Keyence Optical Measuring device

We have the same model.
Sure, if you are talking about a calibration artifact.
Dealing with real parts with imperfections.... no.
The more complex (or imperfect) the profile, the more variation you will see.
Thickness of the part can play a role too because you have to worry about focusing issues.

Here is a user forum, but it is VERY low traffic.
http://www.imusers.com/

Has your Keyence been calibrated to certify accuracy? We have a Keyence VHX-1000 microscope that will make measurements but it’s never been calibrated for that. Some people still use it for measuring. Sometimes a fixture is placed on the stage so a part can be tilted to see the area of interest. The digital camera can also be rotated ±45° to better view the part. These angles are “eyeball” accuracy and are used for alignments to measure dimensions to a micron.

I just use it to photograph magnified areas of parts.

I demoed a 6120 for a while - it's a neat machine that can be very accurate and easy to use. With that said, there are limitations to what a telecentric vision system can do, just like any optical measurement system. What you are trying to measure has a big impact on the performance you'll actually get out of such a system.

The 6120 absolutely shines when it comes to measuring thin, flat stamped parts, such as connector components. When dealing with prismatic or turned parts, you can start to run into issues with surfaces that are parallel to the line of sight. If you set a turned part on a conventional optical comparator with the part center line perpendicular to line of sight, you'll see very sharp, defined edges on the turned diameter, but the flat ends of the part will bloom out into a blur as you move away from the edges. On a telecentric system, such as the Keyence, you won't see a bloom on parallel surfaces but rather only the high points, plus error from any mechanical misalignment of the part/stage to the line of sight. Measuring end-to-end on turned parts is iffy on the Keyence, and measuring side-to-side on tall prismatic parts is very unreliable.

We have one that is used quite extensively. I believe that it is quite repeatable and accurate for distances and diameters. But along with that, any imperfections in the piece will cause some variation. I have found that very small radii, usually under 0.010-0.015", it can be a bit unstable. I try not using it for GD&T too much, especially profile. Any spec of dirt or scratch in the lens, could potentially throw this off. True position has limitations as it cannot do MMB on the datum, just MMC on the feature.

We mainly use it for our flat, stamped parts.

I have found that very small radii, usually under 0.010-0.015", it can be a bit unstable.

That's definitely another area that can be problematic, depending on the application. Vision systems with digital cameras use sub-pixel algorithms to produce measurement resolutions that are easily ten-fold the native resolution of the camera, but they become increasingly unstable/inaccurate on features that approach the size of the native resolution.

We have a few telecentric vision systems that work on the same principal as the Keyence, just with a horizontal line-of-sight that is ideal for turned parts. The largest of them has a resolution scale of about .0024" per pixel, but it can easily measure turned diameters to within .0002" of conventional tools. However, it struggles with, say, a .010" corner radius in a groove since it is trying to extract the feature from a 4x4 block of pixels.

I don' have too many issues with our machine. I find the more attention I pay to ensuring the depth of field is set to the part edges I need to analyze the better the program runs. This also seems to assist pattern recognition. The manual says it will do a half micron in HD or a full micron in wide. Since most of our stuff is turned and we often have burrs to contend with i doubt its good to that level.
Perhaps to 0.001" for some things though.
rickfngp: I do not know if there has been any type of certification. The rep was here a while back and we asked if it required calibration and he said no.
Thanks!

rickfngp: I do not know if there has been any type of certification. The rep was here a while back and we asked if it required calibration and he said no.
Thanks!

A measuring system that doesn't require calibration? That always sounds very suspicious to me.

rickfngp: I do not know if there has been any type of certification. The rep was here a while back and we asked if it required calibration and he said no.
Thanks!

A measuring system that doesn't require calibration? That always sounds very suspicious to me.

A measuring system that doesn't require calibration? That always sounds very suspicious to me.

of course not. it was most likely calibrated at the factory and is similar to a 'lubed for life' bearing......

of course not. it was most likely calibrated at the factory and is similar to a 'lubed for life' bearing......

Well, in theory, there's nothing to the design that is subject to mechanical change or wear - it's a fixed telecentric lens on a simple industrial digital camera, so there are no moving parts. Once the "pixel pitch" is set, there's no reason it should change (short of the machine being disassembled, dropped, etc.) The only other step would be correcting optical aberrations in the lens, which usually achieved capturing an image of an optical chessboard standard and compensating for the distortions, but this should only need to be done once, just like teaching the pixel pitch, since glass isn't actually a liquid.

That said, other systems that work on the principal do come with standards to allow you to "re-calibrate" or reteach the pixel pitch. I believe OGP's SNAP comes with an etched glass standard similar to their normal vision systems, and my OASIS machines have each come with two different Deltronic pins and a sled to allow them to be positioned for teaching the X or Y pitch.

(Shortly after getting our OASIS machines, our calibrations guy asked me how often they need to be calibrated, and I forwarded the question on to the manufacturer. Their reply was, "technically never, effectively as often as you want." Our calibrations guy decided a "Calibrate Before Each Use" sticker was appropriate...)

As for Keyence, their background is process controls and inspection, which - rightly or wrongly - is not footed in the world of traditional metrology that we're used to. Their lineage is, for example, cameras mounted on a process line for inspecting completeness of inkjet printings on food packaging, or continuous visual two-point diameter inspection of a copper wire as it is being drawn (because, hey, if the hole in the form making it was round, the wire should be round, too, right? Sounds like machinist talk.) This is why you don't see any commonly accepted standards to which their machine was calibrated, or even how its OEM accuracy spec was determined. In other words, they are not from "our" world

I never got to where I was realistically considering the purchase of a Keyence IM, so I never got around to badgering them about calibration to ISO/ANSI/ASME standards and whatnot.

Sorry about the long post, I do find this stuff interesting. I've been engineering a vision system for our own purposes, namely locating parts on a conveyer for a robot, that works on some similar principals - http://i.imgur.com/M6MDDj4.jpg