I am trying to understand this. Well most of the sensors nowadays have their electrical board right on top of them. This is how they are manufactured. I happened to know that the ASI071 (IMX071) is one those sensor yet ZWO manages to cool it down pretty effectively. Same thing with the ASI183. But with the back illuminated ASI294 it is problematic based on what I have read here on CN.
So how do you cool your sensor when you do not have unobstructed access to it?
I have never seen the detailed specifications for the Sony IMX071 sensor. Since it seems to have been developed and marketed for the DSLR market, I can see why it might be supplied on a "board" (presumably FR4 / Fiberglass-Resin or derivative).
However, the Sony IMX294 is a different animal. It is supplied as a 268 pin LGA (Land Grid Array) with a ceramic package. That ceramic package has flat pin contact points arranged around the outside. (The pins for an LGA package are supplied by a socket or other mechanism on the electronics card rather than the chip.) The LGA packages like this generally have an open area in the center beneath where the sensor chip sits. The pin contact points are arranged in rows around the periphery of the LGA. There is no FR4 material between the silicon sensor chip and the outside (bottom) of the first level package. A TEC would be able to be connected to the bottom of the ceramic LGA package directly.
See the following link for a information on the IMX294 and its packaging:
Typically, since the LGA package has no pins, the chip would be mounted into a camera using a socket. I could not find a 268 pin LGA socket but did find a 257 pin LGA socket which is probably very close the the one required by the IMX294. See that LGA Socket at this link:
In a camera, the LGA socket would be mounted to the electronics board and then the sensor chip would be clamped into the socket around the periphery. I have not taken my camera apart to see exactly how that is done.
In any case, there would be a rather large open area below the sensor for direct contact with the cooling element (cold finger?) which is attached to the TEC.
All right, but how is that effective cooling. If the peltier cooled unit is not in immediate contact with the sensor itself?
Say that there are sufficient through wholes, whatever they may be, they are not the metal (copper or aluminum I guess) from the cooling element. Thus reducing heat (cold) transfer to the sensor. If it menages cooling only around the edges, wouldn't that result in uneven sensor temp. Either way, matching dark frames will be almost impossible. So, how do they go around this?
I am not aware of any commercial camera where the cooling is in direct contact with the sensor. There are always going to be a first level (silicon sensor to carrier) package, a second level (chip package to electronics card or assembly carrier) package, and sometimes a third level package (assembly carrier to electronics card). Only in custom-built scientific cameras would you be likely to find a Peltier directly bonded to a silicon sensor. The best we can do with commercial cameras is to get a ceramic first level package and connect that to the TEC.
There are some compromises or trade-offs with using a BSI (back-side illuminated) sensor. You generally get an increase in light sensitivity since there can be fewer structures blocking the light from hitting the photodiode area of a pixel. However, depending on the construction of the silicon sensor and its SoC (System on Chip) control circuitry, there can be a thermal bottleneck in the physical mounting of the silicon sensor to its (ceramic) first level package -- the LGA.
The main difference is not really in the absolute amount of cooling that can be delivered. Rather, it is in the time it takes that cooling to "soak through" the thermal path. A sensor mounted via an FR4 carrier is one example of a thermal bottleneck. Another is BGA (Ball Grid Array) Flip-Chip packaging between the silicon and the LGA in some BSI sensors.
In both cases, even cooling is simply a matter of allowing the system to reach equilibrium and then keeping it there. (Ultimately, reaching equilibrium also depends on the overall thermal design of the camera.) If you blindly turn on maximum cooling and start shooting, you may find that the sensor takes a while to start responding consistently. Allowing plenty of thermal soak time after reaching the desired temperature set-point will go a long way to improving results.
In my opinion, thermal stability control has a greater effect on results than the set-point temperature chosen. I am a proponent of running these CMOS (and CCD for that matter) cameras at reasonable TEC Power settings. Think about how much control you have if you are running at 98% of maximum TEC power and the chip warms up slightly. How long will it take to bring the chip back to temperature? On the same note, think about how much control you have when running the TEC at 5% of its maximum power. If the chip cools too much, how long will it take to warm back up? I have always thought running the TEC in the 30% to 70% range of maximum power will give the best (stable) control for cooling.
This post is already too long so I will cut the opinions and conjecture off here. I will just mention that there are some things about cooling in the latest generation of CMOS cameras that bother me. I have an experiment in mind that I need to run to characterize the relationships between TEC Power, Set-Point Temperature, and Ambient Temperature. I just need to get time to run it and take real data. With the amount of rainy / cloudy days we have had for the past 10 months, I cannot image so I may as well experiment.
[Edit] PS: All of our guessing aside, the best way to know how ZWO actually implements its cooling system attachment between the TEC and sensor package is take the camera apart and look. I have had no reason to do that but there may be someone out there with a dead ASI294MC-Pro that document a tear-down and show us what the cooling system involves.
I vaguely recall seeing some images here of an ASI camera that was partially disassembled to replace a vibrating fan (I think). I could not find those images via a search here.
Edited by jdupton, 21 May 2019 - 11:47 AM.