Canon Sensors - Quantum Efficiency

CometGuy · #1 29-01-2009, 09:42 PM

These are the first reasonably precise measurements done of Canon CMOS Quantum Efficiency that I've seen:

http://astrosurf.com/buil/50d/test.htm

Whats it all mean? In a nutshell, the Canon CMOS sensors are 1/5 as sensitive as the best monochrome sensors. This difference is mostly due to the bayer colour filters which cover the pixels. I guess we all knew that but its interesting to see the hard data for once!

Interestingly the new 5D is quite a big improvement over the old 5D.

Terry

rogerg (Roger) · #2 29-01-2009, 10:05 PM

Thanks for the link. I have to admit I presumed they would be pretty much the same QE as single shot CCD cameras such as the QHY cameras, but that appears to not be the case, with some QHY cameras having about twice the QE (if I remember right).

Roger.

Merlin66 (Ken) · #3 29-01-2009, 10:14 PM

According to the data I have, the MX7c - one shot colour has a QE of 0.7 against the Canon 300D quoted at 0.3.
http://www.licha.de/astro_article_cc...e.php?iOrder=2

bojan · #4 30-01-2009, 09:16 AM

Well.. Canon DSLR is NOT an astrocamera. It is only capable of producing acceptable results.

Much better sensors are used in Sigma cameras (www.foveon.com) as they do not use Bayer matrix. However, the prices of those cameras are astronomical...

CometGuy · #5 31-01-2009, 01:25 PM

I've been looking at some of the Foveon sensor specs and I've seen quoted a net Peak Quantum efficiency of 45%, i.e the overall result from adding all the colour channels together. So you can actually use the sensor as reasonably good monochrome camera...interesting.

I haven't seen much in the way of astrophotography done with the Foveon. Has anyone on the forum got a Sigma SD14?

Terry

Gama · #6 31-01-2009, 02:35 PM

These are on par with QE figures we where talking about 2 odd years ago here on the forum, but many didnt believe some of the figures bieng estimated. I hope now they can accept the figures with CMOS, although very good, isnt yet on the same level as CCD technology.
But its just a matter of time before it does, then exceed.

Roger, your correct, the Sony chip has about twice the QE, and the full well is around 25500.

Theo

CometGuy · #7 31-01-2009, 05:32 PM

Theo,

>> Roger, your correct, the Sony chip has about twice the QE

As I also stated 2 years ago I doubted the claim of 60% Quantum Efficiency for the Sony ICX453 (as used in various cameras like the QHY8 and Nikon D70, etc.). Where does this figure originally actually come from ?

If Canon sensors have a peak sensitivity of 35-40% and Sony sensors are at 60% there should be a huge difference in sensitivity that should be obvious in astrophotography. But lets look at some actual comparisions:

******** Comparision1 - Christian Buil's D70 vs 10D ***************:

http://astrosurf.com/buil/d70v10d/eval.htm

Look at the graph "Nikon D70/Canon 10D relative spectral response at the same ISO". Note you have to multiply the Nikon graph by 2.98/2.41 to compensate for difference in GAIN. But the final outcome is the the D70 has very slightly higher Quantum Efficiency. Of course the 10D is several generations old now..

******** Comparision2 - Wayne Cosshall's D200 vs 350D ***************

Not the ICX453, but similiar CCD technology. I asked Wayne Cosshall who was near Geelong at the time to perform near simultaneous equal exposures with his 350D and D200 using the same lens at the same FL/aperture (Tamron 18-200 on both cameras). The Canon was clearly more sensitive than the Nikon. I'd be happy to post raw files if you doubt me (and if Mike will allow me).

http://www.pbase.com/terrylovejoy/image/60876915

******** Comparision3 - Chuck Vaughns 20Da vs Orion Starshoot Pro ***************

NGC891 - Starshoot pro 1170mm focal length, 200mm aperture, 260min exposure

http://astrophotography.aa6g.org/Ast...gc891_ssp.html

NGC891 - 20Da 1180mm focal length, 155mm aperture, 260min exposure

http://astrophotography.aa6g.org/Ast...os/ngc891.html

Even taking into account extra light loss in the mirror system, the Starshoot Pro is working at an advantage (don't forget it is also cooled). The 20D shot shows as much detail and limiting magnitude is very similiar.

Conclusion

From the evidence I've seen either the 60% QE claims for the Sony CCD's are exagerated or Christian's values for the Canon sensors are too low.

Terry

Gama · #8 31-01-2009, 11:16 PM

The QE is stated by Sony, not by me, Elvis or houdini.
Well, maybe Elvis, but anyway, the specs sheet from Sony has the curve, wether they exagerate i dont know. You may well be right, again, the specs sheet shows the QE curve and thats what we go on. Also remember Sony use microlensing, which increases QE.
Your not going to see much difference by just looking at the images.
A sensor with low QE and low Full well will look like its collected more data that a Higher QE with a larger full well, but in reality we know differently. Even the KAF11002 has a maximum of 50% QE, yet again looks like its on par with the canons too.
I'll try and dig up the sony specs sheet somewhere..

The only real way to check CCD specs is to only test the sensor. Not the whole camera. Just like they test IC's, you set up a jig, and test the component on its own and make sure it meets manufacturers specs.

Theo

Peter Ward · #9 02-02-2009, 04:15 PM

Quote:

Originally Posted by Gama

The QE is stated by Sony, not by me, Elvis or houdini.
Well, maybe Elvis, but anyway, the specs sheet from Sony has the curve, wether they exagerate i dont know. You may well be right, again, the specs sheet shows the QE curve and thats what we go on. Also remember Sony use microlensing, which increases QE.
Your not going to see much difference by just looking at the images.
A sensor with low QE and low Full well will look like its collected more data that a Higher QE with a larger full well, but in reality we know differently. Even the KAF11002 has a maximum of 50% QE, yet again looks like its on par with the canons too.
I'll try and dig up the sony specs sheet somewhere..

The only real way to check CCD specs is to only test the sensor. Not the whole camera. Just like they test IC's, you set up a jig, and test the component on its own and make sure it meets manufacturers specs.

Theo

I've never known Sony to quote an absolute QE curve.

Do you have a link to this data? The only Sony data links I have found were for relative QE. (a fairly useless measure)

Gama · #10 02-02-2009, 08:58 PM

Why would "relative" be useless ?.
Anything referencing a relative value is to a reference, which manufacturers state in their Data sheets, so i think you are confusing this with something else.
In fact all electrical component data specs have relative values stated and are given reference to what, including your supplier Sbig, e.g Dark current is relative to a temperature, so go tell SBIG there data is fairly useless. They'll send a firing squad over ...
All the Sony HAD and SuperHAD sensor data is available on the net, so im sure a little surfing can yield your answers. Many manufacturers are using the Sony sensor, so they may have it on their web already. But i'll keep an eye out for my copy, somewhere.

Theo

Peter Ward · #11 03-02-2009, 05:17 AM

Quote:

Originally Posted by Gama

Why would "relative" be useless ?.
Anything referencing a relative value is to a reference, which manufacturers state in their Data sheets, so i think you are confusing this with something else.

Theo

Theo, I have numerous Sony CCD data sheets, and they are all the same.

Sony oddly state the spectral sensitivity of the devices in relative QE terms for the device rather than absolute QE referenced to a flux source.

Telling me that the green sensitivity is 30% better than the blue really doesn't help when trying to work out how the device stacks up in H-alpha or against other sensors, be they Kodak, Marconi, Thompson, Fairchild or whatever.

This is not say the Sony chips are poor in any way, but their "official" spec documentation is cryptic and you have yet to provide a specific web link to a absolute Sony QE curve..... can I take it because there isn't one?

Gama · #12 04-02-2009, 12:49 AM

Manufacturers get different data sheets. In fact you get binders worth, some almost manuals.
So unless you are one (manufacturer), then you really need to dig hard.
But i honestly dont give a hoot. Thats the specs, many manufacturers including Apogee use the ICX-285 in the cameras for instance, so go ask them for it. Do you believe a conspiracy being yarned out by Sony against Kodak ?.
Every manufacturer using the Sony sensors, big or small, color or mono, must have got the data somewhere..
Anyway, enough said.

Terry,
it was a great read, and a great post. These guys spend a lot of time doing this sort of work, and its good to see how they do it.

Theo

CometGuy · #13 04-02-2009, 10:36 PM

Theo,

If you come across any the Sony absolute QE charts please post them. I have no doubt they are high performance low noise devices (PeterM discover 2 supernova with a Sony chipped camera recently).

I am currently doing some research on Foveon devices and once I've got enough information I'll post some findings. From what I've researched so far Quantum efficiency seems exceptional. It seems to be >50% total, which would make them equivalent to the Kodak interline monochrome devices but with the ability to record colour! Biggest bugbear seems to be extremely high readnoise.

Terry