Cold Finger - Test Results

[Stub Mandrel]

I have just tested the 'cold finger' for my Canon 450D.

Here are two images, all stretched as RAWs in paintshop pro by '7'  (roughly 32 times), then a screenshot taken of them side by side and saved as a PNG.

The left hand image is a bias frame at ISO800, the right hand frame is a 601 second exposure at ISO800.

I think the cold finger is a success! Bye-bye dark frames!

 

[MarsG76]

Excellent work.. after full spec modding my 40D, I'm trying to pluck up the courage to freeze my sensor using the cold finger also.

 

[Adam J]

What temperature did you get on the EXIF data? also are you getting condensation?

[Stub Mandrel]

1 hour ago, Adam J said:

What temperature did you get on the EXIF data? also are you getting condensation?

Confusingly the Exif data showed no cooling effect at all, well perhaps two degrees, I think it is measuring well away from the sensor chip.

I have just found this " My understanding is that the temperature sensor is actually built in to the DIGIC processor, which is on the main controller board, not the board on the sensor back. My experiments with TEC cooling the sensor (and mounting a thermistor in the gap between the sensor and its board) found that the EXIF sensor data does not agree at all well when the sensor is artificially cooled. Under typical conditions, it fairly closely tracks with sensor temp. It reports more of an internal body temp, rather than sensor temp. "

That would explain why people using cooler boxes notice a change in the exif temperature, but I'm not seeing much happen. I have the sensor back very well coated in hot melt, which will provide some insulation as well.

With the ambient temperature at 22 degrees a thick layer of hoar frost rapidly formed on the short bare section I left between the cooler and the camera body. I then put a piece of foam in the gap to cover it, so now virtually all of the finger is insulated. It looks like I am getting a similar 24 degrees of cooling to yourself, but obviously I can't prove it.

At the moment I'm putting about 0.05 watts into a nichrome wire around the sensor, which is almost certainly too low, so i fear condensation under filed conditions. I have dearth of power transistors for setting a sensible current level. How much power do you put into your condensation loop?

[Adam J]

8 minutes ago, Stub Mandrel said:

Confusingly the Exif data showed no cooling effect at all, well perhaps two degrees, I think it is measuring well away from the sensor chip.

I have just found this " My understanding is that the temperature sensor is actually built in to the DIGIC processor, which is on the main controller board, not the board on the sensor back. My experiments with TEC cooling the sensor (and mounting a thermistor in the gap between the sensor and its board) found that the EXIF sensor data does not agree at all well when the sensor is artificially cooled. Under typical conditions, it fairly closely tracks with sensor temp. It reports more of an internal body temp, rather than sensor temp. "

That would explain why people using cooler boxes notice a change in the exif temperature, but I'm not seeing much happen. I have the sensor back very well coated in hot melt, which will provide some insulation as well.

With the ambient temperature at 22 degrees a thick layer of hoar frost rapidly formed on the short bare section I left between the cooler and the camera body. I then put a piece of foam in the gap to cover it, so now virtually all of the finger is insulated. It looks like I am getting a similar 24 degrees of cooling to yourself, but obviously I can't prove it.

At the moment I'm putting about 0.05 watts into a nichrome wire around the sensor, which is almost certainly too low, so i fear condensation under filed conditions. I have dearth of power transistors for setting a sensible current level. How much power do you put into your condensation loop?

I dont have one I use a cool box which has been sealed to air and has desiccant inside it. I sealed the front with a CLS clip filter. I have gotten the internal box temperature as low as -15c without any condensation issues. The rear of my camera is removed also and a cold finger placed behind the sensor with a small heat-sink and fan attached to couple the cool air within the box into the sensor. I chose this method instead of direct cooling as its seems more versatile and I have heard of issues with EMI from the cold finger when the TEC is directly mounted also as already noted i dont need a heating loop.  The hot side of the box TEC is water cooled to keep the weight of the whole thing relatively low. I have had several different cameras in the box over time including one that i have mono modified.

[Gina]

EXIF data is indeed measured by the digital processing chip and is different from the sensor temperature with a cold finger.  I suggest a digital thermometer chip placed on the cold finger to read the temperature.  Can't remember the type number mow but it's a DS something and looks like a plastic cased transistor.  It's a 1-wire device and can be connected to an Arduino to read the temperature.  I posted a thread or two about cold finger cooling and control with a DSLR.

[Stub Mandrel]

Tempting as it is to pop a sensor on the cold finger, I think it is likely to be just a distraction. I can always check the temperature with my IR thermometer and assume the temperature difference with ambient will be pretty much constant.

[Gina]

I guess it depends on how much you rely on dark frame subtraction in your processing.

[Stub Mandrel]

I did start grouping darks according to the exif temperature (which has mostly 2-degree steps), but that way madness lies as the temperature can vary by several degrees over a session.

I now group my darks for single-figures, 10-20 and over 20 degrees and that seems to work fine. If I remake my darks on the same basis they should be OK.

[RichLD]

Excellent work! 

[Stub Mandrel]

A couple of pictures. To finish off I'm going to add a support for the wires, screwed to the tripod mounting hole. Will need a different resistor holder when the square replacement arrives.

The heatsink was huge, from a long-forgotten PC. I bandsawed it in half, and removed some fin with a linisher to create space for the wires. The fan mounting plate was a bonus that gives better airflow.

 

[benzomobile]

Very good work, Neil!

I made this mod in past on my 1100D and it performed very well (deltaT° around 31° C). 

To preventing dew, I putted a clear anti-reflective thin glass (from Edmund Co.)  in the position  of lpf filter (obviously I removed it), BUT,  just before  sealing it, I purged the air inside the small gap (glass/sensor surface). I used Nitrogen to purge.  After, I stacked  thirty-three smd 1 ohm resistors - solded in series,  all around the  external surface  of  the optical window, very close to the black plastic frame.

A 3.3 Volt power source  produces current around 100 mA. Dissipation is around  1/3 of Watt.

I find this power was very effective to avoid dew forming, but not high enough to significantly affect sensor temperature.

Beppe

[Stub Mandrel]

I have 10 ohms of nichrome wire, with 7 ohms of it around the sensor.

Let me do some calculations!

I will need about 220mA to get 1/3 watt 0.22*0.22*7=0.339

Voltage across the wire = 0.22 x 10 = 2.2 volts

Voltage to drop from average battery voltage of 13 volts = 10.8

Resistor required = 10.28/.22 = 46 ohms

I've ordered some 100 ohm resistors, with a 100R resistor, I will get  13/110 = 0.12A, = 0.1 watt.

I did try 30R but I felt it was too much power (about 1.3 watts) and the dropping resistor with 4 watts in it got very hot and now have 220 which is probably too little (0.025 watts)

I think my best strategy will be to start at a high value and work down if I get condensation.

 

[benzomobile]

I think your reasoning is correct.   

However,  instead of putting resistors  to drop tension  - producing too much high temperature, I suggest you to implement a voltage regulator like LM317, or similia.

With this chip you can do a linear  control  of  power dissipated by NiCr wire, simply  turning  a standard little linear pot.  

No dropping resistors, no unnecessary waste of energy.

Beppe

[Stub Mandrel]

55 minutes ago, benzomobile said:

I think your reasoning is correct.   

However,  instead of putting resistors  to drop tension  - producing too much high temperature, I suggest you to implement a voltage regulator like LM317, or similia.

With this chip you can do a linear  control  of  power dissipated by NiCr wire, simply  turning  a standard little linear pot.  

No dropping resistors, no unnecessary waste of energy.

Beppe

LM317 is a linear regulator and also works by turning the waste energy into heat! A resistor is simpler and doesn't go wrong!

It's tempting to add PWM control from my GOTO box, but I want to minimise the number of cables which are already multiplying far too fast

[MarsG76]

4 hours ago, Stub Mandrel said:

A couple of pictures. To finish off I'm going to add a support for the wires, screwed to the tripod mounting hole. Will need a different resistor holder when the square replacement arrives.

The heatsink was huge, from a long-forgotten PC. I bandsawed it in half, and removed some fin with a linisher to create space for the wires. The fan mounting plate was a bonus that gives better airflow.

 

That is one great and neat looking piece of work... you're an artist.

 

[Stub Mandrel]

2 minutes ago, MarsG76 said:

That is one great and neat looking piece of work... you're an artist.

Certainly better than my usual bodge - I'll post a picture of my failed cool box so you can see what I normally end up with!

[MarsG76]

47 minutes ago, Stub Mandrel said:

Certainly better than my usual bodge - I'll post a picture of my failed cool box so you can see what I normally end up with!

Let's see it.

[Stub Mandrel]

Well you asked for it... in my defence it hasn't improved from various changes in heatsink etc. then having an outer layer of polystyrene glued on and stripped off.

The peltier fits in the hole with heatsink and fan above, and a copper plate below. A small internal fan glued inside the door helped but I was still only getting about 8 degrees of cooling.

 

 

[MarsG76]

Definitely the cold finger is a lot beater/better and will work a lot better.

 

[jbrazio]

On 7/14/2017 at 17:34, Stub Mandrel said:

The left hand image is a bias frame at ISO800, the right hand frame is a 601 second exposure at ISO800.

I think the cold finger is a success! Bye-bye dark frames!

I'm far from being an expert on the topic but I don't believe we're seeing only pure random noise yet, there seems to be some horizontal banding on both pictures.

[Stub Mandrel]

14 hours ago, jbrazio said:

I'm far from being an expert on the topic but I don't believe we're seeing only pure random noise yet, there seems to be some horizontal banding on both pictures.

The banding goes with the 450D territory, it isn't thermal noise, it's some sort of read noise as it is present on bias frames.