Chinese engineer updates the EOS R5

In a Baidu post, a Chinese engineer takes apart an EOS R5 and finds some curious problems with heat management.

From the images, the CPU and memory, while they get hot, do not have a direct path to the magnesium shell from the top as there is a daughterboard sitting in between the path.  While I understand the need for thermal paste - it's very unnecessary if there's no path for the heat to take.  The pads that are mentioning are more to assist with electrical isolation and to seat the daughterboard then it is to provide thermal conductivity.  There's no point in applying grease or pads to the top of the CPU or memory because the last thing Canon wants to do is have a path of heat from the CPU to the daughterboard sitting above the CPU and memory. 

What the Chinese engineer did is apply thermal paste to the CPU to sink heat away from the other side of the board.  The heat profile certainly did spread the heat out better which would lead you to believe it "works".

But it's unknown as he still hit a 20-minute limit, which some are stating maybe a firmware limitation - however it should be noted, that all he did was transfer the heat to another surface inside the camera.  

With the cameras sealed, heat sinking does very little in the way of good things.  You can't compare this to a ventable system (PC, Laptop) and assume the same for the internal of a camera - a heat sink, in this case, will simply just act the same, the temperature will equalize to that of the ambient air around it.  If anything, a heat sink would be worse - as it would take longer to cool down.

This document as changed a lot since the original form. Partly because it's a living document as more stuff gets spewed onto the Internet.  It's no longer just talking about what the Chinese engineer did.  Not all of it bad, some people legitimately trying to work with this camera and find out its limitations in an unbiased manner.  Reviewers such as nolifedigital and GeraldUndone have done amazing work for the community.

My thoughts

Now to preface, while I'm not practicing, my field of expertise was electronic engineering out of college.

There is a balancing act between removing heat from the components on a device that is handheld.  A personal computer is not held in your hand for perhaps 8 hours a day.  A camera is.  The thermal rules (and laws) are different between these devices.  One does not equate to another. 

I feel i have to stress this because of the complete bullshit that is being written about this camera and the camera's overheating from some sources. 

Let's get into the details.

Camera Layout and Design

This is a sealed camera - I really can't mention this enough.  There's no airflow, no way of getting hot ambient inside of the camera to the outside with a great deal of efficiency unless it's directly attached to the outer magnesium shell.  You can't think of your personal computer and think of all the heatsinks, paste, fans, etc - and think the same in a camera.  It is two entirely separate engineering problems.    Also the size and mass of the camera come into play.  For instance, it would increase both the size and weight to "heat pipe" all 5 components (RAM and CPU) and also force a complete redesign of the daughterboard.  

It's now pretty common knowledge that CFE cards are getting very warm.  There is even a warning when you open the Canon R5's door that stated warning, cards may be very hot.  Very hot cards in a sealed camera?  That wouldn't cause any issues, would it? 

While the Chinese engineer did move heat from the CPU to other sources, it still just simply moved the heat somewhere else inside of a sealed camera.  I don't think in the long run it helped much.

Now Canon can of course increase the mass of the camera (make it 1DX form factor) or add an active cooling solution (which will also increase its size and weight), however, Canon has commented that these were undesirable for the camera design.  Canon places a high degree of importance on body size and weight.

Sensors Can't run hot.  Running hot is bad.

There is an issue of heating up the area around the sensor - while this doesn't dramatically impact video because they only shoot at 10 bit usually and maximum 12 bit for RAW, for stills performance and exacting the most dynamic range from the camera, heat around the sensor increases the noise.  This increase in noise has a dramatic impact on the camera's ability to record a high dynamic range image.  This has to be a major factoring in the internal operating temperatures of the camera.

In this one study called Measured Effects of Temperature on Illumination-Independent Camera Noise that looked at both CCD and CMOS sensors with respect to noise, they observed this.

As you can see both spatial and temporal noise increased as the ambient temperature increased.

For this reason alone, Canon must keep the internal temperatures low, especially around the sensor.  Canon doesn't want the internal temperature increasing too much because this will cripple the camera for stills photography.  The Camera has to be able to go from shooting video to stills without degrading the image quality.

This is a stills camera first and foremost.  The EOS 5D and now the R5 line has always been primarily a stills camera.

Low-Temperature Burns

When you hold an object that has an elevated temperature, over time you can develop what is known as a low-temperature burn.  In lower temperatures (around 44C) you may just get a red skin at the point of contact, however, this can actually cause serious problems if the temperature is slightly more elevated (around 50C). Severe pain can start to occur at 52C.  These burns can actually be quite serious. So you don't want the camera to burn you, at least I don't.  This chart is from ISO 13732-1 which discusses low-temperature thresholds for equipment.

Keep in mind that these temperatures are based upon holding the object for 20 minutes.  Professionals usually hold the camera for up to hours.

Apparently, there are new EU laws regarding low-temperature burning and equipment going into effect, that Canon had to design the heat to minimize the shell temperature as much as possible.  You simply can't design the hardware to run in one region and not another. 

And people are stating this camera should run at 95C?

Electronic Operational Temperatures

While personal computer CPU may spike to 95C (Frankly it shouldn't - if your laptop or computer is hitting 95C - get it fixed.  Now. ), they are actually not supposed to exceed ~68C for even short periods of time. Running over 60C or so dramatically reduces MTBF (mean time before failure) rates of electronic components.  The main problem children are electrolytic capacitors, which degrade by halving the MTBF for each 10C temperature rise from 60C onwards.

This is why so much care and attention usually goes into server equipment and why they usually have massive cooling systems - because to guarantee reliability you can't run them hot.

While a camera should never hit a MTBF hour rating of the electronic components, still running it very hot over time, will cause an increase in the odds of failure. 

Faulty Software or The Canon Cripple Hammer™

While I'm not entirely convinced there is not some software bugs in Canon's operational temperature predictive logic, it is trying to do something that no other camera is doing - which predicts how long you can record for given the current operational temperatures of the camera.  Jesse Evens from FredMiranda has done some "freezer" tests with the R5 and found that the temperature and overheat warnings do come back down quickly while putting the camera in a freezer versus ambient which means it's a temperature-based limitation, and not simply a software timer.

When you have air trapped inside of a camera, and even if you open the doors, it's still trapped inside of the camera.  Air isn't exactly a good conductor of heat, so it's going to take a while for elements internal to the camera to cool down.  The fact remains that tests do show that extreme ambient temperature drops (aka freezer) can greatly reduce the cooldown time.

The software code may be too aggressive, and it also should have some camera thermal settings to allow the user to tailor the camera's thermal envelope to suit their needs. For instance, if you run the camera on a cage, you probably don't care if it runs out.  You are never holding it, and you have an increased mass to help with the heat dissipation.

Conclusions

This problem seems to be three-fold;

• To move as much heat as possible from the inside of the camera from heat-generating areas.
• Don't move enough heat to cause problems with the legislation by making the outer shell too warm
• Keep the sensor zone cool enough not to dramatically impact stills performance.

What we're seeing here I don't think accomplishes any of that.  I'm not even convinced that what Canon did was wrong when you consider all the factors, and not just look at one in isolation.

I think in the race to get video working (or to prove that Canon made mistakes), some are forgetting that this is primarily a stills camera and that temperature thresholds may be lower because an increase in sensor heat decreases your dynamic range.  There are also new operating temperature limits being imposed that may cause problems that prevent a more direct transfer of heat to the magnesium shell.  There's also the case of cards creating heat issues internally of which Canon can really do nothing with - it just is.

I do think Canon has to be more forthcoming on a possible heating solution especially considering there seems to be an issue with cards.  Perhaps if you are recording externally they can cut power to the cards Or maybe they have to make a "ludicrous" heat setting for those people that could care less about stills and just want to drive the camera until it melts. 

There's a lot more to this, take a deep breath, and stay safe out there. 

Edits: Added some more information regarding cards causing temperatures and the fact we know this thing works without cards.  And some musings on the "conspiracy theory" about firmware stopping the video.  I type faster than I think when I read something that truly annoys the daylights out of me.

Added some more about FredMiranda testing and discovery that it's not software-based limitations.

Basically re-wrote the entire thing

Another Edit. Cleared up the heat sinking of the DIGIC and images a bit.

• Baidu
• EOSHD