Canon 550D Tests

[MarkS]

I've put my eBay purchased secondhand Canon 550D through the usual batch of tests to measure its vital statistics.  

Here are the results:

Canon 550D      
ISO   Gain   Read Noise
100   2.24    12.8
200   1.12     6.9
400   0.56     4.2
800   0.28     2.9
1600   0.14     2.4
3200   0.07     2.2
6400   0.035   1.9

Gain is measured in electrons/ADU and read noise is measured in RMS electrons.  I only measured the gain at ISO800 - the others gains were estimated by successive halving and doubling.

The above results are almost indistinguishable from the Canon 600D though the read noise might be very slightly less.  In my mind ISO 800 is the best compromise between read noise and dynamic range for the 550D.

I've added both the Canon 550D and Nikon D7000 to my graphs of dark current.  The x-axis is the exposure number in a 2 hour sequence of continuous 5 minute exposures:



It looks like the 550D runs slightly quieter than the 600D once the sensor has warmed up.

My intentions for this camera are as follows:
I have a Baader replacement filter window on order - for the H-alpha mod.
Remove the mirror because this interrupts the lightpath for scopes at fast F-ratios (F/2.8 and F/2)
I also intend to give it Peltier cooling - I'll decide how exactly to do this once I have it in pieces.

Mark

[MarkS]

For those of you wondering how the dark current figures can be used in practice, here's an example calculation:

At room temperature (20C) a typical Canon in the graphs below stabilises at a dark current of around 0.5 electrons/sec (in each pixel)

So a 300sec exposure on a warm 20C night would collect 150 (=300*0.5) electrons.
Dark current halves for every 6C drop in ambient temperature.
On a moderate night at 14C this would then be 75 electrons (half of 150)
On a cool night at 8C this would be 37 electrons.

The thermal noise is the square root of the number of electrons.
So on a cool night at 8C a 300sec exposure would have thermal noise of Sqrt(37) = 6.1 electrons

But at the beginning of an imaging session, before the sensor has become hotter, the dark current is around 0.12 electrons/sec
So, working through the arithmetic again, the thermal noise in the first 300sec exposure on an 8C night will be only 3 electrons assuming the camera had properly cooled down to 8C beforehand.

Hope that helps.

To put this noise into perspective, although their thermal noise is close to zero, the read noise of a typical cooled one shot colour astro-camera is around 7-11 electrons.

In my next tutorial I will explain how to calculate the imaging noise caused by the skyglow.
You will then have all the ammunition you need to understand what is the dominant source of noise in your sub-exposures  

Mark

[Mac]

I think you need to publish a book.

Or failing that a nice PDF,

Mac.

[MarkS]

I think you need to publish a book.

Or failing that a nice PDF,

The thought had crossed my mind but it is a lot of work - even just to produce a few "essays" in PDF format.  Maybe one day. 

[Tony G]

I think you need to publish a book.

The thought had crossed my mind but it is a lot of work. 

Can you not just collate all your posts and send it to the printers 

Tony G

[MarkS]

I've successfully removed the 550D's mirror!

I took photos of the operation so I'll document the procedure later but here's a hint:
http://www.canonwatch.com/fake-mirrorles-canon-camera-or-how-to-destroy-your-canon-t2i/

The camera still works but but of course I can't see anything through the viewfinder.  Autoexposure and autofocus still work in liveview mode.  And without the mirror slap it is so much quieter when taking photos!

H-alpha mod and cooling still to do.

Mark

[mickw]

Oh dear, there's cut wires and a soldering iron - it will all end in tears 

[Tony G]

Oh dear, there's cut wires and a soldering iron - it will all end in tears 

You can always revert back to basics if anything goes wrong.



But with your expertise and capabilities, I will look after these for my use. 

Tony G 

[MarkS]

You can always revert back to basics if anything goes wrong.

Brilliant! 

I took photos of the operation so I'll document the procedure later but here's a hint:
http://www.canonwatch.com/fake-mirrorles-canon-camera-or-how-to-destroy-your-canon-t2i/

OK, I lied about the hint!   There is no need to dismantle the camera because I have worked out a much easier way of doing it.  Full description here:
http://www.markshelley.co.uk/Astronomy/Projects/Mirrorless/canon550mirrorless.html

It also shows a Takahashi Epsilon flat field before and after conversion.  I really, really wish I had attempted this years ago!!

Mark

[RobertM]

Did you remove the anti alias filter - that's not needed any more.

[Mike]

That's a nice neat solution Mark. Saves taking everything apart.

[MarkS]

Did you remove the anti alias filter - that's not needed any more.

No not yet - I'm dismantling the camera today - in readiness for the Baader replacement filter and to measure up for the cold finger - in case I need to order some copper with a different thickness.  I'll order sufficient for more than one conversion

Mark

[Rocket Pooch]

Small question with the mirror out, how do you stop the LED display in the prism reflecting back into the camera, are you going to blank it off.

[MarkS]

Small question with the mirror out, how do you stop the LED display in the prism reflecting back into the camera, are you going to blank it off.

The red focusing lights and green display in the viewfinder should not be alight during exposures.  But in any case I'm planning to remove the focusing screen and flock it or paint it matt black to avoid any possibility of stray reflections off the focusing screen.  Focus screen removal is an easy operation:
http://www.focusingscreen.com/work/550den.htm

[mickw]

I don't know what capture software you'll be using but APT would use the Live View circuitry.  Would the removal of the focusing screen confuse the electronics and Live View output ?