3d indicator design

[David Jupp]

Three prongs/legs? Or a cunning plan to get all level and square?

PekkaNF

Or turn the whole thing as a profiled disc on an axle (like toy gyroscope) - an infinite number of legs better than 3 ??

[BillTodd]

Yes, the cross-section view can be a bit misleading.

It is, as David suggests, a disc with a torus around the edge.  It might  make a simple edge detector but, because to the non-linearity, it would be only be suitable to a fix sized probe


I haven't abandoned the ball slide idea just yet; While sitting on the throne this morning, contemplating the weightier of matters of state, I started to wonder where the errors are creeping in...

First up is to test the DTI. Then, just to see where the 50um/2thou" on the Z axis has gone, test the upper taper section with out the lower taper or ball in the way.

At the moment, I cannot account for all the missing 80um/3thou" in the XY plane; The play in the vertical slide  is << 1 thou" and by playing with it (pushing and pulling it ) the ball-slide has <10um according to the dial.

[PekkaNF]

Or turn the whole thing as a profiled disc on an axle (like toy gyroscope) - an infinite number of legs better than 3 ??

I think that accuracy needeed for that would be such that very few of us could reach it. On that front my only bet would me to use something massproduced part that has to to have that accuracy, like ball bearings and preloading them such way that clearances don't accumulate on measuring direction....which is ofcourse harder when there are many degrees of freedom on movement.

I'm really tool-envy on that lathe. Nevertheless would it be possible to double check that the measurement method and/or device does not introduce it's own set of variables. Can you affix 3D indicator on the mill or something and test 3D indicator reading agains micrometer anvill movement and possibly to verify it with a micron dial gauge or such? Even better if there were no oilfilm anywhere (spidle/slides/you name it).

PekkaNF

[BillTodd]

Nevertheless would it be possible to double check that the measurement method... Can you affix 3D indicator on the mill or something and test 3D indicator reading agains micrometer anvill movement

That's a good idea Pekka , and I have a 1/10000" micrometer head that would be ideal for it   ATM the lathe is a convenient way to quickly test bits.


I found a hour yesterday to test the DTI etc.

The DTI is OK within a few microns (i.e. within readable limits)

When I tested the Upper Taper alone. I discovered that there is a small difference (~15um) between one direction and the other. That is, going IN to zero on the DTI will give a different reading on the DRO to moving OUT to zero.

I think this is caused by the motion of the ball in the right-angle hole and this maybe due to my use of a non-flat tip on the end of the DTI, so I'll have to make a flat tip and test it.

I'm now fairly sure that the major cause of the other errors is the angle of the tapers: The upper taper shows a error of ~50um over 2mm (middle taper was cut at the same time so will have the same error) .  The lower tape was cut in something of a hurry so is likely to be even worse!

When I recover from this mystery back strain that has been giving me grief all day, I'll set up the top-slide with a sine bar and re-cut them.

Bill

[PekkaNF]

I am soo happy  you are back on track 

It happens me all the time - I'm so absorbed with the problem in my hand that I overlook procedural complications. Once I spent some time trying to figure out where the error comes from. And I was using simple DTI! Long story short: Everything I tought being square was not square and the "balls" where so small that even a little offset from theoretical measurement axis introduced error....the object was straingt but my measurement wasn't.

Spindles, bearings and slides never stop to amaze me. Move them and when you stop them they "sit". You can measure that, but they lift when you turn or mill. How on earth anything gets done with any predictability....I get massproduction, but one-offs!

PekkaNF

[BillTodd]

Thanks for the encouragement Pekka - it is appreciated   

I've fixed IN/OUT the problem with a flat DTI tip. It now repeats within a few microns.

However 

I spend a couple of hours checking and rechecking my taper setting using a sine bar. Finding a stack of objects exactly 3.5355" was a challenge but,  in  the end,  two V blocks & a thrust washer measured to better than 1/2thou".

Then I carefully set the top-slide (the picture below is of the general method used ) and re-cut the tapers - well I say "re-cut" but actually there was so little difference between the existing tapers and the new uber-accurate ones it was hardly worth touching them

So, with my backache getting the better of me,  I set about measuring the top taper in the indicator :

Unsurprisingly; 2.000mm indicated on the DRO read 1.920mm  on the DTI   

So, where is this error? 

I then stepped the DTI 100um at a time while noting the DRO reading (see PDF)  to see it I could spot a jump or anything that could explain a taper (as shown by the difference plot)  of nearly 0.1mm (>3thou") over just 2mm (~3/32")  (that should be almost visible!)

The only place it can be, is in the right-angle hole but that doesn't look too bad from what I can see so....


what am I missing??? Ideas welcome. (remember I'm testing this without the lower taper part and ball-slide so it should be at least as good as the DTI)





 

[PekkaNF]

You are going to nail it big time! Every step gets you closer and closer.

I even can't cut realiably very smal DOC. I'll try to fix last cut around 0.1 - 0.2 mm depth (even though I clock it someting like 0.123 mm indicated to read the desired final dimension), othervice my parts seem to come out worse geomertically! But it's lack of my skill - no doubt.

I'm pretty damn sure your DRO and DTI reads on same axle, angular difference should be pretty large to cause error of that magnitude.

Your graph show classical scaling error that could come from incorrect aplication of metric/imperial correction on engineering of DTI or settings of DRO. I worked sometimes with SONY magnatape system and there were quite a few parameters that could be keyed in or set wrong on DIL's.

I would first check DTI against gaugeblocks to clear it. My bet would be that is the reliable one.
Then I would assume DTI fairly reliable and use same measurement range and DTI reading against lathe slide to see how much deviation is on one single axis of movement. See where DRO measures the axle and double check that you read with DTI same movement that tape/sensor sees.

Many DRO:s have many settings, not only to cobat linear erros, but also for periodical errors and sometimes error tables too, that are wonderful to have, but pain to calibrate.

PekkaNF

[BillTodd]

Hi Pekka,

I've checked the DTI against the DRO directly and both correlate very well (to within the thickness of the needle). I have great confidence in the Sony DRO - I have checked it against standard length gauges etc.  - It was a very expensive piece of kit when new !


I'm now thinking about increasing the bore of the right-angle hole to 8mm (as I have some 8mm bearing balls)  that should compensate for the material removed from the tapers and give me the opportunity to correct any taper (or irregularity) in the bore.

[edit]

I threw caution to the wind and drilled the hole out to 6.5mm   -  worse errors  and on close inspection there was a small step in the bore just  where the ball would be running.

Drastic action...

Dismantled the body parts completely and drilled & reamed another right-angle hole at 90° to the previous one. 

Result - less error but still ~50um short ...

Back-ache stopped play -  retired to the pavilion ...


Bill

[DaveH]

Hi Bill,

I really hesitated to post this, but it seem (to me) the error comes from the 45 deg taper.
The way I'm looking at it the DTI reading is always smaller, so the ball isn't moving as much as it should. So if the taper angle was larger than 45 deg that would cause the ball not to move as much.

Try a new 6.5 mm dia hole opposite the very first one, that should eliminate any slight difference to the accuracy of the main bore and the right angle hole not being exactly at 90 deg.
 
Just a thought Bill
 
DaveH

[BillTodd]

I really hesitated to post this, but it seem (to me) the error comes from the 45 deg taper.

Don't hesitate if you spot something ('cos I'm a blind as a bat when it come to some things)

Anyway you're kinda right ....


Cracked it

Or rather my friend Chris cracked it 

I was showing him the problem this evening and he came up with the brilliant suggestion that the ball might not be running down the centre of the cone . BINGO Indeed it will  try to do everything in its power to avoid running down the highest spot

Any error in the centring of my hole, or play between the ball and the hole, will allow the ball to run to one side of the taper and thus run on a parabola (a conic section) not on the true 45° face.

The fix was simple: replace the ball with a small brass slide that presents a flat plane to the cone thus only following the highest part.

Testing on the lathe now shows the right-angle DTI  tracking to within a few microns (<8um at 2mm) .

I'm back on track

I now have to adjust the fit of the large thrust washer, so it goes a bit deeper into the body (again!) and I'll be ready to test it properly (again!)

Thanks guys,

 Bill

[HS93]

There are some clever people on hear (and there friend's)  nice work. 


                                                              Peter 

                       

[PekkaNF]

Good job!

Hope your back-ache fades away as well.

PekkaNF

[DaveH]

Phew,............ I'm glad you have sorted it out Bill, I was starting to lose my beauty sleep 

 
DaveH

[Fredbare]

Well done Bill, excellent work.

John

[BillTodd]

Quick update:

I have not had to much time to fiddle with this, but following a few tests last weekend I have found the next critical bit is the ball-slide.


The ball-slide has to be tight yet slide freely for the indicator to work accurately. On this version I bonded the thrust bearing together under a  pre-load (~ 50N) using Loctite. This was fine initially but, with use, started to come loose and slightly rough.

To remove the play in the ball-slide while testing, I locked the Z axis motion using spacers and a spring washer (to pre-load the bearing again) .

The tests showed the indicator could work well - e.g. it was really quite accurate measuring the distance between two faces

However, the problem is zeroing accuracy: it does not return to 'zero' properly if moving slowly due the the roughness in the ball-slide.

Today I pressed the slide apart only  to find all the surfaces badly brinelled (dented by the bearing balls)  - I knew this could be a problem with non-hard surfaces, but was surprised because of the relatively light loads on the bearing.

So, now I'm trying to  figure out how to fit standard sized hardened thrust washers to the bearing surfaces 


Bill

[PekkaNF]

The nanosecond when I saw this design, I was eyeballing these:

http://www.arceurotrade.co.uk/Catalogue/Bearings/Thrust-Ball-Bearings
http://www.arceurotrade.co.uk/Catalogue/Bearings/Needle-Roller-Thrust-Bearings

Maybe these nedle roller thrust bearing would donor hard enough discs.

You'll need preload. What kind of washer are you using? Wafer (the wavy one) or the one with fingers/prongs?

I admire your tennacy,
Pekka

[Divided he ad]

Fair do's Bill... You sure give a project your all 



I've missed a few instalments, but have just read through and caught up.... Top job. Hope the hardened bearing plates will cure the (final) Hurdle?   




Ralph.

[BillTodd]

We're back! Welcome to round 2...


I can't find suitably wide (i.e. large diameter with a small hole) thrust washers anywhere, so I've had to redesign the locking ring to hold a standard (30 x 17) thrust washer.

I found time today to make the locking ring and press fitted the washer. seems OK

Hopefully, tomorrow I'll trim the bearing cages to clear the lip (that keeps the washer square )

Bill

[PekkaNF]

Nice to see latest development. Does the washer mount locates only to indicator face (I assume that this surface was trued) and other surfaces are non locating i.e. clearance. Could not spot it from the design.

PekkaNF

[BillTodd]

Hi Pekka,

Yes, the washer sits down on a tiny facet turned true in situ (I used the indicator body to hold the ring while turning the face).

I hope to get on and finish it this afternoon (it's a bank holiday here today).

Bill

[BillTodd]

Well, it's back together. Seems to centre to zero pretty well. There's a tiny bit of dead space ~20um ( from some angles and not others, which is a bit odd ) .

It has just over 3.5mm movement in all directions which is pleasing, since that's more than the previous set-up gave. The 45° slide still feels a little rough (need to be run in).

I'll try to find the time to test it tomorrow .

Bill

[AussieJimG]

This is an enthralling thread. I have been watching in admiration of you Bill, for your initiative in starting this project and your determination in pursuing it.

It is really wonderful to watch you track down the errors and modify the design and to see the interaction and assistance from others, particularly PekkaNF. Thanks to all of you.

Jim

[BillTodd]

Thanks Jim.

This afternoon I just got the thing in the lathe to test it when a mate turned up with a dead motor.

His swimming pool pump had eaten its own bearings (again!)  -

'it smells like the motor's dead to me' -  I said, 

'no, it's been running fine'

two hours later, after making a sleeve to fit the new bearing on the knackered shaft, we had the motor running....

for about three seconds !   

Verdict:  new pump required

Will try again to get the indicator tested tomorrow

[BillTodd]

Between phone calls, I manage to do a quick test to see how well the indicator works between faces...

Hmmm quite well, but consistently 17thou off (measuring an inch)  - I was expecting an error because of the dead spot in the middle (slack in the ball slide & other parts) but not that much   

However, it is consistent, giving the same error in all directions (well the 90, 180 etc. that I tested) so it's a simple matter of adjusting the size of the probe to compensate...

That's where the day peaked 

Had a phone call - "This 'clip' you want made... Is it really nearly an inch thick and half-a-metre wide ???"

DXF conversion strikes again - 16mm gets changed to 16 inches then converted back to 406mm on import   

I make the probe tip, after a lot of faffing around

Then I manage to break the  washer out of the ring while trying to screw the probe in  - Of course,  it will not go back in with out more faffing - Now I'm waiting for the loctite to set... 

Time for more coffee

[BillTodd]

Arrrgh... 

Finally I found time to do some tests.  I clamped a micrometer in the tool post and used the DRO's touch probe to give me an absolute zero face  then adjusted the mic to give a second face at 20mm spacing.

 

While it appears reasonably consistent in any one direction ( i.e. once zeroed in a direction it will return to zero within 10um - even if removed from the lathe),  it will not accurately measure between faces because of the play in the bottom taper slide (I had to re-ream this after I threaded it, so I guess it's now too big )

Maybe it's time to throw in the towel ? (I've lost faith in the design)

Bill