It doesn't look like Nikon would have the same issues from: "these oh-so-revolutionary translucent mirror" - if this is the cause I do not see that the sensor is the issue but the mirror.

Drab nailed it Re: light through the lens. Rob, that's a good point, as well.

What I suspect we're seeing here is that Sony inadvertently crippled themselves a bit with the way they decided to do image stabilization when the entered the DSLR game. Because their sensor is moved instead of lens elements to counter camera shake (Nikon and Canon both use in-lens stabilization), it seems to me that the trouble is probably that their sensor can't have very much mechanical contact with the body/frame to sink the heat from the continuous sensor operation during video. I don't know how or if Nikon fixed the overheating problem, but it would be much easier to improve if you could use the entire back surface of whatever the sensor is mounted on to transfer heat away.

Thank heaven the NEX-5 has in-lens stabilization. You can feel the frame near the sensor heating up during a long video sequence, but if there's heat to get rid of, it's a good sign if you can feel the camera heating up because it means the energy is moving away from the electronics.

edit: looking at the max video length with stabilization on/off, maybe I'm wrong and it's the stabilization system that's generating the heat.

Drab: Very thorough (and interesting) explanation. As someone who majored in English, I tend to leave the scientific stuff to people who understand that sort of thing better than me, which is... well, nearly everyone. :)

krevlingoodspeed said:
I read this as well and it really makes me wonder what Nikon did, if anything, to fix this problem with the D3100/D7000.

I am trying really hard to get funds together for the purchase but a part of me is glad I don't haven enough because I don't know if I am willing to play guinea pig!

Hopefully D3100 people can chime in and say if their video overheats or if Nikon delivers and 20 minute HD video is totally possible.

Well for one thing the D3100 and D7000 don't have a mirror in front of the sensor while recording to preventing the air from circulating in the mirror box. Just that alone would help to keep the sensor cooler.

aslightdelay said:
This might be a better question for JonnyApple, but my (really limited) sense of how light works would seem to indicate that less light lost means more light making its way to the sensor, which would mean it'd probably overheat quicker, no?

Outside extremely bright lighting conditions (and continuous use, such as video) the amount of light which strikes the sensor should play a very small role in how hot the sensor gets.

The much larger factor is in the on-sensor amplifiers, and they would be working harder the less light (and thus higher ISO) which falls on the sensor.

Math: (Let's apply a little logic to the question)
Let's assume a perfect (f/1) pro lens (77mm filters)
Let's also assume 100% of solar radiation is in the bands absorbed by the sensor (no reflection).
Let us also assume you're on the equator, on a cloudless day. (While the insolation values for most of the western world are 40-60% of this.)
Let us also assume (wrongly) that a lens not only has 100% efficient light transmission (T-stop = f-stop), but also that all light which strikes the front element is focused on the sensor.
Let us also assume filter-size = front element size.
Let us also assume that 100% of light which strikes the sensor gets converted into thermal energy (very bad assumption, if it did there would be no charge developed in the wells.)
And, lastly, let us assume a subject albedo of 100% (mind you fresh snow has around a 80% albedo).

These assumptions will create a worst case scenario.

A 77mm front element has a surface area of 0.0046566257 square meters.
Taking the rough equatorial number of a kW / m^2 we get 4.66 watts of light hitting the sensor, worst case (as if you're filling the frame with the sun with a f/1 lens.

A more reasonable worst case is shooting blinding fresh snow (80%) with a f/2.8 lens (1/8th the light or 12.5%) at 30 degrees latitude (75%) and we get .35 watts of light hitting the sensor.

That is negligible heat energy, even IF (magically) all of it got converted to heat.

EDIT: Oh, forgot to discuss the fact that there is an aggressive UV and IR filter over your sensor, there goes another significant percentage of the solar radiation.

TL;DR? If you're on earth ambient light does not heat the sensor significantly.

I read this as well and it really makes me wonder what Nikon did, if anything, to fix this problem with the D3100/D7000.

I am trying really hard to get funds together for the purchase but a part of me is glad I don't haven enough because I don't know if I am willing to play guinea pig!

Hopefully D3100 people can chime in and say if their video overheats or if Nikon delivers and 20 minute HD video is totally possible.

This might be a better question for JonnyApple, but my (really limited) sense of how light works would seem to indicate that less light lost means more light making its way to the sensor, which would mean it'd probably overheat quicker, no?

Don't worry, It's translucent. :)

More light lost than SLR.

http://www.engadget.com/2010/09/29/sony-a55-a33-video-recording-limited-by-overheating-sensor/

Saw this article on Engadget about the A33 and A55 sensors overheating under increased use. I hope this doesn't carry over into our Nikons. Though it seems to be related to video only which isn't a big deal to some.