Gantry mill

[jsheerin]

I did some more thinking about my variation this morning, and I think I'm already doing really well on variation. I'd say I'm doing pretty good if I get each reading within +/- 0.2 arc seconds. But because the height measurements those readings translate to are cumulative, that means I could be anywhere between +/- 0.000068" after 70" of measurements. If I then reverse and measure back towards zero, that would potentially double the variation to 0.00014". So the fact that I'm already getting less variation than that is really pretty good. This also helps explain (I think) why on my last set of measurements where I went from near to far twice instead of down and back I saw more variation towards the far end of the straight edge.

I think the conclusion from this is that improving the resolution of setting the autocollimator crosshairs to the target could ultimately be necessary to get to my goal, and that may or may not be possible. Reducing the variation at each position towards this level is still my primary goal, but I'm already there part of the time.

[ad_bfl]

Hi

Have you tried taking readings from the other side/direction? Meaning switch ends and see if it mirrors the readings.

What about running it against one of your other straight edges and see if you get the same variance?

Also, since you have FEA capability, have you ran a FEA sim where you add a specific weight to the straight edge and see if the simulation matches measured?

Pittsburgh would be a bit of a drive. BTW: you a product of Carnegie? Me thinks your a bit over the edge doing this in your home shop

and I mean that in a very positive way....

Best Regards
Al

[jsheerin]

Thanks for the good suggestions, Al. I will give all of those a try, although the other straight edge might have to wait a while.

Thinking about turning the straight edge around, I also thought that maybe my support at that end is preventing the straight edge from moving as if it was simply supported. The support clamps the top of the straight edge a bit (not actually by design). So I wonder if as I move the mirror along and straight edge deforms ever so minutely if the support is preventing it from movingly freely and then it unsticks itself suddenly and gets a lot of movement. I wouldn't think that would cause large enough errors to matter from my previous fea, but maybe...

I grew up in Indiana and went to Purdue. I work with a lot of people who went to CMU though.

John

[ad_bfl]

One more crazy suggestion, do the same measurements except with the straight edge lying horizontal to negate gravity.

Make up some supports to hold it horizontal, gently clamp the retro-reflector to the edge and a known spot on the straight edge surface to understand what clamping does to the system. Then see if it repeats.

Question, what do you think is the "working" resolution of your system? I have read the posts and wonder what you feel the effective resolution will end up being now that you have had some time to play with the tool.

Best Regards
Al

[jsheerin]

Al, check out post 279. I simulated supporting the straight edge on its side. Gravity still has an effect without being able to completely support it like you would on a surface plate.

I'd say being repeatable to 0.2 seconds is definitely possible at the moment. I'm going to try to quantify this a bit better the next time I get to play with this - probably this weekend. The most precise way to use the equipment is in a setup where you don't move the autocollimator or the mirror once in position, for example measuring the rotation of a rotary table. The mirror obviously still moves relative to the autocollimator, but you're not actually picking up the mirror and repositioning it. As far as resolution, I can definitely read the micrometer to 0.1 seconds, and probably to 0.05 seconds. There are marks every 0.2 seconds, and it's easy to judge when you're centered between two marks. I've been recording 0.05 second increments when the reading is not on a mark but also not in between two marks. This is also pretty easy to see. But I do need to quantify a bit better how precisely I can set it. It might be that I'm just fooling myself about the resolution by recording down to 0.05 seconds.

[ad_bfl]

Hi John

Any results? I am expecting a full report tomorrow.

BTW: I am attempting sarcasm so please do not be offended by this, not many people do what your attempting in your house.

Best Regards
Al

[jsheerin]

Hi Al,
I didn't do much this weekend, but I did try measuring deflection with a known mass (the mirror base). It seems like I'm off by a factor of about 2 somewhere. Putting the mirror at the end of the straight edge, I measured 0.00018" of deflection above the handle closest to the mirror and 0.00024" at the end of the straight edge. From fea I get that the SE should deflect 0.0000018" above the handle and 0.000026" at the tip, but this does not include the support I have under the SE. So if I assume that at the handle I'm seeing my support for the straight edge compress and subtract out the difference between my sim and measurement, I get that the tip of the SE dropped 0.000062" without the effect of the support, or a bit more than 2x what the sim predicts. I've been looking for a mistake in my sim to account for the factor of 2, but I haven't found one so far. I've also checked all my units, as that's also close to 2.54, the conversion between inches and cm, but nothing. I'll have to let it bounce around in my head some more and take some more measurements tomorrow.

[jsheerin]

I did some experimenting with my autocollimator today. I am not as repeatable as I thought at setting it. I took 30 measurements without moving the mirror, and my measurements were in a range of +/- 0.775 seconds. However if you apply a bit of statistics (most of which I've forgotten, so anyone who's good at this feel free to chime in), I'd be +/-2.23 seconds for a Cpk of 2 which would give a very high confidence that I'd always fall within that range. I thought I could improve things if I could zoom in more on the target pattern, so I modified my camera mount a bit to get it a bit closer so the image would fill a bit more of the TV screen and readjusted the lenses a bit. I didn't get zoomed in much more, but a little bit. I also changed my method to zero on the top of the first circle instead of the top of the center dot. I took 30 more measurements and my results improved to +/- 0.55 seconds, or +/- 1.79 seconds for a Cpk of 2.

I haven't crunched these numbers any more than this, but going from what I said in post 289, I'm doing pretty good to get the repeatability I am when measuring the straight edge, so I might be stopping at this point. I'll have to think about it some more.

[jsheerin]

To follow up my previous post, if I can only repeat to +/- 0.55 seconds, that would translate to +/-0.00019" of error after 70" of measurement. However, I seem to remember from school that more typically, you'd see 0.707x that error due to the random nature of the error (sometimes it's +, sometimes it's minus, and it's unlikely to be all + or all - which would be necessary to get the error I just quoted). So that would give me +/- 0.00013" after 70" which is about what I get.

The one possibility I see for getting better repeatability is to increase the magnification more. Currently I can't visually discern exactly where I'm setting the cross hairs with more precision than what I'm getting now. I know just about nothing about optics, but I might try playing around my camera a bit more to see if I can find a setup that does give me more magnification. I know Davidson does make a 24x eyepiece and a 12x eyepiece for my autocollimator. I'm assuming I have the 12x, but I don't actually know that. That's one of those things where I'm afraid to ask the price though...

[jsheerin]

I spent a little time researching eyepiece design today. I found a method to measure the magnification and focal length of an eyepiece here: Check your Eyepiece Focal Length with a Ronchi Grating (long) - sci.astro.amateur | Google Groups
So I did that with the eyepiece off my autocollimator using a piece of plastic that I marked two lines on with a fine point sharpie and a flashlight. I measured my lines being 0.097" apart on the plastic, 0.37" apart on the wall when in focus, and the plastic being 4.77" from the wall. That gives me a magnification of 0.37/0.097=3.814x, and then the focal length is 4.77/(3.814+2)=0.82" or 20.8mm. The focal length of the autocollimator is given as 508mm by Davidson on their website, so from here: Eyepiece - Wikipedia, the free encyclopedia the total magnification is 508/20.8 = 24.4x. So I guess I have the "24x" eyepiece instead of the 12x one. I would have taken a picture - it was kind of cool to do - but I already had my hands full with the three parts plus a set of calipers. So from M=fo/fe, if I change fe to, say, 5mm instead of 20.8, I'd get about a 4x increase in magnification. So I bought some eyepiece parts to try to do that with from here for about $20: Surplus Shed. If I can make something work with this and no other sources of measurement error are present (unlikely, but I can hope), then I could get close to the repeatability I need to flatten my straight edge enough to make it useable for making hydrostatic bearings with the gap I currently have planned. And by the way, in case it's not clear, that is the whole point as to why I'm trying to do all of this tweakery with my straight edge. I'm shooting for a bearing gap around 0.0003". This means I need a rail that's flatter than that, ideally by 10x but I'm targeting better than 4x flatter. To measure the flatness of such a surface, once again I'd ideally be able to measure to 1/10 that level but I will be (very) happy if I can get to 1/4x.

If I can't do this, then I'll be forced to do something different for my bearings. I was looking at larger bearing gaps and oil for the fluid this past week. Another option would be to make conventional sliding ways, but that would limit me a bit on speed and acceleration. There would also be wear.

[jsheerin]

For anyone following my design, no, I haven't stopped working on it. I've been taking a break to get my motorcycle ready to ride this summer. It needed new timing belts plus a new valve shim which I made on my lathe and then hardened and polished. More recently I've been working on building a data logger to record engine parameters so I can do some engine tuning. I'm getting close to finishing that up and then will get back to the mill.

I found a cheap load cell supplier today: Only Weigh Scales
These are $89 compared to ~$400 from Omega. I'll probably pick one these up to help with measuring the stiffness of my bearings. My electronic indicators have an output on them, so I'm thinking I might be able to record displacement and load signals together to get a nice set of load vs displacement data.

[jsheerin]

Here's an update for anyone interested. After I finished working on my motorcycle in April, my car broke. I rebuilt the cylinder head and did a lot of other work on it through the summer, finishing a few weeks ago. Then I got a new job which required me to pack up all my stuff and move across the country. It will be another month or so until I get all my stuff back and then probably another few months until I've sorted through things to get room to work. That said, I probably will not build as ambitious a machine as I was originally planning here. I still have all the parts, but I am now a bit constrained on size, noise, and portability so I'll probably try for something more compact. Other than that, I haven't thought a lot about it recently other than to protect all my equipment and parts for the move.