[wjfiles]

I have built a probe for use on my router.
I am using Mach3 and the wizard for probing.
Is there a simple circuit to amplify the signals therefore minimizing missing steps.
wjf

[pastera]

Can you give us more info on the probe?

Aaron

[wjfiles]

My probe is much the same as the one Mr Bean posted.
I am using Mach3 and the wizard.
I am of the opinion that unless the original is complex and can not readily be measured, it is much easier to use CAD CAM software. I had not realized that I do not have the skill or software to create tool paths from a point cloud.
Most of my work can be handled quickly with the AutoCad,Mach wizards and the conversations.
wjf
I can see the advantage for complex models or carving.

[CJL5585]

I have built a probe for use on my router.
I am using Mach3 and the wizard for probing.
Is there a simple circuit to amplify the signals therefore minimizing missing steps.
wjf

I am having a very difficult time attempting to understand what you mean by "amplify the signals therefore minimizing missing steps."

Imagine the X and Y axis coordinates as being the part being probed. The Z axis is the probe. The machine will move to position X1, Y1. The Z axis then lowers until the contacts open in the probe (touched the surface of the part). In normal configuration (This is a very simplified explanation for positive probe logic) the probe is supplied with 5 volts DC through a resistor to one side of the probe contacts. The other side of the contacts are wired to the 5 volt ground side of the supply. So, in normal operation the 5 volt power supply to the probe is shorted with the inline resistor limiting the current in the circuit to a safe value.

When the probe touches the part with enough force to open the contacts, the short circuit no longer exists, and the power supply voltage at the contact instantly goes to 5 volts dc. The software senses this voltage change as a transistion from a logic 0 to a logic 1 TTL level. At this point in time, the software writes to a file the X position, the Y position, and the value of the Z position. There is nothing to amplify, unless you are speaking of having a higher voltage applied to the probe which would give a 0 volt (logic 0) or a 12 volt dc (logic 1) output. This output would probably smoke the logic chips if they are ttl level.

The probe itself does not have ANY interaction with the positioning of the machine or any missed steps. It only gives the Z position of the axis when the contact(s) make or break internal to the probe per the design.

Now if you are referring to contact bounce, where there are glitches in the voltages as the contacts open and close, ......causing multiple readings ......that is a whole different problem and solution.

jlc

[wjfiles]

Yes I am aware of the principal that a contact probe works.
The reason for my query was because I have seen a circuit somewhere that makes sure the positive voltage is always 5 Volts.
What my probe is doing is probably sticking.
It will scan for maybe an hour and the misses steps continuously.
It does not even come all the way down to touch the part..???
If I flick the probe it will start again and run OK for a while.
When I put a meter on the contacts I get varied resistance readings even if the contacts appear to be closed.
I have a new design I have done that only has 3 contact points instead of 6 balls. This should require less precision to ensure a positive contact.
Thanks for your reply.
WJF

[CJL5585]

Sounds as though you do have some problems with the probe. The resistance readings should not vary other than the open/close resistance. If the contacts do not close when raised above the scanned object, the next cycle in the process cannot occur, and one would have the problems you mention.

[pastera]

Use two gates from a 74HC14 between the probe output and the port. This will clean up the signal. You need two because the gates are inverting.

Also, what is your current limiting resistance in the probe? If you keep the current under 10mA (> 500 ohms), your probe could have a closed resistance of up 40 ohms before you would have problems. A good value would be from 1-2k.

If you're good with a soldering iron, use a NC7WZ14 in the SC70 package. It's only 2x2.10 mm for a 6 pin package so it would fit inside your probe.

Aaron

[turmite]

http://www.cnczone.com/forums/showth...ighlight=probe

This link is about probe building and addresses the missed point, sticking etc.

Mike

[wjfiles]

Thanks for the replies, Pastera you came up with what I was looking for.
I am almost certain that my problem is lack of precision.
I mentioned earlier about an Idea of using only 3 contacts.
I have attached a rough sketch of what I had in mind.
I am interested if anyone has an opinion
wjf

[wjfiles]

Sorry, forgot the image.

[pastera]

It looks like you have a simple switch-like setup. Unfortunately, this will not work reliably in a noisy environment. You need to pull the input in both directions, high and low.

You could put a 470 ohm resistor to ground on the kead to the control board. This would pull the input low when the switch opens. A floating input, which is what you currently have when the switch is activated, will exhibit strange behaviour or simply stay high for long periods of time.

Aaron

[gar]

071002-1552 EST USA

wjfiles:

I would suggest 50 millionths gold over nickel plating on your contacts.

Connect the fixed contact via a wire to the common (ground) of your logic circuitry. Maybe a 220 ohm resistor from the moving contact to your +5 V supply. This assumes the moving probe contacts are not at ground when contacting the part. Otherwise interchange the wires, isolate the fixed contacts, and make this your signal output.

If the input impedance of your logic gate is sufficiently high (CMOS), then at the logic gate iput put a 1000 or 10,000 ohm resistor from your non-ground contact to the gate input. A 0.1 mfd capacitor from the gate input to common. This gives you a low pass input filter. The RC time constant at 1000 ohms is 0.1 millisecond, and at 10,000 ohms it is 1 millisec. It won't hurt to have a Schmidt trigger as your input logic.

If you need the opposite phase put an inverter after the first gate.

The signal source resistance before probe contact with the part is 0 ohms and after is 220 ohms.

.

[wjfiles]

Thanks for the responses.
I should have explained that I am using the HobbyCNCPRO board and there are pullup resistors in the circuit. I also use Mach3.
I had thought of using some very small SMT tactile switches but the the life of these is only about 100000 cycles, not a lot when probing.
wjf

[gar]

071002-2101 EST USA

Martin:

The debounce circuit you suggest is good. but as structured requires a SPDT non-shorting switch.

Really it probably is not necessary because the software detecting that the probe touched effectively detects the leading edge and goes off and does whatever is required. It does not come back to do another test for a substantial time. What is important is to minimize the possibility of a trigger before probe contact. So a normally closed switch connecting the input to common (ground) until probe contact provides a very low impedance at the logic input until probe contact.


wjfiles:

I have no idea what a HobbyCNCPRO board is and thus what its input circuit looks like.

At low voltages and low currents there are only a small number of materials in air that provide long term reliable electrical conduction. Silver is not one, but gold is very good in general.

.

[martinw]

071002-2101 EST USA

Martin:

The debounce circuit you suggest is good. but as structured requires a SPDT non-shorting switch.

Really it probably is not necessary because the software detecting that the probe touched effectively detects the leading edge and goes off and does whatever is required. It does not come back to do another test for a substantial time. What is important is to minimize the possibility of a trigger before probe contact. So a normally closed switch connecting the input to common (ground) until probe contact provides a very low impedance at the logic input until probe contact.


.

Dear gar,

You are absolutely right about the SPDT stuff as regard the circuit refered to in my post. Also, I was not aware that software debouncing was available "further downstream". Apologies to all.

Best wishes,

Martin

[gar]

071003-0623 EST USA

Martin:

The reason there is fundamentally a software debounce is because the typical computer is a sequential device, or operated with time delays.

When looking for the probe to make contact you are in a continuous loop testing that input, or you use an interrupt to detect it.

Assuming you do not want to do noise filtering, then the first detection of the contact will cause an exit from the test loop to go service (process the operation) whatever is to be done upon this contact. Probably save the current location of X, Y, and Z. There will be instructions to move to a new location (path) to again test for contact. All this will generally take more time than the bounce time of the switch. Had an interrupt method been used, then upon receiving the interrupt it would be disabled until you were done processing data and ready to test for another contact. Either way the software is doing the debounce.

.

[martinw]

Dear wjfiles,

I do not know the details, but it will do you no harm to put some de-bounce logic on your switch contacts if you can. Here is a link that quotes the book by the awesome "Horowitz and Hill"

http://www.eng.uwaterloo.ca/~tnaqvi/...Debouncing.htm

Best wishes

Martin