Optical shop testing
Optics have to be tested before you can send them to coating services or mount them in your telescope. We have to verify that the wavefront conforms to the requirements defined by the optical design of the telescope. Taking into consideration the tight tolerances and the fact that we are working with fractions of a wavelength of light, it becomes obvious that we cannot measure the shape of the optical elements by mechanical means.
Over the past centuries several indirect optical test methods have been developed in order to be able to accurately confirm the shape of the optical substrate and in consequence the correctness of the shape of the wavefront. The test methods mainly used by amateur telescope makers are the Foucault test and the Ronchi test, but more and more mirror makers also make use of interferometry. On this page I will show the setup I use for those three test methods. For a more theoretical background I recommend the books: Optical shop testing, Understanding Foucault and Telescope Optics Evaluation and Design. You can find links to those books in the right sidebar of this page.
We differentiate essentially between quantitative and qualitative test methods with Foucault and Interferometry belonging to the first the Ronchi test belonging to the latter. Quantitative meaning that the test returns numerical results that we can use to model the exact shape of the optics while qualitative test methods allow us by visual means and representation to judge the shape of the wavefront instantly without the numerical exactitude of the qualitative results. Evidently, each test method has its own range of application and I like to contrast their results by using them concurrently.
The testing of optics might very well be one, if not the most interesting aspect of Amateur Telescope Making, as we are dealing here with the very nature of light at such miniscule scales of measurement.
- Test stand
- XYZ testing stage
- Foucault/Ronchi tester
- Bath interferometer
- Webcam adaptation
Universal mirror-test stand:
Growing tired of makeshift mirror stands manufactured specifically for each mirror, I decided to construct this universal mirror test stand. The main advantage of this type mirror stand is, that you can place the mirror on the test stand horizontally and then move the testing platform into testing position. This obviously is much safer than awkwardly trying to place a 16 Kg mirror blank in 90° position on the test stand.
The stand doubles as storage space and can easily be wheeled out of the way whenever it's not needed.
- My mirror test stand can be used for mirrors up to 18 inch. The design however can be scaled up to any dimension. The holes for the different mirror diameters have been predrilled and include T-nuts on the backside of the wood board.
- The edge supports you can see on those photographs are nylon discs. In the meantime I have migrated to using brass discs with a narrow rounded over edge. I found that the nylon discs deformed ever so slightly under the weight of the mirrors and introduced unpredictable aberrations in the test results.
An easy to build spherometer:
A Spherometer is a very important tool during the process of grinding a mirror. While we cannot use it for testing the completed mirror, we need it to control the process of grinding a spherical shape into the glass and for confirming the depth of the Sagitta in order to be able to arrive at the correct focal length.
There are many different types of spherometers, the main difference being the number and type of support/contact points. The highest accuracy of measurement can be achieved with a three point spherometer but even a two point spherometer with outrigger as it is presented on my page is sufficiently accurate when calibrated carefully. My instrument is based on the design presented on Bob May's web page and is fabricated from standard aluminium parts and standards screws.
For calibration I used to glass discs which have been fine ground against each other and therefore are matching counterparts. Measuring the deviation from flatness of those discs allows us to calibrate the spherometer for that particular offset.
- Those images show the smaller of the two spherometers which I made. It is made of standard aluminium square and round stock and features a Mitutoyo Digital dial indicator for readout. The effective measuring length of this spherometer is 10 cm.
- My second spherometer has an effective measuring lengths of 19 cm and is otherwise similar to the smaller model.
- The feet of the spherometers are composed of 6 mm steel balls which have been glued with superglue to the ends of the set screws. To provide a larger gluing surface the ends of the set screws had been slightly hollowed on a drill press with a 6 mm drill bit. The diameter of the steel balls should not be too big and its value is factored in the formulas that are used to calculate the actual sagitta and focal length.
XYZ fine motion testing stage:
Building an XYZ translation stage usually presents a major obstacle as it turns out to be quite difficult to construct such a contraption with common tools that allows for precise positioning and at the same time is stable enough to avoid introduction of vibrations and flexure.
I chose to build mine from plywood and aluminium, as I had those materials available in my shop.
The main difficulty consists in providing movement in the 3rd or "Y"-axis which we need to adjust tilt in testing with the interferometer. For Foucault and Ronchi testing it is usually sufficient to provide movement in X- and Z-axis. Commercial units are commonly built as equal modules that can be attached to each other and I adopted that approach for my model as well. It consists of 3 similar stages that once assembled can be mounted to a sturdy tripod.
Initially this XYZ stage was meant to be an experimental prototype, soon to be replaced by a refined and prettier version. However, it works so well that until now there was no need to replace/rebuild it.
- This image shows the completed XYZ stage completely assembled with Foucault/Ronchi and Bath interferometer testers on top of my heavy duty tripod. All 3 stages are clearly visible and you can also see the Mitutoyo digital dial indicator attached to the Z-axis stage. The dial indicator is used only for Foucault tester operation.
- Detailed views:
- Each stage is made up of plywood base and carriage, 2 stainless steel rods and 2 aluminium end pieces, M6 threaded rod and turning knob.
- 2 stages are directly connected via carriage/base. The 3rd stage is attached with a angle-bracket
- The carriage is driven with the M6 threaded rod and moves quite smoothly along the 10mm steel rods. Future designs might replace the plywood carriage with Teflon™ blocks!
- I opted in my setup for a configuration where the Z-axis stage forms the base and the X-and Y-axis stages are coupled and attached at 90º to the base stage. This was done mainly for practical considerations and to allow more space for the dial indicator. In terms of stability, it might be beneficial to couple X- and Z-axis instead. However, as you can see on this photograph, despite the cantilever the assembled stage has a lot of stability and the use of 20mm plywood dampens out most of the vibrations.
Quick change Foucault/Ronchi tester
I assume the reader is familiar with the basic theory of Foucault and Ronchi tests. There is a lot of literature available and I recommend in particular Optical shop testing and Understanding Foucault for further information. You can find links to both books in the right column of this page!
Slitless Foucault tester:
Using a slit instead of a pinhole for the knife edge test has several advantages. One of them is better illumination of the mirror in comparison to the pinhole. But the slit-test requires the knife edge to be exactly parallel with the slit. This can be difficult to achieve.
The slitless tester overcomes this problem. Instead of producing a very fine slit with two razor blades, a round light source (super bright LED for example) is half covered by a razor blade or utility knife. The half-moon shaped light is reversed by the mirror and thus produces in combination with the razor blade a slit of variable width.
If we arrange in the setup for a sight hole above the light source that is half covered by the same razor blade that covers the light, we eliminate the need for a separate knife edge and by that the problem of getting the knife edge parallel with the slit. Moving the tester sideward until the mirrored half-moon is almost covered by the razor blade, we can produce a very small slit. Moving the tester even further sideward we simulate a knife edge penetrating the light-slit. Since the virtual knife edge is at the same time the object that cuts the light source, it will be automatically parallel to the slit.
By changing the distance from the light source to the mirror we are able to measure the longitudinal difference between the focal points of predefined mirror radii. That is basically what the knife edge or Foucault test is all about.
We only have to take in consideration for the data reduction that the knife edge moves along with the light source in our slitless tester .
My Ronchi tester uses the same light source as the Foucault tester since Ronchi grating and knife-edge (razorblade) are mounted to a rotating support. This setup allows for multiple testing configurations like gratings with different line density.
While I earlier used to look directly through my testers, I now exclusively use a webcam for monitoring the Foucault- and Ronchigrams. The webcam is mounted permanently above the light source and as close to the knife-edge/grating as possible. Different webcam lenses are used depending on the focal length of the mirror being tested.
- The Foucault/Ronchi section of my tester:
- The knife-edge and grating(s) are mounted to a plastic disc, thus allowing for quick change of tester configuration. The disc is a cover disc from a CD-drum.
- As light source serves a bright white LED with some frosted glass as diffuser.
- The webcam is mounted very closely to the light source.
While Foucault and Ronchi tests give you a lot of information about the mirror being examined, both test arrangements depend on the interaction of the test person to provide usable results. For the Foucault test this means that test results will only be as accurate as the readings that have been taken and the interpretation of the Ronchi test depends heavily on the experience of the person executing the test.
Interferometry is different in that respect, as once you have taken care of the proper test setup and data acquisition procedure, the analysis of the acquired data is entirely done by the computer. Of course there are still many variables that have to be considered for test setup and data acquisition and I will talk about them in detail later, but with careful arrangement and diligence during execution of the test, we will be able to achieve highly accurate and repeatable test results.
I will not go into details about the theory of interferometry, as plenty of information about this topic is already available on the web. One excellent starting point for finding out more about interferometry is the Interferometry WIKI, which also includes a dedicated section about the Bath interferometer.
In interferometry we compare the wavefront of the optics under test with the wavefront of an existing reference optic. In order to do that we have to overlay the two returning light beams — the beams being called "test beam" and "reference beam" — at the light detector which can be any kind of camera or WebCam, depending on the setup.
The Bath interferometer (named after its inventor Karl Ludwig Bath) falls into the class of common path interferometers which means that both reference – and test beams follow the same optical path. This greatly simplifies setup, which is one of the reasons why the Bath interferometer is so popular among amateur telescope makers.
Interferometry is highly sensitive — essentially we are zooming into the scale of a wavelength of light — and the test setup is very susceptible to air currents and vibrations. For that reason it is very important to establish a repeatable test procedure and get to know our testing environment to be able to eliminate most or all factors that could invalidate our test results.
Jan van Gastel incorporated an attractive testing tunnel in his design with good results.
Vladimir Galogaza wrote an excellent manual about building and using a Bath interferometer. It can be downloaded/viewed here.
The Bath interferometer is a fairly simple device consisting of the following parts which can be purchased at Surplusshed for a very reasonable price:
- Surplusshed item No: M2122
Any cheap laser will serve the purpose. It is best to remove the housing and mount the laser module in a custom-made fixture that permits rotating the laser in order to improve contrast of the interferogram because the laser beam is partially polarized.
- Surplusshed item No: L2046D
Beamsplitters tend to be expensive items, but the non-polarising 50/50 item available from Surplusshed generally yields very good results. Since we want to keep the distance between the two parallel rays as small as possible, it is convenient to choose a small beamsplitter size. The 15 mm wide unit works perfectly and should be the first part to be mounted, as all other components will have their location defined in relation to the beamsplitter. Additionally it should be mounted allowing for rotation around its Y–axis.
- Small mirror:
- Surplusshed item No: PM1012
There's nothing special about the small mirror. Its purpose is to produce a parallel beam at the smallest possible separation.
For that reason again we select the smallest item available.
When mounting the small mirror we should allow for rotation and longitudinal movement along the side of the beamsplitter.
- Diverging lens:
- Surplusshed item No: L4438 or similar. There is no direct link available. You have to search for lens specs via "Lensfinder".
As with the other components I recommend to follow the indications of the interferometry wiki which recommends a double convex lens not bigger than 10 mm. I used a 6 mm lens with 4 mm focal length which results in a very wide light cone.
The lens mount has one side cut off in order to provide space for the parallel light beam. To aid testing setup, the lens mount can be moved out of the light path and like the other components it can be rotated around its axis.
- I use a WebCam for data acquisition and over the years I have tried several models and found that the single most important issue apart from CCD/CMOS chip quality Is the configurability of the WebCam driver. In particular we need to be able to control exposure manually. Unfortunately most drivers of cheap WebCams will not allow that.
However, until now I have had my best results with a five year old LIDL camera that at the time cost 10€.
- Camera mount:
- One of the big advantages of using a WebCam for data acquisition is that we can achieve a very compact layout for our Bath interferometer that can be mounted permanently and is therefore available for testing instantly.
Of course, the original lens that comes with the camera can not be used, as those lenses are wide-angle lenses and we need lenses with longer focal length in order to fill the maximum chip area with the image of the optics under test.
Depending on the focal length of the optics we want to test, we choose the focal length of the camera lens to make maximum use of the chip area.
I use simple plano convex 10 mm lenses from surplusshed mounted into lens/camera holders that I turn on my lathe (see next chapter on this webpage for more details).
Commercial lens mounts are also available from UKAoptics.
- Optional power supply:
Variable Power Supply for Laser
It is beneficial and in my opinion highly recommendable to build a dedicated power supply for the laser. Not only will it help to save on batteries but in order to achieve maximum contrast on the interferograms it is very important to be able to regulate the brightness of the light source. There are many simple circuits available on the web to that end and I provide one for reference.
In relation to the above mentioned circuitry Bruce Griffiths recommends:
A simple modification to limit the maximum output to 3V: Change R1 to 150 ohms and R2 to 200 ohms.
If you check Farnell's range carefully you will find much cheaper cermet pots than that recommended for R2.
Adjust the value of R1 slightly to set the range and compensate for tolerances in R2.
Use a 1% metal film resistor for R1.
Do not use carbon resistors for R1.
- The basic layout of the Bath interferometer can be seen on those photographs. Evidently the tester occupies very little space. I mounted everything on the small board of Plexiglas simply because that was what I had available in my shop.
All components have been mounted to small pieces of aluminium which in turn are attached to small screws that allow for fixing and moving around the base board.
- As I mentioned earlier an interferometer is extremely sensitive to exterior disturbances like air currents or vibrations.
This video shows the effects of a spinning washing machine while testing a mirror with the Bath interferometer. Washing machine and interferometer are separated approximately five metres and the flooring is solid concrete. The interferometer is supported on a tripod equipped with antivibration rubber feet.
While I'm testing in the basement of our house I can usually make out if someone is walking on the second floor by simply looking at the interferogram.
Webcam adaptation and lens mounts:
In order to be able to use a WebCam with Foucault, Ronchi or interferometry, we have to provide the camera with a custom lens.
The focal length of the camera lens is chosen depending on the focal length of the optics that have to be tested and the physical size of the CCD/CMOS chip so that the entire image of the mirror fits on the active part of the image sensor. If you are using a 1/4" image sensor then the imaging area is 3.2mm x 2.4mm and the lens focal lenght will need to be flens = 2 × FOptic under test × 2.4 to fit the image within the sensors active area.
Example for an F5 mirror: flens = 2 × 5 × 2.4 = 24 mm
Another thing to consider for the use of WebCams is the loss of resolution on colour CCD/CMOS chips due to application of monochromatic light sources:
Quote from the QCIAG mailing list:
A colour 640 x 480 pixel CCD is covered with a Bayer mask. On 4 pixels there is one covered with red, 2 with green and one with blue.
If the laser is blue or red, only 1/4 of the pixels will get some light. If the laser is green, only 1/2 of the pixels will get some light. The monochrome CCD will get 4/4 ! If we use monochromatic light sources like laser or monochromatic LEDs in our testing setup, then it would be beneficial to use only b/w WebCams!
- The lens mount body and lens retaining ring.
The lens mount is turned from solid brass and the retaining ring is turned from mild steel. The thread is a standard M12/0.5 mm pitch WebCam lens thread.