Jump to content


Photo

Zambuto Criteria #4 on Strehl

  • Please log in to reply
318 replies to this topic

#1 Alan A.

Alan A.

    Mariner 2

  • -----
  • Posts: 268
  • Joined: 08 Feb 2007
  • Loc: San Francisco

Posted 10 October 2013 - 11:38 AM

There is extremely interesting information posted on Carl Zambuto's website here on how a high strehl can be misleading:

ZOC#4

I assume since professional institutions rely on the interferometer, that if the testing interferometer is set up correctly and sensitive enough, it could measure all scales of ripple quite accurately. However, is Zambuto on to some deeper point here about the limit of the interferometer, or that perhaps most interferometers aren't able to measure all scales of ripple?

#2 Starman1

Starman1

    Vendor (EyepiecesEtc.com)

  • *****
  • Vendors
  • Posts: 23482
  • Joined: 23 Jun 2003
  • Loc: Los Angeles

Posted 10 October 2013 - 01:29 PM

There is extremely interesting information posted on Carl Zambuto's website here on how a high strehl can be misleading:

ZOC#4

I assume since professional institutions rely on the interferometer, that if the testing interferometer is set up correctly and sensitive enough, it could measure all scales of ripple quite accurately. However, is Zambuto on to some deeper point here about the limit of the interferometer, or that perhaps most interferometers aren't able to measure all scales of ripple?


IIRC, interferometers take measurements at a finite number of points on the mirror. The larger the number of points, the more accurate the overall figure is revealed. But, to get to the really micro-ripple CZ talks about, a prodigiously larger number of points has to be sampled or another test performed. There is an inherent limitation here in that the errors measured have to take into account the errors in the testing equipment and astigmatism introduced by the testing conditions.
Plus, how many points per square inch need to be taken into account to reveal the type of error you're talking about? 100? 1000? 10,000?

So other tests, such as the Foucault (pronounced Foo-Ko), or the Lyot (pronounced Lee-Oh) phase-contrast test, which can reveal ripple down to the near-molecular level (even very smooth mirrors appear bumpy in this test) are typically performed. Or the Ronchi (Ron-kee) test.
A combination of tests can reveal in slightly greater detail what the interferogram shows.

Questions arise:
--are what other tests (like the Lyot) reveal significant if the interferogram shows a good figure? CZ would argue yes.
--How can a mirror with a high Strehl ratio not be a smooth mirror?
--Why doesn't the interferogram reveal all we need to know about the mirror?

It's interesting to note that Wolfgang Rohr, in his mirror tests, displays results from Foucault, Ronchi, Lyot, and interferogram tests to reveal the truth about the tested mirrors. Carl's Seven Criteria reveal his more stringent approach to mirror making.

#3 Alan A.

Alan A.

    Mariner 2

  • -----
  • Posts: 268
  • Joined: 08 Feb 2007
  • Loc: San Francisco

Posted 10 October 2013 - 02:12 PM

If you tested a mirror with Rhonchi, Foucault, and Lyot and all three looked excellent, would an interferometer add anything then?

#4 Sean Cunneen

Sean Cunneen

    Let Me Think

  • *****
  • Posts: 3271
  • Joined: 01 Aug 2007
  • Loc: Blue Island Illinois

Posted 10 October 2013 - 02:27 PM

An average telescope mirror is figured to a precision many times finer than what the most capable of equipment can consistently measure. That is why you have to use a variety of tests to ascertain the true quality of the optic. Numbers on this scale fail.

That article takes things a step further by showing an extreme case where the optic tricked the interferometer into a high reading. Considering most mirror makers wouldn't rely solely on an interferometer in fact most would use Foucault and Ronchi with star testing, all of which would indicate machine ripple well before you would toss the mirror under an interferometer for final testing.

Unfortunately having one number as a tell-all is a perfect marketing tool and many of us fall victim to paying based on strehl alone.

#5 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 4142
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 10 October 2013 - 03:16 PM

If you tested a mirror with Rhonchi, Foucault, and Lyot and all three looked excellent, would an interferometer add anything then?


Yes, if done correctly it would add a full surface estimate of the error (that an interferometer can see). Rhonchi isn't that accurate quantitatively, Lyot isn't really quantitative, and Foucault although capable of high accuracy across single diameters at a time, is a sparse sample and quantitatively blind to gross errors of revolution like astigmatism.

A mirror that tests well with IF and shows a smooth microsurface under more sensitive tests is likely to perform better in practice. Its almost impossible to quantify what that "better" is though - although plenty of experience tells me that smoother surfaces perform noticeably better in practice, and this makes it worthwhile to produce those surfaces and test for them.

Best,
Mark

#6 wh48gs

wh48gs

    Surveyor 1

  • *****
  • Posts: 1662
  • Joined: 02 Mar 2007

Posted 10 October 2013 - 04:03 PM

Ripple below 4-5mm in width generally are not worth measuring, except for some specialized instruments. The p-v/RMS error scales roughly with their size, and microripple (up to 1-2mm) have worst-case scenario of about 100 wave RMS on the wavefront. Plug that in to the Strehl formula and it gives you the corresponding worst-case Strehl degradation factor 0.996.

If roughness on the first mirror is, say, 1/30 wave RMS, and I wouldn't expect it is more (it's kind of hard to tell, since that photo seems to be manipulated differently than the one at right, and blurred up quite a bit), that still degrades Strehl by 0.96, or so. Still at the very limit of perception for general observing.

The other mirror has the entire roughness range within less than 1/100 wave p-v (1/300-1-400 wave RMS). That really does nothing.

Similarly, 1/4 wave p-v turned edge at 95% radius will reduce the Strehl by 0.96. Half as wide ege, by 0.98; half as much p-v by 0.99. Sure, 90%+ of mirrors are not as good as claimed, and none is perfect. But both these mirrors will perform very well.

BTW Strehl is not "the measure of large scale surface". For one, it is an indicator of wavefront quality, which may and may not be proportional to the surface error. Second, how "small" it gets depends only on the number of sampling points. Given sufficient number, it is as good as wavefront map used for its calculation.

Vla

#7 Alan A.

Alan A.

    Mariner 2

  • -----
  • Posts: 268
  • Joined: 08 Feb 2007
  • Loc: San Francisco

Posted 10 October 2013 - 04:09 PM

Thanks Mark, as usual - very informative.

Since we don't get these four types of test results from our optician, for those of us interested in understanding how well an optic performs and who don't have the equipment Herr Rohr has, I suppose it would have to be through star testing.

I wonder how they test for smoothness on small scales in large mirrors like the 8 meter mirrors at Steward if you can't do it with an interferometer?

#8 Arjan

Arjan

    Mariner 2

  • -----
  • Posts: 222
  • Joined: 21 Jan 2009
  • Loc: Netherlands

Posted 10 October 2013 - 04:17 PM

every type of test has its own merit. you don't use a micrometer to build a house, or a yardstick to adjust a spark gap.
foucault can also fool you if used improperly, even about smoothness.

#9 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 4142
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 10 October 2013 - 05:39 PM

Yeah, it's far too easy to be fooled, or to fool yourself. Foucault works well for smooth mirrors. IF works well for smooth mirrors. Star test works well for everything and should of course agree with other tests if you're doing them right.

I think one of the problems that arises here is how smooth is smooth, or more exactly, how regular and consistent is the smoothness? How free is the mirror from small scale regular patterning (aka ripple) that could scatter significant light? I suspect that simply by producing them so that they don't show any visible roughness under grazing illumination and a KE you get the benefit of removing all the doubt. But I don't know exactly what roles certain sorts of errors play in practice, except in the broad strokes. Vla makes the good argument that below some level roughness can't matter. So long as it's not periodic that has to be true.

The mirror in the first example is a good illustration. Yes, it looks kind of horrible, but after you've seen enough examples of machine made mirrors you get to recognize the symptoms. It would have worked, and probably pretty well, given the overall correction from the IF - although you can just detect some of that roughness there as well. It might not satisfy a critical user. That's usually why you end up seeing examples like this anyway. :shrug:

Best,
Mark

#10 Ed Jones

Ed Jones

    Mercury-Atlas

  • *****
  • Posts: 2731
  • Joined: 06 Apr 2004
  • Loc: Sin-sin-atti

Posted 10 October 2013 - 07:57 PM

Its almost impossible to quantify what that "better" is though


Good point. "functionally non-existant, as in nearly undetectable to the eye" are not quantifiable terms whereas the interferometer shows the roughness and quantifies it.
  • fjh likes this

#11 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2625
  • Joined: 20 Jan 2011

Posted 10 October 2013 - 09:41 PM

There is extremely interesting information posted on Carl Zambuto's website here on how a high strehl [sic] can be misleading...

It all depends on how the software is used to manipulate the data. Software reduction programs allow you to remove surface "noise" (artificial smoothing) and thereby increase the Strehl ratio.

You can also artificially block off a desired amount of the edge which explains how come synthetic i-grams show no edge and Focualt does. Effectively blocking off most of the TDE will also increase the Strehl score.

Likewise, you can increase the qualy of your results by running Fast Fourier Transformation (FFT) analysis instead of the regular fringe analysis which has many time fewer points and less accuracy.

Mike Lockwood also has a section about commercial interferometry (IF) reports, which he summs up rather spot on that the results of an interferometer are only as honest as the optician/supplier allows them to be.

But, data manipulation and artificially improving the results is not limited to operators of IF reports only. The same can be said of many who use other tests, such as the Ross Null, the Double-Pass Autocollimcaiton Test (DPACT), and not to tlak about the fantastic Foucault measurements (remember that repeatability is not accuracy, but precision, and that just because your micrometer reads 0.002 it doesn't mean it's 0.002 without proper calibration of the measuring tool).

That's why everything defaults to, and hinges on a single factor: honesty (honest mistakes notwithstanding)! Some people have agendas to sell or promote their product and wish to make it look better than it is. Zambuto's and Lockwood's sites are full of such examples of other manufacturers.

Vla made a good point as regards the old adage: that just because you can measure something it doesn't always mean it's significant. Besides, what's the point of having a 0.997 Strehl ratio primary if the secondary is going to be 1/5 wave PV? Or what's the point of miniscule micro ripple when the atmosphere is less than optimal?

Zambuto is right, however, in that reports should be honest, and that's the point of his comment. But, as Vla also observed, the quality difference of images provided for his mirrors vs the "bad" one is so blatant that one gets the impression Zambuto is doing the very thing he is accusing others of doing!

regards,
Mladen

#12 Arjan

Arjan

    Mariner 2

  • -----
  • Posts: 222
  • Joined: 21 Jan 2009
  • Loc: Netherlands

Posted 11 October 2013 - 05:12 AM

And you can even fiddle with Foucault to smoothen the image. Small scale surface errors usually show up only when you pinch off the returning light or use a small pinhole or slit lightsource.

#13 kfrederick

kfrederick

    Gemini

  • *****
  • Posts: 3026
  • Joined: 01 Feb 2008

Posted 11 October 2013 - 07:06 AM

Carl is trust worthy and gives you his best work . That is what matters . No worries on the optic . Might be over kill but a mirror lasts many lifetimes .When the seeing is perfect perfect optics shine .

#14 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2625
  • Joined: 20 Jan 2011

Posted 11 October 2013 - 08:56 AM

And you can even fiddle with Foucault to smoothen the image. Small scale surface errors usually show up only when you pinch off the returning light or use a small pinhole or slit lightsource.

Yes, exactly! Unfortunately, the slitless testers, which have become the "new" ATM norm relatively recently, are notorious in the "fiddle" department because the "slit" is not maintained mechanically but by the position of the eye, and the easiest one is when the "slit" is wide open.

Anyone who's ever done a Ritchey-Common test would know that the smaller the light source is the more sensitive the test becomes. The same holds true for the Foucault test. Ideally, your light source/pinhole should be close to the Airy disc diameter, which is a few microns. The slitless LED source is 5 mm!

Mladen

#15 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2625
  • Joined: 20 Jan 2011

Posted 11 October 2013 - 09:08 AM

Carl is trust worthy and gives you his best work . That is what matters . No worries on the optic . Might be over kill but a mirror lasts many lifetimes .When the seeing is perfect perfect optics shine .

People who are willing to spend the kind of money on Carl's mirrors are also likely to spend money on good accessories and will not cut corners in that department. They will buy top-notch secondary mirrors, and often have parts profesisonaly manufactured for them, whereas those with limited means may try to improvise and cut corners. Clealry the two groups will not have the same quality telescopes for reasons other than just Zambuto's mirrors.

Two mirrors can objectively be compared only if all other factors, parts, accessories, etc. are the same and the images of both are observed side by side simultaneously.

regards,
Mladen

#16 Arjan

Arjan

    Mariner 2

  • -----
  • Posts: 222
  • Joined: 21 Jan 2009
  • Loc: Netherlands

Posted 11 October 2013 - 01:29 PM

Carl is trust worthy and gives you his best work.

Sorry, didn't mean disrespect to Carl Zambuto!

#17 Mike Lockwood

Mike Lockwood

    Vendor, Lockwood Custom Optics

  • *****
  • Vendors
  • Posts: 499
  • Joined: 01 Oct 2007
  • Loc: Usually in my optical shop

Posted 11 October 2013 - 02:00 PM

But, as Vla also observed, the quality difference of images provided for his mirrors vs the "bad" one is so blatant that one gets the impression Zambuto is doing the very thing he is accusing others of doing!

I can't comment on the vendor, but I can comment on the images in the example - as I see them, the images for both mirrors are about as equal in quality, exposure, contrast, etc. as they could possibly and reasonably be. Therefore I think that your statement/accusation above is completely wrong and groundless.

Yes, exactly! Unfortunately, the slitless testers, which have become the "new" ATM norm relatively recently, are notorious in the "fiddle" department because the "slit" is not maintained mechanically but by the position of the eye, and the easiest one is when the "slit" is wide open.

That's completely wrong, too - the slit is formed when the reversed image of half of the light source passes by the knife edge, resulting in a virtual slit. Eye position has nothing to do with it whatsoever, it is only a function of knife and source position.

As a bonus, the virtual slit is always aligned perfectly with the knife edge, so there is no "fiddling" with the slit to get it parallel to the knife. Even better, the nasty diffraction effects from the narrow slit itself are pretty much non-existent.

I consider it an improvement over classic Foucault with a slit, for those and other reasons.

Ideally, your light source/pinhole should be close to the Airy disc diameter, which is a few microns. The slitless LED source is 5 mm!

Completely wrong three. What passes the knife is very narrow, and results in shadows with good contrast that show roughness readily.

And you can even fiddle with Foucault to smoothen the image. Small scale surface errors usually show up only when you pinch off the returning light or use a small pinhole or slit lightsource.

Well, here are some "un-fiddled" images from my slitless tester (which I use daily for figuring mirrors), showing surface roughness and even cleaning residue quite easily with good contrast:
Slitless Foucault images, 18"
Slitless Foucault images, 16"

Please, before throwing a useful, accurate, good test under the bus, get your facts straight.

#18 Mark Harry

Mark Harry

    Vendor

  • *****
  • Vendors
  • Posts: 6215
  • Joined: 05 Sep 2005
  • Loc: Northeast USA

Posted 11 October 2013 - 02:02 PM

It's not the test used; whether IF, AC, Foucault, etc.

IT'S THE REP(UTATION) AND ABILITY OF THE MAKER!!!!

And I GUARANTEE any side-by-side comparison of any of the mirrors with RIPPLE will prove out to be far deficient in sharp focusing ability.
******
Another thing to ponder-
Numbers arent everything. SEE-ing is.
M.

#19 Mark Harry

Mark Harry

    Vendor

  • *****
  • Vendors
  • Posts: 6215
  • Joined: 05 Sep 2005
  • Loc: Northeast USA

Posted 11 October 2013 - 02:11 PM

I also agree with Mike on this one.(since I just missed reading his post) Regardless what the math says STREHL should be taken with a grain of salt, period--- as what the actual Criteria#4 was stating in substance, if read to the end.
****
This is a classic instance of subjectivity missing the actual truth of the matter!
M.

#20 mark cowan

mark cowan

    Vendor (Veritas Optics)

  • *****
  • Vendors
  • Posts: 4142
  • Joined: 03 Jun 2005
  • Loc: salem, OR

Posted 11 October 2013 - 02:52 PM

.When the seeing is perfect perfect optics shine .


Amen, brother. Welcome to the church of the holy hyperboloid. :waytogo:

Best,
Mark

#21 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2625
  • Joined: 20 Jan 2011

Posted 11 October 2013 - 04:02 PM

Post deleted by MKV

#22 Mark Harry

Mark Harry

    Vendor

  • *****
  • Vendors
  • Posts: 6215
  • Joined: 05 Sep 2005
  • Loc: Northeast USA

Posted 12 October 2013 - 06:42 AM

Mladen, I wish you would
-STOP-

the never-ceasing arguementative attitude you have displayed every time someone who has vast experience and knowledge corrects your misconceptions.
PLEASE!
M.

#23 MKV

MKV

    Mercury-Atlas

  • *****
  • Posts: 2625
  • Joined: 20 Jan 2011

Posted 12 October 2013 - 08:30 AM

There you go, Mark, I deleted my post. See, I value your opinion. But I am still looking for answers, so please answer these three simple questions:

(1) do you think Vla's statement was "wrong and groundless"?
(2) do you know how wide is a virtual "slit" in a slitless tester (and therefore how sensitive is the test), and how do you determine that width?
(3) do you think it's okay to throw interferometry under the bus?

Thak you.

regards,
Mladen

#24 Pinbout

Pinbout

    Cosmos

  • *****
  • Posts: 8302
  • Joined: 22 Feb 2010
  • Loc: You can't see me...

Posted 12 October 2013 - 09:14 AM

I read this on the zambutogroup from cal's doc file

Criteria for a GOOD mirror
The M-L envelope that was described in other
postings would be the applicable criteria for most cases, and
that is for accuracy, only. That covers the four items you have
mentioned above. An estimate on the factors above given the
mirror is in the envelope and reasonably smooth would be
something such as the PV wavefront better than 1/8 wave, the
RMS better than 1/30, the Strehl ratio in the mid-90's or better,
and the RTA at less than 1.0.

But we can't just stop there. We must also consider the quality of
the polish, the edge condition, the surface roughness, whether it
has zones, or "rings" that don't necessarily affect the
measurement, and how much astigmatism is present. All these
factors would be evaluated and communicated with the
customer to determine their level of need. Based on that the
recommendation would be made.

I have measured commercial mirrors that are accurate by the
numbers, but had rings, or a turned edge, or a rough surface, or
incomplete polish. Those would require a judgment call
combined with what the customer is after. But for accuracy, if it
falls within that envelope, that is typically accurate enough to
where further work will have diminishing returns. Then only if it
had other problems would it be reworked.



#25 Ed Jones

Ed Jones

    Mercury-Atlas

  • *****
  • Posts: 2731
  • Joined: 06 Apr 2004
  • Loc: Sin-sin-atti

Posted 12 October 2013 - 11:34 AM

(3) do you think it's okay to throw interferometry under the bus?


Hardly. The phase shifting interferometer did a good job of quantifying the surface roughness (as a good case of measles on the wavefront map) where the KE couldn't. They might have also done a surface profile to show the roughness. I'd rather have the original Igram than synthetic fringes however. Carl would be well served switching to IF but it's rather costly, after all he uses IF to test his flats and the same rules apply there.






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics