Adventures with DSLR cooling

[12dstring]

I've decided to dedicate my 350D to astrophotography, so have been making a few changes.

The IR filter replacement was done a few years ago, and a few months ago I added in an amp-off modification to remove the lower-right amp glow. The next step is peltier TEC cooling.

I took piccies of the internals of the amp-off and cooling as I was doing it but won't be able to get to them until next weekend, will post later...

But for now here's how the outside enclosure looks at the moment:

It's fitted into an IP65 box, and I'll plug up all the holes I make so I keep the humidity low enough to prevent condensation on things that shouldn't have condensation on them.

Instead of cooling the whole box, I've opted to cool the sensor directly, there's an L shaped piece of copper that attached to the back of the sensor and passes through the side of camera and box. There's two 2.5v 3.5A peltier modules attached to that and a 5v laptop heatsink and fan.

Inside the box will be a small microcontroller to monitor everything, a humidity sensor to make sure the temperature can't go too far below the dew point, and a TEC controller. The lid of the box will have a small LCD display to display temperature, internal humidity..etc as well as temperature setting controls.

The front just has a hole for the T-adapter, and a hole to poke the lens release button.

Whilst the camera is taking a picture, it's actually consuming 220mA at 8V. I'm not sure what it's doing, but it ends up as 1.75W of heat being produced inside the camera.

Whilst cooling the entire camera enclosure certainly helps, as you're cooling the cameras from the outside whilst the camera heats itself from the inside, it's not as efficient as cooling the insides directly. Also as the cooling path is less impeded it will cool down a lot faster (as it doesn't have to cool down the whole camera first), as well as responding to changes in the cooler power quicker - which will hopefully allow fairly stable temperature control.

I ran a few tests to see how well it works.

The green line shows how the sensor heats up with no cooling (pre-modification, no copper attached yet). After just 20 minutes of imaging it's already 7°C hotter (roughly doubling the noise) and after half an hour it was still going up.

The blue line is with the camera off and cooling at 80%. After 15 mins it is 22°C cooler that ambient - not bad.

The red shows the effect of having the camera on. As the camera is heating itself up now it can't get quite as cold, but still closes in on 20°C below ambient.

Now taking into account that without the cooling the sensor wouldn't be at ambient, but heating up, here's a graph of the difference in sensor temperature over time due to the cooling:

Just 15 minutes in and it's already 25°C cooler than it would have otherwise been, reducing noise by over 90%.

There's still a bit more insulation to add around the TEC itself and the camera which may improve things slightly. Once done I'll do a longer retest at 100% (I don't intend to run it normally at full power, the main aim is cold but stable).

For comparison, Gary Honis has a graph of the temperature drop using his camera cooler here: Peltier Cooling of Modified Canon Digital Rebel XSi (450D) - Version III -by Gary Honis

The temperature final drop is similar, but it takes quite a bit longer to cool down, and with a larger TEC (48W vs 15W).

[Gina]

Thank you - that's very interesting I'm thinking of trying Peltier cooling in time.

I've noticed the noise with long exposures on my Sony A200. Now this very day I have received a Canon EOS 1100D body which I bought for dedicated AP use. Not had a chance to try it yet but that will be the one I shall experiment with once I've got used to it.

[12dstring]

One things that soon becomes apparent after cooling is how intrusive the amp-glow becomes. As with modified webcams the glow is caused by an amplifier on the sensor that heats up, and the glow is causes gets worse with exposure length but doesn't respond to cooling.

Luckily as with webcams it can be eliminated the same way, by reducing the voltage supplied to the sensor whilst it's actually imaging, before returning it to the full voltage during readout.

All inspiration for my modifications came from Mike Kudenov (http://glogg.jupiter-io.net/300d/amp_off/amp-off.asp) and Cas Wilders (http://members.chello.nl/c.wilders/350D/350dmods.html).

One good thing about the 300D and 350D is that there is an onboard signal that can be tapped into to trigger the voltage drop, so the whole thing can be self contained. The circuit itself is pretty simple; a 3.6V low noise regulator, a pair of mosfets, a couple of resistors and 3 capacitors to lower noise. Mine is a variation of Cas's circuit, with one less mosfet.

I chose the LP2985 as a regulator, and a SI4532 which contains both mosfets to save space. Total cost to build <£2.

R1, R2: 4.7k

C1, C3: 10uF

C2: 10nF

They circuit was squashed onto a pcb about 12x8mm, easy enough to hide inside the camera. There are four (electrically connected) 5v lines going to the sensor, which must all be cut first. I chose to do this at the ribbon cable connector, but if I could go back I would do it at the sensor legs (the pin spacing is much larger).

The red wire is 5v, blue is ground, and orange the amp-off signal. The isolated 5v pins on the sensor were then connected together, a couple of extra 0.1uF capacitors soldered to the pins for luck (interference is never fun), and the four wires soldered to the little pcb.

As said before the amp-glow is much more visible when the sensor is colder, but here is a dark comparison before the camera was cooled (sensor at ~30C) with amp turned on and off. Heavily stretched 15 min exposures:

The modification proved a success as can be seen. The horribly annoying glowing semi-circle has disappeared completely from the lower right, and glow I hadn't previously noticed around the edges is also reduced.

[Tim]

There's some work gone into that!

Have you any pictures yet, astro ones that is?

Cheers

Tim

[Gina]

I've noticed amp glow with my Sony A200 DSLR at long exposures - of an amplitude similar to the noise level at ambient temperature. I haven't tried my Canon 1100D yet.

[12dstring]

I've not taken any astro pictures with it yet since starting these mods, but the temperature control electronics will take a while so I expect I will before that's done. Just having the copper as a passive heatsink ought to keep the noise down by drawing away the heat produced.

The 1100D should be fine for amp-glow. I believe all models after the 350D efectively have an amp off circuit included.

[Gina]

Dark frame tests of my 1100D show no sign of amp glow

Do you have any info/pictures of the camera mod to put the cold finger onto the sensor? I intend doing the filter removal mod to my 1100D myself and when I get right inside to the sensor I may be able to see if the cold finger method is viable.

I was planning to make a cold box for the camera but a cold finger direct to the sensor is clearly much more efficient. And while the camera is it bits for the filter removal would be a good time to add the cold finger.

[MarsG76]

I know that this is an old post, but one question with the peltier cooling. I'm about to start work to add a cold finger to the back of my 40D sensor and I'm wondering if you're getting any frost or dew over your sensor?

I added some peltier cooling to a DMK41 in the past and even though there was moisture dripping off the case there was no moisture on the CCD but what about a cooled DSLR ?

 

[tooth_dr]

5 minutes ago, MarsG76 said:

I know that this is an old post, but one question with the peltier cooling. I'm about to start work to add a cold finger to the back of my 40D sensor and I'm wondering if you're getting any frost or dew over your sensor?

I added some peltier cooling to a DMK41 in the past and even though there was moisture dripping off the case there was no moisture on the CCD but what about a cooled DSLR ?

 

I would be very grateful if you would keep me posted about your developments. I would really like to cool my 40ds

[MarsG76]

1 hour ago, tooth_dr said:

I would be very grateful if you would keep me posted about your developments. I would really like to cool my 40ds

Yeah, of course... hopefully I wont be posting that my 40D is dead!!!!

As far as any dewing issues, I'm thinking of wrapping a eyepiece heat strip, that I have in my case, around the t-adapter. Hopefully that will give just enough heat to stop any dew but the cold finger will keep the sensor as cold as possible.

[Gina]

What you really need to do is have the sensor in a sealed enclosure with a desiccant bag to absorb moisture though this isn't easy with a DSLR.  OTOH I did manage it several years ago with a stripped-down DSLR.

[Stub Mandrel]

On 07/02/2012 at 01:34, Gina said:

Dark frame tests of my 1100D show no sign of amp glow

Just noticed this thread is six years old...

My cooled 450D shows no amp glow either, for the record. This is a 5-minute master dark, stretched brutally far beyond what would ever happen to a sub.

Looks like the later sensors have the amp more distant from the sensor (the plaid pattern is an artefact of the masterdark being saved as grey so the bayer pattern is visible).