16
\$\begingroup\$

Sensor cooling is a common technology to get less noise. Why isn't it available on high-end D-SLRs ?

(see http://www.andor.com/scientific_cameras/ikon-m_cooled_ccd/ for an extreme example)

I'm not saying D-SLRs should be cooled with liquid nitrogen. Just some cooling system !

\$\endgroup\$
7
  • 1
    \$\begingroup\$ It has been done, but I think this will demonstrate some of the downsides: youtube.com/watch?v=W4QYPIlMnVQ astro-fotografie.blogspot.com/2009/04/… astro-fotografie.blogspot.com/2010/05/… \$\endgroup\$
    – ESultanik
    Jun 20, 2012 at 18:09
  • 1
    \$\begingroup\$ The astrophotographers do it now. It does not need to be as large as they do it. I want to try it sometime. Most people who answered have missed your point. Very small Peltier on rear of sensor wold do. Ideally you'd have it in a dry air chamber to avoid fogging. \$\endgroup\$ Jun 21, 2012 at 0:17
  • 1
    \$\begingroup\$ @RussellMcMahon - Saying you can't do it for energy reasons is still valid, even if you scale the technology as small as possible. It's basically impossible to perfectly insulate something. As such, you will have to continuously expend several watts of power to keep a CCD/CMOS of any decent size cool. There is a reason astrophotographers cart around giant car-battery sized batteries to run their cooled cameras. You need the capacity. \$\endgroup\$
    – Fake Name
    Jun 21, 2012 at 8:20
  • \$\begingroup\$ @FakeName - Energy reasons were well down my list of erroneous responses. The 1st link had a 'noise fee' half hour exposure. That's not a minor improvement and many shots would benefit in selected cases. Then ... \$\endgroup\$ Jun 21, 2012 at 8:32
  • 1
    \$\begingroup\$ @RussellMcMahon - The noise reduction from cooling a sensor is largely proportional to the length of exposure, as cooling a sensor mostly serves to reduce the dark current. As such, it's really kind of pointless unless you're doing long exposures. Furthermore, there are many products on the market that do use the tiniest cooler they can get away with, and even so, they still use lots of power. See my answer for more. \$\endgroup\$
    – Fake Name
    Jun 21, 2012 at 8:49

4 Answers 4

24
\$\begingroup\$

Cost. Every price raise results in fewer sales.

Size. Cooling has to fit somewhere, those handgrips are already full of batteries...

Weight. There's a reason P&S are popular and not lugging around a brick is one of them =)

Battery Life. Cooling costs energy, lost energy means fewer shots in each battery pack.

Minor Improvement: only shots pushing the envelope would even benefit.

Condensation: artificially lower temp + humid air = water. Water + electronics = brick.

Heat Sink: all that heat has to go somewhere, in this case probably your hand.

Complexity: one more thing to go wrong in the field.

\$\endgroup\$
1
  • \$\begingroup\$ I just saw one in the busted equipment pile at work, and was like WTF. I'd never seen a liquid cooled camera before. An old guy told me they paid like 50K$ for the thing new, with pump, in the 1990s. Its kinda weird, it only has one line in for coolant, with ethyl glycol. I thought about grabing it but the pump was too big to carry easily. \$\endgroup\$
    – j0h
    Oct 1, 2016 at 15:18
9
\$\begingroup\$

It boils down to power, and lack of market demand.

There are specialty cooled-sensor cameras out there. They're generally just used for astrophotography.

The cooler that is used in almost all cooled cameras is what is called a thermoelectric cooler, commonly colloquially called a "Peltier" or "Seebeck cooler".

Generally, you will need a fairly chunky peltier to keep a image sensor cooled. For example, the Orion StarShoot G3 draws 12V at 1A to keep a 1/3" image sensor cooled to -10°C. That's 12 watts!

To calculate battery size needed, you multiply the current draw times the running time. As such, you would need a 1 Ah, 12V battery to run a cooled sensor for just one hour. As a comparison, the common Canon LP-E6 battery (as used in a Canon 5D2) is just 7.2V at 1.8Ah. Even ignoring the voltage difference, that's less then two hours of runtime with the camera on, for a much smaller sensor.


Furthermore, cooling a sensor is unlikely to do much to reduce ISO noise! Cooling a CCD/CMOS sensor largely reduces the dark current. However, the effects of dark current are purely a function of exposure time, so it only really helps with long exposures. High-ISO exposure noise is as much or more of a function of the CCD/CMOS sensor read noise then the dark-current noise of the sensor.
Readout-Noise is not affected by cooling the sensor, so high-ISO levels will be noisy, even with a cooled sensor.


Basically, there is really no reason to bother cooling a image sensor other then long-exposure. It offers very few benefits, and requires considerable additional system complexity, and massively increased power draw. A cooled system has to run continuously for the duration in which one expects to take photographs, as the cooling system will likely take many minutes (10-30) to cool the sensor down, and for the temperature to stabilize.

Furthermore, thermoelectric coolers are highly inefficent, and dissipate all the transfered thermal energy as heat. As such, a 5W peltier will dissipate 5W + any energy removed from the image sensor. This will almost certainly require active cooling, as the cooling efficiency is directly related to how cool the "hot" side of the peltier is.

It's actually common for high-end cooled image sensors to use liquid cooling, and they can dissipate many tens or hundreds of watts of heat.

\$\endgroup\$
3
  • \$\begingroup\$ FN - My continued, and still, disagreement with what you say relates to power levels, not to other photographic related aspects. Just because people do things really badly at present it doesn't mean it must be so. Peltier coolers are marvellous tools, but horrendously inefficient, as you note. You are unlikely to see anyone producing Stirling cycle coolers for 35mm domestic market sensors, but they would vastly exceed Peltier efficincies and are actually "quite easy" and could be quite low cost if eg Sony decided to use them. Cool down times ... \$\endgroup\$ Jun 21, 2012 at 19:44
  • \$\begingroup\$ FN ... When eg a Stirling cold-finger is attached to a sensor cool down time to Liquid Nitrogen order of temperatures can be a few seconds. Some guided artillery shells have Joule Thomson expansion nozzle cyrocoolers used to cool the IR sensors which guide the shells to their targets(!). The cooling cycle is triggered by the firing impact, the sensor is cooled to LN temperatures as the shell rises and by the time it tops its ballistic arc the IR seeker is ready for work. That they go to this effort for real time motion related use suggests substantial gains for short exposure IR use. \$\endgroup\$ Jun 21, 2012 at 19:49
  • \$\begingroup\$ @RussellMcMahon that's unique to (far) IR. Warm objects emit IR, that includes the IR sensor itself. So, you'd like the sensor to be as cold as possible. \$\endgroup\$
    – derobert
    Jun 26, 2012 at 20:52
6
\$\begingroup\$

Most likely because it would be bulky, and have a very high energy consumption.

Most cooling for electronics is for bringing it down closer to room temperature, but that wouldn't do much for a camera sensor, as it is mostly used for fractions of a second, so it won't heat up much. You would need a cooling element to get the temperature down, so it would essentially be a mini fridge or a mini AC.

This would need an element on the outside of the camera to dispose of the heat, which of course would be very inconvenient. The batteries needed to run all this would further add to the size.

So, what you get is a huge, heavy camera, with a hot surface, and a long startup time. Too impractical to make up for the reduced noise.

\$\endgroup\$
5
  • \$\begingroup\$ Hasselblad cameras are huge. And in low light conditions, you have all the time you need. (stars, animals in the forest with moon light) \$\endgroup\$ Jun 21, 2012 at 8:38
  • \$\begingroup\$ @SkippyFastol: Yes, you would only put up with the camera when used for things like night fotography, and the people who would buy an extra camera just for that is too few to get it down to a price where anyone would buy it. \$\endgroup\$
    – Guffa
    Jun 21, 2012 at 8:53
  • 1
    \$\begingroup\$ @Guffa - There are many manufacturers that make cooled cameras. They're just non intended for consumers. They're generally targeted at scientific imaging and astrophotography. They're also many thousands of dollars to many tens of thousands of dollars. Google "cooled ccd" to have a look. I count 6 different manufacturers on the first result page. \$\endgroup\$
    – Fake Name
    Jun 21, 2012 at 8:57
  • 1
    \$\begingroup\$ @FakeName: Yes, but the question was why they aren't use in D-SLRs. \$\endgroup\$
    – Guffa
    Jun 21, 2012 at 9:03
  • \$\begingroup\$ @Guffa - Good point. I guess I generalized the question to "Why aren't there hand-help cooled-sensor cameras out there. You are correct. \$\endgroup\$
    – Fake Name
    Jun 21, 2012 at 9:21
3
\$\begingroup\$

I've been making digital cameras since 1994, continuously. We currently have a line of security cameras. Up till recently, there wasn't a particularly good reason to add a cooler to a moderate range security camera. They have $10 sensors in them. Low light isn't a reasonable expectation at higher resolutions.

But lately the trend in new cameras is going two ways, due to lost cost camera pressure from China. To outdo the low cost cameras, everyone wants smooth 4K@30fps, or super low light @1080p.

I'm working on a super low light, using a new 2/3 inch sensor from Fairchild. It's an unheard of size in a security camera.

It's amazing. I'm tempted to put a cooler on it, because it's rated best at 68F. At that temp, you can take full color photos in a dark parking lot, and be able to clearly read license plates, without IR illumination.

But at normal California temps, and inside that sealed 4W camera body, the sensor is going to run around 140F, and so there's hundreds of hot pixels once the exposure exceeds 1/160th second.

A peltier is very tempting. Trouble is, they're highly inefficient. Adding one would increase the wattage of the camera to 10, which makes it a poor retrofit camera. Retrofits need to use existing power, which typical doesn't exceed 5W per camera.

Result: software hot pixel repair instead of what's needed, a cooler.

\$\endgroup\$
1
  • \$\begingroup\$ Isn't building a heat dissipator with sealing in worth it? \$\endgroup\$ Oct 5, 2016 at 20:09

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.