Debayering a DSLR's Bayer matrix.

[Gina]

The sensor is fine Next job filter glass removal...

[Gina]

First try - hot nichrome wire, red hot where it wasn't losing heat to the glass and frame. Not the slightest sign of anything. I have the glass covered except a mm round the edge, with a sticky pad. Only sign of anything at all was the singing of the sticky pad at the corner where I was heating. I had already gone round the glass with the tip of a craft knife (like an X-Acto) and removed some glue.

I'll try more pressure with the knife tip otherwise it looks like the glass cutter.

[Gina]

The knife tip would only break off tiny pieces of glass so I attacked it with the glass cutter. Certainly no chance of getting the glass off intact. The diamond cutter dug into the glass, no problem, and I cut grooves around the inner edge of the bond, going right through the glass at a couple of places in the middle of the ends (short sides) away from the wires. I was then able to get the tip of the craft knife under the glass though clear of the sensor and break pieces of glass out. Extreme butchery! I haven't got all the glass off yet but I'm going to test the sensor. The wires appear perfectly intact. Here's a photo.

[Gina]

Now I'm getting an Err 70 but I can't see anything wrong

If this is a sensor error I can see no way of getting the cover glass off without stopping the camera from working. This is with the camera and kit lens without computer connection - just using the camera as designed with the back on and SD card. I could try with the netbook but I would expect the same result.

This is the last chance - I've run out of sensors available for this purpose with the possible exception of the camera I can't find.

Unless someone comes up with a reliable way of removing the cover glass on the 1100D I will have to admit defeat at this point. I've given my absolute best to this project and I have failed. I do manage to succeed at most of the things I tackle but this is one ultra extreme mod too far for me. And that's that.

I believe I could remove the CFA successfully using the filter frame for protection but the cover glass has defeated me for the 1100D. I believe so much that the 1100D is the right DSLR for astro uses, being so much better than it's nearest rival, that I see no point in trying with any other model.

I shall continue to read this thread with interest.

[Luis Campos]

Hello all,

Now that I have several cameras modified and debayered ( 350D MONO full spectrum with no cover glass, a 450D COLOR full spectrum modified and another 450D mono full spectrum I did for a friend) I did some comparasion shots with the 3 camaras, these were 350D MONO full spectrum, 450D mono full spectrum with the cover glass only and my 450D color sensor but full spectrum modified...please bear with me

First let me tell you two words that matter for mono mods: PIXEL SIZE!

The 350D mono is the MOST sensitive of the batch...and it's not a small diference, I say it's roughly 25 - 30% more sensitive then the mono 450D

The second most sensitive...and I'm a bit in shock here, is the COLOR 450D by a small amount but it is more sensitive then the debayered 450D I did for my friend.

So, this tells us a simple story, if you are going for narrowband work...stick with the 350D, cooled and if possible amp off moded, this is by far the best camera for this type of work, if you do simple "L" work, the 450D mono is OK, just expose a bit longer then with the stock color filter and you are good to go, if it's cooled even better.

If you have a fast optical system and don't mind doing 15 or 20 min. subs I guess the 450D will be ok too, the noise levels are very low and as long as you have a good tracking you are fine.

So here's what i do, glad i took this route, I will keep my 350D mono for "L" and narrowband work, and keep the COLOR TEC cooled 450D for chrominace, then blend under PS the data L-RGB or Ha- RGB from the mono 350D and 450D COLOR.

Please take a look to the following shots (Fox fur nebula) LOL:

350D MONO:

450D MONO:

450D COLOR:

Well, next time I'll do a serious test under the stars to get some "real life" data, but I think it's clear too see that the oldie 350D is the winner...sensitivity wise

Cheers,

Luís

[Luis Campos]

well done Luis

way to go

when you say you scraped the edge with a knife, do you mean there was glue outside the glass edge that you scrapef away?

I must get myself a 450d. seems to be the best compromise for everything

14 bit, 12MP, decent pixel size, possibility of cooling, relatively low cost.

and I thought best to get two.

one for mono and other, remove the ir glass, and you'll have a spectrum enhanced camera for rgb data with rgb filters already on it

so rather than spending on a filter wheel and lrgb filters, just switch cameras for rgb data

and since the imaging scale is the same, easy to stack.

look forward to your images luis.

tonight is the first clear night for us in a month, so will try my debayered 350d and qhy8l

Cheers Alistair,

Thanks mate

Very happy with this

And happier to finally overcome the dredfull cover glass problem, yes you must get a small thin bladed x-acto and gently cut the glue all around the cover glass, just stick the blade point betwen the glass and sensor housing and take your time, you'll se some "threads" of glue once you pass the blade on the joint, apply some more presure at each pass and you'll see the glass becoming white as you pass the blade, eventually it will turn white all around and the glass will pop out, begin lifting the corners gently and start from there, don't force the corners though as they are the weak spot, too much and crack...

Yes, it's what I do the 350D mono for "L" and the 450D color for "RGB" then blend in PS...works great!

Cheers,

Luís

[Gina]

That's interesting Luis Do you think it possible that the mono 350D might be more sensitive than a mono 1100D?

[Luis Campos]

Yes Gina, yes it is the 1100D as the same pixel size as the 450D so I guess it will be the same story

But both 450D and 1100D will work fine as long as you keep your subs long enough, with cooling it will be fine

Cheers,

Luís

[Luis Campos]

Whoooo....I meant, the 1100D as the same pixel size as the 450D , a bit confused there...

[Luis Campos]

Hello all,

Now that I have several cameras modified and debayered ( 350D MONO full spectrum with no cover glass, a 450D COLOR full spectrum modified and another 450D mono full spectrum I did for a friend) I did some comparasion shots with the 3 camaras, these were 350D MONO full spectrum, 450D mono full spectrum with the cover glass only and my 450D color sensor but full spectrum modified...please bear with me

First let me tell you two words that matter for mono mods: PIXEL SIZE!

The 350D mono is the MOST sensitive of the batch...and it's not a small diference, I say it's roughly 25 - 30% more sensitive then the mono 450D

The second most sensitive...and I'm a bit in shock here, is the COLOR 450D by a small amount but it is more sensitive then the debayered 450D I did for my friend.

So, this tells us a simple story, if you are going for narrowband work...stick with the 350D, cooled and if possible amp off moded, this is by far the best camera for this type of work, if you do simple "L" work, the 450D mono is OK, just expose a bit longer then with the stock color filter and you are good to go, if it's cooled even better.

If you have a fast optical system and don't mind doing 15 or 20 min. subs I guess the 450D will be ok too, the noise levels are very low and as long as you have a good tracking you are fine.

So here's what i do, glad i took this route, I will keep my 350D mono for "L" and narrowband work, and keep the COLOR TEC cooled 450D for chrominace, then blend under PS the data L-RGB or Ha- RGB from the mono 350D and 450D COLOR.

Please take a look to the following shots (Fox fur nebula) LOL:

350D MONO:

450D MONO:

450D COLOR:

Well, next time I'll do a serious test under the stars to get some "real life" data, but I think it's clear too see that the oldie 350D is the winner...sensitivity wise

Cheers,

Luís

I forgot to mention that all shot were with exactly the same settings: 100 ISO, 1/125 th second, f/8 at 55mm f/l with the stock 18-55 lens.

Cheers,

[Gina]

Thank you Luis I just wondered if the newer processor and 14bit data depth of the 1100D made it better.

[pixueto]

yep I like this workflow as well but rather than using the chemical, why not try gina's method with ajax and the cleaner? seems to do a clean job. maybe you could use a soft wood tip rather than a sensor swab. the wood should soak up the liquid and dispense with slight pressure.

I'm very wary of applying chemicals which can get to the sensor. Maybe if some people experiment further with dead sensors... a mean, this project is getting expensive!!!

Edited September 2, 2013 by pixueto

[pixueto]

Pixueto, this could serve as a guide for exposure lengths for uncooled mono dslr's.

• Take 2N-1 exposures of t_ORN/2 duration.
  

In other words, if you determine the sky limit exposure time t_ORN to be 10 minutes, instead of taking five 10-minute exposures and median combining, take ten 5-minute exposures and use the Sigma Clip combine method. And be sure to dither the exposures. This will result in excellent noise reduction from all possible sources (assuming you take a sufficient number of dark frames, and taking a flat field or three couldn't hurt).

http://starizona.com/acb/ccd/advtheoryexp.aspx

Thank you. What free software is there around for measuring the sky glow in the formula? If any.

[pixueto]

Cheers Alistair,

Thanks mate

Very happy with this

And happier to finally overcome the dredfull cover glass problem, yes you must get a small thin bladed x-acto and gently cut the glue all around the cover glass, just stick the blade point betwen the glass and sensor housing and take your time, you'll se some "threads" of glue once you pass the blade on the joint, apply some more presure at each pass and you'll see the glass becoming white as you pass the blade, eventually it will turn white all around and the glass will pop out, begin lifting the corners gently and start from there, don't force the corners though as they are the weak spot, too much and crack...

Yes, it's what I do the 350D mono for "L" and the 450D color for "RGB" then blend in PS...works great!

Cheers,

Luís

Congratulations on your mod Luis. If there are some scratches they don't seem to come up once you enter the control frames in IRIS as you tought me but it's nice to have as clear a sensor as possible.

As for the glass, I hate being the bearer of bad news but, according to my experience, the 450D has one of the easiest glasses over the sensor to remove -only the 1000D is easier. I reckon the 1100D and the 350D are different kind of beast. Has anybody managed to remove the glass intact in a 350D yet? I'm deciding when I should have a go at it inspired by your Eagle nebula H-alpha picture (wonderful). Still thinking whether or not I should apply heat to remove that glass!

In the meantime, I've ordered a nikon D80 sensor for £40 but it will take ages to arrive. I want to have a go with the solvent suggested by JTW. After all the work shown in this thread; your experiences and mine , I've decided that I'm going to have a go with the 350D and the nikon D80. I kind of suspected that the CFA on the 350D was the easiest to remove without scratching the sensor but this last information from Luis just confirms the 350D is the best choice.

Edited September 2, 2013 by pixueto

[Gina]

I've had the 350D apart to see about cooling and including a filter wheel.

Cooling looks alright as there is a reasonable distance between the sensor and the main board screens. The cold finger can come out of the side. Furthermore the thickness could be increased once away from the sensor to improve conductivity. A thermally insulating material could be wrapped round the cold finger to improve efficiency too.

Adding a filter wheel is a different matter. Unlike the 1100D, the circuit boards and other components come up almost to the lens attachment ring. A debayered 350D would probably best be used with clip filters and lenses. It might just be possible to arrange a slide in and out very slim filter drawer taking the 36mm unmounted filters - there wouldn't be room for mounted filters.

[Kbramley]

Just been looking around for sensor specs to see if the 300D is a viable option (nice and cheap and larger pixels than the 350D), found this webpage which appears to be a goldmine of info for a whole range of sensors. I'm tempted to try a 300D if I can get one for say £20-£30.

http://www.clarkvision.com/imagedetail/digital.sensor.performance.summary/index.html

Keith

Edited September 2, 2013 by Kbramley

[alistairsam]

hi

interesting to see those test results Luis, but I'm sure you'll all know this, pixel size has to be taken into account with your scope as well. work out the image scale and it all depends on your scope as well.

smaller pixels are needed for high res imaging with a fast system and large pixels with slow systems. All depends on what you want to image.

so whilst small pixels, along with the loss of microlenses will yield an inferior result to larger pixels, it all depends on your scope at the end.

For getting details with galaxies on an F4, you need between 4 and 5 micron pixels to stay between 1 and 1.4 arcsec/pixel.

so a 450D would be very good on an F4. it might need slightly longer exposures but you will get some high resolution imaging compared to larger pixels cause the scale would be around 1.2 arcsec/pixel and its not a waste to oversample.

for example, this was taken on an 8inch F4 with an sxvr-h694 with 4.5micron pixels to get details on the galaxy

http://www.iceinspace.com.au/forum/attachment_browse.php?a=141917

if you are undersampling, you can try using the "Drizzle algorithm" to recover data if it has been dithered.

so don't worry too much about pixel size on its own and loss of sensitivity.

Keith, that website is great, I've used that info in my spreadsheet to work out which canon would be ideal for my F4.

i've added read noise and range for a few sensors as well.

worth checking the camera specs along with your scope for the best results

https://www.dropbox.com/s/ea2jurl3lglh2hv/dslr-canon-sensors.xlsx

the 14 bit of the 450d will make a significant difference compared to 12bits for nebulae.

hope this helps.

Alistair

[alistairsam]

Just been looking around for sensor specs to see if the 300D is a viable option (nice and cheap and larger pixels than the 350D), found this webpage which appears to be a goldmine of info for a whole range of sensors. I'm tempted to try a 300D if I can get one for say £20-£30.

http://www.clarkvisi...mary/index.html

Keith

Keith,

read noise for a 350d is 3.7 e- and for a 300d, its 10e-.

as a result, dynamic range is a lot lower for the 300d even though they have similar well depths of 40k plus.

whilst this is clear in theory, it'll be interesting to see practical results and what all of this means.

no harm in trying it out, they should be quite cheap on ebay.

Alistair

[Kbramley]

Good points Alistair, asyou say you would certainly need to consider the whole imaging system (and the seeing too I guess) to optimise the system. I think the 300D may only be an option for those wanting to get into DIY mono astro for as cheap as possible, it would be a good starting point and for about £30 you could get a large format mono sensor which would produce some reasonable results, and with cooling probably some very good results.

I'm following a few on ebay currently and may just have a dabble (time permitting)

slightly off topic, I'm picking up a PS3 Eye today I got for £4, I plan to de-bayer this and experiment with fast FPS mono planetary work. There are some windows drivers available for it which allow RAW streaming which I need to work on though.

Keith

[Luis Campos]

hi

interesting to see those test results Luis, but I'm sure you'll all know this, pixel size has to be taken into account with your scope as well. work out the image scale and it all depends on your scope as well.

smaller pixels are needed for high res imaging with a fast system and large pixels with slow systems. All depends on what you want to image.

so whilst small pixels, along with the loss of microlenses will yield an inferior result to larger pixels, it all depends on your scope at the end.

For getting details with galaxies on an F4, you need between 4 and 5 micron pixels to stay between 1 and 1.4 arcsec/pixel.

so a 450D would be very good on an F4. it might need slightly longer exposures but you will get some high resolution imaging compared to larger pixels cause the scale would be around 1.2 arcsec/pixel and its not a waste to oversample.

for example, this was taken on an 8inch F4 with an sxvr-h694 with 4.5micron pixels to get details on the galaxy

http://www.iceinspac...se.php?a=141917

if you are undersampling, you can try using the "Drizzle algorithm" to recover data if it has been dithered.

so don't worry too much about pixel size on its own and loss of sensitivity.

Keith, that website is great, I've used that info in my spreadsheet to work out which canon would be ideal for my F4.

i've added read noise and range for a few sensors as well.

worth checking the camera specs along with your scope for the best results

https://www.dropbox....on-sensors.xlsx

the 14 bit of the 450d will make a significant difference compared to 12bits for nebulae.

hope this helps.

Alistair

Exactly Alistair, sensitivity is not everything

Also I believe that the better performance of the 450D will make up for the difference (14 bit and much lower noise levels) so if you expose a bit longer you will end up with very, very good results

Weather permiting, I'll do some basic test tonight on my back yard, both in "L" and "Ha" with the 350D and 450D mono, will post results asap.

These two I will test are not cooled yet, but will be in a near future

Cheers,

Luís

[Gina]

It's no good - I can't put these experiments down Just bid for and won a 300D with bent CF card pins which I reckon I should be able to fix (£23 + £10 P&P). Then I can play with that. Wouldn't mind having a go with the350D - thinking about that. But the 1100D is out until I find a way of getting the cover glass off. Last time I used heat I destroyed the sensor and then destroyed the final one trying a glass cutter.

[alistairsam]

hi Luis

yes some ha or L testing will show the difference

see if you can get a large galaxy or a large neb

Alistair

[Kbramley]

lol that was the one I was watching, missed the bid as i was working agreed its just a bent CF pin and may work over usb still anyhow so an ideal candidate for modding. Hope you have better luck this time Gina

Keith

[alistairsam]

It's no good - I can't put these experiments down Just bid for and won a 300D with bent CF card pins which I reckon I should be able to fix (£23 + £10 P&P). Then I can play with that. Wouldn't mind having a go with the350D - thinking about that. But the 1100D is out until I find a way of getting the cover glass off. Last time I used heat I destroyed the sensor and then destroyed the final one trying a glass cutter.

I know the feeling Gina

I'm constantly. looking for 350 or 450's, not sure why

I'm going to try my 600d soon.

I'll try scraping the edges first and then mild heat around the edges

Alistair

[Gina]

Thank you Keith

Edited September 3, 2013 by Gina