[Mecanix]

Good morning all,

I'll have to reverse engineer quite a few non-prismatic parts this autumn (3D contours, radius, etc) and gearing up for a cloud point system.
I do have a budget for this, however, and judging from the extremely HIGH cost of some relevant equipment, I find it hard to justify the expense for what I need to scan.
Another problem is the constraints I'll have to work within... e.g. probe's OD can't exceed more than 31mm on a critical part and most (if not all) of the probes within my price range are rather let's say bulky, some literally over-sized and heavy, others [no comments!].
That leads me no choice but to DIY my own 3D measuring probe (/end of short-story-made-long lol)

My question is: what sort of accuracy are you getting with the classic DIY 3 balls/3 poles type probe. I'm aiming at 0.01mm~0.02mm (which is by far the maximum my milling machine can cope with when up to temp).

Realistic? Share your numbers!

[dharmic]

Mine's rough but repeatable to about 0.05mm. If I built another one now there's a few things I'd pay much more attention to:
- Material selection: I used a carbon fibre tip and I suspect it's a lot more elastic, so it deflects more before it trips;
- Spring selection: I used whatever is handy and I think it's a little stiff - there's a sweet spot where it's just stiff enough to ensure re-engagement but any more and it takes too much force to trip (hence deflection in the probe);
- Concentricity. I can dial in the tip but that puts the probe shaft on a bit of an angle so I can only test at the tip. Not what you need on some side engagement tests;
- Size of triangle. Wider = more accurate because levers; Mine's about a Ø45mm circle because I had Ø60mm delrin stock to turn the enclosure for the case. It seems about right, I'd be leery of getting smaller.

I think your 31mm constraint might be your biggest drama but care in your build and using a commercial probe shank and tip should easily get you in that range of accuracy.

[RCaffin]

My CNC has a basic single-step resolution of 0.98 microns.
With a Z-axis probe I can see that in a profile of ground stock.
With a 3D probe the resolution is a bit larger - maybe a few microns.
I made both the 3D units and the Z-axis units (several of each).

Do you need a full 3D probe, or would the XY abilities of your CNC combine adequately with a far more simple Z-axis probe?
You might also like to read my thread on MYOG stylii, at MYOG and Repair Touch Probe Styli . Some of my stylii use ruby balls, while other use CZ balls or even balls from a race. I had to drill them myself.

Cheers
Roger

[Mecanix]

@dharmic - way (way) much more useful information than asked for, I appreciate the time you took to help out with this. Thanks man!
I can sorta live with a huge max of 0.02, which is, well, what my machine can cope with on a very good day. I was afraid to get 0.05 which would not work for me though...
I'll try to make two types; a lightweight micro-size one (OD30mm) with a light comp spring and, a heavy-duty one with weight and see what comes out winning. Only way to find out in-the-real-world (CAE aside) is to make them both :/ what a pita lol
For the stylus; you've convinced me to go for the ceramic shank and SS mating screw type, so I'll see how this performs first. Thanks for the heads up

@RCaffin - I had a look at your thread and found it quite good and informative, thanks! I will certainly give this a try for when I'm fed-up of breaking my tips due to my often (recurrent) over-travelling mistakes ;-)
In my case I need the full 3D capability unfortunately. I'll try to document the built process... I'll shoot you a note when I do so you can keep me on track!

[__Britt]

Is what you are probing electrically conductive?

On small machines (benchtop sized) what's sometimes done is to ground the part, and detect when the probe tip gets grounded out... the method has it's drawbacks, but it is compact...

Not that it helps you... but... something that I've thought about on occasion is not using a switch closure, but detecting stylus deflection using a more 'analog' approach... perhaps capacitively, similar to how digital calipers work; inductively with something similar to a LVDT; or optically by a photo-interrupter sort of arrangement. All of these would require creating a custom PCB and doing some micro-controller programming...

[dharmic]

I was going to mention the strain gauge style probes which measure force applied to the shaft in terms of much, much smaller deflections than the three pairs of balls and give better accuracy too, but (like the cap sense and others) they add orders of magnitude more complexity to the probe which would make me baulk from attempting a home build - and I'm an electronics engineer by training. If the simple ball switch arrangement will yield the required accuracy and I believe it will, I'd keep it simple.

Good point on the electrical side though __Britt - a simple conductive touch probe would make things a lot more compact (and simple), even if you were to drop a spring and rocker plate in there to save tips in case of overruns.

[maxspongebob]

A while back I ran across this probe on the web and, if I was buying one, I would really like to try this one.
The videos on the site are very informative and I learned a lot about probing. The price is higher then the average 3 switch probe, but no where near the cost of some.

https://hallmarkdesign.co.nz/probe

[Mecanix]

@Britt - a whole lot will be composites and anodized surfaces. I actually wanted to bring-in a conductive shank of a known diameter once... to touch off work offsets (vise's clamp?) but that idea of a divorce with my Haimer/3Dtaster crunched that project real quick ;-)

As you and @dharmic mentioned, there are far better tech avail but its all down to time-vs-money isn't it. Decoupling/filtering those noisy environments aka 10/12/16bits resolution ADCs, C code for the SPI, or whatever bytes those new ICs needs to be interfaced with... I can see a whole lot of R&D work which would then justify getting the $5k Renishaw solution and just call it a weekend.

You guys just reminded me... that is a NC switch those tri-balls circuit :/ Looks as if I'll have no choice to throw in a custom pcb to invert the signal anyway (got a tool setter on the machine that I'd like to keep). arggghh...

[RCaffin]

Renishaw use strain gauges these days, which explains both their sensitivity and their price.

I did look at the idea (and I do have the strain gauges and the necessary electronics workshop), but eventually I found better things to do. In fact, 99% of the time I use a wobbler of known diameter for zero alignment and rely on the balls screws for the rest.

BUT: any good ideas????

Cheers
Roger

[Mecanix]

The wobbler is by far the winner in this category. Properly adjusted and spun at the right rpm its just always spot on. Simplicity often makes perfect! Too bad he doesn't do Z or I'd have mine back.

It terms of ideas; the capacitive sensor would be the cheapest and easiest to implement. Although would require semi-serious mechanical & electrical engineering and most likely precision machining and processes also, which would then, again, shoot you up close to the Renishaw price range once completed (i.e. time). Strain gauges in those sizes and orientation aren't all exactly cheap to my knowledge (requires custom-made, ideally). In all cases we are talking about very low voltage (µV) so... lots of work at the interfacing part also :/

As for myself/my req; I'm sticking to the tri-balls for now. I'll crack on with a OD30mm along with what appears to require a custom PCB now, and see how that one goes.

[mactec54]

The wobbler is by far the winner in this category. Properly adjusted and spun at the right rpm its just always spot on. Simplicity often makes perfect! Too bad he doesn't do Z or I'd have mine back.

It terms of ideas; the capacitive sensor would be the cheapest and easiest to implement. Although would require semi-serious mechanical & electrical engineering and most likely precision machining and processes also, which would then, again, shoot you up close to the Renishaw price range once completed (i.e. time). Strain gauges in those sizes and orientation aren't all exactly cheap to my knowledge (requires custom-made, ideally). In all cases we are talking about very low voltage (µV) so... lots of work at the interfacing part also :/

As for myself/my req; I'm sticking to the tri-balls for now. I'll crack on with a OD30mm along with what appears to require a custom PCB now, and see how that one goes.

If you have trouble getting the accuracy you need then you could buy this probe https://www.kurokesu.com/shop/TPA2 which has a 2um repeatability

[dharmic]

Could also use an inductive or even hall effect sensor maybe, to measure tiny deflection.

As far as the NC/NO is concerned Mecanix- I just used a couple resistors and a cheap FET (2N7000?) inline on the cable from the probe socket in the electronics enclosure back to the BOB.

[Mecanix]

Could also use an inductive or even hall effect sensor maybe, to measure tiny deflection.

As far as the NC/NO is concerned Mecanix- I just used a couple resistors and a cheap FET (2N7000?) inline on the cable from the probe socket in the electronics enclosure back to the BOB.

Good call on the hall sensor. I'd need to dbl-check this but I'm sure I've seen new ICs that detect linear motion and rotation (magnetic).

How wonderful! and its just now you tell me I could have simply used a fast(enough) switching fet to invert the signal LOL Just joking, just upset I had to spend 3+ hours finalizing the PCB an hour ago (not bigger than a coin, OD26mm). Had to throw in a TH Led (for boot/live testlight mostly) and a 4mm MCU in the kit for calibration and sensitivity adjustments (I plan on measuring the tri-balls independently, e.g one set of balls = one mcu pin). Decoupled but no room left for a ferrite bead I'm afraid. I'll deal with noise at the control I guess...

[RCaffin]

Um ... having been down that path several times - just how do you intend to attach the balls to the PCB holes? You will need to use very hard balls: brass need not apply imho.

Cheers
Roger

[Mecanix]

Um ... having been down that path several times - just how do you intend to attach the balls to the PCB holes? You will need to use very hard balls: brass need not apply imho.

That's an excellent question. The whole idea of using Vias on the PCB is to have the pcb manufacturer to tin the wall of those 6 holes. I guess the plan is to secure those precision R2 copper balls with solder paste from under the PCB. That way I can hot-air gun the kit and precisely adjust those balls on the flat granite table. A friend recommended me to drill the 6 balls (?!) and attach them to the body of the probe with a set screw. I doubt I'll go this/his route as thermal exp would ruin this sort of set-up overtime. If you have any suggestions by all means, now or soon could be a really good time to share your ideas... I'm just about to send over the gerber to my pcb mfg and run these solids in the lathe ;-)

Understand this is not a final design; more or less a proptotype to validate the data I've pulled out from several thermal, linear & non-linear analysis. Once I have this proto in hands, and validate how it behaves I'll re-do it with fatter this & shorter that, etc...

@mactec54 - I'm doing the smallest 3D Probe in-the-World! Will crunch out 1um or less when done I appreciate the link up, unfortunately the advertised probe is way too big for some of the parts I'll have to angle & scan on my table.

ps. quick preview of what came out from the cad side so far. I'll resume this tomorrow and hopefully turn a prototype this coming weekend - if the weather is bad and worth hiding in the shop lol

[mactec54]

That's an excellent question. The whole idea of using Vias on the PCB is to have the pcb manufacturer to tin the wall of those 6 holes. I guess the plan is to secure those precision R2 copper balls with solder paste from under the PCB. That way I can hot-air gun the kit and precisely adjust those balls on the flat granite table. A friend recommended me to drill the 6 balls (?!) and attach them to the body of the probe with a set screw. I doubt I'll go this/his route as thermal exp would ruin this sort of set-up overtime. If you have any suggestions by all means, now or soon could be a really good time to share your ideas... I'm just about to send over the gerber to my pcb mfg and run these solids in the lathe ;-)

Understand this is not a final design; more or less a proptotype to validate the data I've pulled out from several thermal, linear & non-linear analysis. Once I have this proto in hands, and validate how it behaves I'll re-do it with fatter this & shorter that, etc...

@mactec54 - I'm doing the smallest 3D Probe in-the-World! Will crunch out 1um or less when done I appreciate the link up, unfortunately the advertised probe is way too big for some of the parts I'll have to angle & scan on my table.

ps. quick preview of what came out from the cad side so far. I'll resume this tomorrow and hopefully turn a prototype this coming weekend - if the weather is bad and worth hiding in the shop lol

30mm dia is not even close to the smallest in the world :wave: Renishaw TP6 basic model is 25mm dia and they make smaller than this

1um yes anything is possible in the right environment

[Mecanix]

30mm dia is not even close to the smallest in the world :wave: Renishaw TP6 basic model is 25mm dia and they make smaller than this

1um yes anything is possible in the right environment

I was hoping you'd noticed the blink at the far end of my sentence

Yea, I know about that TP6... received a quote 2 or 3 weeks ago from the local re-seller here and nearly had a heart-attack while falling off my chair. Still traumatized today... so so so much money for something so dumb that will eventually(most likely) smash in the side wall of your vise.

Wait! we'll do the Tri-Compact model in the OD24.5mm with a 16bits 'something' not long after the one I'm prototyping now (I'm in the mood LOL)

[mactec54]

I was hoping you'd noticed the blink at the far end of my sentence

Yea, I know about that TP6... received a quote 2 or 3 weeks ago from the local re-seller here and nearly had a heart-attack while falling off my chair. Still traumatized today... so so so much money for something so dumb that will eventually(most likely) smash in the side wall of your vise.

Wait! we'll do the Tri-Compact model in the OD24.5mm with a 16bits 'something' not long after the one I'm prototyping now (I'm in the mood LOL)

There is nothing wrong with being in the mood, a lot of amazing things can be made, I say go to it keep it flowing :cheers:

Did you look on Ebay there you will find a TP6 starting at $325 with the probe, you can find some great deals for precision measuring instrument's if you have a need, .02 is a long way from 1um but a good starting point if that is all you need

Tooling only gets smashed into something if you don't know what you are doing, I have some probes that have been used daily for 20 years and are still as good as the day they where manufactured

[maxspongebob]

Um ... having been down that path several times - just how do you intend to attach the balls to the PCB holes? You will need to use very hard balls: brass need not apply imho.

Cheers
Roger

To attach the balls, apply the solder past to the pads on the PCB, place the parts, then preheat the balls to just above the reflow temperature of the solder paste. Place the balls on the pads. They should melt the solder. Then put the whole thing in the oven and let the rest of the solder reflow.

Do a normal cool down.

Done.

BTW, you want to make sure that your balls are completely clean before you heat them.

[RCaffin]

To attach the balls, apply the solder past to the pads on the PCB, place the parts, then preheat the balls to just above the reflow temperature of the solder paste. Place the balls on the pads. They should melt the solder. Then put the whole thing in the oven and let the rest of the solder reflow.

Heh heh heh...
Now try doing that with case-hardened steel balls. Without disintegrating the PCB in the process.
It is much easier if you can get some gold-plated steel balls - finding them and paying for them might be more difficult though.

I used gold-plated rods in place of the 6 balls, and gold-plated ball-ends in place of the 3 rods, for some of my creations. But I happened to have such parts left over from other R&D work (not from my budget!).

On the other hand, you could try the jewellery channels:
https://www.ebay.com/itm/3Pair-Set-G...4AAOSwZQRYc9X3
But a caution here: the gold 'plating' on all these jewellery bits is extremely thin and when you try to solder to it the gold 'plating' just melts off the steel surface. You would have to go for a mechanical connection, which can be a bit unreliable, or spot-weld to the shaft. They are stainless steel, not case-hardened, but with low forces they can work.

Your next problem will be the stability of the PCB material. That is only medium. It can be subject to slow creep during use. Using a ceramic replacement material is more stable, but you have to find the ceramic and be able to machine it (and pay for it). Macor, which is a machinable glass, is a possibility here IF you have the budget. But it is really NOT cheap!

Sad experience shows (me) that one spends more time realigning the probe tip (every time) than actually using it.On the other hand, one can make a Z-axis probe with micron resolution quite easily, and one can equally make a single axis single-direction sideways probe without too much hassle.

Does all this mean I (and many others) have spent ages experimenting with making 3D probes? How did you guess? Fwiiw, I have 3 - 4 3D probes of various sizes and several Z probes and Z-touch plates on the shelf (and more bits too). They work, very well too, but they are also too much bother in practice.

Cheers
Roger