Fibre-optic star field for scope alignment

[PeterW]

The light will tend to spray out over a reasonable angle, check the numerical aperture of the fibre, so critical perpendicular alignment t shouldn’t be needed.
£2!!!!!!!!!!!!!!!! We used to pay about 10x that (OK about 20yrs ago!). For light launching a fixed array of SC connector seems good, maybe you could find a higher intensity collimated light to illuminate. (A crazy long distance LED torch for instance). Your idea of poking FC-UPC through drilled holes should work well, much more robust than bare fibre.

Peter

[symmetal]

5 hours ago, discardedastro said:

That's an excellent set of result images you've got there!

The ST connectors are a bit oldschool but should work great for that application - as you say, the round hole is easy to make and you get a free clamp. The other option I'd consider is LC-LC alignment sleeves with flanges - these can be secured with wire/bolts but are a rectangular cut-out which is a bit trickier.

The polishing materials you've got there are meant to be used with a connector - the only way to make a "clean" end on a bare fibre is score-and-tension cleaving. You also need an alignment puck to hold the connector flat on the polishing sheet (which should be on something nice and flat - surface plate in a lab, piece of flat glass is the common field option). I would well imagine that on your 9um bare you'll not have a clean end. If you have a cheap USB microscope that'll actually do alright at showing the end.

Thanks discardedastro. We used ST connectors a lot at BBC TV Centre from around 2000 when long digital video cable runs changed over from coax to fibre optics. As they look similar to the familiar BNC people weren't so wary of them. 😁

The polishing sheets did come with an alignment puck but as you say without the connector to hold it perpendicular it would be a bit hit and miss. The sheets were cheap so not a great loss. Bare 9um fibre would need each end processed so a second hand cleaver would be a better bet than just polishing. Leaving both ends terminated is probably best. 50um fibre seems OK though with just a plain cut.

I initially chose the black foam board as its 5mm thickness would hold the bare fibres in position relatively easily, but if using the ST adapters any mounting board type will do. If the board does have to be further away to avoid the problems shown above, and needs to be bigger, then 2m patch cables may be a better bet. I'll hopefully manage a test at a longer distance this weekend to see how it reacts. The A2 board with the 618mm FLT98 and ASI6200 fills the frame at just 10m as it has a 3.3 degree FOV.

3 hours ago, The Admiral said:

I've been trying to follow this but optical fibre technology is well outside my knowledge base. When it's all sorted and working I'd appreciate a summary of the setup, with a DIY parts list, please. 

I imagine that diffraction of the light coming out of the fibre end will enable the light to still be visible slightly off axis, so that the whole star field will be visible.

One other question, could you check and adjust the back-spacing with the artificial stars in daylight?

Ian

No problem Ian, I'll do that if tests are successful. 🙂

I'd imagine that having a light source spread over a larger area rather than a smaller point will give more spread of the light cone coming out of the fibre end. The properly terminated ends do have a larger exit viewing angle than the bare cut ends so I don't think that will be a problem. Looking by eye about 10 degrees off axis still gave a useable light output.

In daylight the light source would need to be very bright to give a good contrast from the background. My first test using a 10W high power white led (ooh my eyes) mounted on a necessary heatsink, gave clipped stars at 0.001ms exposure so had to turn the brightness well down. The video light I used later could be a bit brigher to enable bahtinov focusing. I've ordered a number of  cheap 1W white leds from ebay to arrange in an array to see how they fare. However clipped stars still exhibit elongation problems etc as well as unclipped so perhaps it doesn't matter so much. CCD Inspector only reads unclipped stars I believe so that may be an issue.

Alan

Edited July 25, 2020 by symmetal

[symmetal]

3 hours ago, PeterW said:

The light will tend to spray out over a reasonable angle, check the numerical aperture of the fibre, so critical perpendicular alignment t shouldn’t be needed.
£2!!!!!!!!!!!!!!!! We used to pay about 10x that (OK about 20yrs ago!). For light launching a fixed array of SC connector seems good, maybe you could find a higher intensity collimated light to illuminate. (A crazy long distance LED torch for instance). Your idea of poking FC-UPC through drilled holes should work well, much more robust than bare fibre.

Peter

Thanks Peter,

Yes, prices have thankfully come down. 😀 I don't know whether a more collimated light source will affect the star aberration results. It'll give a narrower exit angle compared to an area light I imagine, but is that a bonus in this case? We'll have to see. 🙂

I chose LC over SC in the end as you can get 24 LC fibres in the panel frame verses 12 SC fibres and ST patch cables seemed to be more common to LC.

I was wondering if the different LC panel adapters specified for OS2, OS4, single/multi mode, UPC/APC etc are physically the same and just the colour is different to identify its use, or are they actually different and an LC cable of one specification won't necessarily fit in another? Same with SC adapters. 🤔

Alan

Edited July 25, 2020 by symmetal

[Davey-T]

Thanks for doing all the leg work on this Allan, greatly appreciated 

Dave

[symmetal]

Good news. Took some test images down the hall tonight at a distance of 6m compared to the previous tests at 3m and the elongated stars seen on the 3m corners have gone apart from the top right which agrees much more with the actual star image. 😀 So it looks like the flattener incident light angle at 6m is close enough to the infinity light angle for it not to affect the outcome to any significant amount. At least with the ZS61. Here's the test images composite:

Now the not so good news. At 6m the 9um stars are undersampled only illuminating 1 pixel fully. This leads to the surrounding pixels getting some light the spread of  which seems to vary on successive frames in a random pattern which can tend to appear as a slight star elongation in a random direction. As seen on the centre star on the corner test. The Top right elongation is greater though and is consistant between successive frames.

At greater than 6m the stars will be more undersampled so I'm thinking that 50um fibres may be a better bet. The errors on the 50um stars in the 3m tests were just as apparent as the 9um stars at 3m. At a distance of 15m the 50um stars will be the same as the 9um at 3m.

50um fibre is multimode compared to the 9um single mode but that shouldn't matter in this application. Multimode vs Single-mode.

Here's the 50um LC-ST multimode duplex OM2 patch cable that should be suitable.

Here's the LC multimode duplex OM4 adapter panel

I assume the OM2 cable will fit in the OM4 panel OK. ST terminations are only available in OM2 cable. OM4 is just a more expensive higher bandwidth cable it appears. Following on from my previous post query would the LC-ST OM2 multimode patch cable fit in the OM2 single mode patch panel anyway. The only difference is the colour? Awaits reply from the two experts. 😁

Alan

[Davey-T]

What about binning them ?

Dave

[symmetal]

9 minutes ago, Davey-T said:

What about binning them ?

Dave

No, I'll keep on trying Dave. Oh...., that binning. 😁 That would give even more undersampling as your pixels are 4 times bigger. If I bin my flt98 images the poor corner stars aberrations almost disappear.

Alan

[symmetal]

I could always set up a heater in front of the board to simulate seeing and get bigger stars that way, but I think 50um fibres are an easier solution. 😁

Alan

[Davey-T]

Still trying to get a usable image out of the Sharpstar 150 2.8 HNC, having rebuilt the thing and fixed a number of sloppy production issues at the moment there doesn't appear to be enough forwards movement on the secondary to centre it under the focuser, all the adjustment is in the other direction, may have to take a hacksaw to it 😂

Dave

[symmetal]

3 minutes ago, Davey-T said:

Still trying to get a usable image out of the Sharpstar 150 2.8 HNC, having rebuilt the thing and fixed a number of sloppy production issues at the moment there doesn't appear to be enough forwards movement on the secondary to centre it under the focuser, all the adjustment is in the other direction, may have to take a hacksaw to it 😂

Dave

Is that the main reason you would like a useable star field simulator Dave? For the price you would have hoped those issues wouldn't arise. Unless you rebuilt it wrong. 😉

Alan

[Davey-T]

Yes could do with something to use indoors, there seems to have been a complete lack of quality control from the finder scope bracket being drilled so [removed word] I had to open up the holes with a rat tail file before I could line up the RDF, all the primary mirror cell screws loose, that's 12 screws not just finger tight but actually not even done up, primary mirror flopping around in the cell, had to stick pads in front of it, mismatched adjusting and locking screws on primary so to enable the locking screws to be below the rear rim as stated in the blurb meant the adjuster screws were nearly out of thread, the secondary has a separate ring screwed onto it for the adjusting screws to push on, this simply kept coming loose and unscrewing itself when trying to centre the mirror so had to Loctite it on.
Blacked all the nice shiny screws inside the tube.

Think that's all, just need to get it collimated now 😂

Dave

[PeterW]

Multi mode gives more area to collect light into, brighter stars. I would guess the OM2 are the same. My experience was for FC where UPC and APC were different colours to stop you trying to grind an angled end into a flat end.

 

peter

[The Admiral]

8 hours ago, symmetal said:

but I think 50um fibres are an easier solution

Now I'm more confused . I thought that you were heading down the route of 125 micron fibres, using the LC - ST duplex 1m cables you mentioned a few posts ago? I must do some maths and work out the pixel resolution when projected at the artificial star working distance and see what it equates to. Complicated by the fact it is a colour sensor too (in my case). That might provide a limiting size for the fibre.

Your proposed working distance is 16x the FL, so with a FL of about double that in my case, with a 0.8x reducer-flattener, I'd need to be looking at 12m. Hmm, that'll probably need the garden! May as well set up at night and look at real stars .

Ian

Edited July 26, 2020 by The Admiral

[The Admiral]

8 hours ago, symmetal said:

but I think 50um fibres are an easier solution.

Doing some quick back of the envelope calcs, and assuming I've got it right (a big assumption ), I work out that each pixel of your ASI1600 when on the ZS61 subtends an angle which equates at 6 metres distant to just 63micron! The data I've used I've got from astronomy.tools, i.e. ZS61, ASI1600, gives a resolution of 2.17"/px. OK, I'm sure aberrations will broaden that a bit, but perhaps it's a bit close to call? I am fully prepared to be corrected!

It would be less demanding for the FLT98 with the ASI6200, which has a resolution of 1.25"/px, except of course with a longer focal length you'd probably need the artificial star field further away.

Ian

Edited July 26, 2020 by The Admiral

[symmetal]

8 hours ago, The Admiral said:

Now I'm more confused . I thought that you were heading down the route of 125 micron fibres, using the LC - ST duplex 1m cables you mentioned a few posts ago? I must do some maths and work out the pixel resolution when projected at the artificial star working distance and see what it equates to. Complicated by the fact it is a colour sensor too (in my case). That might provide a limiting size for the fibre.

Your proposed working distance is 16x the FL, so with a FL of about double that in my case, with a 0.8x reducer-flattener, I'd need to be looking at 12m. Hmm, that'll probably need the garden! May as well set up at night and look at real stars .

Ian

Hi Ian,

125um is the diameter of the cladding surrounding the actual optical fibre core. In the pursuit of a genuine point source it was considered that 9um would be best. I was using 9um core in the latest tests and were the items listed in the shopping list you mentioned but as I found out last night the 9um core is too small when used a useable distance from the scope as it illuminates insufficient pixels to determine its actual shape. I think the 50um cores will therefore be better as the star should cover several pixels. It'll also be brighter and have a wider viewing angle. In proper usage 9um cores are usually illuminated by a laser while 50um cores are illuminated by a LED. Also for your colour camera the larger star would be helpful as the 9um stars may just illuminate a single random colour pixel.

I would think the working distance is dependant more on the focal ratio rather than just distance so a slower scope could be tested with the board closer I imagine as the light cone is not so broad to start with.

As you say it may need to be set up in the garden to get the distance, but I can go weeks without seeing any stars and the only 3 nights I've had where stars were visible since buying the ASI6200 were taken up trying to get the spacing right as it's so sensitive to distance. If I could have got it right or very close using the fibre star field, on the cloudy nights I could have had 3 nights of imaging. 🙂

Alan

[symmetal]

6 hours ago, The Admiral said:

Doing some quick back of the envelope calcs, and assuming I've got it right (a big assumption ), I work out that each pixel of your ASI1600 when on the ZS61 subtends an angle which equates at 6 metres distant to just 63micron! The data I've used I've got from astronomy.tools, i.e. ZS61, ASI1600, gives a resolution of 2.17"/px. OK, I'm sure aberrations will broaden that a bit, but perhaps it's a bit close to call? I am fully prepared to be corrected!

It would be less demanding for the FLT98 with the ASI6200, which has a resolution of 1.25"/px, except of course with a longer focal length you'd probably need the artificial star field further away.

Ian

At 3m distance the 9um fibre, terminated both ends, only fully illumated 2 pixels while the 50um fible, bare at the board end gave a fuzzier edged star with about 5 pixels diameter fully illuminated. With a proper terminated end this should be cleaner edged and a bit wider I would think. I chopped the only 50um cable I had so can't test it terminated both ends.

Alan

[symmetal]

Can discardedastro or PeterW (or anyone else who knows 🙂) say whether the ST multimode cable fits into the ST-ST Single-mode coupler as the ST-ST Multimode coupler will take 2 weeks to arrive as it isn't in stock locally. It appears the guide hole inside the coupler is just for the end ferrule which is specified as 2.5mm diameter for the ST connector. Is it just the end cap colour that's the difference. If it is I'll order Single-mode couplers to get them sooner. They're the same price. I don't know why FS are so low cost, £0.60. I thought I could get some from ebay for less but the cheapest there are £3.50 each. Farnell £6.00 each. 😲

Alan

[discardedastro]

On 25/07/2020 at 16:14, symmetal said:

Thanks discardedastro. We used ST connectors a lot at BBC TV Centre from around 2000 when long digital video cable runs changed over from coax to fibre optics. As they look similar to the familiar BNC people weren't so wary of them. 😁

The polishing sheets did come with an alignment puck but as you say without the connector to hold it perpendicular it would be a bit hit and miss. The sheets were cheap so not a great loss. Bare 9um fibre would need each end processed so a second hand cleaver would be a better bet than just polishing. Leaving both ends terminated is probably best. 50um fibre seems OK though with just a plain cut.

I initially chose the black foam board as its 5mm thickness would hold the bare fibres in position relatively easily, but if using the ST adapters any mounting board type will do. If the board does have to be further away to avoid the problems shown above, and needs to be bigger, then 2m patch cables may be a better bet. I'll hopefully manage a test at a longer distance this weekend to see how it reacts. The A2 board with the 618mm FLT98 and ASI6200 fills the frame at just 10m as it has a 3.3 degree FOV.

I'd imagine that having a light source spread over a larger area rather than a smaller point will give more spread of the light cone coming out of the fibre end. The properly terminated ends do have a larger exit viewing angle than the bare cut ends so I don't think that will be a problem. Looking by eye about 10 degrees off axis still gave a useable light output.

In daylight the light source would need to be very bright to give a good contrast from the background. My first test using a 10W high power white led (ooh my eyes) mounted on a necessary heatsink, gave clipped stars at 0.001ms exposure so had to turn the brightness well down. The video light I used later could be a bit brigher to enable bahtinov focusing. I've ordered a number of  cheap 1W white leds from ebay to arrange in an array to see how they fare. However clipped stars still exhibit elongation problems etc as well as unclipped so perhaps it doesn't matter so much. CCD Inspector only reads unclipped stars I believe so that may be an issue.

Alan

As an ex-TVC-adjacent bod (across the road at Centre House) for a few years the fact that ST connectors found a home in BNC panels does not surprise 🙂

Both ends terminated is definitely the simplest way to get a good exit pupil with a reasonable NA. Definitely better to get the exit pupil of the point source as small as reasonably possible I think, the flipside is your entry pupil is also tiny and so getting light in is tricky. I'd maybe consider looking at cheap laser diodes you could align and glue or otherwise secure onto the far end of the patch cords, but you'd end up testing at a fixed wavelength rather than broadband.

How do the point sources look defocused?

3 hours ago, symmetal said:

Can discardedastro or PeterW (or anyone else who knows 🙂) say whether the ST multimode cable fits into the ST-ST Single-mode coupler as the ST-ST Multimode coupler will take 2 weeks to arrive as it isn't in stock locally. It appears the guide hole inside the coupler is just for the end ferrule which is specified as 2.5mm diameter for the ST connector. Is it just the end cap colour that's the difference. If it is I'll order Single-mode couplers to get them sooner. They're the same price. I don't know why FS are so low cost, £0.60. I thought I could get some from ebay for less but the cheapest there are £3.50 each. Farnell £6.00 each. 😲

Alan

Yes, the couplers mechanically work on the ferrule outside diameter only so you're safe with that.

FS are well known in industry for being cheap as chips basically by virtue of volume on a restricted product portfolio and a 100% CN-sourced portfolio. Their quality control is okay, and for most passives they're okay. We use them for stuff where optical quality isn't critical and where we can tolerate failure (short links in datacentres etc).

[symmetal]

32 minutes ago, discardedastro said:

Yes, the couplers mechanically work on the ferrule outside diameter only so you're safe with that.

FS are well known in industry for being cheap as chips basically by virtue of volume on a restricted product portfolio and a 100% CN-sourced portfolio. Their quality control is okay, and for most passives they're okay. We use them for stuff where optical quality isn't critical and where we can tolerate failure (short links in datacentres etc).

Excellent, thanks for that info. I'll order single mode couplers. This project isn't critical so FS will be fine for the bits. 😀

44 minutes ago, discardedastro said:

How do the point sources look defocused?

As part of the test yesterday I tried defocusing to see whether that would effectively give larger stars but it actually gave more star distortions as well as larger stars so it looks like optimum field flattener distance is only specified for focused stars. It seems defocused stars would need a different flattener distance to appear round. Here's the defocused test.

Alan

[PeterW]

We used to use ceramic lined  uniters for high power stuff as they were more precise than the metal lines ones... when you don’t want any light to “miss”... remembering you are lining up 8micron diameter “pipes”. Replacing “burnt” connectors is a pain, blowing fibres up “mission impossible style” is fun, but expensive!

if the LED launch is working i’d stick with them, if you get fibre coupled lasers, be careful to ensure the power levels are kept down and don’t get them too close to your eye (or eye view with any optical instrument)... that way the light will spread out such that your eye shouldn’t be able to take in and focus a dangerous amount (be careful with LED as hey can damage retinas like laser can if powerful enough).

Having bought some “5mW” pointers only to measure tham at 10x that I’d advise caution.

 

peter

[symmetal]

33 minutes ago, PeterW said:

We used to use ceramic lined  uniters for high power stuff as they were more precise than the metal lines ones... when you don’t want any light to “miss”... remembering you are lining up 8micron diameter “pipes”. Replacing “burnt” connectors is a pain, blowing fibres up “mission impossible style” is fun, but expensive!

if the LED launch is working i’d stick with them, if you get fibre coupled lasers, be careful to ensure the power levels are kept down and don’t get them too close to your eye (or eye view with any optical instrument)... that way the light will spread out such that your eye shouldn’t be able to take in and focus a dangerous amount (be careful with LED as hey can damage retinas like laser can if powerful enough).

Having bought some “5mW” pointers only to measure tham at 10x that I’d advise caution.

 

peter

Thanks Peter, yes I think leds are safer, and cheaper. 🙂 Light efficiency and loss isn't really an issue in this project and the increased light spread from the exit fibre if fed from a wide light source probably simulates a real star better after having gone through atmospheric disruption.

Put the order in today and FS has said it's been dispatched. I ordered this LC Multimode panel as I'm using multimode cable but suspect I could have used a LC Single-mode panel instead and saved £5. If you or discardedastro can confirm they are mechanically the same, or not, that would help when I compile the list of materials needed, and possibly save a bit of money for those wishing to make one. 😀

Alan

[symmetal]

FS order arrived today. Got enough fibre stuff to make 25 stars in a 5x5 array for the moment. 😀 The maximum thickness board on which the ST couplers can mount is 3mm. 3mm black foam board doesn't seem to be available in A2 and would be a bit fragile in any case. I've ordered a 3mm A2 sheet of black perspex, and a matt black A2 3mm sheet of Foamex Foam PVC, to see which is more suitable. Both similar price of around £20 each. 😲 The perspex is gloss black but that may not be a problem as there should be no actual light shining on the backing board.

For the leds I've ordered several COB led panels in different sizes and a couple of constant current LED driver buck converters from ebay. All fairly cheap. The COB led panels will need mounting on heat sinks.

Alan

[symmetal]

Made the light box and drilled the perspex board. Had an old unused metal project box, probably 30 years old or more. Chassis looks like galvanized steel.

Mounted 2 fibre panels on the front and 2 12V, 20W, cob led array panels on the back panel. Each fibre panel hold 24 fibres so for the minimum 25 stars needed two panels. I'll add extra stars if it's successful to give a fuller field. A constant voltage / constant current buck converter powers the led arrays and I removed the constant current adjustment multi-turn trim pot from the module and wired a standard rotary pot on the front panel in its place to quickly adjust the current / led brightness. The converter is rated at 20A so is just ticking over in this project and stays cold.

As I didn't want to exceed the maximum 1.6A to each led panels I wired a meter in the led supply. A 0R47 2W resistor is put in series with each panel in order to balance the current between them. As the leds get hot at full power they need to be mounted on a metal panel/heat sink. As it is, after 2 minutes at full power the rear panel is too hot to touch. High brightness is only needed for bahtinov focusing and not star shape assessment, and focusing won't need full power anyway I suspect.

I used the perspex panel in the end as it's stiffer than the foamex though the foamex would be quicker to work with as it can be cut with a saw and drilled easily. It's robust and waterproof though quite bendable so would need bracing across the back if used as the star board.

I did drill the edges of the perspex to fix some aluminium angle for added rigidity though it probably isn't necessary. I should have made the stars the same distance apart H and V but drilled the A2 as a 5 x 5 array without thinking. A 6 x 5 or similar would have been better. I did drill extra holes between the 5 x 5 array to add extra stars if I need to.

Fibre couplers fixed to panel. They're the red protective boots on the end in the photo.

Just need to connect up the fibre cables now and take it for a spin. 😀

Alan

 

[Davey-T]

On 09/08/2020 at 23:40, symmetal said:

Made the light box and drilled the perspex board. Had an old unused metal project box, probably 30 years old or more. Chassis looks like galvanized steel.

Any results yet ?

Dave

[symmetal]

4 hours ago, Davey-T said:

Any results yet ?

Dave

Hi Dave. I actually did some tests last night. First time out for several weeks as it's been either rainy and/or windy. I did get some extra fibres to fill in the gaps on the board and here it is set up in the afternoon. Yes, I know the grass needs cutting. 😁 The red dust caps on the board are removed before use. I put the scope on my HEQ5 with an extra weight to balance. It's a bit overloaded but as it's not tracking it was OK. The board is 17m away and the focuser is racked out about 2/3 distance.

I had recently fitted a Gerd Neumann M48 tilt adjuster as the standard tilt adjusters with the push pull screws were too much of a hassle, and the one I had introduced a lot of tilt even when fully screwed in. I hadn't tried it since fitting the new tilt adjuster so only had a measured FF spacing to start with.

I found the led panels even at minimum brightness (150mA) were too bright and the stars would clip at 0.05s exposure. When used indoors I had to run at high brightness as the background was much brighter than outside. Outside, at 0.05s the background was pure black, just set by the offset value. This meant that the auto-stretch previews on SGP severely clipped the stars as they tried to raise the background. Manually adjusting the stretch in SGP wasn't much better as there was too much contrast between the stars and background. The only way to see the star shapes properly was to convert the images to tiff and load them into photoshop and use curves to greatly stretch the background without clipping the stars which was a palaver each time. Much dimmer stars are needed so that a longer exposure which gives some background level would be better.

Bahtinov focusing was easily done at 0.2s exposure at current minimum led brightness. This showed one side had different focus to the other so moving one side of the panel closer by about 2cm made the focus more even though the outside stars all had a different focus point compared to the centre showing the field wasn't flat. Was this just because the centre was closer to the scope than the edges by a few millimetres. I don't think so as to get that amount of focus shift between the centre and edges needs about 3cm movement of one edge. Also, moving one side closer to even out the focusing, means I've adjusted for any tilt in the imaging train which isn't what you want really, though it could still be used to get a more correct FF spacing. An easy way to get the panel orthogonal is needed for tilt adjustments. 🤔

Here are two images showing the bahtinov focusing and a star shape exposure stretched in photoshop. The focus across the field isn't too bad, but the star shapes at the corners are quite bad. The bahtinov image also shows smearing in the corners. They are full size and you'll need to view them that way to see the detail. This is the full frame ASI6200 on the FLT98.

It's possibly clear for a bit tonight so I may be able to put it on my AZEQ6 on a pillar, to get an actual star image to compare. Until I do that I can't say whether the artificial field is realistic.

Bahtinov focus test:

Star shape test

Alan