Chip tilt test jig.

[gorann]

Why not just screw two pieces of round (or square) wood next to the hole and press the camera up against them by hand while you rotate the camera. That should keep it centered. However, I have not gone to IKEA yet to get the base for the jig so this is just speculation in my inner theatre.

Edited September 14, 2021 by gorann

[tooth_dr]

10 minutes ago, gorann said:

Why not just screw two pieces of round wood next to the hole and press the camera up against them by hand while you rotate the camera. That should keep it centered. However, I have not gone to IKEA yet to get the base for the jig so this is just speculation in my inner theatre.

Wood was my second choice, but my printer arrived a day early so it’s currently running off its first print 👍🏼 

[symmetal]

Made my version of the jig and it's a great success. 😊

Cut up an odd piece of 18mm ply for the box and 3D printed a camera holder to fit a 2" nosepiece, or if you're lucky a 2" diameter 48mm extension spacer as the one on my FLT98/ASI6200MM was exactly 2" outside diameter. Also made a 3D holder for the laser pen angled at 10 degrees. I tried one at 13 degrees but it was more critical in laser placement and the 10 degree one worked fine. A clothes peg makes a handy on/off lock. 😀

Here it is with the 6200, filter wheel, tilt adjuster and a 16.5mm long 48mm spacer fitting nicely in the 3D printed adapter. The filter wheel unbalancing it had no effect on laser pointing. 🙂 The 3D adapter centre hole is 51.00mm across. I did make one with 50.90mm diameter but it was rather too stiff to turn.

Here's two initial images with the camera rotated around 180 degrees between them. As you can see it was way off. There were 3 laser dots near the centre. One appears to be aligned with the pixel grid, the opposite one was much brighter, with a faint dot between them. I assumed the brightest one was the sensor cover but it didn't matter in reality as when the tilt was adjusted to optimum, all three dots stayed in the same place while the camera was rotated. I needed to put a lot of tilt adjustment with a gap on one side around 0.5mm 😲 as can be seen on the picture.

The filter being in the path didn't seem to affect it and reflections from it weren't visible. 

It was clear this evening so went for a test image and have never had the left side stars so good. The far right hand side shows rather bad coma 🙁 but the star centres are sharp. Not bad for a full frame sensor. The coma was there before adjustment, but didn't appear so bad as the stars were more out of focus. The CCDI images look good and corners pretty similar with zero tilt reported, (first time ever). CCDI seems to ignore coma and just use the star centre for the test.

Here's the corner crops of the test image. If not for the coma I'd be very happy but I don't have to keep fiddling with the tilt anymore as I know it's optimum. 😀 I tried another image rotated 90 degrees and it was very similar.

I'll have to live with the coma but it's only on the far right, and binned 2x2 it won't look so bad. Wonder if Es Reid can align the scope to improve it, as I did buy the FLT98 from FLO. 😁

Alan

Edited September 14, 2021 by symmetal

[gorann]

12 hours ago, symmetal said:

Made my version of the jig and it's a great success. 😊

Cut up an odd piece of 18mm ply for the box and 3D printed a camera holder to fit a 2" nosepiece, or if you're lucky a 2" diameter 48mm extension spacer as the one on my FLT98/ASI6200MM was exactly 2" outside diameter. Also made a 3D holder for the laser pen angled at 10 degrees. I tried one at 13 degrees but it was more critical in laser placement and the 10 degree one worked fine. A clothes peg makes a handy on/off lock. 😀

Here it is with the 6200, filter wheel, tilt adjuster and a 16.5mm long 48mm spacer fitting nicely in the 3D printed adapter. The filter wheel unbalancing it had no effect on laser pointing. 🙂 The 3D adapter centre hole is 51.00mm across. I did make one with 50.90mm diameter but it was rather too stiff to turn.

Alan

Great, informative and convicing post Alan! Thanks a lot👍 Just one question: is there any risk for light leaking in through the gap in the ASI  tilt adjustment plate?

[symmetal]

1 hour ago, gorann said:

Great, informative and convicing post Alan! Thanks a lot👍 Just one question: is there any risk for light leaking in through the gap in the ASI  tilt adjustment plate?

Thanks gorann. I've always put black tape around the outside rim of my tilt adjusters to avoid that possibility. Taking flats or darks during the day in particular would be risky having  a visible gap on show.

I'm surprised the filter in the path didn't seem to give a visible reflection. Maybe they are really good coatings. 😀

I think it's beneficial putting as much of the imaging train as possible on the jig to also eliminate any potential tilt errors from all the adapters and couplings. If you swap cameras a lot it's worth checking each camera on its own and then through the imaging train once on the jig just to check the train is good.

Alan

[ollypenrice]

On 14/09/2021 at 17:40, wimvb said:

I wonder if a well collimated laser collimator would work to adjust the sensor tilt. It might need a considerable extension to work. But the idea is the same as ordinary collimation; get a reflected light beam at the correct location, and stay there when rotating the optical elements.

Hmmm. That's one to think about. The difference is that you're putting the laser beam on axis. On that basis the return beams from chip window, cover slip and pixel would, in a perfect world, also be on axis, though the pixel reflection might be offset? But what if the chip window is not in quite the same plane as the cover slip? How will you know which is which? It's the cover slip that needs to be tilt-free. Does the off-axis laser give you a better chance of identifying the cover slip reflection? 

Olly

[wimvb]

1 hour ago, ollypenrice said:

How will you know which is which?

Whatever is fixed to the front cover window/plate, will not tilt when you adjust the sensor. Ideally only the sensor will tilt. So it will be the only reflection changing. If everything is perfectly aligned and parallell, all reflections will end up on top of each other, and into the hole of the collimator. It should work in theory. The question is; does practice agree with my theory?

If the sensor cover slip is not parallell, it will still move together with the sensor, because it is fixed to it. You have the same problem with figuring out which reflection is which, independent of the angle of incidence.

[symmetal]

An interesting development with different cameras. 🤔

ASI071MC. The pixel dots are much closer together and only one bright spot is visible at the assumed centre of rotation. It was slightly out so adjusted with the built in tilt plate very easily.

ASI1600MM with filter wheel, M42 to M48 adapter and M48 tilt adjuster. Wide spaced pixels like on the 6200MM but this time many more dots to choose from. I counted 6 around the centre, though some were very faint and while the camera was rotated they changed from all being roughly coincident to being spaced apart when rotated 180 degrees. Here it is before adjustment showing the large tilt error again. I chose the brightest spot which wasn't coincident with the pixel grid but this may not be correct and will have to do a star test to confirm. When adjusted to centre this spot the other spots still went from roughly coincident to widely separated as before. This spot was the same brightness as the pixels this time. The Baader L filter in the filter wheel may have been adding some visible reflections this time as each pixel dot had several faint spots associated with it.

Maybe the Beyer matrix somehow made the OSC camera pixels appear close together for some reason. 🤔 I'll try my ASI178MM and ASI178MC cameras to see if they exhibit the same phenomena.

Edit: I redid the 1600MM choosing the bright spot closest to the centre pixel spot as the one to centre on as the further bright spot is probably the front window. As they change spacing depending on rotation could imply that the front window is tilted wrt the sensor.

Edit again: The 178MM and 178MC produced the same wide grid spacing though there is also a faint dot between the bright pixel dots so the Beyer matrix doesn't affect it.

Alan

Edited September 15, 2021 by symmetal

[wimvb]

If this is a diffraction pattern then larger pixels = closer spacing of the light dots. And smaller pixels = wider spacing of the light dots.

[symmetal]

10 hours ago, wimvb said:

If this is a diffraction pattern then larger pixels = closer spacing of the light dots. And smaller pixels = wider spacing of the light dots.

The 071 has 4.78um pixels and the 6200 and 1600 are similar at 3.76 and 3.8um. The 178 had a similar pixel spacing to the 1600 but has 2.4um pixels. Although the 178 seemed to display a faint dot midway between the pixels too. All very interesting, particularly the central spots behaviours between cameras. No sign of clear weather for the moment so can't confirm whether my assumption for the cover slip location on the 1600 is correct.

Alan

[wimvb]

On 15/09/2021 at 19:51, ollypenrice said:

Hmmm. That's one to think about. The difference is that you're putting the laser beam on axis. On that basis the return beams from chip window, cover slip and pixel would, in a perfect world, also be on axis, though the pixel reflection might be offset? But what if the chip window is not in quite the same plane as the cover slip? How will you know which is which? It's the cover slip that needs to be tilt-free. Does the off-axis laser give you a better chance of identifying the cover slip reflection? 

Olly

I did a quick & dirty test of my idea, but I had too much wiggle room for it to work. The idea may work if the rig is stable enough, but it won't be easier or cheaper than the setup in the original post.

[gorann]

Then my jig is finnished. I found a bed-side table on sale for about 20 Euro that forms the basis of the construction. Good news was that I could not see any wobbling of the bright central spot in any of my three ASI2600MC, but it felt at the same time a bit frustrating to have built the thing just to find that I did not need it🥴

Cheers, Göran

[wimvb]

5 minutes ago, gorann said:

but it felt at the same time a bit frustrating to have built the thing just to find that I did not need it🥴

That's a bit like insurance; you pay dear money for something you may never need. But better to not need it than to find out you have tilt.

Edited September 19, 2021 by wimvb

[gorann]

5 minutes ago, wimvb said:

That's a bit like insurance; you pay dear money for something you may never need. But better to not need it than to find out you have tilt.

You are right! Thanks Wim, I feel much better now😁

[symmetal]

2 hours ago, gorann said:

Then my jig is finnished. I found a bed-side table on sale for about 20 Euro that forms the basis of the construction. Good news was that I could not see any wobbling of the bright central spot in any of my three ASI2600MC, but it felt at the same time a bit frustrating to have built the thing just to find that I did not need it🥴

Cheers, Göran

What price for peace of mind though. 😉

Alan

[gorann]

 

Yes, but one less thing to blame if I get odd stars in the corners.....

Edited September 19, 2021 by gorann

[GalaxyGael]

Neat ideas for classic square pattern diffraction to solve sensor tilt. Since the rejected array of dots is an inverse space representative of the pixel grid, the further away (higher order) dots represent smaller dimensions, more accuracy. 

You can probably use the basic diffraction relation with a wavelength of approx 633 nm or 675 nm (common red diode laser wavelengths) for order m (m=1) for the dots closest to physical in line retroreflection point to confirm they are related to a sensor with a pitch equal to the 3.76 um for asi2600 and accordingly for other cameras. I'll have a test of this at work with our angle resolved reflectance setup and see can it be used this way. 

In the mean time, as per a suggestion on this thread, those of you with the baader laser kolli can use the planar diffusing grid to make checks. Must test its accuracy though as a function of rotation/tilt.... 

 

[wimvb]

1 hour ago, GalaxyGael said:

Neat ideas for classic square pattern diffraction to solve sensor tilt

In this case, it's the reflection of the main beam that is used for alignment. But the diffraction spots could be used as well, just a little more cumbersome.

[GalaxyGael]

True, and its a nice simple jig for it, that box. I'm going to try the diffraction option using a known flat ccd sensor in the lab, to look at tilt in two dimension simultaneously, like tilt examination of a star field using ccd inspector. I'll see if anything useful comes of it... 

[wimvb]

25 minutes ago, GalaxyGael said:

I'll see if anything useful comes of it... 

my initial guess would be that the front plate, and the screen need to be abolutely parallel. With the reflection metod that is not a necessary requirement, since one is only interested in getting the reflection stationary during rotation along the camera axis.

[GalaxyGael]

19 minutes ago, wimvb said:

my initial guess would be that the front plate, and the screen need to be abolutely parallel. With the reflection metod that is not a necessary requirement, since one is only interested in getting the reflection stationary during rotation along the camera axis.

What I am testing is somewhat of an overkill scenario, but the thread reminded me the idea as something I wanted to test . The system I use was built for an entirely different reason (electron holography of crystals to <0.1 nm spatial precision with a very precise interferometric control over tilt, a position sensitive detector on the nm scale, much small that a fraction of  sensor pixel). It is extremely precise for optics alignment so the confidence in an absolute reference is possible vs a commercial sensor with coated glass cover plate.  It will measure the relative intensities of all diffracted points from the sensor as a map of the accuracy of the pixel pitch and the tilt overall. The go-no go aspect is whether I can fit the entire camera into the sample holder . That might put the kibosh on this.

[wimvb]

That's definitely past the IKEA level shown here.

[GalaxyGael]

But it is nowhere near as accessible/sensible. 

[wimvb]

And here was I thinking that using x-ray diffraction to measure the sag of silicon crystals under their own weight was advanced. Did that in the early nineties.

Now, to get back on the IKEA track. I thought of measuring the sensor tilt by using  0 degree incidence and a laser collimator. That didn't work. But I did find a way to collimate a laser collimator with a 1.25" drain pipe. How's that for low tech? 😋

[GalaxyGael]

When xrd and knowledge of what really affects intensity line shapes is done right, it's very powerful. That instrument we have is very specific but also an engineering marvel. But low tech works gresy when cleverly thought out. 

When service costs get too high for a scanning election microscope, we had a garden house, fridge freezer, a cooking pot and a diffusion pump circulating cooling water through the instrument and it worked perfectly. Even the (insert Japanese electron optics name) engineer thought it was good. 

I was going for a similar approach to your normal incidence method with the baader, looking for off axis intensity indicative of tilt. Path length could give more sensitivity but moving the collimator around the cover plate was awkward and not straight forward enough to bother working out a solution. I like the rotating option, simple and effective in the main.