Spin Polishing - Making a Perfect Edge - extracted from Cloudy Night's thread of same title

Posted 22 July 2008 - 01:29 PM

This subject appears to need its own thread. In the recent past both here and elsewhere (Zambutomirrorgroup posts starting at 6989 (July '06) and then again at 9553 (Dec '07)) I've described the method I use to prepare large fast mirrors that are polished out for figuring. At least a couple of people have picked up on this and used it. :)

From an earlier description (see previous thread from a few months ago) I'll repeat the description. 
Basically, inside the sweet spot (SS) the curve goes shorter, outside the SS the curve goes longer, at the SS the curve stays the same. Assuming excellent contact with the tool or lap to the mirror, that is. Grinding or polishing, the principle is the same.

For any given sub-diameter lap that's about 55% or more of the mirror diameter there is indeed a SS position. If the lap is too small you will not be able to cover the center of the mirror while offset. The smaller the lap % the smaller the overhang % is at the SS, and vice versa. 
I typically get spheres using pure static-position polishing ("spin polishing" at the "sweet spot") that null out uniformly, right off the machine. Granted, I do some simple lap prep and follow a particular protocol for time, which I've detailed a couple times over on the Yahoo Z******MirrorGroup, mostly about how to obtain a perfect edge. Overhang % and excellent lap contact is key. 
For spin polishing (or grinding) OTOH the lap (or tool) is free to follow the curve of the glass, but other than that its pivot is just fixed in space above the rotating mirror.

A picture being some K better, here's one of a mirror that's undergoing final polishing prep before figuring today. This is a fixed quill (static position) polishing machine (Derek) doing edge preparation on a 14.7" f/4.4 fused silica mirror. The aft end of the overhead channel pivots via an automobile ball-end joint. The fore end is clamped onto a reinforced support table, and that's what takes the force load (which can be quite high). The table spins clockwise at about 42 rpm, the lap is AccuLap medium on an epoxied concrete base, and there's 25 lbs of weight on it. 

to be continued

Attached Thumbnails

Posted 22 July 2008 - 01:32 PM

And here's a cropped Foucaultgram of the above mirror now ready for figuring - this image is with the KE well advanced and using a 50 micron pinhole source, it shows the condition of the edge. An image like this is a definitive test for edge condition, with the KE advanced a little beyond the center of the mirror. 

Note particularly how the first diffraction band (which is about 2/3 wave from the actual KE shadow) just tracks straight to the edge. The small scale ripples right at the edge are diffraction ringing off the edge. The mirror doesn't need to be spherical to start figuring, very smooth to the outside with shallow correction (any conic b slightly less than 0) will do, which is what this is. ;) 

It shows around 1/4 wave, I get maybe -.07 for the conic. There's a certain amount of turbulence in this photo, it's an average of several video frames taken over about 30 seconds, in the dark with a pretty dim source. Most anything imprinted by the tool towards the inner half of the mirror vanishes during figuring (but see next post). Note that this is not true of defects near the edge! If you don't start with a smooth edge without artifacts, you won't end with one, if using normal sub-diameter parabolizing techniques (from the outside in). 

For those curious, this parabolized mirror will have 3.54 waves of correction, compared to a sphere! 

to be continued

Attached Thumbnails

  • 2534648-edge finish.jpg

Here's an update 02-09-18:  

DIFFRACT  simulation (almost) of the preceeding image - it won't go quite as far in from ROC without aliasing as the actual image but it's very close.  Brightness and contrast is adjusted to more closely match the actual image, and you can see all of the edge effects present there coming out of the simulation as well.  The conic in the first image is obviously greater than the -.07 I used in DIFFRACT though.

DIFFRACT simulation of previous image

Posted 22 July 2008 - 01:47 PM

Detail on how to do the lap prep for spin polishing:

Spin polishing, I use a 75-80% lap most of the time but it works with smaller or larger laps. The lap is held in a fixed position so it can only rotate with around 15% overhang (diameter). This is the easiest way to grind by machine and it works just the same for polishing as well.

You need excellent contact while you're doing this and it's mostly useful to finish a mirror that has the typical rolled edge that comes from polishing out with a spinning lap over long sessions.

Wire brush the lap under hot water to get a roughened surface. You don't need much in the way of faceting for this, as it works mostly from the scratches. I channel the lap while avoiding the center and angling each channel so it doesn't put zones on the mirror. The pic is of an AccuLap lap made that way, but channeled for extended polishing. For finishing the channels would be narrower and allowed to almost close.

With a generous shot of slurry on the mirror, press it briefly while the lap is still hot - use a lot of force, and then check there's good contact. If there isn't do it again until there is.

Put it onto the machine. I use a lot of weight and try to set it up with the overhang and start it going quickly. That's so there's no chance for a groove to set in from the overhang position - this will kill the mirror's edge. Run it 15-30 minutes, less time as you get closer to a good edge. Very important point - the lap must turn just a little slower than the mirror, which shows proper contact. Faster than the mirror or a lot slower, it's not working, so take it off and try again.

Once the time expires, take it off and redo it. You can check it if you want, but it will push a good sphere slowly out to the very edge with anywhere from 2-5 reps in most cases. It's the repeated establishing of the good contact that does it, and the lap only spins that way reliably until the microfaceting from the brushing gets smoothed out.

No particular mystery about it but you really have to experiment a bit to see what happens and then work with it.

Caveats and more blather:

I need to emphasize that I'm talking about a very specific technique in the final stages of spin polishing. To produce a perfect edge I'm not talking about just putting the lap on and letting it run for hours, I'm talking about a carefully prepped lap set at the overhang position where it raises the typical TDE coming off of polishing out or extended polishing. And you must check the results with the KE to see when to stop applying it, but that's not hard. Be aware that it's only the combination of these factors that gets to the results documented in the earlier thread, and I've tried to be careful to describe it in detail each time I've described it. 

A summary sort of:

Perfecting the edge is a perennial issue in prepping large mirrors that everybody encounters (see this post. ). I'm not sure I can say much more about it, the method is very simple and reliable. It does what it's supposed to and it does it extremely well.

Attached Thumbnails

  • 2534694-periodic spin polish example.jpg

Posted 28 July 2008 - 09:15 PM

I'll see if I still have a corresponding Foucaultgram for the f/4.4...no, sorry, only that one edge image. It was well inside ROC BTW. 

Here's a series of photos including uncropped Foucualtgrams on a 14" f/4.66 .8" quartz mirror that I'm making for myself. The mirror was partially figured last year but it had a problem with a lagging edge and I started over on it. I ran it on the polisher to remove the correction (spin polishing) until it was mostly a decent sphere with a rolled edge (this again is typical). The next few steps take care of that condition, and will raise the edge if continued too long. 

This is the surface after the first 20 minute session (in all of these there are reference Sharpie spots used to align the video Foucautlgrams in Registax). You can see swirl marks from the lap:

Attached Thumbnails

  • 2547509-1st session 20 minutes surface.jpg

Posted 28 July 2008 - 09:22 PM

Here's the surface after an additional 30 minute session as detailed in the first part of this thread:

Attached Thumbnails

  • 2547516-2nd session 30 minutes surface.jpg

Posted 28 July 2008 - 09:23 PM

Here's the edge of the mirror after the 2nd session:

Attached Thumbnails

  • 2547517-2nd session 30 minutes edge.jpg

Posted 28 July 2008 - 09:37 PM

It's looking pretty good, without any swirl marks, and the diffraction edge shows well in the Foucaultgram, but it still has somewhere around 1/8th wave TDE in the last quarter inch. The mirror could be figured from this state with little difficulty, but I'm too picky. So I ran it again and this time did an experiment with a new half-micron cerium sample (SRS426) but it showed no difference. Went back to the usual compound with the usual brief lap prep ran a 25 minute session - same lap speed all the way through, just slightly lagging the turntable (I think I mentioned this is important, as that's what telling you it's enforcing a sphere on the mirror). Here's the result:

Attached Thumbnails

  • 2547537-4th session 25 minutes surface.jpg

Posted 28 July 2008 - 09:41 PM

You can see marks about 1/4 way up from the bottom that were left by the rinse water as it dried, and a few mild zones about halfway out. These are of no consequence for figuring and will disappear after a session or two. The image is taken at almost an exact null. Here's the corresponding edge image. It shows a slight undercorrection corresponding to the previous Foucaultgram, the amount of that measured by the first fringe (which lies about 3/4 wave from the shadow) is about 1/10th wave:

Attached Thumbnails

  • 2547542-4th session 25 minutes edge.jpg

Posted 28 July 2008 - 10:01 PM

You can see in the shadow edge what looks like a slight hook. That close to the edge diffraction ringing off the edge itself adds spurious detail. If you examine the first fringe instead it tracks straight, and I've spent time in the past trying to eliminate that apparent defect. 

I'll run another session on it just to see what happens :mrevil: 

Keep in mind (please!) I'm not suggesting people adopt this method, nor am I making any claims about it being superior or smarter or more efficient or pretty much anything else. I'm documenting something that works and works reliably for me in a production environment, and that has been adopted by a few other people I know of who had trouble getting a better edge. 

Also note that the way I'm working overall I don't care about or want correction put on the glass when it comes off of polishing prep. It's easier to verify a sphere than any other surface, and using a figuring technique that works over the whole mirror has no surprises when starting from spherical. For my work and robotic methods, figuring is an entirely separate process from polishing prep, with different laps, different methods, different machines, different strokes and different controls. The polisher only does one thing, polish. If I had problems with the polishing prep I would look for alternative methods to produce the original surface... 

Glad some people find it interesting. :grin: 

Posted 15 November 2009 - 03:35 PM

OK, returning to an old thread. A couple comments above I don't recall seeing before...

Which brings me up to the present, and a new example to show. For the thread vault here's a link to a YouTube video showing what spin polish finish prep looks like on a 14.7" f/4.5, with a 12" polisher under 25 lbs weight (1/3 lb per square inch working pressure). 


This sequence ran essentially unchanged for about 30 minutes, after which the surface shown in the next post (if there is one) resulted. 

Posted 15 November 2009 - 04:37 PM

Here's a shot of the edge of that mirror - actual size as taken (though cropped) - it's with the KE past center and well inside the paraxial COC as discussed in Texereau, illumination from a 50 micron precision pinhole. Not much to not like. ;) 

Attached Thumbnails

  • 3449947-spin example.jpg

Posted 17 November 2009 - 02:17 AM

I just worked with the results from the process as it evolved - knowing that there was a sweet spot that keeps the ROC the same without strokes was the starting point. ;) 

Note the picture of the polisher faceting early on - this works against zoning, which I'd be concerned about with concentric rings - although I think I see what you're saying. I have found that whenever the facets tend to line up concentrically a zone, usually narrow, appears at that (radial) spot on the mirror, and I work to minimize those. To me the faceting is easy, and a small price to pay for the simplicity of the overall process. 

PS Slightly edited for clarity...

Posted 16 February 2010 - 08:27 PM

Some time in this thread the question arose as to whether the spin polishing method works for long f/ratio mirrors as well. Yes, it does. Here's a shot of the edge condition of a 14.7" f/6.67 (FL=99") mirror after polish prep - the mirror isn't clean, but you can see the important details. Same technique used to illuminate it as before.

Attached Thumbnails

  • 3629237-14.7inch_f6.67.jpg
Posted 22 August 2014, 09:07 PM

Here's an update based on recent work...

This is a 20" f/6 that needs about one more 20 minute session to be ready to figure.  Here's a handheld shot of the edge condition, with air currents and all, take with a 50 micron pinhole illuminating a KE that's pulled well ahead of ROC, as previously described.  These diffraction fringes are about 2/3 wave apart. Although the detail doesn't compare to the more elaborate image capture earlier, the mirror appears as a near perfect sphere with only a tiny amount of TDE left to correct.  The last session will always do that. 
Here's what the polisher looks like, it doesn't actually need to be scored again, so I won't do that now.  I adjusted the image to show the score marks on it easier.  They are done using a razor blade, spaced across the surface and angled at 120 degrees.  The polisher has no facets at all to start with, it's standard grade Aculap on an epoxied cement base. 
Once scored (or microfaceted) it lasts for several 20-30 minute sessions.  Each session is preceded by wire-brushing under hot tap water (hot enough to soften the pitch slightly).
Looks like this when ready, wire-brushed in a couple of directions.
And on the machine, with the overhang set at neutral ROC.  Only a small polish charge is needed for each session, and if you use too much the AcuLap can load up and stop working correctly, requiring resurfacing or melting down and repouring in order to restore the correct action.
At neutral ROC (the overhang) the curve stays the same on the glass.  What is more, by repeatedly prepping the polisher and running it with good contact for 20-30 minutes, the entire mirror takes on exactly the same curve overall, starting on the inside and gradually working out to the extreme edge.  It's always rewarding to see the sphere develop from such a simple process.
The polisher needs to have low rotational symmetry in order to leave no zones or ripples on the glass.  There will almost always be a small residual error right near the center, where the inner edge of the tool sits, but it's of no consequence in practice.  Thanks to the prep on the tool and the method the mirror ends up with excellent smoothness and regularity for figuring.  I've used the same technique from f/3.5 to f/7 with consistent results for edge and sphere prep.

Posted 05 November 2015, 05:35 PM

Here's another example in my occasional updates, this time on the short f/ratio side - 18" fused silica f/3.57, 5 sessions as described in general tek, jail bar straight Ronchi, perfect balanced diffraction edge.  


This is a hand-held shot of the KE inside focus cutting across axis, as described earlier:


Happy polishing!

Mark Cowan

Veritas Optics

last updated 13mar17