My astronomy project:
Adapters and modifications
1: Overview wide field and planet configuration
Certainly more than I have struggled to get together the mechanics so that the optical parts end up at the right distance. This is my first draft that and I will try it out, thinking it surely are of interest to many so I took some photos to show and wrote some text to it.
Here is an overview of the parts, from left to right:
adapters has been selected to handle the full frame camera (which will be a big disappointing later as you will read).
Achieved optical length is 73.2mm, 0.3mm difference from the optimal, but some day the camera will be IR filter modified, the optical distance is changed so that margin is good to have. If necessary I put shims between the parts until exact distance is achieved.
Unit no. 2 may be replaced with a more compact version that is only 1mm thick in case more accessories has to be mounted. Riccardi field flattener with reducer with 2.5" diameter is in the smallest limit to be used with a full frame camera, optical image circle is about 42mm. But the border is not as sharp so I think it will work pretty good.
Here you have all the parts assembled, feels quite sensibly, have used threads as much as possible because it is more stable than the 2" pipe coupling.
Here is the optic package mounted on the telescope. It has not very much space for the guide camera, but the distance will be sufficient if the calculations are correct. Everything is relatively steady, the only thing that's a little bit weak is the bayonet connection to the camera. With the bolt protruding, the entire package can be turned relative the telescope.
Overview of the parts, from left to right:
I chose this Pentax tele extender, not sure if the optical quality is good enough, but provides a good mechanical stability. I already have them so natural to try these first, moreover planet photo is not the main purpose.
Here are the parts assembled, without T2 to M42 adapters that I will use later. T2 and the Pentax M42 threads is quite similar, differing only in the pitch. But you must have an adapter between them, otherwise the threads will be destroyed.
Here camera and tele extender attached to the telescope. Steady and good fitness, also possibly to add an additional tele extender.Back to contents
2: Calculation of the optical system
Here is the theoretical calculation of the optical system above:
Apparently, the system is heavily under sampled with the telescope's relatively short focal length and the camera's large pixels. The system is more optimized for low noise rather than high resolution. With a more modern camera as Canon 6D it would be improved and even more preferably a monochrome camera and in combination with Drizzle the optical resolution can be fully utilized without any optical changes.Back to contents
Advertisement / Annons:
3: First light test
How good was it now with the field flattener and the sensitive distance between the camera and flattener?
In the evening with no clouds came the opportunity to take some test shoots of stars. There are thirty images of 30 seconds each stacked together. No image processing more than a gamma curve. Taken from the balcony of the light polluted suburban.
How does it look in the center of the image? Focus ended up probably not quite perfect, the telescope became much colder after the focus was set. 1000x1000 pixel from a 12Mpix sensor, approximately 25% of the image width.
Here is the left corner at the top. Is clearly visible that the stars are elongated. The distance camera - corrector is set to 73.2 mm, the goal is 73.5 mm according to the Riccardi data.
This same corner but now the distance has been increased to approximately 74.7 mm, it got worse, may find a thinner spacer to try later. A rather complicated process to try out the distance needed for optimal result. And it's not only the distances, it should be orthogonal also to the optical axis. Do not know how good this Riccardi flattener is, but it's told to be of good quality.
The optical image circle of the field flattener is about 42mm, slightly too small for a full frame sensor. This Riccardi has a diameter of 2.5", 3" had been better and additionally resulted in less vignetting. The field flattener is also a reducer with power 0.75x, hence an extra optically demand.
But overall, quality is much better than what i had with my earlier Pentax telephoto lens 500mm f4.5, it was not an APO construction. The downside is that this refractor is much more expensive and much heavier. If I had chosen a fixed Canon 300mm or 400mm ED and f4 it had become cheaper and easier, and maybe also could have a smaller mount like the EQ5. No focus motor has to be built because it's already there built into the lens, even the field flattener. But one thing had not been possibly to implement with this solution and that is the off-axis guidance, it would have required a separate guide telescope. Though most fun is to build something together by myself and that I have 683mm to shoot with, good, right?Back to contents
4: Investigating vignetting problem
Now the telescope and adapters are ready for test and I can start picking up the first astronomy photos. I already in the first test noticed that it had some vignetting problem and I will now investigate what caused that.
The image in the example had just been dark subtracted and color channels where scaled, red 150% and blue 190% so they normalize at the same level. There is a stack of 53 x 30 seconds at ISO 800. If the signal is about 200 ADU in the center and one reads of 140 at the edge, the signal is thus reduced by 30% due to vignetting.
Here it's clear that the system vignetting heavily in places where it should not be a problem. Note that vignetting is not rotationally
symmetrical but rectangular in parts, the two circular vignettings comes from limitations in the adapters.
The rectangular vignetting can not be interpreted in any other way than the camera body is too narrow. Have written about it
before but now I have measured it so it can be better seen what happens. With a worse field flattener in 2" mount this had not been seen so clearly
then instead the field flattener inherent vignetting had dominated and it had seen more like a circle. NOTE: There is a dark frame along the edge because it is
photographed with dithering, but it is barely visible on this scale.
Such large vignetting like this is very difficult to handle with flat field calibration. You have to crop a big portion to get rid of it.
You may have read in my thread of thoughts about the Sony A7 series mirror less camera. Canon has a back focus of 44mm Sony A7R II has only 18mm,
Let us study it in more detail:
Here is an line graph done from the top down to bottom (vertical). The sensor is 24mm in height and the signal strength of the edges drops dramatically. It is the camera's mirror in the up position that obscures the light cone. The two vertical lines are the stars I happened to cross when I drew the line. The signal drops to 140 units here at the edges, should instead remained at 160-170 with this field flattener. Extend only the line that goes to the edge without the sharp bend that becomes close to the edge.
One thing that puzzled me at first is why the signal is higher on the right side and not symmetrical, furthermore distinguishes between the different wavelengths. But the side is higher on the downhill. Interprets it as a gradient starry sky and downside has more light pollution. Se also the sharp bent to the right, it is the top of the camera body where we have the mirror.
Here are the corresponding line graph horizontally. Apparently, the signal is much higher here at the equivalent distance of 24 mm or +/- 12 mm (+/- 1400 pixels) from the center. Compare the level of the corresponding distance in height you can see that they have the level of 170 units and not 140 as it was in the previous line graph. This is the level you have from the corrector itself, the narrow width of the camera body and small openings in adapters limiting the signal farther to the edges. Very irritating to loosing signal due to this.
Here it is instead the diagonal that the signal is measured along. The second peak from the left side is a galaxy. Here one can see that the signal should be around 160 units at the edges. Thus, the center 190, 12 mm out of 175, 18 mm out of 170 and edges 160 units, very good values and thus little vignetting. So what to do with the camera body restrictions on the cone of light into the sensor? This was a problem I had not taken into account, have actually never heard it mentioned, but now when studying camera body closer it is so obvious.
I was aiming at the M106 galaxy, have not really had time to check where it ended up in the picture. Perhaps it is that galaxy we see in the upper left corner. Also I have done a line graph to show how the Galaxy stands out from the background. Galaxy core has signal strength about 300 and the disc is around 150, from this subtract the background so the disk might have only one signal contribution of 5 to 10 units. The background signal 150 has an internal noise of about 12 units, greater than galaxy disc. Therefore, it's not good to shoot astronomy photos from my balcony but far out of town so this background signal and noise disappears. Alternatively, the narrow-band techniques, on nebula.
Finally, I have detailed studied how the dithering technique has worked in the hot pixels (red here). As you can see dithering spread out over several pixels at a weaker level as a consequence, note this is an extreme case with a hot pixel of perhaps 100%. However, there have been two groups, the left and right. Something has probably gone wrong with the automatic stacking process out so that reference star had been shifted to some other. Explain the part of the elongated stars, but still depends mostly on the optics, the distance between flat field corrector and the sensor is wrong, and yes, with some fine adjustment it is possible to get it better.Back to contents
Advertisement / Annons:
5: Vignetting analyzes
Devoted time to take flats to calibrate the images. Used data monitor and a cardboard disc diffuser and sets the RGB color of the monitor so that the signal becomes 50% on all color channels of the camera in center (raw), that's the way I usually do. In the camera histogram window this level is on the right side, it will be so because the camera shows histogram with log scaling. The signal level must be checked with the appropriate program in the computer, I use Fitswork and rgb gain are set to = 1 for the RGB channels, it is important when doing analyzes to do like this. I did not succeed to remove all the dust on the sensor, and unfortunately it is not the dust when the currently used test image was taken. Has changed little on the mechanical adjustments since I took the picture I do test on here which is visible in the calibrated image. But curious as I'm I couldn't wait.
As I keep on testing with different distances between field flattener and the sensor becomes easy to see that the different parts have rotated between trials. This is visible in the upper part of the image, how off-axis adapter ended up in a different angle and the flat calibration is therefore incorrect. This can be calibrated away with flats, but also believe that the prism to off-axis adapter can be pulled out a millimeter or two. When all this is finished it is important to secure the position of all the mechanics so it does not move and thus ruin the flat calibration. At the center of left edge there is a visible little darkening, it's the heel that stands out in the camera body that mirror resting on in folded down position.
Strangely, not visible to the right, it seems as if the sensor is not centered to the middle of the optical axis which can be seen in other pictures too. For what reason? The darkening in the bottom in the picture is a real serious problem. The camera's mirror in the up position blocking parts of the lights. More or less such problems are for all DSLR cameras with full frame sensors. I have, however, contrast enhanced it significantly, it's also that my background light is 10 times higher here than it's on the countryside, hence that's why it stands out so clearly. There are also two concentric rings, that effect are from the adapters that are not wide enough.
All together at this situation my useable sensor area isn't much larger than a APS-H sensor. I'm very happy with the off-axis adapter and the opportunities it offers so here I do not want to change too much. The telescope, I am also pleased with although it was very much clumsier than the 500mm f4.5 lens I had before, but less than the 10" Newton. The camera I am pleased that there is a full frame sensor. But the house is so crowded in a DSLR as it causes such problems is not good. Solutions I must consider to get rid of vignetting problem:
A full frame sensor without mirror, like the Sony A7 series would otherwise be an ideal solution. Additionally 26mm more in back focus is attractive. But there will probably be many surprises and problems here too, what you see right away is that the bayonet entrance hole is too small, only 46mm. The latest I have read about Sony is that they do some ugly things to the raw files, very bad. (Update: The later Sony A7R II has much less compressed raw files).
Alternatively, would surely be a monochrome CCD camera, but such a full frame sensor camera costs a lot of money, from 4000 Euro with filter wheels and then it's still a pretty simple sensor. On the whole, it feels that Kodak's old sensors are well antique today compared to what Sony accomplishes. Quite properly outside the budget too.Back to contents
6: Why do we get vignetting from the DSLR camera house?
How it come that the mirror in a DSLR camera can block the marginal rays of the light cone, it can not be seen with ordinary camera lenses?
DSLR camera mirror must sit very close to the sensor to let the viewfinder to see the whole sensor when the mirror is folded down, 98% of the field is normal. If you put it further away then it causes to have a larger mirror and then the distance between the mount (optic) and the sensor is increased. It's not practical to increases the size of the camera body, larger mirror also becomes slower.
Study the ray trace above I made. Telescope-like optics with big opening gives a large diameter on the incident cone of light towards the sensor. Seen quite clearly in the picture what is happening. In my case with the Canon 5D it looks like 30% of the cone of light is blocked by the mirror of the light ray that's directed to the sensor edge. Canon has a fairly large opening in the bayonet, 54mm, bigger than most other DSLR cameras. For a normal camera lens the light cone is not that big and nothing like this phenomenon occurs. A little rough estimate, maybe you can appreciate that when the rear lens element exceeds 30mm and at full aperture, this should be a problem. Mirror less cameras for example. Sony A7 would solve this problem if just the camera body entrance had been large enough, now, unfortunately, Sony bayonet have only 46mm free opening, but the back focus is short, only 18mm. Off-axis adapter causes a similar problem, it is placed further out from the camera, the distance is about 50mm from the sensor. The pickup prism sits normally along sensor's long side and the sensor on a full frame is 24mm in height. The edge of the sensor is thus 12mm from the optical center. But at a distance of 50mm the light ray cone diameter is quite large.
It gives a diameter of the optics of 200mm. Close to the sensor the light cone is almost 0 mm in diameter. At 50mm spacing is 50/1000 * 200 = 10mm. The center of the cone of light will also slightly closer to the optical axis, 2.2mm. We do not monitor the center, it's the edges that's important. The sensor height is 12mm outward from the optical axis is calculated. The distance to the off-axis adapter prism must therefore be at least (24 + 10-2.2) /2=15.9mm out from the optical center. All that is in the way in a radius of 16mm will block light rays. Unfortunately prism of normal off-axis guiders are in this area. In the other direction the sensor is 36mm, it gives a radius of 22mm must be free. I have set up an excel sheet to calculate this. There is a limit how far away we can put the prism from the optical axis, optics image circle will set a limit even if the mechanics would allow it. The light must come into the off-axis pickup-prism, if the image circle is too small on the optics then the off-axis field is dark at greater distances. With a shorter focal length and a less opening problem it becomes easier to solve. Or an APS-H sensor camera.Back to contents
Advertisement / Annons:
7: Calculation of free opening to avoid vignetting
Here I have made a detailed excel sheet.
Here you can now easily calculate forward valuable information on his telescope / camera system. One can see the minimum clear opening required at different distances from the sensor to the example, adapters and filters. If there is not space enough for circular openings maybe there is a chance to have rectangular openings, dimensions in the table. In addition, there are calculated as the minimum distance an off-axis guides must be to not give vignetting. Assumes that it sits along the long side where the sensor is narrowest.
I have now measured my optical system where I have vignetting problem. It looks like this:
o.a. = distance from optical axis, ADP = adapter.
This will a problem to solve. There is only space for a thin off-axis adapters (max 11mm) and they are not in the larger design than M48. Those with 55mm through hole builds on too much in length to let a DSLR camera be able to connect. Does a mirror less Sony A7 solve the problems?
How would the foregoing appear together with a Sony A7:
Significantly better, but no 3 is still a bottleneck. Unfortunately, this adapter from M63 to M48 is very special. In principle, I could grind up the hole in it to 55mm. Since grind of the off-axis adapter's M48 thread and then screw or glue them together. This would improve the passageway through the adapter off-axis, but not completely.
This off-axis adapter has sufficient through hole, 55mm:
I have thought a lot about it, but unfortunately it is built with a back focus of 24mm compared to 11mm that my current TS off-axis adapter has. Perhaps it would be possible make it a little thinner, but not enough. Another disadvantage of this tube "hole" up to guide the camera is very small compared with the TS model. But once again, together with a Sony A7 series camera it can maybe be solved.
But very much suggest that life in full frame world would be a lot easier with a mirror less Sony A7, but then we have the problem with the raw files of this camera and how to control the camera from computer?Back to contents
8: Modification of the adapters
Return to the workshop and problem solving.
As you read earlier, I have problems with vignetting in the adapter between the field flattener and camera, and then the camera body itself. Thought about selling my Riccardi flat field corrector and buy one that can handle larger sensors. But it's really unnecessary, already this in 2.5" size handle 42mm image circle. Full frame is 44mm and one mm losse in the corners are not so important. If I buy a field flattener in 3" design with matching off-axis adapters, and some adapters rings, I will have a cost of 1200 Euro or more. Not really in my budget. And as I wrote earlier, it will just give me other kind of problems to solve.
So back to the problem, it's the original adapter that the main problem lies in that it's the one you should go for. One may well wonder how they can manufacture an adapter that makes it impossible to utilize the expensive Riccardo corrector fully and causes all these problems. For the ones who is satisfied with APS-H format, it is not a problem but then you need of course not so expensive corrector like this one.
So it looks like this:
The adapter holes on the input side is 45mm and the corrector lens on the output side is 48mm. The light cone is slightly conical, but it shall of course also reach out to the corners of the sensor. Means that the free passage must be at least 48 mm of the hole if it's circular. Should it fit with flock paper needed further a few mm. I should be able to modify it to nearly 52 mm so that part can be solved. As a bonus, the off-axis guider will get more light and thus it becomes better of the guidance to weaker stars.
The off-axis guider and Riccardi adapter are paired with a M48 thread, it will be grind away completely. Feels a bit sorry to do this on this new adapters, they cost close 300 Euro, but as they are now they are not usable today so it is not so much choice.
When I lathes away the M48 thread, this gap of 2.5 mm disappear, maybe it will be to weak and I keep a part of it. It admit that I have a little more adjustment range of the critical distance between the corrector and the sensor.
Here are the incoming side, this is where the hole should be taken up to 52 mm. To take up the hole from 45mm to 52mm means that lot of material disappear, fortunately, there is a lot to take of here.
The off-axis adapter's output side. Here should six 3mm hole be made. Three to assemble off-axis adapter with the Riccardi adapter and three to the stop screws which I will fine-tune the distance between them. It's crowded here, several passages of holes for other purposes as they do not allowed to collide with.
On the output side of the off-axis adapter is an another adapter, this is the M48 thread. Now we are closer to the camera's sensor and has not quite the same requirements. A hole of 46 mm should make it. Original hole in the adapter are here 44 mm, it may be possible increase it, but there will be no room for flock paper. An alternative solution is to grind up a rectangular hole with rounded corners. It appears to be feasible. This adapter has a different problem, it tapers in the bracket and locked with the three set screws. It is not so mechanically stable and my camera is pretty heavy, but I keep the solution so long.
Today, I use a Canon EOS full-frame camera, but like any full-frame DSLR cameras have narrow inlets to the sensor which make problems with telescopes with big openings. In my case, a 130mm F/5.3. For this adapter do I make no modifications at the end to the camera bayonet. Maybe it will be a mirrorless Sony A7 serie camera in the future, it assumes that Sony removes the compression they put into the raw format. Maybe a option of a real full-frame astronomy cameras show up, but it is not so likely because of the cost.
Now it has come so far in my planning that I can take up the holes needed in the Riccardi adapter before the modification of it.
The aluminum can be a bit difficult to drill, it is easy to have the drill stuck and it breaks. All my tools I had disappeared when I moved so I had to buy new ones. In Jula, a Swedish company, I found this drill kit with Titanium coated drills.
When drilling I took my time so that no drilling accidents occurred. Continuous lubricating is necessary when drilling. It now hold nine holes, where three is 3 mm consistently and meet up in the second adapter part where 2.5 mm holes are drilled, the latter shall be threaded M3. Furthermore, there are additional three 2.5 mm holes, they shall also be threaded M3 and here shall the stop screws be installed to enable fine adjustment of the distance between the adapter halves.
Here is the adapter as seen from the other side, the recesses are for fastening screws trail heads.
The next phase will be to thread the holes for M3, for that purpose I bought a bunch of simpler set up, also from Swedish Jula. The 15 Euro gave a lot, hope the quality is sufficient.Back to contents
Advertisement / Annons: