Gothic Arch Ball Race Profile

[[Ch...]]

Hi everyone!,

I would love to hear thoughts /ideas on this one...
We have many parts with this configuration.
Each arch is offset from "CL"(found with a gage ball self centering) the basic amount - in this case 0.0044" and also offset a basic amount (0.00762) from CL of gage ball, so every part will have different dimensions. I've put the formula into a torus, and I'm scanning 4 small circle paths eq. spaced. - see attachments...
What I'm curious about is ... I output the nominal torus points (includes formula for bsc offfsets) and vectors to .txt, then I manually create CAD FFS from this file...
I can tolerate the profile this way, however I'm fairly certain, that each CNC run, this FFS doesn't get 'updated' or re-drawn on the CAD, and that is what I need to compare the actual measured points to...
Any thoughts /advice is appreciated. I'm starting to visit the forums more each day...

I think this can be done 'manually' after every CNC run, but with the volume of these parts, and knowledge of our operators, it would not be a good solution.

Is there any PCM that could create CAD FFS ( with proper alignment - and filter, etc.), then create a profile characteristic too?

Thanks everyone, ttyl !

Chris

CAD race feature.PNGBtm Race Torus Scan.PNGrace dimensions.PNG

[[Cl...]]

I'd create several (maybe 4-6 equally spaced) secondary alignments around the torus, translating 4.7780 + .00762 from the center of -A- & then the .212 + .0044 from the other surface. Now you can probe points/circles inside the torus relative to their theoretical exact location, then recall those points into a FFS for evaluation. You can also use the "Radius Measurement" to report the actual location of the torus.

[[Br...]]

Why don't you recall the torus into a freeform surface? I'm not sure I follow what the significance of the gauge balls are. This Profile tolerance has no datum references. The profile of that surface only needs to be in reference to itself. What I would do is scan those as toruses like you have, then recall all the points into a freeform surface(make sure you select the surfaces used.). Then report a Profile of surface characteristic using that FFS, and for the datum reference frame select "No Datum Reference Frame."

[[Ch...]]

Brett,



Thanks for the insight. Yes that is the correct way, however; per our Eng Dept: The basic dimensions will are coming from edit: CL of gage ball and the end face, these will change part to part, so the program has to be dynamic.
I use a gage ball specified ( not on this drawing (Ø .2087) to self center and find the .211-.213 "CL", this is used for the torus Z height variable per part. I then offset from gage ball CL (0.00762*2 added to torus D1) for it's variable per part.

I've tried recalling the torus into FFS, however... If I select the CAD entity, that is based on a basic nominal value, I need a new CAD entity to compare to every measurement....

Trust me , I understand this may be confusing.

In simpilest terms: The FFS I created from file...I'm not sure this get updated or "redrawn" in the CAD everytime. The measured points have to be compared to a different size arch/race; or torodial segment, every run/part.
I suppose I can experiment, by doing a CNC run and exporting the CAD entity, to see if it actually changed each run.

This is probably beyond PCM capabilities.

Thanks for the input.

[[Is...]]

The Self-center is correct, then you have tu generate a curve based on the center of your gage ball.
if your gage ball is not accurate enough you can use the Feature circle in contour fit and avoid the self center.

you don't need PCM if you extract the nominals from the CAD model..

[[Ch...]]

Israel.

Thank you for the advice...
So you're saying to create a 3D Curve of the torodial race with gage ball Centerline coordinate reference? Would this be from 'export points'?
In the past here, other Engineers and programmers have completely overlooked profile and made up their own tolerances.
Ideally I need to and would like to measure all the way around this race (across machining tool lines), so I would measure 4 to 8 equally spaced small radii scans, to do that with a 3D curve might take a lot of slicing.

The nominals from CAD will not be accurate for me, as I mentioned the nominal each of the 2 gothic arches will be different with every part; unless I'm missing something.

I understand this is a tough one guys...

BTW, I did put a ticket into Zeiss, waiting to hear back.

[[Ch...]]

Please sign in to view this quote.

Thank you Clarke.
They've done something similar in the past, except not using FFS. My concern in that once I create the FFS, the recalled actuals points will always be different but does the nominal data of the FFS ever really get updated?, because we have to "select surfaces used", etc. I would be comparing to the same incorrect CAD entity everytime.

This may need to be done manually (create FFS from points + Profile) every CNC run, which would be an obvious pain.

I suppose if we could do profile of a torus, that would do the trick. In theory a torus should be able to have a profile band around it...
Hmmm....

[[De...]]

Its frustrating not knowing exactly what is necessary, if the engineers want this profile related to the end face then they should really call it out in the profile callout to avoid ambiguity. Hopefully my answer is not missing something that is not actually stated on the drawing.

I do not think this is going to be possible with FFS short of manually creating a model based on the parts for each part so I would likely attack this using 2D curve as follows.

First measure the distance from the end using the self centering point you described, however I would create a pattern of self centering points so that you can then recall them to construct a circle to use for the center of the 4.788 C/L of balls location. This should allow for creating an alignment to the 'groove' even in cases where there is axial runout to the end face or radial runout to datum A. (Keep in mind for this to work, the self centering points should contact in the groove rather than on the edges of the groove / datum a to avoid radial runout adversely impacting the results.) You will also probably want to create the alignment with an offset to the base alignment so location values will equal in the base alignment and the newly created alignment prior to measuring the self centering points.

Once the circle is constructed it can be used as the spatial rotation and X Y and Z origin to create an alignment.

You can then create a 2D Curve from the model that will look like the section view below and then pattern that around the alignment you just created. Because this nominal geometry is created from the model it will not include the proper offsets based on the height, to allow for this we assign it the alignment we just created from the circle (since the coordinate systems are matched we do not need to worry about the geometry being transformed until after the points are actually measured). Because the 2d curve will be in relation to the alignment created from the circle, the nominals will be updated each time the program is run based on the results of the self centering points.

It does not use FFS but because you have multiple 2D curves around the periphery of the part aligned to the self centering points, so long as they are all measured using line profile to the alignment the check will meet simultaneous requirement and therefore should give you the result for profile that you need.

[[Ch...]]

Derek,

Good thinking. I agree, it would be nice if Engineering would dimension this better,
Yes, I explored the option of the 2D Curve, as well, however since I'm unable to achieve the profile of a surface on the report, I'm not satisfied with it.

If I use the alignment you mentioned, how would we account for the 0.00762 radial basic offset for each 2D Curve?

Thanks for the ideas, I like them, I feel its getting me closer to a miracle solution or to exhausting all avenues, etc.

No word from Zeiss yet...

[[De...]]

Because the initial 2D curve would be created by sectioning the model geometry, so long as the model was created based off the print, the 0.00762 would be 'baked in' when the 2d curve was created via sectioning the model.

By making both coordinate systems (the base alignment and the new alignment) you would not see any variation in position of the 2D curves, however if you measured them and then recalled the points into a second 2d curve that used the base alignment, you would see the geometry was in fact shifted based on the position of the groove.

[[Is...]]

Please sign in to view this quote.

Yes Gage ball center line, other easiest way is just bestfit, but this could ignore some nominal values that you have in the design (maybe better to consult about the functionality with engineering)

Zeiss shouldn't have Problems to help you on this, I have discussed this specific topic in Oberkochen more than once and the support was OK..
if the CAD model is not accurate enough then the other option is generate the points with PCM code or generate a text file with the nominals using other software (excel for example).

this is the format of the text file:

xNom    yNom    zNom   uNom    vNom    wNom
-5.7630589093    7.9500000000    0.0000000000    0.3479852727    -0.9375000000    0.0000000000
-5.8015688560    7.9359298441    0.0000000000    0.3383577860    -0.9410175390    0.0000000000

there are other optional columns:

xNom    yNom    zNom   uNom    vNom    wNom  lTol  uTol maskEval  maskBF

and then you can read it in the curve feature:

[[Ch...]]

Bump.

Can someone with Calypso 2018 or 2019 let me know if you can select a torus in profile of a surface?

We have 2018, just not installed yet.

Thank you.