The "One Penny" Touch Probe

History

It all started about six months ago when some CNC tinkering friends and I were Skyping away late one night and came up with the idea to collaborate on a touch probe project. As is often the case, this resulted in MrBean doing 95% of the work and 6 months later he sent me his working prototype, based on Graham Stabler's Renishaw probe copy here (indoor flyer). The details of MrBean's project and plans for it can be found on this cnczone thread.

The main problem with the homemade Renishaw ‘ball type’ probe which both MrBean and I banged our heads against, was their propensity to stick in the open state. The principle of these probes is to have 3 radially placed conductive pins resting in the nooks of 3 pairs of ball bearings, the pin forming a contact between each ball pair. The probe fires when any of the pins breaks contact. The problem is, you have 6 ball bearings and 3 pins that you need to get in pretty sensitive alignment. Perhaps with a highly accurate machine and or much manually tinkering and adjustment you’ll get them running reliably. I could get it running fine for a while, then the probe would start to stick and I would swear most heroically.

New Design

Rather than 6 ball bearings, how about just 3? By it’s very nature this would be inherently stable. Instead of round pins, a flat disc would be needed to sit on the 3 contacts. Then I started to ponder how best to hold the balls down and make electrical contact. This turned out to be trickier than I first though so I considered alternatives.

The new probe design I came up with consists of 3 pins (of the "steal them from your wife’s sewing box" variety) arranged around the outside of a cylinder of plastic (I drilled a piece of acetal on my mini lathe with a 10mm hole), with the heads of the pins inside the cylinder. A thin plastic disk is screwed to the end of the cylinder to hold the pins in place.

Here is a very brief CamBam tutorial on the steps involved in making the end disk with a central hole and radial mounting holes (a common sort of task).

CamBam mini tutorial. To make the end disk

1. Insert circle, change diameter to 26mm.
2. Insert 2.5D profile. Tool Diameter = 3mm, Tool Number = 1, Target Depth=5mm, Step=1mm
3. Insert->Point List->Centers
4. CAM->Drill (Tool Diameter = 3mm, Tool Number=2, Target Depth=5.2mm)
5. Insert another circle diameter 20mm
6. Insert->Point List->Divide Geometry (into 3).
7. CAM->Drill (Tool Diameter = 3mm, Tool Number=2, Target Depth=5.2mm)

For the contact disk I was looking for something hard, smooth and conductive that wouldn’t tarnish too easily. I considered cutting a section of hard disk platter out but instead as an indication of how confident I was with the new design I decided to invest hard currency into the project and turn down and face a 1p coin. This was done by drilling a centre hole and tapping that to M3, then attaching it to an M3 shaft which was chucked in the lathe. Given a choice between having the Queen's head or traitors gate still visible, I opted for the gate so then defaced (or faced?), the Queen’s boat race away to as shinny a surface as my rickety lathe would permit. I’m sure this involved committing at least one criminal offence.

An M3 screw is then screwed all the way onto the disk to make the probe shaft. The disk sits on top of the 3 pin heads (lying on their sides), like a bronze age Dolmen. A spring pushes the disk down onto the pin heads and a mounting shaft is attached to the other end of the probe. I made my mounting shaft by turning down a piece of M10 threaded rod to 6mm to fit my spindle collet. By screwing the 10mm threaded part into the body, the spring is compressed, thus making a very simple sensitivity adjustment. To prevent the disk sliding around the pin heads, a bead of silicon is run between the probe shaft and the outside of the end disk hole. This also seals the internals from dust. Different probe tips can be easily screwed onto the end of the M3 probe shaft.

This is the prototype so excuse the appearance. Later versions will be much shorter and more elegant.

Electronics

Each pin is pulled high by a resistor and the disk is connected to ground. In the resting state, the disk touches all 3 pins so they will be at 0v. As soon as the disk breaks contact with a pin it gets pulled high. The three pins need to produce a single output so each pin is fed into a NOR gate. A simple way to do this is to use a single transistor. This does mean running 3 wires to the probe, but the signal should be much more noise resistant and best of all, you can put a LED on it. To really go to town, you could also put a transistor triggered LED on each pin, so you can see which pin is firing.

ERRATA! The previous documentation had a wrong value for R1-7. The correct value should be 12K.

Software

At the moment I am using Art Fenerty’s digitize plugin for Mach3. This is simple and effective but mind numbingly slow. Apparently Art has a new and improved probing application in Beta so look out for that. I’m keen to have an experiment writing my own probing routine but that may have to wait until next week.

I am currently viewing the resulting point cloud files in CamBam, using a very simple plugin I knocked up to convert the point lists to a polyline. To use the plugin, copy the dll file into the CamBam plugins folder. Improved point cloud and 3D support is planned for CamBam in 2007.

Result!

Here is the (nearly) finished result, a New Zealand 20c piece that had mysteriously appeared in my pocket somehow. This has probably increased my criminal offence count to 2, but one of those is in New Zealand so I may get away with that one. This is a 30mm square grid scan at 0.1mm increments (90k test points and zero glitches!!!). The Z resolution seems to be around 0.01mm. After this run I plan to try a sharper probe and scan a small area at 0.01mm X and Y increments.

I can’t believe how easy to make this probe was and how reliable it is turning out to be. Having a lathe helped, but I’m sure the design could be adapted to use only hand tools or parts cut using a CNC machine.

MrBean and I have been coming up with some improvements and plan to make some second prototypes soon. Watch this space for new updates.