◄ 14.1. Early telescopes   ▐    14.3. Observatory telescopes ►

14.2. ATM TELESCOPES

Most ATM telescopes are of known designs, but their makers can customize them to their liking. That makes them sort of unique, and adds to the satisfaction of accomplishing, sometimes very challenging, creative projects. With more complex systems, it all starts with a design created and optimized through raytracing, as the basis for elements' fabrication. And some of ATM telescopes are truly unique systems, out of common knowledge of the amateurs and professionals-alike.

Marc Baril's 7" f/8 modified Sigler-Maksutov-Cassegrain

Making Maksutov-Cassegrain of any kind is not for the faint of heart, primarily due to the full aperture Maksutov lens corrector, whose "easy" to make spherical surfaces have to remain within tight tolerances for the surface itself, the two surfaces relative to each other, and both relative to the meniscus thickness. A good example of it is Maksutov-Cassegrain made by Marc Baril, an amateur astronomer with professional background. 

Main modification to the Sigler's design was shortening the tube length which, according to Baril, did somewhat lower system's axial correction, but still allowed a high-level performance. Here is raytrace of the prescription given in the article.

Design shows great field correction, with near-zero coma and astigmatism (it could, in fact, have 2-3 inches - or even more - smaller corrector separation without perceptibly affecting coma correction). Correction barely changes going from the field center to 0.5 degrees off axis. 

One thing to notice is that axial monochromatic correction of the starting design is a bit off (top left). It looks like undercorrection, but it is suboptimally balanced 6th/4th order spherical, with the lower order (undercorrection) being nearly twice larger than the (overcorrected) higher order. With the RMS wavefront error of 0.043, it is at the level of just a tad stronger than 1/7 wave P-V of lower order spherical. Baril mentions having a slight undercorection when arriving at the testing stage, which he removed by working the first corrector radius with a petal lap. While he thinks that what it did was slight aspherising, even just a slight strengthening of the first radius - going from 10.3507" to 10.344" - would do the trick.

After taking care of undercorrection, we come to a design with 0.982 Strehl in the e-line, 0.927 in the F, and 0.926 in C. The polychromatic (430-670nm photopic) Strehl is 0.964. If we minimize the chromatism by bringing the colors tightly together strengthening corrector radii to 10.05" and 10.481", the two lines go to 0.972 and 0.976, respectively, with the e-line at 0.98. The polychromatic Strehl goes to 0.977.

The Strehl figures are actually somewhat improved due to the 38% central obstruction, which took out the central deformation of the wavefront caused by spherical aberration. But the central obstruction did, of course, more damage than good at the level of system's PSF. Still, it's negative effect is commonly exaggerated. Baril states in his article that the biggest disadvantage of this design is its large central obstruction. How big a disadvantage it really is?

The fact is that it reduces the wavefront error in all wavelengths - the larger wavefront error, the more so. If we would take it out, these same optical surfaces would produce wavefronts of lower quality. The central line, for instance, would go from 0.183 wave RMS and 0.982 Strehl, to 0.0225 wave RMS and 0.972 Strehl. At the same time, obstruction causes 0.86 drop in the relative central intensity of the diffraction pattern, with the second 0.86 reduction factor for the energy contained in central maxima - 0.73 reduction in total - coming from the reduced central maxima size. Due to the size reduction, central maxima is not only smaller, allowing higher limiting resolution, but also significantly brighter than what 0.73 degradation factor implies. In this case, central maxima is only 7% less bright than in a perfect aperture, in this respect corresponding to 0.93 degradation factor.

So, while the nominal degradation factor is 0.74 (rounded from 0.73*0.982/0.972), it is also accompanied by 13% higher limiting resolution, and higher contrast transfer in the high frequency range. MTF compilation below illustrates the magnitude of both, loses and gains.

Here we see that the worst part of the central obstruction effect is lowering contrast transfer to the level of 33% smaller aperture in approximately 0-0.3 range of frequencies. In the 0.7-1 range the transfer increases to the level of 16% larger aperture, while over the 0.3-0.7 range we have transition from lower to higher transfer, extending all the way to the resolution limit of 16% larger aperture. Overall, nominal view favors obstructed aperture, which has contrast transfer advantage in nearly as much of the frequency range as the unobstructed, with extra high frequencies added beyond the resolution limit of unobstructed aperture.

However, to better understand the implications, MTF frequencies should be associated with magnification. With the limiting resolution at cutoff frequency λ/D in radians (times 206,265 in arc seconds), magnification at which a general detail is magnified to 5 arc minutes (likely more near threshold of detection), needed for the average eye to recognize its shape is 300D/206,265λ, or 2.64D for the aperture diameter D in mm and λ=0.00055mm (67D for D in inches). For details of Airy disc size, 2.44λ/D, corresponding to 0.41 frequency, it is 1.1D, and half as much for twice larger detail (0.2 frequency). So we could say that high magnifications start, quite approximately, at 0.3 frequency. If so, the bad part of central obstruction effect falls to the range of mid and low magnifications, while the good part covers most of the high magnification range. That said, should be noted that the largest contrast drop for this obstruction, some 25%, is at about 0.4 frequency - somewhat deeper into the high magnification range - while diminishes toward the mid-to-low magnification range (<0.3 frequencies), down to 12% at 0.2, and 5% at 0.1 frequency.

There are direct implications of different contrast levels on the limiting resolution. Taking approximate minimum contrast requirements for different object types from Rutten/Venrooij, we get that the obstructed aperture will have appreciably higher resolution limit not only for the conventional MTF limit (two bright lines, nearly identical to that for two near equally bright stars), but also for bright contrasty objects, like the surface of Moon (1 and 2, respectively). On the other hand, unobstructed aperture has higher resolution limit for bright low-contrast objects (3) and dim low-contrast objects (4).

Obviously, we can't compare apples and oranges, so whether the effect of central obstruction is good, bad, or insignificant, will depend on the specific use.

In all, the effect of central obstruction is a mixed bag, and certainly is not only bad. For photography, considerations are somewhat different, but the effect is generally considered less of a factor than for visual use.

Summing it up, while it is not possible to execute the actual optics to its theoretical optimum, it is not even necessary. No mortal soul can see the difference of one or two 1/100s in the Strehl ratio. From Baril's article, it is very likely that his commitment to making this telescope as close to its design marks as he could, did bring it to the level where the difference vs. theoretical optimum didn't matter.

Sure, performance level would have been even better with a system with slower primary, such as the original Aries' Maksutov Cassegrain, without significant difference in the tube length. But if one wants a faster system, there is a price to pay.

Clyde Bone Jr. 30-inch f/5 Mersenne

In an S&T article from 1999 Clyde Bone (San Angelo, Texas) describes his 30-inch f/5 Mersenne-type telescope, for which he used Tele Vue's 140mm f/5 Petzval apo at the eye end (there is also an article on the Tele Vue site). The telescope was built with the highest standards, using, for instance, astro-Sital mirrors with 96% reflectivity - 1/16 wave primary, 1/10 wave secondary - and Pyrex flats with dielectric coatings (98.5% reflectivity). The two mirrors are confocal paraboloids, producing collimated beams free of aberrations (except Petzval image curvature, which is here -836mm). An image-forming system is required and, if the completed system is to be a visual telescope, it has to consist of an objective and eyepiece - i.e. telescope. Image below shows raytrace of such a system. Since no information on the Tele Vue refractor was published, a Petzval system that should be similar to it in its optical output is used to illustrate the combined performance level. It is a borderline "true apo" based on the errors in the four unoptimized wavelengths (F and C balance out at about 0.80 Strehl, and only the violet g-line is little over the required 1/2 wave P-V), but not according to its 0.917 polychromatic Strehl (photopic). This is mainly caused by keeping the violet g-line error nearly as low as possible, causing the blue and red - particularly the former - to be suboptimally corrected. If the F-line error is minimized, its Strehl increases to 0.856 (w/0.898 C-line), g-line drops to 0.183, but the poly-Strehl is up to 0.945 (all needed to accomplish this is to change R5 i.e. 9 in this prescriptionto 520mm, and R8 i.e. 11 to -237.7mm; that produces f/5.15 system, rescaling it to f/5 would lower the poly-Strehl to 0.934).

The actual secondary is six inch in diameter, but in order to avoid stopping down the aperture, the reflected beam width has to be no more than 140mm (here, it is slightly less, which produces an effective f/5.1 system). Note that refractor alone is evaluated at f/5, and the setup used only stop at a distance of the Mersenne system exit pupil (located at -k+k² from the secondary, toward primary's focus, in units of primary's focal length, with k being the marginal ray height of the axial pencil at the secondary, in units of aperture radius). Prescription with the complete system given with the image is a compilation of two prescriptions (OSLO Edu has 10-surface limit). This means that the nearly flat field here in the actual telescope corresponds (approximately) to the Petzval image surface, somewhat lessened by astigmatism induced by lens objective processing diverging pencils from the secondary (shown on astigmatism plots). Also, the 0.5° field radius in a 700mm f.l. refractor corresponds to 0.09° in the 3889mm f.l. system (will be explained ahead).

The field is limited to 0.5° i.e. 0.09° radius, since off-axis aberrations are already significant. It is a consequence of the widely separated aperture stop (also, as the box right from prescription shows, vignetting at the objective lens becomes significant). Part of it is that the incident ray angle is magnified after reflection from the secondary. With infinite secondary magnification, incident angle is magnified by a factor 1/k; with k=0.18 here, the incident angle magnification is 5.56, i.e. 1° angle at the primary becomes 5.56° at the lens objective. Hence, 0.5° at the lens objective is 0.09° in the actual system (linear field height is identical in both, since the two respective focal lengths differ by the same ratio). The central obstruction is 20% linearly, or little more. Since its effect on off-axis aberrations here is negligible, it is omitted to keep interferogram patterns intact. Also, to show the original polychromatic Strehl of the refractor, which would be superficially increased in the presence of obstruction (specifically, it would increase the poly-Strehl to 0.934 - and to over 0.95 with the F-line error minimized - as a result of small wavefront error decreases in all wavelengths - the larger error, the more - hence only in part reflecting better chromatic correction; on the other hand, that small gain is paid for with a 0.96 drop in central intensity - to the actual 0.897 - plus another 0.96 ratio energy loss due to the smaller central maxima, for the final 0.87 central maxima energy ratio on axis).

Wavefont map for the 0.5° point shows that most of the P-V error comes from raised edge on one side. Because of that, both P-V and RMS error are somewhat misleading, implying larger than the actual error. The main wavefront component is in the blue area, less than 1/3 of the entire 8 wave P-V error. Zernike analysis gives as dominant terms primary coma (#6), astigmatism (#4) and defocus (#3), followed by secondary astigmatism (#11), coma (#13), primary spherical aberration (#8) and trefoil (#9). However, the primary spherical aberration term indicates 0.32 wave RMS, much more than what is present on axis; this means that most of this term is actually lateral spherical aberration, a secondary Schwartzschild aberration of the same form as primary spherical, but increasing with the square of field height. Also, raytrace shows that coma-like ray spot plot grows significantly faster than in proportion to field radius, indicating presence of other "odd" secondary aberrations, such as lateral coma (secondary Schwatzschild aberration, of the same form as primary coma, but increasing with the cube of field radius), and probably some others. Diffraction simulation for the 0.5° point (e-line) shows some elongation, but still well formed central maxima, embeded in a tight, bright ring (mostly due to 0.01 normalization); however, 80% energy radius is 0.0286mm (large black circle), much larger than the Airy disc (not sure how adequate is this number, since it could be based on PSF centroid location, which is here by some four Airy disc diameters - or 0.025mm - from the PSF peak, indicated by the dense ray area above the Airy disc circle on the ray spot plot). The bright central oval is similar in area to the Airy disc, but the peak intensity is only 0.40, indicating energy comparable to that with 1/2 wave P-V of spherical aberration (note that the peak intensity is much higher than central intensity, due to intensity shift from the PSF center). Box bottom right shows corresponding ray spot plots for the refractor alone, for the same 0.5° radius field; evidently, all off axis aberrations are a consequence of the large stop separation.

There is a discrepancy between the 0.5° wavefront error magnitude and the corresponding PSF. Even if limited to the blue area on the wavefront map, it is as much as 2-3 waves P-V, not agreeing with still quite compact central maxima. Taking a closer look, we see that OSLO default wavefront map is for "minimum RMS OPD" (image below, left).

The corresponding interferogram shows that most of the blue area is tilted, with a small depression going into the raised upper edge. With the fringe-to-fringe separation being 1/2 wave OPD (optical path difference), the tilt of the blue area is little over two waves. Since tilt itself doesn't degrade point image, merely moves it in the direction of tilt, all needed to find the wavefront corresponding to the best image is to center it on that point (the raised edge adds another 5-6 waves, but it cannot be optimized tilt-wise simultaneously with the remaining, larger portion of the wavefront; its energy is, therefore, widely spread, contributing very little to the central diffraction maxima). The second OSLO option, "best focus" displays wavefront that combines tilt with a significant defocus over the blue area (middle). It actually has lower both, P-V and RMS error, with only a slight difference in the vertical shift (Y; the defocus shift along Z axis is not shown). In order to correct for the tilt alone, the third option, "reference ray intersection" needs to be used. Entering 0.002 Y-shift to the default coordinates, brings out the wavefront shape over the blue area for the best image proximity (right). It shows that most of this area is a low-aberration region, outlining a two-circle-intersection shape, oriented horizontally (the nominally higher overall P-V and RMS are the consequence of a larger error in the raised area, which is of little importance since it was already mostly wasted with respect to its energy contribution). This explains both, vertical elongation of the PSF central maxima and its integrity. Unfortunately, OSLO Edu has no option to calibrate PSF output with other than default data (it produces the best diffraction image decentered, in the process of generating PSF simulation for a wider area). However, entering -0.044mm defocus produces interferogram similar to that on the left, only with somewhat less of tilt in the blue area. For this point, PSF peaks at 0.52, suggesting it is the best focus (image below). Note that the actual PSF peaks are lower by 0.96 factor, or little more, due to the effect of central obstruction.

Note that even with OSLO making vertical shift adjustment, it is not the wavefront centered on the PSF peak, since it was looking for the "minimum RMS" error, and found it at an image point with still significant tilt error in the main, lower portion of the wavefront. If we manually enter -0.0002 shift - which should be equivalent of +0.0078 shift with respect to the default - we see interferogram nearly corrected for tilt and other aberrations (except a roughly 1/4 wave zone) in an area somewhat larger than with the "minimum RMS" focus corrected for tilt, which should be the reason for the higher PSF peak. But the P-V and RMS wavefront error are again larger for this point. This illustrates why both, nominal P-V and RMS wavefront error become less reliable indicators at large error levels. The "best focus" option is supposed to find what it promises, but it evidently doesn't always work as planned. Contrast transfer plots below show the defocused image at 0.5° as having appreciable higher contrast transfer than the "minimum RMS" focus. Note that these plots do include the effect of 20% linear central obstruction, omitted above.

The contrast advantage results in a significantly better limiting resolution of the defocused "minimum RMS" image for bright low-contrast (BLC) details. For comparison, dashed plot shows contrast transfer for 1/2 wave of defocus (unobstructed aperture).

Tolerance to miscollimation between the mirror and telescope axis is comfortable: 5mm decenter will lower the central e-line Strehl to just below 0.95, while 0.5° tilt at the objective - implying 4.4mm axis-to-axis deviation 0.5m behind - lowers the Strehl to 0.96. However, 1° tilt already induces over 0.4 wave P-V of all-field (even magnitude across entire image) astigmatism, with the Strehl dropping to 0.73. Miscollimated secondary would induce combination of tilt and decenter. Considering large separation, it is the most sensitive part of the system. As little as 0.1° of tilt would translate to 5.5mm decenter combined with 0.2° tilt.

Replacing the above Petzval refractor with two others (FPL53/ZKN7 and FPL53/BK7; the former has 0.999 central line Strehl, similar error magnitude in the g-line, but tighter rest of them, resulting in 0.954 polychromatic Strehl, while the latter has the central line Strehl already below 0.95, due to more of higher order spherical residual), all with different limits to spherical aberration correction show that off axis aberrations in this setup mainly result from uncorrected spherical aberration in the refractor: the more of it, the larger off axis aberrations induced (uncorrected off axis aberrations also are a factor when present). Since in this demonstration the refractor limit is at 0.018 wave RMS, or less than 1/16 wave of primary spherical, the actual units would produce more of off axis aberrations, and significantly so. For instance, the FPL53/ZKN7 combination has only 2.4 wave P-V wavefront error of the same form at 0.5°, and we are talking 1/51 vs. 1/16 wave P-V of primary spherical aberration (the FPL53/BK7 combination, with 1/8 wave P-V of primary spherical equivalent, has as many as 22.3 waves P-V at 0.5°). Considering that this field size (0.18° diameter in the actual telescope) is already within its medium-high magnification range, there is no room to absorb further field definition worsening. It is wise not to push too far with the objective speed: all else equal, an f/6, or f/6.5 could do noticeably better in this respect than f/5.

Terry Platt's 318mm f/21 Buchroeder "Quad-Schiefspiegler"

Described in detail in Tonkin's "Amateur Telescope Making", this tilted-mirror system stems directly from Kutter's Schiefspiegler, with the third mirror added to cancel out astigmatism once the primary and secondary are arranged to cancel out coma (in the book it is called "quad-schiefspiegler" but since reflecting flat does not influence system output, it is a three-schiefspiegler). Reflecting flat is added in order to create two alternative focus locations, as indicated below. However, since it doesn't influence raytracing, and the telescope is fully operational without it, it was omitted from consideration. Primary mirror is aspherized (-0.55 conic) in order to minimize axial aberration, with spherical secondary and tertiary. Field angle is chosen to create image that fills out 1.25-inch barrel.

Astigmatism plot reveals significant image tilt. At the field edge, defocus due to the tilt is over 2mm, or over 1 wave P-V. However, what matters is how much of accommodation it will require in the image formed by the eyepiece. Axial correction is much better than in the Schiefspiegler (considering three times larger aperture), with the design limit better than 0.95 Strehl. According to Zernike wavefront decomposition, nearly 90% of the axial aberration is trefoil, with some residual coma. Longitudinal astigmatism of 0.85mm at the top edge (somewhat less at the bottom) implies 0.44 wave P-V wavefront error of astigmatism, which makes it dominant aberration in the outer field, with small contributions from trefoil and coma. Field aberration is not radially symmetrical, but it is relatively insignificant since at the very edge of the field central diffraction disc remains well defined.

The maker of this telescope states that the image tilt - which is over 9° - does not significantly affect observing. It shouldn't be a surprise, considering the slow f-ratio, but it can be informative to assess it specifically with raytracing. Using 30mm f.l. 2+1 Koenig design, ray spot plots are shown for a perfect telescope (bottom left) and the actual telescope (bottom right). Note that the field angle is too large for this particular eyepiece design, requiring lenses larger than 1.25 inch barrel, and resulting in a sharp increase in the field edge aberrations (the corresponding undistorted field of view is over 53 degrees which, with 15% positive distortions produces 61° AFOV; a bit too much for a 50° AFOV design). Except the perifferal field, telescope aberrations dominate, which was to expect with this slow f-ratio. The effect of image tilt only shows toward field edge, increasing eyepiece aberrations on the bottom, and decreasing them on the top (the two field sides should have edge astigmatism at the level of the eyepiece in a perfect system). Accommodation required for the field edge is less than 1.5mm which, for 30mm f.l. eyepiece i.e. diopter equaling 0.9mm, translates to nearly 1.5 diopter of accommodation (infinity to 0.67m), positive on the field top, negative on the bottom. Without excessive edge astigmatism of the eyepiece, accommodation required would have been yet smaller.

In all raytrace confirms that the telescope, as a design, offers well corrected field despite nominally significant image tilt. That, however, assumes well collimated system. The 0.80 axial Strehl tolerance for tilt is 0.025° (less than 1/14mm deviation at the edge) for the primary, 0.05° for the secondary, and 0.035° for the tertiary (for the aberration induced by tilt added to the existing design minimum). Since this is with the "tilt and bend" choice "on" - meaning that change in tilt of the primary automatically adjusts axial position of the secondary and tertiary (and change in the secondary's tilt adjusts position of the tertiary) - the actual tilt tolerances should be tighter. On the other hand, secondary and tertiary are relatively insensitive to decenter (up to a few mm still leaves the Strehl well above 0.80).

Bratislav Churchich's 146mm f/3.9 Wright camera

Much less well known than its older cousin Schmidt camera, the Wright camera is not as well corrected, but has features that can make it preferred choice for some applications. In relatively slow systems, correction difference becomes less important, and its better compactness and inherently flat field can overcome the disadvantage of having to make an oblate ellipsoid for the primary and twice stronger Schmidt-type corrector. The latter apparently wasn't an obstacle for Churchich, who in order to make the camera more compact decided to go with even more strongly aspherized primary (nearly 1.5 conic), which also required as much stronger Schmidt corrector. As he describes in the same Tonkin's book, he stopped a bit short of the optimum primary conic (1.5) preferring to keep an exceptionally smooth figure. All this makes his Wright camera different from the standard. The corrector was made with 70.7% neutral zone, being the easiest shape to fabricate, but it also has the lowest spherochromatism (86.6% neutral zone has the smallest blur, but over 2.5 times larger wavefront error). Table below describes the perfect Schmidt profile needed for this instrument. Any deviation from it generates wavefront error by a 0.517 ratio (e-line), since the light speed in the BK7 glass vs. air is 1/1.517:1.

It shows that its maximum depth is less than 10 waves of 550nm light, i.e. that corrects mirror wavefront error 1.517 times smaller, around 6 waves P-V. The diagonal mirror is 55mm in diameter, taken as the size of central obstruction. Raytrace of this system below assumes perfect surfaces.

Due to the relatively slow focal ratio, chromatic correction is excellent, comfortably within the "true apo" requirements. The ray spot plots show residual coma (about 0.80 Strehl large at 1° off), but it is low enough not to significantly alter the shape of diffraction image (for fully canceling coma, either primary conic of 1.5, or corrector separation of 460mm are needed). The diagonal is not quite large enough to fully illuminate 2° field (it is raytraced w/o offset, but in the actual telescope is offsetted to be centered under the focuser). At the field edge, vignetting is about 10% (taken as that on the horizontal ends of the diagonal, the 3rd from left), which is within acceptable. Note that 711 out of 917 rays is what remains after central obscuration.

Since the main purpose of the instrument is imaging, what really matters is size of the circle of energy. In this case, the camera places 80% of energy into 8-micron circle on axis, 10-micron circle at 0.7° and 14.8-micron circle at 1°. Angularly, it corresponds to 2.9, 3.6 and 5.4 arc seconds. The actual star image, as the diffraction images implies, is a bit larger, some 17 microns 1° off (of course, before seeing-caused bloating). The Airy disc for an f/3.9 system is 5.2 microns in green light (546nm) but these numbers are for the five wavelength from 436nm to 706nm with even sensitivity. Since neither photo emulsion nor CCD have flat sensitivity in that range, the actual circle sizes are somewhat smaller. In this case, they were smaller than the typical photographic blur, despite the unplanned, not negligible spherical aberration residual. If the system was built as the standard Wright, with a weaker primary and corrector, the 80% energy circle 1° off axis would have been 13 microns, not much smaller than 14.8 microns in Churchich's customized camera (if optimized by going for 1.5 primary conic, the circle would measure 13.6 microns).

Dan Cassaro's 11.4-inch f/4.3 Slevogt-Schmidt camera

Although still in making, this ATM camera posted at the CloudyNights forum is both, interesting and challenging. In order to obtain flat field with the Schmidt corrector and two spherical mirrors, corrector needs to be widely separated from the mirrors. The alternative is the Write-Schmidt, less than half as long, but with significantly more field astigmatism, and requiring a hard to make oblate ellipsoid as the main mirror. Its advantage is significantly smaller central obstruction, but if field quality is main concern, Slevogt-Schmidt is the way to go.

Home  |  Comments