Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Lenstip drops a bombshell
And an expensive failure of one at that...


Here they have published a new article on the sony A99 mirror:


They go into several areas they would have been all the wiser to stay out of... first they have a test flat, about 6" diameter I would wager, and they ruin it for lab use by placing something directly on top of it.  This is a $10,000 part...  Certainly I assume they will continue to use it, but no good lab would ever use it because it is no longer flat due to scratches on its surface which will perturb the fringes.


They also go on to note fringes seen in the mirror with the flat... yes they would exist!  You can place two test flats on top of each other and get such fringes even if they were polished to 1/500th wave roughness.  The minor changes in the airspace produce fringes regardless of how flat the surfaces are.  There is spherical aberration as well as astigmatism in the fringes of the mirror placed on the flat... it is clearly at least somewhat warped and also not perfectly flat on the test flat.  In essense, that observation says nothing about the roughness and it certainly is not on the order of 1um - such a high value would appear very noticeably matte/opaque and not at all "clean" or sharp as the mirror actually does.  1um roughness corresponds to ground, not polished glass.


Their (homemade!) interferometer appears to be on a sheet of plywood, thus any alignment is not scientifically valid at all.  It also is not in a controlled room and vibrations as well as air currents will disturb the results. 


Their reference field contains fringes and very few of them, so the interferometer is useless for precise measurement.  Anything finer than 5 waves or so I would say is invalid using it. 


The disagreement between horizontal and vertical tilt in their data also would invalidate their findings.  Any axis of tilt should produce equal results for a flat, certainly not a difference of a factor of 2. 


I would also question who and how designed their diffraction limited ultrafast lens that covers such a large field.  Such a feat is not impossible, but is very difficult.


The article is useful for generalizations only, it is not scientifically accurate.  Please tell anyone who shares it with you to not take it as gospel, the test is magnifiscently flawed.

Howdy all
I am just nev below, along with Let me point out howdy to all aged users: ).
Hi Sythels,

                  Frankly, the whole thing went over my head technically  (no surprise to you guys), what did surprise me however, was for all that techno, the conclusions reached were not anything that many non scientific reviewers had already stated, I was expecting real world resolution comparisons or some tangible discovery!


  However I like Lenstip's style!



Howdy DanPr, welcome!

It's really quite a shame, but understandable.  I just question why go through acquiring some optical lab equipment (namely the HeNe laser, beam expander, staging stuff, beam splitters, etc) the beam for which would be $35,000+ and set it up on plywood.  The Air-ride tables we use to isolate experiments from vibrations are more expensive than that in and of themselves, but 1/4"-20 4" breadboards large enough to set this experiment up are under $1,000 and would improve the accuracy (and legitimacy) of things tremendously.  The air conditioning and filtering necessary to clean up thermal interference would be far outside their budget, but many of the things they have require contact/ going through a supply chain instead of anonymous ordering - I question why not ask for some counsel before spending so much only to get 25% of the way there.


It's still interesting, but it would have been useful to do a better test.


I mean, renting lab time and having access to an 'actual' interferometer like this puppy: would have been so much more useful.  They could have measured the surface roughness to the nearest 1~5nm depending on the lab/interferometer, quantified the aberrations from the mirror, and even tested mirror+lens vs a reference flat and then computed out the spherical aberration induced by testing non-flat systems on a flat.


But, I guess they have begun stocking some lab equipment now... there are still hundreds of thousands of $ left to go before they can produce a legitimate setup, however.

Post removed



Hi Scythels

      I replied to a spurious  "sales pitch" post which appeared from nowhere here which Klaus "removed", that left my comment floating in mid air, hence "post removed",  no connection with your valuable contribution! 

Ah.  I was quite confused for a moment.

I was tempted to mumble stupid things about Eastern technology, but their Sojus rockets still flew when American space transporters had to stay down. So I better don't say stupid things?


Hard to avoid. Anyway, I digested once a lenstip article about Otus lens and there were so much bad samples that I didn't click their site until your contribution, Scythels, made me curious. And now I stay absent again until somebody manages to wake curiosity again Smile

I have no problems with lenstips' general lens evaluations- I believe that for that their methods are sound.  Depending on the mount they are testing they use a D3X or IIRC a 1Ds 3 which is fine.  The problem with the sample images is that they are all shot as in-camera JPEG files with the sharpening turned all the way down.  This makes their samples look soft when really they are just poorly processed by the camera...


Forum Jump:

Users browsing this thread:
1 Guest(s)