problems than if the probe is moving faster.
If I keep my probe above 5 ipm it's happy - below 5, I tend to get enough switch bounce that I get false trigger pulses.
I've had other people tell me they experiences similar problems - so slower is not always better in this game.
Dave
-
You know I read that in your guide and was suspicious of that concept. During the testing with a 10ipm fr and a 0 for the slow speed. I rarely got an error. So maybe its right? It doesnt seem right. I am still tinkering actively in pursuit of me understanding it or it working correctly. Probably a little of both. I am getting closer and it seems to be more consistent though occasionally it triggers or stops with no apparent reason. I say that because I dont see it travel the xy distance before stopping. Also during some tinkering without connection to my controller, just running the MSM in general. I could watch the dro's keep going in search of a face and go well past the 1.5" (" not 'LOL) and sometimes not even fault/error. I will watch closer, try to use faster probe rates and see. I have not experienced any overrun that caused any problems or any touches that were disobeyed. Mostly it just stops short of the target which seems to be within the specified parameters. I did try a few different debounce settings with no real conclusion yet. 0 was not possible it seemed but 50 was getting there.
-
FYI - If you initiate a probing operation (any G31 command, not just a MSM probing operation) on a PC that is running mach in Sim mode (I.e. there is no motion driver layer present), you will see the DROs run until the end of time in the direction you issued the G31 for.
This is because mach can't do G31 with out a motion layer present...
Mach starts the G31 motion and then waits to be notified that either the motion reached the target coordinate or that the probe triggered. Since the motion layer is not present, mach just keeps waiting for an event that never happens.
Dave
-
FYI -
I separated this thread from the "how to get started with MSM probing" thread so that people looking at the "how to" thread would not need to read this particular discussion thinking it was part of the "how to" info.
Dave
-
Understood and again thanks.
-
Dave, upon a little more testing today. I can get the debounce down to less than 250 and be relatively consistent with the various operations. I am understanding most of it better. I do have a question about noise. Presently I am not running any caps on the inputs from my probe. i have read that using a .01 -.1uf cap can help with the noise issue. Is this the same as using the debounce or does it quell the noise some other way? Would this be an advisable way to help the situation in your opinion? If so, which value cap would you use of those 2 mentioned?
-
Well I went ahead and placed a .1uf cap between the probe pin and ground on the controller and reduced the debounce to 0 (was working ok at 50). So far so good. I still get a stop occasionally but it is operating much better. But I dont know if it was the cap or just a better understanding. I think the cap must be doing something as it was very hard to complete a sequence with the debounce on 0 before. I can repeat measurements after retracting the probe and placing it back and remeasuring it. The measurements come out the same. If they do vary it isnt til the 3rd digit and mostly only the 4th, way accurate for anything I am doing.
-
Welcome to the wonderful world of herding electrons - I'll skip the EE discussion and keep things empirical.
The cap is acting in conjunction with a resister in the BoB to create a simple filter circuit.
The higher the cap value, the larger the amount of filtering added.
Adding the cap is an electronic way of doing what the mach software de-bounce setting is doing.
So the effects from adding the cap and the debouce setting will be additive - and this matches what you are seeing - with the cap you need less software filtering.
If you are curious to learn more, you can google "RC time constant".
Here is reasonable tutorial I spotted from the google results:
http://www.electronics-tutorials.ws/rc/rc_1.html (http://www.electronics-tutorials.ws/rc/rc_1.html)
Dave