What's new
What's new

Accuracy of Comparators

Matt@RFR

Titanium
Joined
May 26, 2004
Location
Paradise, Ca
Hey guys. I'm getting into more and more parts with tolerances that I just can't measure with what I have. For example, two .2500"/.2504" holes tied to a .7500"/.7505" hole to .001" true position. I can get Deltronic pins for the sizes, but the true position has me worried.

My question is if a good quality comparator is accurate enough to check both size and .001" true position? Having never used one, I just don't know. And I can read manufacturer specs all day long but wanted to get opinions from those of you that have actually attempted this.
 
Hey guys. I'm getting into more and more parts with tolerances that I just can't measure with what I have. For example, two .2500"/.2504" holes tied to a .7500"/.7505" hole to .001" true position. I can get Deltronic pins for the sizes, but the true position has me worried.

My question is if a good quality comparator is accurate enough to check both size and .001" true position? Having never used one, I just don't know. And I can read manufacturer specs all day long but wanted to get opinions from those of you that have actually attempted this.

I wouldn't trust it for those tolerances.... The comparator is totally reliant on getting the part square to the light source...if it is off .001 in square then your readings will be off. Best bet is a CMM at those tolerances.
 
Thicker parts are tough to measure with a comparator... Post above is spot on,especially when measuring holes, just cannot get a good focus. I could not get by without mine, but they do have limitations..
 
That's what I was afraid of. :) All these parts are low quantity so it's pretty hard to justify buying something like that, although I'm just guessing at what they cost.

Is there such a thing as a CMM without driven axes? Something like a manual mill with cranks and a readout? That would be just fine for my needs as long as the accuracy is there.
 
In general terms an optical comparator makes most CMMs look like tape measures.
It must be set up correctly, checking axis squareness.

Magnification is the big one, I'd be looking at 50x here.
Any gauge needs to get a R&R run but outside of the 100K plus machines I'd take a comparator any day if the part fits in the envelope and the measurements are on one plane.

At .001 between two holes you are pushing lower cost CMMs pretty hard and you will need to cut your tolerance band in half to make up for the measuring machine's errors.
It has always amazed me how much people "trust" a CMM without hard R&R numbers.
Bob
 
Why not inspect the part on the machine that you made it on, using a probe? Of course, you'd spend more $$ having the machine calibrated and tested fairly frequently, but isn't that only worthwhile in the long run?

Where else are the positioning discrepancies going to come from, apart from variations in material hardness that cause tool deflection (and which probing should show)?
 
Why not inspect on a surface plate with jo blocks and a good indicator? You do have a good set of blocks and a known good angle plate? Doing a manual check on a surface plate was, back in the day, basicly assumed correct unless bad technique or uncal'ed instruments were used.
 
In general terms an optical comparator makes most CMMs look like tape measures.
It must be set up correctly, checking axis squareness.

Magnification is the big one, I'd be looking at 50x here.
Any gauge needs to get a R&R run but outside of the 100K plus machines I'd take a comparator any day if the part fits in the envelope and the measurements are on one plane.

At .001 between two holes you are pushing lower cost CMMs pretty hard and you will need to cut your tolerance band in half to make up for the measuring machine's errors.
It has always amazed me how much people "trust" a CMM without hard R&R numbers.
Bob

When it has the name Zeiss on the side ....I do not doubt it at all. It is the end all be all answers /sarcasim




Why not inspect the part on the machine that you made it on, using a probe? Of course, you'd spend more $$ having the machine calibrated and tested fairly frequently, but isn't that only worthwhile in the long run?

Where else are the positioning discrepancies going to come from, apart from variations in material hardness that cause tool deflection (and which probing should show)?

....that won't tell you anything? You can not check true position like that on the machine it was cut on. I dont care if you use a renishaw probe or an indicator.


You can get Manual CMM's like the one I have shown here pretty cheap now a days. I think this one is worth around 5-10k in good shape. The key is getting master gauges to double check. The Mit measures within .0002-.0003 of the Zeiss that I have...
http://www.creationsultd.com/quality.html
 
I would argue that it is pointless to even take on a job where your known machine accuracy is not less than the requirements of the job, unless you like to 'test' your machine accuracy by making random articles for customers, and then measuring them.

Now the customer spec may mean he doesn't know if you are laying out holes with a center punch, they have to be so and so in results. But if your cnc is in good condition, then inspecting the product on another machine that is only of comparable accuracy is an exercise in futility, plus you have the setup errors to deal with when moving the part.
 
I would argue that it is pointless to even take on a job where your known machine accuracy is not less than the requirements of the job, unless you like to 'test' your machine accuracy by making random articles for customers, and then measuring them.
I have allready done exactly what you're talking about. I do have a Renishaw spindle probe and tried this very thing on some .003" true position holes awhile back. I made the holes by ramping an endmill to size, then a spring pass, also ramped in. I also unclamped the part, then reclamped extremely lightly to make sure it was in as free a state as possible. What made me nervous is that the probe reported perfect results on every hole location. Not even .0001" error. There's no way in hell they were that close... not in my opinion. It's a nice tight machine, but really?

On Edit: I guess I could have someone with a jig borer make a bolt hole circle and certify all sizes and locations of holes to some datum (or multiple datums, preferably), that I could use as a master to calibrate the machine/probe. Anybody have a jig borer and want to do an experiment in Haas machine / Renishaw probe accuracy?
 
I have allready done exactly what you're talking about. I do have a Renishaw spindle probe and tried this very thing on some .003" true position holes awhile back. I made the holes by ramping an endmill to size, then a spring pass, also ramped in. I also unclamped the part, then reclamped extremely lightly to make sure it was in as free a state as possible. What made me nervous is that the probe reported perfect results on every hole location. Not even .0001" error. There's no way in hell they were that close... not in my opinion. It's a nice tight machine, but really?

On Edit: I guess I could have someone with a jig borer make a bolt hole circle and certify all sizes and locations of holes to some datum (or multiple datums, preferably), that I could use as a master to calibrate the machine/probe. Anybody have a jig borer and want to do an experiment in Haas machine / Renishaw probe accuracy?

Basically it means that your machine is repeatable but doesn't mean that the part is perfect. .001 true position is not a walk in the bark for most brand machines. Even on our NH5000 I wouldn't expect probing on the machine to qualify as inspection. Maybe if you had glass scales?? Even still...

If I was trying to get .001 true position I would spot, drill, endmill undersized hole and then bore the holes with a boring head.
 
For low volume, occasional parts I would check them on a surface plate using a angle plate and a height master (Cadillac gage). Greater tolerance features could be measured directly using a height gage, the tight tolerance holes would all be transferred to the height master. There are also single axis cmm that are basically a height gage with height master accuracy. Its not a trivial exercise, but if you have the skill to make the part it won't take long to acquire the skill to measure it.

Machinistrrt's comment about Jo blocks is another method but you can spend all day swapping blocks when you are trying to work to tenths. And then you end up trusting your indicator for the last few tenths, which muddles up the accuracy and traceability of your work.

As far as your question about manual machines built as a cmm, yes, Moore built a line of them. The measuring machines were made larger than the bores and grinders. IIRC there was one based on the grinder/bore frame and a couple of larger gantry machines. Someone on this site tried to sell one for to much money a few years ago, and then had to come back and gloat about scraping it later.
 
I might be thick, but I think true position comes into play when you must move the part, by reason of the way you have to machine the part, and are forced to disturb the initial setup. Suppose you must counterbore a hole from both sides, and both counterbores must be within some small spec of true centerline of the main hole, then true position has some value. But just true position of a bolt pattern, why is it going to be any worse than your machine accuracy? And if you discover a problem what are you going to do, fix the machine (as you should have) or write a special program with features 'off position' to try to make them correct in true position?
 
write a special program with features 'off position' to try to make them correct in true position?
That's exactly what is needed, or possibly adjust the fixturing or tool wear, or even the whole method of machining that feature. Think about a hole tied to an outside corner of the part. The center of the hole must be within .xxx of the theoretical corner of the part. If the outside corner and the hole are done in the same setup, you still have tool flex, part flex, tool size (wear offsets), machining methods (spring passes, etc. on the corner and hole machining methods) and clamping pressure possibly distorting the part so when it's in a free state (sitting on the bench), it may measure differently than when it was clamped in the machine. Then there's still parallelism, cylindricity* and straightness to deal with on a hole like that. I hate it all, but it is what it is, as they say.

*I think I just made up a word. :)
 
Concerning the height master suggestions, I have one problem with that: They cost $3,000 from Mitutoyo, where a Mits QM height gauge that has a listed accuracy of .00011" is roughly the same price and performs a myriad of operations and calculations to speed things up. It's not as accurate, but in my shop in particular, which is heated and cooled but not remotely precisely, .0001" accuracy is plenty.

And just for the record, I'm going to no-quote the parts with .001" true position hole callouts. This discussion will definitely help me in the future when it comes time to dive into parts like that though.

Thanks guys.
 
One should never confuse machine accuracy and repeatability.
Machine tools are very repeatable , they are never accurate even when calibrated.
There is a reason measuring machines are built differently than metal cutting tools.
Two very different design worlds and I would hope most here would get this.
Real dimension measuring of a part on the machine that cuts it is garage shop level work.
Bob
 
Last edited:
Matt, you may have already gotten this but there are a fair number of older CMM's that were manual only. They had switches to turn on or off the axis air bearings and were moved by hand to touch off. I have no idea how accurate they would be but for folks like you and me who dont make things in large numbers I would think they would be more useful than one that runs off of software. If you only wanted to measure a few points or basic relationships like you have discribed I am sure that would be the way to go.

I know you dropped this job but if you ever get one let me know how it works out for you. I have been thinking about them for a while but as I have no prior experience with them and dont know anyone here who has one I am a bit reluctant to take a chance myself.

Charles
 
Charles, Mitutoyo actually makes one: Crysta-Plus M443/574/7106 Series 196-Manual-Floating Type CMM

I have no idea what the cost is, but I have a strong hunch it's more than I'm willing to spend! Probably like you, it would take me many, many years to hit ROI on something like that.

After spending a lot of time today poking around, I really think one of the advanced height gauges would be the most universal purchase for the parts I tend to make.
 








 
Back
Top