Measuring Sensor Temperature by Hot Pixel

[avandonk]

It occurred to me that by measuring a hot pixel at various temperatures you could then measure the sensors temperature directly by just measuring the same hot pixel.

I took a ten second dark at 400 ISO straight after turning on the camera at ambient 23.7C. The camera was then turned off and the fridge set to 15.0C. When everything equilibrated took another dark at the same settings. Repeated this for 10.0C and 5.0C.
Raw frames converted to linear tiffs and digitally developed in IP.
Below is a graph of fridge temperature vs pixel value measured from the same pixel obviously.

Now the interesting bit. If you then just turn the camera on and leave for about thirty minutes to equilibrate. The temperature of the sensor rises by 2.2C.

Set the camera taking 8min darks and when constant take another 10 sec 400 ISO dark. The sensor temperature has actually risen by 16C.

So when the Peltier cold side is -5.0C the environment in the fridge is -0.4C. The sensor when taking a continuous series of exposures is at 16.0C.

While the camera is exposing

Fridge Temp 8.8C pixel value 30,131
Fridge Temp -0.4C pixel value 19,664

This needs further exploration.


Bert

[h0ughy (David)]

never would have thought of doing that Bert - interesting to see how this pans out

[Dennis]

Hey Bert

You’re revealing the scientist in you! Great investigation and a nice set of data. I look forward to the peer reviewed article – let me know the science journal in which it will be published!

Cheers

Dennis

[sheeny (Al)]

Science?! Bert! How could you?! What's science got to do with astronomy?...





Al.

[avandonk]

I am retired Dennis and will never again go through the pain of peer review. This is most probably not very original. There are many people working in this area.

Although what is really interesting if you could make a 'hot' single pixel by ion implantation say you would have a temperature sensor with a tiny thermal inertia.

I just want to know what my system is really doing so I can get better images. Anything I discover will be passed on to others, gratis.

It took a while for the penny to drop after using the fridge for some time. Hindsight has 20 20 vision!

Yes h0ughy and Al it is all about Science. Science is the only means we have for understanding the Universe. All the rest is superstitious nonsense.

Bert

[avandonk]

This was also posted by me on CN. It explains a bit better what is happening.

It takes more than 40 min of continuous exposures for the sensor temperature to level out. From a cold camera start at -0.4C it takes about five 8 min exposures for the sensor temperature to reach equilibrium. The size of the out of camera JPG file will give you an indication of this.

I always turn the camera on before starting fridge cool down. This stops any possibility of condensation on the sensor. When the fridge gets to 5C I then set it taking dummy exposures so that the sensor is eqilibrated for then collecting real data. I normally image with the fridge cold Peltier side set to -5.0C in summer and -10.0C or -15.0C in winter. The PID temperature controller can keep the cold side of the Peltiers to within 0.1C of set temperature.

The main purpose of the fridge is to have a constant temperature environment so darks match lights exactly. The fact that the camera is in an environment some 20C to 25C below ambient is a bonus.

Temperatures can drop by up to 15 or 20C over a night in Melbourne so making it almost impossible to get matching darks. I used to use ICNR but the fridge allows me to collect almost twice the data in the same time.

Bert

[Terry B]

Interesting concept. My only comment is don't use a very hot pixel .They tend to be more unstable and will give variable readings regardless of the temperature. Hopefully there are very few unstable pixels on your sensor but there will be 1 or 2.

[avandonk]

Originally Posted by Terry B

Interesting concept. My only comment is don't use a very hot pixel .They tend to be more unstable and will give variable readings regardless of the temperature. Hopefully there are very few unstable pixels on your sensor but there will be 1 or 2.

You are correct Terry there are a few unstable pixels on my sensor. It is easy to find them by just scrolling through a series of darks taken at the same temperature. I had thought of giving them names but you cannot use naughty words on this forum. Lets just say b1, b2, etc.

The calibration curve would soon show any unstability of a pixel.

Bert

[rally]

Bert,

By coincidence I was reading this today.

Its a technical note by Defence R&D Canada. (pdf format)
Accuracy of Dark Frame Subtraction

Its a method developed to extract faint signals of satellite smears below the normal Signal to Noise ratio of a CCD (below 3db !!) - so its of interest to the AstroPhotographers here.

http://pubs.drdc.gc.ca/PDFS/unc67/p528754.pdf

Whilst not intended to be used to measure the CCD temperature it may be able to be applied to this purpose although you would need some reference data to start with - the issue is the correlation between dark frames subtracted from each other having a temperature differential and the resulting residual gradient.
They are talking about 0.1°C differentials here and the fact that they can better determine CCD temperature than the the temperature readouts of the sensors attached to the chilled CCD cameras due to lags between actual real CCD temp and the measured temp from a sensor located on the chiller or elsewhere.

From Page 14 Point 3.
"The real temperature indicator can be retrieved by calculating the mean dark frame intensity value."

For the astrophotographers its a method to determine if your darks are actually at the same temperature (because they say, they are not always, despite what the camera says, within a range that is able to to resolve signal below this lower threshold.

Have fun.

Cheers
Rally

[tornado33]

very interesting discussion this. Its a very sound idia of measuring temperature by hot pixel.

Because my camera isnt temperature controlled, the software i use Iris has the ability to do "adaptive dark subtraction. I draw a small box covering at least 10 x 10 pixels or more on a light frame on an area free of stars. Iris measures the dark noise there and compares it to the dark frames. if there is a difference, it accounts for it. That way if a dark frame is hotter or colder then the light images, a good subtraction will still happen. Of course such a process is not 100% perfect, and I still try and take my darks right after the light images.

It would probably be fairly easy for a software programmer to write a small piece of software to measure the dark noise, and read the ISO and exposure time, and from that calculate a chip temperature.

I thought of a way to calibrate it by measuring the chip temp direct. I have a non contact thermomenter, that reads temperature by sampling IR ratiation from the object. I might take some long exposures and point the thermometer in towards the sensor and see how far above ambient the sensor gets
Scott

[kinetic (Steve)]

Hi Bert, all,

Interesting stuff Bert,
what does one use if they have no hot pixels?

Seriously, sounds like a good thing to follow up and as
you mention, it does take a while for a CCD to settle
even after capturing for an hour or so in stable ambient!

I notice this when I do about an hour series on a cold night
with the DSI II.
Darks at the start don't match darks at the middle and
darks at the end even with an only minimal ambient change.
I always 'normalise' the darks anyway and make a bad pixel map.
The application of the BPMap is always different even on
almost identical nights.
Sometimes 100% works, sometimes 160% works.
This says to me that using a pixel as the basis for any type
of temperature measurement and indicator of how a session
may turn out(for me at least), is way too variable.

Steve

[avandonk]

What all this tells me Steve is to take dummy images to get your sensor up to temperature and then take your lights and darks. If your ambient changes your sensor temperature will change by the amount ambient changes albeit slowly. In the case of my Canon 5DH it is 17C above ambient once equilibrium has been reached. I am afraid that an uncooled CCD heats up even more as they consume more energy.

The aim of the post was to show how sensor temperature had no real correlation with ambient as it depended on the exposure history. My carefully controlled experiment just gave me an insight into what was really happening. I do not think it can be used to control sensor temperature as the hysteresis between ambient and sensor temperature is too great.

Here is a much better graph

http://www.iceinspace.com.au/forum/s...ad.php?t=43805

Bert