Brent I believe the temperature is sensed at the Digic processor chip but could be wrong, I will do some research on where It's coming from.
Haha I just ran across a thread on the Astromomy forum where It's states the Digic processor temp is what is reported in EXIF and It's primarily there to provide info on temps around the processor and overheat due to prolonged use of Liveview - location is not on the sensor, sensor board, at all rather on another board - likely the rear one. The funny bit is that you were part of that thread discussion.
Ok back again. I have been researching exactly what the EXIF temperature is all about and it seems it has nothing, absolutely nothing to do with the CMOS sensor operating temperature. It is built into the DIGIC processor and reporting the temperature of the processor. The DIGIC processor gets hot during sensor readout and processing, and seems to be responsible for most of the heat generated inside the camera.
There is an interesting thread on this on Cloudy Nights located here:
Scroll down to Spokeshave's post (#8) which explains what is going on inside the camera and the role of the DIGIC processor. Spokeshave seems to know this topic well.
Hopefully people can get away from quoting the EXIF temperature in relation to camera cooling functions - if it does cool off at all it would only be a side benefit of cooling and is not directly related to optimal image sensor noise reduction.
It is reasonable to conclude that only direct reading of the CMOS sensor or the cold finger adjacent to the CMOS sensor, can give an indication of CMOS sensor operating temperature.
An interesting aside comment from my reading would be that longer sub times would seem to mean less frequent readout and processing by the DIGIC processor so less heat buildup inside the camera body. Typically cooldown intervals that we program between subs are really cooldown periods for the DIGIC processor. So logically the longer the subs the cooler the camera should be compared to more numerous short subs that require more frequent readout and processing.
As an adjunct to all this, electronic noise comes from all the parts involved so cool down of sensor, processor etc would all have some effect on total noise output of the system. We've tended to focus on the sensor as being critical but any of the analog to digital conversion components probably contribute at least.
The fact that it managed to hold ambient on a longer exposure multiple image run has some value. I must pull off the darks it created and take a good look at them. I would assume the sensor does get a lot colder, I had dew on the finger edges outside. Getting a proble in somewhere handy is now an imperative.
This started off as such a simple idea ....
I am researching where the A/D converter and the Digic processor are located. The rear 'L' shaped board is known as the Main System Board and the Digic may reside there but seems to be un-marked in the 450, as it is in later models. The A/D converter I believe is on the daughter board in back of the sensor, It would get cooling from the cold finger. If anyone can find a block diagram or schematic of the camera it would really help.
OK mystery solved. The large chip under the shield on the sensor daughter board (above the CMOS sensor) is a R2A450022FT, aka a CCD Driver, or CCD Row Driver, and not associated with A/D conversion or DIGIC processing (thus not a major heat source, and even if it was it would be cooled effectively by proximity to the cold finger).
The DIGIC III chip resides on the main system board on the side facing the front of the camera hidden under the heat shield. I pryed mine off on the mono camera to have a look. From my reading, the A/D conversion is taking place in the DIGIC III processor as well, so this is a major internal heat source - but well away from the CMOS sensor and encapsulated in a aluminium shield which is chassis earthed.
Photos attached. FYI the two ELPIDA chips to the right of the DIGIC III are 512M DDR SDRAM chips.
Well done that man ... I assume temp sensor will be built into the Processor ?
Wonder why we've assumed for so long that reported temp was from the sensor ? Mind you, Canon has not been very forthcoming with model specific info.
I imagine that a cooled dslr is adequately cool not to be a problem. Still, interested to see what transpires
Yes but understanding the temp\noise relationship is important to both maximise the result and avoid icing up the system. Ideally around 5* is the easiest target but we'd need Glens argon bag or something else to go much further. Accurate data is therefore imperative.
Having generated and analysed a whole bunch of darks for my experiments, my dark noise does correlate with the EXIF reported temperature to a certain extent. I've also compared my "synthetic" darks (taken with the cold finger and/or being placed in the fridge/freezer) with "real" darks taken out in the wild and they correlate also.
I should add that the on a cold dark night, the EXIF reported temerature tracks ambient plus 5-7 degrees. My cold finger passes between the daughter board and main board off to the right of the camera to escape, and the metal shielding around what is likely the Digic chip applies pressure to the (insulated) cold finger as a structural "feature"
Incidentally, my experiments have shown that at a stable EXIF the darks are noisier for longer exposures...meaning 10 minute subs show more noise than 5 minute subs. Having run a bunch of experiments lately, I'm content with 5 minute subs at ISO1600 for most of my imaging, since they show no more noise than 10 minute darks at ISO800, except of course on tricky targets like M42 where the core is so easily blown out.
The Law of Diminishing Returns kicks in eh ? Just have to find the sweet spot. As I said before noise comes from any electronic device so coolling generally must help.
I am planning to try and locate the temp sensor along the top edge of the CMOS assembly. Should be about as close and as accurate as can be expected. Just hope to probe head is small enough. Shown below as the green line.
The Law of Diminishing Returns kicks in eh ? Just have to find the sweet spot. As I said before noise comes from any electronic device so coolling generally must help.
I am planning to try and locate the temp sensor along the top edge of the CMOS assembly. Should be about as close and as accurate as can be expected. Just hope to probe head is small enough. Shown below as the green line.
Cooling the camera internals with a cold finger should have some ("cooled by association") affect on the Digic processor even without direct cooling of it- having a 0C piece of copper running through is going to extract some heat from the enclosed space. Even if it just holds at ambient or closer too it, it is preventing heat build up. I don't know that I would support a direct correlation theory, but generally were all happy with the improvements in dark quality.
Of course, and importantly, the camera used and the DIGIC processor setup within that model, will have a bearing on heat and noise generation. Some Canon models have duel DIGIC processors (like the 1D, 7D,5DS, etc), and later generations of DIGIC have vastly larger feature sets than the DIGIC III (and very different heat generation profiles I would suggest). Many of the recent chips (Digic IV and above) provide better noise reduction and can support higher ISO speeds, but menu choices that turn on camera noise reduction are generally not recommended in astro camera setup. The point here being that correlation cannot be established across the model range, and any study of processor heat relationships should be done with identical models and processors (IMHO).
Saw some noise data on the 450D and it was the best at the time compared to some of the 500 series cameras and the 1000D. The 1100D and 1200D weren't in the test group but I gather they are of a similar low noise performance.
My other cold air idea for the 1200D may have some value in cooling all internals, we'll see when I get to that project.
Found a miniature temp sensor in amongst my 'bits and pieces' box. I already have an analogue temp meter, gets down to zero C so ran a few tests.
Got the probe tucked down the back of the finger close to the camera body and held by the insulation. 5V did very little, slow drop of a few degrees. 12 v on the fan and TEC got me a 12* C drop over ambient of 22* C so down around 10* C.
A small exposed edge of the finger was dewing up and felt quite cold so I have concerns about moisture inside the camera. If my order from China doesn't arrive this week I'll open the 450D up and get this wee probe right inside and see how it goes.
I'm a little disappointed in the drop, I was expecting more but as has been mentioned before consistency is more important and anything down closer to 0 is a good move.
I know I'm not hard core like you guys as I only cool down to 12-14C (as reported by EXIF), but the subs come out pretty decent at that temperature and all I wanted was to be able to still image in the QLD summer....
But anyway, I was out the other night and it was warm and humid, even at 3am, but the seeing settled down quite nicely and my camera ran like a champ. No fritzy images this time around. What did I do differently? I wrapped the camera in cling film
I really must stop trying to image on any available clear night (of which there have been too few!) and make some progress on v2...
Dunk I am sure you know this already but for other readers, the EXIF temperature data is actually the temperature of the DIGIC processor, or processors (some models have two). The DIGIC processor is on a separate board to the sensor. On the 450D the DIGIC 3 processor is on the front side of the main board under a metal s;hield - near the battery compartment and SD slot, well away from the sensor. The procesor chip is the component that really puts out the most heat.
Yeah so I've read, but the exif temperature with the cold finger correlates well with subs I've taken in the cold dark of winter, by statistical analysis of the darks. Also I use my Mk 1 to look at the subs and try to find a dark corner of the frame...not always an easy task
Brent what TEC and heatsink/fan combo are you using?
2706 (?) I think. Or something like that. It says 50 watts at 14 volts. 40mm on a side. The heat sink is my 'Tower' finned radiator, rescued from a server. Lightweight but efficient. I suspect though I could do with a much smaller radiator as there is very little thermal capacity in the finger. I know it's pulling heat from the camera but with the surface area involved and location behind the sensor I'd say that was somewhat minimal. Also the big heat sink looks like it will get in the way as the mount comes round.
The fan is a small 2" muffin, again out of a scrapped PC.
I LIKE the cling film idea !!! Brilliant !! It's going to be the next accessory for the observatory. Strap a couple of silica gel bags in as well. Purge it with argon !! Nice one...
Last night did some imaging, got the 450D alongside the scope with a lens loaded. Got it cooled down to 5*C, ambient about 20*C. The idea was to geta few first IR shots but no go. I forgot the fact that the 450D has lens communication problems which is why I got it cheap in the first place. I might be able to get it to work if I switch it over to manual focus but was getting late and I was getting the new PC to work with BYE and PHD2 so other priorities.
Nice one...
