[Muzzer]

No thank you very much!

See excessive thermal expansion between 20dC~30dC (10dC diff.). The worst materials you can possibly use to engineer a probe, or any precision measurement tools, are composites.

Errr, isn't FR4 a composite? Perhaps I misread but it seems you intend to use your PCB to mount the balls. Could be why Renishaw use a (presumably high grade) plastic for their internals. You might look at high stability plastics rather than crappy Delrin if you believe normal plastic isn't up to scratch.

[Mecanix]

Yes, plastics have a significant thermal expansion, but it would be symmetric around the central axis. I don't see (yet) why a plastic centre should affect the accuracy much?

Cheers
Roger

I don't know, I haven't tested it in a real-life setting so I can't confirm Roger. Just kept failing the initial sim validations here so I didn't bother and omitted the plastic all together. In fact I still have those sims on hdd, let me see if I can convert all of those 1D objects (RB2/3, spiders, nodes, etc) to a readable 3D mesh format so folks can make sense of them and I'll post them up for you guys to understand what's going on with those plastic parts in there (pom,abs,poly,pvc,pp,pe, etc all the same really). I also have the spring/damping/force simulations data coupled with a motion sim, which I'm sure you'll be interested to see.

Errr, isn't FR4 a composite? Perhaps I misread but it seems you intend to use your PCB to mount the balls. Could be why Renishaw use a (presumably high grade) plastic for their internals. You might look at high stability plastics rather than crappy Delrin if you believe normal plastic isn't up to scratch.

FR4 (glass) is indeed a composite, per definition. By far one with the lowest thermal expansion coefficient also, hence the main reason it is used widely for micro-size circuit/traces (and holding probe balls). Well documented everywhere, I'm not inventing any of that stuff and very much likely this is going nowhere within +/-10dC ;-). And I do not know about Renishaw's tech and/or their materials, so I can't comment. I'm sure they know what they're doing and have engineered the correct mat comps/formulas for their particular application.

[jkkmobile]

I don't think you should be measuring anything on a space with 10dC temperature variations..

[Mecanix]

I don't think you should be measuring anything on a space with 10dC temperature variations..

lmao... very true! Unfortunately my super duper non-high-tech shop here is not all that environmentally controlled. And this probe is prepping to launch for a 24/7 digitizing journey into point cloud space! Some parts might actually take a few days to scan (i.e. reverse-engineered). PS: Among those parts is a full size car dash and console to scan down to +/- 0.02~0.01mm (to put things into perspective).

We do get pretty bumpy morning/afternoon and night/day temp variations in the autumn here... so no excuses to not have it designed half-decently from the start ;-) Just can't see myself re-calibrating this probe 3x per day.

[RCaffin]

@Mecanix

What matters most is any asymmetric temperature distribution. That is, if one side is hotter than the other side. This should not apply to your probe.
I predict that a 10 C change in temperature will have negligible effect on the X&Y calibrations, and very little (a couple of microns?) on the Z.
In fact, the temperature coefficients of the aluminium body and the stylus shaft will be far more significant for the Z axis.

My worst-case was a PE mounting jig for 10 small aluminium parts spread along 220 mm. The jig was made in summer, with 20.00 mm spacing between parts. It was off by a measurable amount in winter, but the temperature shift would have been about 30 C! No problem: I simply adjusted the between-unit spacing from 20.00 mm to about 19.8 or 19.9 mm in the definition file, and kept going.

Cheers
Roger

[Mecanix]

@Mecanix

What matters most is any asymmetric temperature distribution. That is, if one side is hotter than the other side.

Was told by all those professors, engineers, mat scientists & techs, that what matters most is a selection/combination of thermally stable materials having low or equal CTEs (Coeff of Thermal Exp.) for a given application/requirement. Plastics-vs-metals assembly matting seems pretty extreme in that context lolll Confused now Roger, just don't know who to believe anymore ;-) But that doesn't matter anymore cuz its done already! Not going to go all too crazy as I can't source nor engineer stable mats or nano-particles anyway. Did my half-best with what I have access too, commercially.

What matters most right now is that they are holding onto my PCB/parcel longer than expected. I should have received this today as I have plans out of town for tomorrow and Saturday and won't be able to play with my balls!

[Muzzer]

FR4 (glass) is indeed a composite, per definition. By far one with the lowest thermal expansion coefficient ....

The CTE for FR4 is nothing special at all - and almost identical to Delrin for what it's worth (about 1E-4 per degree C). Even standard glass filled Nylon 6.6 isn't much different / worse.

Where have you seen it used to hold probe balls before?

[Mecanix]

My manufacturer's FR-4 is rated 14ppm/k, while acetals(delrin?) are over 100ppm/k. By all means use delrin if this is what you think works best for you, I'll be just fine with what I've selected and signed-off on already (validated).

Attaching a research paper extract of various engineered plastics' CTEs, those common ones that are commercially available to us. Chose wisely!

[Mecanix]

Tada! Voila, we can finally quit that CTE, continue building-up the electronics and writing that C code for this AVR!
Quite relieved to find out that my pcb mfg didn't sabotage my 10P too much, turned up quite nicely considering the size. In perspective, one pin is only 0.16mm in width, a pad is 0.2mm wide so not all that bad (pass).
I'm off work and out of town tomorrow so the plan is to work those late this weekend.

All good fun ahead! As always, here goes the visuals...

[RCaffin]

Was told by all those professors, engineers, mat scientists & techs, that what matters most is a selection/combination of thermally stable materials having low or equal CTEs (Coeff of Thermal Exp.) for a given application/requirement.
Academic approach, a shade divorced from real life engineering. (Btw, I am an academic!) Lacking matched CTEs, go for symmetry of design.

What happens if your PCB expands uniformally by, say, 5%? That will not alter the alignment of the rods on the balls, will it?
The same Q for the core of a symmetrical spider: the spider will not twist around and misalign, will it?

Er - I had better hedge my bets there: if the plastic used has not been tempered and thermally aged, then twisting IS possible. I had some acetal bar from two different vendors. I drilled and reamed some 8.00 mm holes in this plastic. The holes stayed at that size in the material from one vendor over a year - as tested with calibrated gauge rods. The holes in the other brand grew from 8.00 mm to about 8.15 mm ID. This stuffed up the O-ring shaft seals in the bore, which was unfortunate as the O-rings were handling propane gas and liquid at high pressure. Leaks are 'not good' with that stuff.

It turned out the second lot had not been cured properly, and the inside of the bars was subject to tension stress. When I drilled the holes, that stress was relieved by the holes growing in size. This is a known problem with some plastics, but it has zero to do with the CTE. A domestic oven at 80 C for a few hours works fine for fixing this.

Cheers
Roger

[Muzzer]

In fact FR4 has significantly different CTEs depending on whether you are talking in plane or perpendicular to plane. The "in plane" CTE is good due to the layers of glass fibre and very little to do with the epoxy. The Z axis CTE is usually at least an order of magnitude worse - almost entirely because of the epoxy. But it's of little practical consequence - I only mentioned it because you claimed it was somehow critical to the design.....yet now it isn't. In the same way that getting the balls absolutely flat was somehow critical....yet in fact it isn't.

The force on the tip will vary according to the direction of contact. This is something to do with the fact that you have the legs at 120 degrees yet the axes are orthogonal. It really isn't an issue.

As I understood it, "validation" is proving the result by passing the tests with the finished product. I don't think we've got there yet?

I'm almost getting high on the rich, self-fertilising aroma here but in many ways enjoying the experience!

[Mecanix]

Saturday's plans scrap'ed due to bad weather forecast so... what's better to do with that free time. You've guessed right, boyz machines and toys!
All up and running at this end; mcu, connector, clearances, cabling, anodized shaft, programming... all seems to work flawlessly.
I am seriously impressed with how sensitive this probe is... slight touch and it triggers. Plenty satisfied with the result (until now anyway!).
All it need next is mostly cosmetic apart from a spring really. Got the enclosure and spindle mounting shaft done already, so I guess we are now moving to the spring tension/damping work!

As always... here goes so more visuals.

[Mecanix]

This probe doesn't seem to require a whole lot of force to be triggered (why the spring exactly?!), considering the force generated from the mainshaft/pins/probe assembly's mass of 0.023856291 kg (24grams?) over 9.8 m/s2 gravity (our earth's, I think).

Anyway... within my initial validation I need to define what sort of force (N) is require to disengage the pins from the balls while imposing a limit on how much the shaft is shifting in X/Y, which is what occurs when there is no pre-load tension (no spring). For that validation I need to know what sort for feedrate (roughly) you guys are probing at so I can have a baseline velocity. Embarrassed to ask, understand I've never had one of those tri-ball style probe so....

At the moment I'm simply plotting the force data required all around in 10deg increments (see grooved gauge), sort of finding out where are the highest F(N) point(s). We'll define the required spring force later on the sim...

So any of you who runs those probe, sharing that feedrate/velocity you use when probing would be useful at this stage. Thanks

[maxspongebob]

In using Mach 4, there are 2 rates for probing. The probe cycle uses a fast rate to find the surface and then a slow rate to measure. For my home made probe I use a fast rate of 20 ipm and slow rate of 10 ipm. For my toolsetter I use a rate of 10 and 5 ipm.

[Mecanix]

In using Mach 4, there are 2 rates for probing. The probe cycle uses a fast rate to find the surface and then a slow rate to measure. For my home made probe I use a fast rate of 20 ipm and slow rate of 10 ipm. For my toolsetter I use a rate of 10 and 5 ipm.

Top man, Bob, appreciated. Thanks, that's all I needed really to start-up.
Another to consider are those dials indicators' spring force... data seems missing from most manufacturers' datasheet so I'll have to start-up with mitutoyo's.
All about giving that ideal pre-load without affecting the accuracy too much isn't it, 'at speeds' and making sure that my dial indicator is able to calibrate it lolll
Nut-thing ever simple :/

[jkkmobile]

Similar ones have:

axial 3 to 5 N

radial 0.5 to 1 N

[jkkmobile]

Stylus contact forces for CMM scanning probes

SP600

[https://www]

The relevant spring rate is 1.2 N/mm in the X, Y and Z axes so the force on the component will vary with the amount of deflection. The SP600 measurement range is ±1 mm with a 50 mm stylus.

Therefore, the*maximum force between the stylus and workpiece for a 50 mm stylus will be 1.2 N (120 gf).

SP25M

[https://www]

This probe works on a different principle from SP600 and the force exerted on the component varies*with the stylus length. In the XY plane, the force at the component varies at a rate of 0.6 N/mm when using a module's shortest specified stylus and 0.2 N/mm when using a module's longest specified stylus. The measurement range is ±0.5 mm so the max force at the component will*be 0.3 N (30 gf) when using a module's shortest specified stylus and 0.1 N (10 gf) when using a module's longest specified stylus. In the Z-axis the force varies at a rate of 0.5 N/mm.

SP80

[https://www]

For this probe, the spring rate is 1.8 N/mm in the X, Y and Z axes, with a measurement range of ±2.5 mm. The maximum contact for is therefore 4.5 N (460 gf).

REVO

[https://www]

The RSP2 REVO probes work on a target deflection range of 50 microns to*200 microns. The actual force on the component is dependent on the particular stylus holder being used because they*have different stiffness. It*should be in the range of 0.05 N and 0.25 N (5 gf*and 25 gf)

https://www.renishaw.com/cmmsupport/...-probes--16566

[Mecanix]

Stylus contact forces for CMM scanning probes...

More info than I've ever dreamed for, you guys awesome. Thank you very much. Got a feeling that'll be easy now
Back to work for me... need to get this rolling before next week... jeez takes longer than anticipated this 3Dprobe thingny.

Cool forum, thanks again

[RCaffin]

I'm simply plotting the force data required all around in 10deg increments
It is well-known that plotting the deflection force and the sensitivity of this sort of probe does create a sort of triangular map - due to the three arms of the probe.

Cheers
Roger

[Mecanix]

I'm simply plotting the force data required all around in 10deg increments
It is well-known that plotting the deflection force and the sensitivity of this sort of probe does create a sort of triangular map - due to the three arms of the probe.

Cheers
Roger

I did looked around but couldn't find any of that data, just a few mentions from folks throwing out random numbers like the bingo ladies does.
Do you happen to have that plotted data handy or know where I could find it? I'd really like to have something to compare to the one I have.

ps. the plot is not triangular (but I know what you mean). Looks more of less like a 6 petals flower when mapped onto a radar chart. Correct spring rate and <0.05mm travel fix this though...