What's Important in Optical Quality to Amateur Observers (And what's not)
By John Lightholder
This article is not primarily about mirror making, but I have to start it off by mentioning that I recently made a 14.5" F/4.5 mirror -- mentioned in the "mirror making" sidebar, the latest and biggest in a series of 8 mirrors, and was extremely pleased it came out diffraction limited, which is "about as good as it gets." (OK, I'll stop bragging now!) I can't say it was easy, but this and previous mirror-making experiences have convinced me that high optical quality is not extremely difficult to attain for an amateur. I'd encourage any ATM who has an interest in the optical area to make at least one mirror. Chances are you will complete it and will find it to be one of the most rewarding experiences of all your ATM pursuits. Let's face facts -- few ATM's are willing to make their own telescope mirrors. Most of us, therefore, will approach the optical quality issue in the role of consumers, as I once did. We benefit, as informed purchasers, to have a good understanding of the standards and test methods used by optical vendors. This is the central issue I want to address, not so much to resolve it conclusively for all time, but more importantly, put it in perspective where I believe it belongs now. I'm going to refer to primary optical quality throughout this article, but this isn't intended to give the impression the primary's the only determiner of the system's output and what you see at the eyepiece, on the contrary. It's less than 1/2 the total system in terms of sources of aberration that affect what you see. So keep this thought in mind as you read on. It's inherently the responsibility of the supplier of any product to establish a standard, and assure the quality of those manufactured supplies, including telescope mirrors, to that standard. If a supplier advertises certifications of quality, the certification in my opinion should contain sufficient data, preferably quantitative as well as visual, to back up the certification. I like the ADMIR charts and output reports, because they are minimal, concise, and provide all the data recommended by Texereau in his book How to Make A Telescope (Willmann-Bell). ADMIR is a companion program to the book "How to Star Test Astronomical Telescopes" by H. R. Suiter (also Willmann-Bell). Mirror vendors start with advertising a commitment to a certain quality criterion. In ascending order of stringency, these normally will be expressed as either (1) 1/4-wave wavefront peak-to-valley (pvwf) or 1/13.4 wave rms; (2) Danjon-Couder criteria; (3) Diffraction-Limited. See discussion herein for a more detailed description of these criteria. The vendor will specify a method of test, e.g., null test, Foucault (knife-edge) test, Ronchi test, interferometer, or a combination of methods. Some vendors might use null testing during figuring, to determine which areas are high relative to the desired parabolic curve, and need to be reduced by polishing action, and then to quantify the slope and wavefront error by using one of the other methods. There is no problem with doing this, provided the methods are clearly described and the output results are understandable in p-v and RMS wavefront error and (if the method can supply it) diffraction disk size. These are the checks recommended by Texereau, in How to Make a Telescope, 2d Edition (Willmann-Bell). Looking through a lot of telescopes containing mirrors made by the small companies that advertise "premium" Newtonian mirrors, it appears that these mirrors are usually very good to excellent, judging from star images and planetary performance. It appears that quality in this area has been uniformly good with few reported exceptions, along with an improvement in the ability of amateurs to discern performance quality. It's my impression that these suppliers universally test and correct each mirror at numerous steps in a prolonged figuring process, and the results are far more likely to result in a high quality level and real performance capability, than a mass production environment involving less individual attention to each mirror. Among the manufacturers of small Dobsonians, the optics and tube assemblies of which are reportedly mass-produced at economy prices, the quality, in my opinion, is not likely to be as good (nor is this expected) and occasionally unsatisfactory by any optical guideline, judging from star tests of some collimated telescopes. The average level of quality is OK nevertheless and a satisfactory value considering the prices. Factual optical detail in their ad copy and less hyperbole in describing these scopes' optical quality would improve their credibility. If you want a go/no-go test for your telescope, the simplest and most conclusive method is the star test. Various, simplified methods of star testing are available to quickly determine whether your mirror, when properly collimated and supported, is performing basically as advertised. Do not expect identical intra/extrafocal images, even for diffraction-limited optics, as this test is very sensitive to extremely minute wavefront error. The Ronchi test with a 2X - 3X Barlow will also help to identify optics with significant RTA or wavefront error. If you use a Ronchi grating, a minimum lines-per-inch (lpi) should be 100, and maximum 150 lpi. I prefer 133 lpi. The only bad thing about the Barlow is that poor seeing wipes out the bands more readily at the higher effective focal length. Incidentally, don't just stop with bands displayed across the image of the primary -- take it closer to focus so the mirror "grays" -- with experimentation you see whether the primary winks out uniformly, and if doesn't, where the zones are that are out of phase with the remainder of the mirror's surface. NOTE: Barlow lens, like all other optical components, can introduce an error in the star test in the form of under or overcorrection and other aberrations too. Don't panic if you see what appears to be over-/undercorrection -- use a second (and a third, if necessary) Barlow to contrast any apparent optical anomalies before you conclude all the error is in the primary. If you have a few dollars and a few hours to invest in buying/constructing a knife-edge tester, and a few dollars more for software, you can quantitatively rate your mirror in RTA and wavefront p-v error. This should not be necessary if your mirror performs acceptably under the star test. However, it doesn't take nearly as much skill to use a knife-edge tester as it does to figure a primary mirror, and the test is fairly definitive when the tester's error of measurement is taken into account. The results of these tests can indicate what to expect from your optics and finally, what you can see with your 'scope. Other factors come into play also, besides spherical aberration, in determining optical quality: Surface roughness (primary ripple and micro-ripple) scatter light out of the disk. The effect is very subtle for micro-ripple and would look like very dusty optics. That dreaded "turned down edge" or "rolled edge". With a rolled edge the mirror surface doesn't come out to a sharp bevel at the edge and will scatter light. A turned edge (up or down) refers to a zone at the edge of the mirror, which focuses long or short of the mirror's average focus. A turned-down-edge (TDE) has a drastic effect on images and should be masked out. Even professionally made mirrors tend to have a vestige of a "low" edge, not nearly as bad as a TDE, but hard to avoid. Do not (realistically) expect the 1% linear edge of the mirror, e.g., 1/10" width for a 20" mirror, to be as good as the inner 99%. If a diffraction ring is visible at the edge on the knife-edge side during Focault testing, AND if the measurement for that zone's center is within the ML tolerance, AND there is not an obvious demarcation, do not mask the edge. Check the afocal images with a Ronchi grating, and only mask out the part of an edge that shows an extreme "edge hook." Holes, hills, raised or depressed rings; all help degrade the image somewhat proportional to their slope and alter an otherwise good wavefront. Out-of-round error is an asymmetrical high or low error that is not distributed in the rings as is typical of most errors. It can hide out and be averaged out by shadows, but can affect the fidelity of images with its aberration at the wavefront. (This type of error is distinguished from classical astigmatism as a localized asymmetry in the mirror, as opposed to general astigmatism, where one axis of the mirror has a different depth and focal length than the remainder of the mirror. It won't likely be prevalent enough to create the 90 degree elongation of star images typified by classical astigmatism.) These factors' effects usually accumulate, seldom cancel each other, and transfer their accumulated visual noise to the image. The Worst Omission. The overwhelming and most significant omission among ATM's is neglect of collimation. Trailing some distance behind is overemphasis on accuracy of the primary and size of the secondary, usually making the latter too small without regard to the negative consequences. Over-reliance on surface and or wavefront peak-to-valley or even RMS numbers, and failure to consider transverse aberration as descriptors of optical quality are shortsighted. These seem to be the most prevalent errors of perception among our peers in the ATM community. The second most significant factor affecting the performance of the primary mirror is the secondary mirror. There is a far higher percentage of flawed secondary mirrors than primaries and it is unfortunate that ATM's don't insist on the same rigid testing and test data for this ONE HALF of the optical system. After collimation, there is no doubt more telescopes suffer in performance from bad secondaries than any of the other common concerns. Some people seemed to be more concerned about the wavefront peak-to-valley error of primary mirrors than about real-world performance issues. It seems ironic that some people insist on a peak-to-valley 1/20 wavefront on the mirror they buy, yet frequently won't take the most basic steps needed to eliminate collimation error. When you think of it in that way, it does seems to be case of "penny-wise and pound-foolish" to spend as much on a primary mirror as a used subcompact car, for a near-perfect mirror, but not to take the pains to realize anywhere near its potential through the necessary collimation process. It should be noted that you cannot "eyeball" collimation in on any mirror less than F/10, and the only way you can achieve optical and mechanical collimation is with collimation aids, which as a minimum includes an accurate sight tube and cross-hair. If you make the sight-tube, measure its dimensional conformance, at the peep-sight end and at the cross hair end, to the optical centerline within .0025" using a high-power magnifier, along with enough non-spiraled shims so it fits snugly in the drawtube. This piece is critical enough that I'd prefer to invest in a commercially made version rather than chance biffing it myself. The primary is only a part of the total system. We astronomers as a rule seem to put far too much emphasis on optical accuracy of the primary and somewhat on the secondary mirrors, but not nearly enough on the controllable factors, such as collimation, support, ventilation, baffling, and so forth. We sometimes loose sight of the fact that everything we build, construct or envision has imperfections and errors, including all the components that make up a telescope. Remember that EVERYTHING in the SYSTEM has inherent errors in it, all with potentially higher magnitude of error than a good-to-excellent primary mirror. There is a point of diminishing returns on the primary mirror's precision where you are throwing away money and optical work. I believe this occurs once the mirror has been worked to an accuracy of slightly better than 1.0 Relative Transverse Aberration (RTA) across its entire diameter, where the mirror is diffraction-limited. At this point the mirror exceeds the Danjon-Couder (D&C) criteria, and I don't think you could ask for more from any mirror larger than 12.5" or less than F/6 focal ratio. For an F/ratio faster than F/4.5, 1/4 wave peak-valley (about 1/14 rms) is a more realistic criterion, even with the degradation suffered below a diffraction-limited threshold. It is generally very difficult and time consuming to make a large F/4 mirror diffraction limited, as defined by the milies-la croix tolerance. As a general rule, wavefront peak-valley wavefront is such a variable, inconstant quantity that it's largely irrelevant as a quality descriptor, except as the second term of the D&C condition. The RTA value, on the other hand, is ALWAYS a relevant expression at its minimum value and can be used to compare one mirror's performance with another that has also been rated using RTA. Bottom line: If possible, vendors, opticians and ATM mirror makers alike should focus on a diffraction-disk related estimate of optical quality, and subordinate the use of wave ratings at the same time. In the absence of a way to rate diffraction-limited conformance due to test method limitations, I suppose the vendor would still need to offer estimated wavefront numbers, preferably with RMS equivalents, although I'd still much prefer to know RTA at best focus, second-best would be RMS, and least desirable, peak-valley wavefront error. The Strehl ratio is yet another criterion that is useful to summarize quality in a single ratting. It's downright unfortunate that wavefront and surface fractional numbers, whether peak-valley or rms, are still used exclusively as the common coin for estimating, expressing and comparing optical quality, when in fact it can be very misleading. This is still true even after we have precisely defined what we are referring to as to surface, wavefront, unilateral/bilateral variation or whatever. As elsewhere stated in this article, 1/4 wave p-v at the wavefront is usually NOT diffraction limited. See the panel on 1/4-wave expressions, borrowed from H. R. Suiter's Telescope Making (TM) 32 Article. In the same TM 32 article, Mr. Suiter depicted 6" spherical mirrors of various focal lengths, along with graphs indicating shaded areas where the mirror would be diffraction-limited and also 1/4-wavefront p-v. In the graphs, 1/4 wave accuracy occurred for a spherical mirror at F/8.2, diffraction-limited performance at about F/10.2, and they began to overlap at about F/11. The purpose of defining the overlap areas is to find where both D&C conditions are simultaneously met. As a rule, only rarely, will a 1/4-wave mirror be diffraction-limited, and a diffraction-limited mirror will nearly always be better than 1/4 wave. As mirrors get larger and of shorter focus, this rule-of-thumb assumes more validity. For a mirror to perform virtually perfectly, both D&C conditions need to be satisfied simultaneously at a given focus position. You may want to refer to this TM #32 article, as it has some excellent information in it. It's long, but practically every sentence and paragraph in it has a valuable gem of wisdom or two. Wavefront focus for minimum peak-valley/RMS error and minimum RTA does not occur at the same point. Fortunately, there are several computer programs, such as ADMIR (copyright R. Suiter/Willmann-Bell) and GENMIR.EXE. I believe GENMIR may be available as shareware on several web sites. These programs calculate the intersects for minimum/maximum peak-valley to obtain the optimal balance for both factors, i.e., the "Overlap Areas" for any mirror. (Within this "overlap area" the minimum p-v value and corresponding RTA is valid and D&C criteria are met.) But there are a lot of apparent inconsistencies when we compare wavefront peak-valley with RTA. For all practical purposes, the wavefront peak-valley error can be ignored in larger mirrors with F/ratio less than F/6 if the mirror's RTA is equal to or less than 1.0, as its p-v amplitude will usually be far less than 1/4 wave, and no more than 1/7 wave. Actually, the user of the scope has the most influence over how his mirror performs, more so than the mirror's limitations expressed by its absolute error. I'd even go so far as to state that a well-collimated 1/2-wave peak-to-valley wavefront (pvwf) mirror/system will perform better than a casually collimated one with a 1/20-wave peak-to-valley-wavefront (pvwf) primary mirror. I've known people who were absolutely amazed at how well their system performed when they got serious about their collimation. If you consider how slight a component misplacement or optical path difference results in a 1-wave error (which is only about .2 ten-thousandths of an inch), then you realize how quickly poor collimation can ruin your fine mirror's images. And if you have a lesser quality mirror, it suffers even more from de-collimation. Having an good-to-excellent mirror gives a slight but significant advantage when seeing is marginal but not quite "over the edge." Under such conditions, an excellent mirror will do much better than a mediocre one. Even though seeing may be such that the excellent mirror can't realize all its potential, it will be less constrained and affected than a lesser mirror by the poor seeing. Errors accumulate -- it's that simple. Even so, optics that were used for very serious research for years were later found in some instances to have serious optical errors. For example, I recall having heard a story somewhere about the Lick 36" refractor objective having been discovered, upon refiguring, to have 1/2 wave of spherical aberration, although it had apparently been used for many years with no complaints by the numerous pro's who had used it for double-star observing. What the future holds for the amateur astronomer is a vast as space itself. Improvements in quality and cost have brought computer technology and CCD's into our hands and enables amateurs to accomplish what was once considered "professional" quality work only a few years ago. Although I don't care for astrophotography or CCD personally (at least not yet), I understand the feelings of accomplishment and personal satisfaction gained from the processes involved. But for me the joy of just observing is fulfilling enough. New concepts in design and new ways to use materials are growing fast and allow more people access to the sky. I'm reminded of a printed observation years ago to the effect "a telescope of 16 inch aperture would require a mount in excess of two tons." Things have sure changed. I would hope that the trend continues into the material sciences, for there may be a "honeycomb graphite something" available soon that could make for a light and stable material on which to deposit aluminum. Seeking out faint fuzzies and pushing personal limits would thus be all the more easier and enjoyable. The most fundamental aspect of observing, the FUN of it, is an observer looking through the telescope at an object. (And, hopefully, the telescope won't get in the way!) A telescope that gets used provides limitless hours of in-depth, four-dimensional observing experience.
Originally in "Amateur Astronomy" Issue 18