Machsupport Forum
Mach Discussion => General Mach Discussion => Topic started by: sivtek on November 20, 2011, 05:44:42 AM
-
Is there a drawing and description of how to make my own Probe,- for probing x,y&z axises?
(Have the drawings made by Shaun Wainford, but there is no description).
Thanks
Johannes Line
-
Did you download the PDf file? The file provides general info on the probe and a set of drawings.
Study the drawings and the 3D pictures some to understand how all it all fits together.
You will need to use a lathe and a mill to make the pieces.
RICH
-
"Did you download the PDf file? "
Yes I have, and I have a lathe and a mill.
I have studied the drawings but is unable to quit understand it. I think that by touching you are making metallick contact in the y and x move, but what in the z direction? Any electrical connections?
I have a mill and have installed a simple probe for x and y works the way the movement stop when the probe hits the metall (mill).
Johannes
-
Yes you are making contact just like the your simple probe but additionaly the tip has some give to take care of the Z as it is spring loaded.
There are contacts which are made depending on the direction of tip movement. Look closely at the pictures.
RICH
-
Johannes,
I don't know if this helps but....
By the method of assembly, all the balls are in electrical contact with the PCB and each of the silver steel pins electrically connects one pair of balls. The, two wire, probe connection is taken from the 'gap' in the PCB track, which under normal circumstances (with balls and pins fitted) is now a short circuit. Any probe contact with an object (no matter from which direction) will open the circuit as at least one pin will fail to bridge the gap between one pair of balls.
It is the software which determines the direction of probing and the probe just indicates when contact is made by breaking an otherwise continuous circuit.
Tweakie.
-
RICH & Tweakie, thanks both of you.
Now I think I understand the working principle. The two wires go through the hole on the side of the body and connects to the small 1,20mm hole at the Fibre Glass Board and short the circuit as you explained. Mowing the ProbeTip the circuit is broken, this is the opposite of the simple probe. I mean that when the simple probe touch ground Mach3 is activated (axis stop moving). Can this bee sett in the softwhere- "high or low"?
Another thing I was wondering about,- how accurate is this Probe? Surly there have to bee some play/movement of the ProbeTip before the circut is broken and a signal given. Can this be compensated for by the software?
Johannes
-
Can this bee sett in the softwhere- "high or low"?
I suppose you could use a 'pull up resistor' for the circuit to be 'active high' and a 'pull down' resistor for the circuit to be 'active low' then configure Mach accordingly.
Another thing I was wondering about,- how accurate is this Probe? Surly there have to bee some play/movement of the ProbeTip before the circut is broken and a signal given. Can this be compensated for by the software?
If it is constructed properly and the spring yields before anything bends then it would be extremely difficult to measure any play/movement in the system. Consider that the tip is essentially resting in the 'V' between 6 points of contact, it will either be making contact or it will not, there is no delay or contact bounce as you may get with, for example, a conventional switch.
When probing there is 'overun' between the contact point and the axis coming to rest but the stored position is the actual point of contact not the final axis position so compensation is automatic really.
Hope this helps.
Tweakie.
-
Thanks, think I will manage now. you have been to great help. Thanks!
Johannes
-
Just to add a little to Tweakie's excellent explanation. These type of probes are normally set as active high in Mach - i.e. whilst the contact is made the input is held low (by being grounded) and when a trip occurs, the circuit is broken thus allowing the internal pullups to take the input high (active).
Also there actually is a bounce issue to be aware of but it's not a big deal. As Tweakie says, the instant the circuit is broken, Mach (if using G31) stores the point of contact and then decelerates to a stop. HOWEVER - if the probe contacts are re-made and then re-broken during that decceleration phase, Mach will RE-WRITE that point as the point of contact. This is more likely if the trip is caused by "glancing" something to the side of the direction of travel for example. This only really becomes an issue if your machine accel/deccel is "poor" however. Whilst on the subject of "poor" decceleration, care must be taken to ensure that the probe overtravel capability is large enough to accomodate the slow down phase, otherwise.... CRUNCH. A useful addition to a home made probe is to add in an overtravel contact to trigger a limit - your probe gets to survive where others are assigned to the bin.
Ian
-
HOWEVER - if the probe contacts are re-made and then re-broken during that decceleration phase, Mach will RE-WRITE that point as the point of contact.
Thanks Ian - this is something I had never considered (or even thought about happening).
Tweakie.
-
HUM I have never seen it double write the trip point here. Once the initial trip takes place the G31 is no longer actively waiting for a trip it is already into the record and stop sequence of the code. At least that is how Art explained it a long time ago.
There IS switch travel to account for in the calibration of the simple probes. It may be small with some probes BUT it is there and measurable so you really need to account for it. ALso with a 3 legged circiuit the switch travel may vary around the radial movement of the probe.
Normally in calibaration you add the switch travel into the tip radius as a total tip radius. Mach3 DOES have a tip diam comp built in and it ONLY comps the Var values of the G31 function.
A normal calibration function would be the ball and post method. It sets the Tip comp AND the axis offsets in the same test.
(;-) TP
-
HOWEVER - if the probe contacts are re-made and then re-broken during that decceleration phase, Mach will RE-WRITE that point as the point of contact.
HUM I have never seen it double write the trip point here. Once the initial trip takes place the G31 is no longer actively waiting for a trip it is already into the record and stop sequence of the code. At least that is how Art explained it a long time ago.
Perhaps this point needs to be investigated, at least as it applies to the latest lockdown version - does it re-write or not ??
Tweakie.
-
HUM I have never seen it double write the trip point here.
Probably because you have decent acceleration and therefore don't notice it. Wind down your accel to a minimum and try it. As it's decelerating after the first trip, trip it again (several times if you like) . The appropriate var will be set to the position of the LAST trip. I've tested this a gazillion times when doing my probing routines with 3.42.20 - can't speak for other versions. Your milage may vary (arf ;D).
Ian
-
Ian,
With version R3.043.022 and using G90 G31 any deceleration settings in motor tuning are totally ignored when the Z Axis comes to a halt.
I am still testing but so far I can find no evidence that the value retrieved from Get Var (2002) is over-written by any subsequent contact, it is just the first trip point that is recorded.
Tweakie.
-
any deceleration settings in motor tuning are totally ignored when the Z Axis comes to a halt.
Not sure I understand what you mean Tweakie.
Anyway it seems my memory is not what it used to be - time to give it up and grow veg I think. I've mis-described the issue. I'll try again. If the trip (or as many trips as you like as it's decelerating after the first trip) is/are momentary i.e. bounced as per my original description of a "glancing blow", then the value stored is where the axis stops after decelerating and NOT where the trip occurred. HOWEVER if the last trip of several remains made (i.e. does not bounce) then the FIRST trip is what is recorded.
Ian
-
Ian,
Probably because you have decent acceleration and therefore don't notice it. Wind down your accel to a minimum and try it
With version R3.043.022 and using G90 G31 the Z axis comes to a halt after first contact without any regard to acceleration settings, winding down the Acceleration has no effect.
If the trip (or as many trips as you like as it's decelerating after the first trip) is/are momentary i.e. bounced as per my original description of a "glancing blow", then the value stored is where the axis stops after decelerating and NOT where the trip occurred.
This is not what I am finding from the tests I have done so far. As mentioned the value retrieved from Get Var (2002) is the first contact point not where the axis finished up.
Could you perhaps describe a test method that you have used to support your re-writing of the value please.
Tweakie.
-
Correction to above ....
By increasing my probing feedrate to a much higher rate than I would normally use the acceleration settings do have an effect on the axis deceleration after a trip has occurred but I can still not see a second contact during that period making any changes to the Get Var (2002) value.
Tweakie.
-
As you've found, you have to set your feedrate high and your accel low otherwise you can't see what's happening. (It's still happening - you just can't see it). Here's the settings I'm using to show this issue. I use the X axis but it doesn't really matter because it's best done offline if your machine can't actually take the velocity value below.
Max velocity 15000mm/min, Accel 62.5mm/s/s (the lowest I can get Mach to take at that vel). Then run this:
F15000
G31 X5000
M666 (just displays var 2000 i.e. message getVar(2000))
M30
Now if I let it accel and get to approx X=1000 and then trigger AND HOLD the probe, Mach will start the decel ramp and eventually stop around X=1500 or so. The value in var 2000 is correct i.e. approx 1000 i.e. where it was first triggered.
Now do the same again but at approx X=1000 trip AND RELEASE the probe. Mach will again start it's decel ramp but when it stops the value in var 2000 is approx 1500 i.e. the same as the X DRO i.e. where it's come to a stop.
You can do this with no machine "attached" and you can either use a real probe, a switch or even with the keyboard using probe emulation.
Ian
-
Something must have changed in the G31. In early testing and argueing with Art about which was the best probe cycle type 1 or 2 it did not do that.
I tested every combination we could think of to test the accuraccy of the cycles. Art ending upbeing right about the accuracy but wrong about the "best" cycle. But it never got added(;-)
Ian do you have a debounce setting that may be blocking the first trip???? I do know the testing I did was before the debounce was added in.
Perhaps the debonce routine that was added played havic on the original code???
Just a thought(;-) TP
(;-) TP
-
Thanks for the info Ian.
My machine is busy for the next couple of days but it will be free for the weekend - I will do some more testing then and report back. :)
Tweakie.
-
Hi Terry - don't think debounce is the issue. Don't want to get too bogged down in this - it's something I've noticed and thought I'd mention. It's NOT really a problem I think if your accel/deccel is decent. Anyway FWIW...
See attached pic.
It's meant to represent a downward probe (i.e. in Z). In both cases the probe trips on the step and then decelerates and comes to a halt at the position shown (obviously exagerated). In the RIGHT hand case all works as you'd expect. The stored point is where it hit the step. However in the LEFT hand case the recorded point is where the probe stops and NOT where it tripped.
Ian
-
HI Ian Good to hear from you(;-) I was about to email you to check on you.
I agree on not getting bogged down on the subject BUT it is something that Brian needs to check out as it should NOT do that.
Maybe Andrew can check it out for Brian to test.
(;-) TP
-
Hi all.
I now have made the probe and installed it to my mill and everything works well.
Attach a picture of the probe and the wiring diagram I have used.
Sivtek
-
Trying to attach the picture
Sivtek
-
Excellent work Sivtek really neat job.
Thanks for posting the picture and wiring diagram.
Tweakie.
-
Very good job sivtek !
Looks very professional.
Did you use a commercial tip ?
(I shrunk your pic to fit better, hope you dont mind)
Russ
-
Thanks for nice words.
No the tip is madi in my lathe of silversteel, but the ball is a 3mm ball from a ballbearing. It is first softened, glued to a holder and a hole 1,7 mm was made. Then it was hardened and glued to the shaft.