Machsupport Forum
Mach Discussion => Video P*r*o*b*i*n*g => Topic started by: Bob La Londe on January 27, 2010, 12:10:22 AM
-
I bought one of zarzuls cameras with the shaft mount. I had a heck of a time trying to adjust it last night. Never could get it as close as I wanted. I studied it a bit after work today, and decided to make the two side screws set a little more uniformly off to each side of the camera. It took me about 15 -20 minutes after that to get it zeroed in pretty darn good. It bisects the tiniest construction lines I could draw on a piece of paper as perfectly as I can see to do at a range of about .375 from the work surface. I can see spatter of the graphite on both sides of the cross hairs at 0, 90, & 180. Looking at the gap in my caliper and comparing it to my lines I'm guessing my line width right at or a touch over .001. That would make this setup within .001 plus machine lash and spindle runout.
Its pretty cool.
Then just for the heck of it I setup some clamping hardware on the mill table. I set a pair of perpendicular stops on the table at one end using my old square and rock method. Then I set my small vise on the table using the camera to zero it. Wow its fast. Set one and and put the cross hairs on one end of the fix jaw. Set soft zeros at that view. Zip to the other end of the jaw and adjust the vise position to match. Hit goto zero and if its still on the mark its square. I had to go back and fourth about three times total, but it was super fast. Way faster than using a mechanical edge finder. Atleast for me.
It also showed me I need to surface my fixed vise jaw. LOL.
Its pretty cool as an inspection camera too. Shows all the imperfections and irregularities that you normally think are perfect even when using a magnifying glass.
-
Compensation for the video camera's axis alignment is something the software ought to do to make it easier. You should be able to set the cross hairs on a feature. Raise or lower the Z a determined amount, rezero the cross hairs by jogging, press a button, and be done. The geometry is pretty darned simple and would be easily compensated by the software.
I'm supposed to be a beta tester for Tormach's upcoming video prober. It has some real special software with it, and that software will also be available separately as I understand it. I don't know if it will do the alignment compensation or not, but I'll be on the lookout and I'll write about the whole thing on www.cnccookbook.com when the beta test is done and they've made all their improvements.
I'm really looking forward to it.
Best,
BW
-
I didn't see that in Mach 3. Am I missing it? Mine is mechanically aligned with the spindle now, but I can see it changing and needing to be rechecked periodically with normal bumping and bouncing.
-
For squaring it wouldn't even need to be aligned. Just for accurate edge finding and point location.
-
No, I don't think the software does do it, I'm suggesting it'd be a worthwhile addition.
BW
-
Hmmm... I was just re-reading your comments and I'm not sure I understand how that procedure could zero it mathematically. It could tell you if your Z-axis post is straight (perfectly perpendicular) in relationship to the table though. If it was an auto focus camera or it was a little easier to refocus than the cheap CMOS board cameras usually used for this sort of thing it might a great tool for Z-axis alignment to the table
For aligning the camera to the z-axis you rotate the camera. You might be able to do it with software by zeroing to a refference mark, rotating the camera 180 degrees rezeroing to the refference mark and then dividing the distance and moving the crosshairs. I do not see how you could do that by moving the camera vertically though. Maybe I misunderstood.
-
The camera is looking along some axis that is not concentric with the spindle. Therefore, as you raise and lower the Z (and let's assume it is concentric or we'll go crazy--the machine needs to be properly trammed), the image is going to shift. It's pretty simple trig to figure out from the direction and magnitude of that shift what the axis the camera actually points to is. Once you know that, you can compensate for it at any given Z height.
Cheers,
BW
-
Seems llike rotating and splitting the distance would be a lot simpler assuming the spindle post is true.
-
Well but you don't know how much you rotated or in which direction. Whereas moving the Z you know exactly how far it moved so your compensation can be both automatic and very accurate.
-
Well but you don't know how much you rotated or in which direction.
You will if you look at the cross hairs or look at the spindle.
Very similar principle to spinning a DTI in a hole, or or you saying you cannot centre a hole using that method ?
Phil
-
I'm supposed to be a beta tester for Tormach's upcoming video prober.
Were you able to test the video prober? I couldn't find any mention of it on your web site and see that Tormach is taking orders for them now.
Mike
-
Mike, I haven't gotten to it yet due to that little hard disk crash. It's coming. I will likely evaluate how well it is able to map and dimension my Colt 1911.
Cheers,
BW
-
Hey - can you post your camera mount and modifications?
Maybe a picture or something.
I have both a web cam and a USB Microscope that I'm going to try to create a mount for.
Thanks for any help,
Joe
-
Don’t know If anyone is interested. This is the result using a 208c cmos cam and a usb easy cap. pic is a 0.004 feeler gage.
-
A
Don’t know If anyone is interested. This is the result using a 208c cmos cam and a usb easy cap. pic is a 0.004 feeler gage.
Actually that is a very useful photo. It might also be a good suggestion for how to use a video edge finder for greater precision. Use a known thickness gage like that to get the scale of the projected line before proceeding. This way you know the tolerance you are setting for your camera to your part. Obviously the part edge and the gage edge would need to be close to the same height from the table.