Teardown: The Canon EOS R5 gets an autopsy

Apr 25, 2011
2,502
1,878
Actually, there is a metal layer between the pad and the second board on top.
An RF shield?

So, the 2 pads ARE thermal pad. The guy who open it even added some thermal paste to touch the processor also as Canon did not.
It's a very weird choice indeed.

The worst is that the recording time did not increase despite a actual thermal measurement completely acceptable for any electronic component.
I don't see why it should have increased. I'd say it's lucky that it didn't decrease.

The most acceptable explanation is that Canon
...has better understanding of thermal management of their cameras than some random "Chinese engineers".
 
  • Like
Reactions: 1 user
Upvote 0
How expensive is the LI-9000 (let's ignore the costs to implement it into a body)? And how big would the tile need to be at 5mm thickness?

Then you would need to factor in multiple copper pipes (CFE, Digic, Memory, Not sure whether you can for the sensor given IBIS), so at least 3 copper pipes)?

Plus the CFE generates a lot of heat from what others are saying as I don't think it has any other choice but to heat up it's metal casing.

And out of curiousity, if the Silica tile is exposed to the outside, does it dissipate any quicker? If Canon used it in the bottom of the body, would that help dissipate quicker? If I understood you correctly, it wouldn't be hot to the touch, but would increase the bottom of the case by 5mm? Is the tile sturdy in all other respects?

Finally, your solution is perhaps a little high end. The R5 is not intended to be a 8k video workhorse, I think many would be happy to shoot stills, shoot HQ video, shoot stills and the only limit is the max record times. I'd be fine with that. Is there anything you know which may not be as costly as the ceramic tiles?


===

Okie Dokie! I did some further asking of our mechanical and materials engineers.

The LI-900 silica heat sink manufacturing costs in Canadian dollars at our own facility requires $75,000 for cryogenically-hardened Tool Steel machined mold that has a 10 by 10 part matrix (i.e. 100 heat sinks in one mold) which takes about 48 hours to CNC-machine properly to 1/10,000th of an inch tolerances (2.5 microns) and then we cryogenically surface harden the tool steel mold by dipping the mold into liquid helium for a specified period of time.

The ceramic matrix interior composite which is a very poor conductor of heat BUT is able to ABSORB a very high number of Joules per hour per cubic millimetre costs us $30 CAN ($23 US) per stamped/baked part. The thin outer heat shield layer is silicon/boron composite which prevents heat from radiating from the interior is another $10 to bake it on at a 1.5mm or 2mm thick layer.

Then you need to run solid copper heat transfer pipes from the parts being cooled (i.e. the CPU, RAM, memory cards, battery, etc) into the middle of the ceramic matrix composite thermal absorbent compound and ANOTHER separate set of heat pipes (i.e. with a gap in between the other set of heat input pipes) that runs from the interior out to a 2 cm square copper plate (or other thermal transfer metal plate material) that is at the bottom of the camera so it dissipates the ABSORBED heat over a long period of time to the outer camera surface.

For strength reasons (i.e. the LI-900 tile is quite weak structurally), we encase the entire tile in high strength heat resistant HDPE (High Density Polyethylene) plastic which melts at about 270 Celcius. Since the magnesium body of the Canon R5 camera has a DIFFERENT thermal expansion coefficient than the ceramic heat sink, you also have to "float" the entire heat sink within the R5 body usually via attaching it to the internal body via flexible silicone rubber gaskets.

AND....after a bit more in-depth of a look-see at the R5 camera tear down, the copper heat transfer pipes MUST be a minimum of 3mm diameter going from the copper plates attached to the CPU's and other hot parts AND that means there MUST be at least 3 mm thick surrounding encasement of heat-keeping ceramic matrix composite on all sizes. The key point is that the ceramic tile and copper heat pipe creates a large thermal differential between the CPU's (and other hot parts!) which gets ABSORBED into the interior ceramic matrix composite of the heat sink tile which continuously absorbs heats but POORLY radiates it.

That heat slowly and eventually finds itself moving over and transferring to the second set of embedded ceramic-gapped output copper heat transfer pipes which then moves the heat out to the 2 cm square metal heat output block on the bottom of the camera.

The thermal mass of the interior ceramic is such that almost 5 million joules of heat (about 60 watts) can be kept within and NOT pass through the outer 2mm thick heat shield layer for up to 24 hours! By covering BOTH sets of copper input and output heat transfer pipe with a thin layer of heat shield ceramic (i.e. a one-mm thick layer of silicon/boron), the heat STAYS in the pipes as it transfers from the hot CPU into the ceramic interior of the heat sink for a long period of time ranging from 12 to 24 hours.

That absorbed heat eventually saturates the heat sink and that thermal energy MUST SLOWLY migrate over to the second set of embedded heat pipes which then moves the heat outwards to the bottom-of-camera radiative thermal plate while the camera is working AND when it is shut off.

Along with the outer heat shield layer, it means a 10 cm by 6 cm by 15 mm thick (total weight is around 200 grams worth of ceramic and copper pipe) of LI-900 tile could easily ABSORB the 160,000 Joules per hour the Canon DIGIC CPU's and other hot parts tend to give off as heat. AND ... it could absorb it for about 12 hours to 24 hours for my estimate at FULL DCI 8k 30 fps and DCI 4K 120 fps continuous recording before the camera would start to get really warm to the touch.

The Total Cost for LI-900 tile-like material plus copper plates and embedded copper heat transfer pipes and ceramic heat shield outer layer plus HDPE casing is $120 Canadian per part or about $90 US at a run of 5000 heat sink parts. The cryo-hardened molds tend to lasts for about 20,000+ stampings so that's two million camera heat sinks. Not too bad at all for a mere $75,000 worth of tool steel and CNC machining/cryo-hardening!

If Canon does a run of 100,000 heat sink parts per order, that cost drops STEEPLY to about $60 US per heat sink tile and copper piping because you can buy copper and ceramic matrix powder at a discount at those volumes!!!

SO YES! Canon could EASILY do this! They don't even need to use expensive silica ceramic composite matrix as the heat absorbent material for the tile. Just use the SAME ceramic material used in electrical resistors that you buy for 25 cents! With some Canon Research and Development scientists and elbow grease, they could bring down the cost of adding a SUPERB 12 to 24 hour endurance super heat sink into an R5 or R1 camera down to less than $25 US!

I would be more than HAPPY to pay an extra $25 to get a slightly thicker Canon R5 and R1dx with such 12 to 24 hour DCI 8K / 120 fps 4k video recording heat wicking capability AND its weather sealing kept intact!

GO DO IT CANON !!!

You've GOT the technology AND the scientists!

YES YOU CAN DO IT CANON !!!

V


P.S. I just realized you also NEED encase the INTERIOR of the battery compartment AND the insides of the memory card slots with uncoated copper or aluminum thermal transfer plate so as to wick lots of heat away from the heated-up battery and memory card via direct contact heat transfer into the heat sink tile using copper heat transfer pipes that are round or flat ribbon in form AND have been coated with silicon/boron heat shield layering. Canon can do some thermal transfer modelling in CATIA CAD/CAM/FEA software (or whatever they use for 3D CAD!) to ensure that the heat transfers into the heat sink quickly, stays there and transfers only slowly out to the bottom of the camera.

V
 
Last edited:
  • Like
Reactions: 1 user
Upvote 0
Jun 27, 2013
1,861
1,099
37
Pune
True but thermal paste can dry and loose its transfer ability over time. Not good to build in a maintenance item that would require a trip to the service facility in 5 years. High content silver paste requires many heat/cool down cycles before it becomes most effective. Head pads are the least invasive method to allow a good contact medium inside a camera.
Canon could have used Graphite pads with improved conductivity instead of the thick ones that are seen right now. Though in that case they would have had to increase thickness of the mag body just enough to make good contact with the heat generating components.
 
Upvote 0
===

Okie Dokie! I did some further asking of our mechanical and materials engineers.

The LI-900 silica heat sink manufacturing costs in Canadian dollars at our own facility requires $75,000 for cryogenically-hardened Tool Steel machined mold that has a 10 by 10 part matrix (i.e. 100 heat sinks in one mold) which takes about 48 hours to CNC-machine properly to 1/10,000th of an inch tolerances (2.5 microns) and then we cryogenically surface harden the tool steel mold by dipping the mold into liquid helium for a specified period of time.

The ceramic matrix interior composite which is a very poor conductor of heat BUT is able to ABSORB a very high number of Joules per hour per cubic millimetre costs us $30 CAN ($23 US) per stamped/baked part. The thin outer heat shield layer is silicon/boron composite which prevents heat from radiating from the interior is another $10 to bake it on at a 1.5mm or 2mm thick layer.

Then you need to run solid copper heat transfer pipes from the parts being cooled (i.e. the CPU, RAM, memory cards, battery, etc) into the middle of the ceramic matrix composite thermal absorbent compound and ANOTHER separate set of heat pipes (i.e. with a gap in between the other set of heat input pipes) that runs from the interior out to a 2 cm square copper plate (or other thermal transfer metal plate material) that is at the bottom of the camera so it dissipates the ABSORBED heat over a long period of time to the outer camera surface.

For strength reasons (i.e. the LI-900 tile is quite weak structurally), we encase the entire tile in high strength heat resistant HDPE (High Density Polyethylene) plastic which melts at about 270 Celcius. Since the magnesium body of the Canon R5 camera has a DIFFERENT thermal expansion coefficient than the ceramic heat sink, you also have to "float" the entire heat sink within the R5 body usually via attaching it to the internal body via flexible silicone rubber gaskets.

AND....after a bit more in-depth of a look-see at the R5 camera tear down, the copper heat transfer pipes MUST be a minimum of 3mm diameter going from the copper plates attached to the CPU's and other hot parts AND that means there MUST be at least 3 mm thick surrounding encasement of heat-keeping ceramic matrix composite on all sizes. The key point is that the ceramic tile and copper heat pipe creates a large thermal differential between the CPU's (and other hot parts!) which gets ABSORBED into the interior ceramic matrix composite of the heat sink tile which continuously absorbs heats but POORLY radiates it.

That heat slowly and eventually finds itself moving over and transferring to the second set of embedded ceramic-gapped output copper heat transfer pipes which then moves the heat out to the 2 cm square metal heat output block on the bottom of the camera.

The thermal mass of the interior ceramic is such that almost 5 million joules of heat (about 60 watts) can be kept within and NOT pass through the outer 2mm thick heat shield layer for up to 24 hours! By covering BOTH sets of copper input and output heat transfer pipe with a thin layer of heat shield ceramic (i.e. a one-mm thick layer of silicon/boron), the heat STAYS in the pipes as it transfers from the hot CPU into the ceramic interior of the heat sink for a long period of time ranging from 12 to 24 hours.

That absorbed heat eventually saturates the heat sink and that thermal energy MUST SLOWLY migrate over to the second set of embedded heat pipes which then moves the heat outwards to the bottom-of-camera radiative thermal plate while the camera is working AND when it is shut off.

Along with the outer heat shield layer, it means a 10 cm by 6 cm by 15 mm thick (total weight is around 200 grams worth of ceramic and copper pipe) of LI-900 tile could easily ABSORB the 160,000 Joules per hour the Canon DIGIC CPU's and other hot parts tend to give off as heat. AND ... it could absorb it for about 12 hours to 24 hours for my estimate at FULL DCI 8k 30 fps and DCI 4K 120 fps continuous recording before the camera would start to get really warm to the touch.

The Total Cost for LI-900 tile-like material plus copper plates and embedded copper heat transfer pipes and ceramic heat shield outer layer plus HDPE casing is $120 Canadian per part or about $90 US at a run of 5000 heat sink parts. The cryo-hardened molds tend to lasts for about 20,000+ stampings so that's two million camera heat sinks. Not too bad at all for a mere $75,000 worth of tool steel and CNC machining/cryo-hardening!

If Canon does a run of 100,000 heat sink parts per order, that cost drops STEEPLY to about $60 US per heat sink tile and copper piping because you can buy copper and ceramic matrix powder at a discount at those volumes!!!

SO YES! Canon could EASILY do this! They don't even need to use expensive silica ceramic composite matrix as the heat absorbent material for the tile. Just use the SAME ceramic material used in electrical resistors that you buy for 25 cents! With some Canon Research and Development scientists and elbow grease, they could bring down the cost of adding a SUPERB 12 to 24 hour endurance super heat sink into an R5 or R1 camera down to less than $25 US!

I would be more than HAPPY to pay an extra $25 to get a slightly thicker Canon R5 and R1dx with such 12 to 24 hour DCI 8K / 120 fps 4k video recording heat wicking capability AND its weather sealing kept intact!

GO DO IT CANON !!!

You've GOT the technology AND the scientists!

YES YOU CAN DO IT CANON !!!

V


P.S. I just realized you also NEED encase the INTERIOR of the battery compartment AND the insides of the memory card slots with uncoated copper or aluminum thermal transfer plate so as to wick lots of heat away from the heated-up battery and memory card via direct contact heat transfer into the heat sink tile using copper heat transfer pipes that are round or flat ribbon in form AND have been coated with silicon/boron heat shield layering. Canon can do some thermal transfer modelling in CATIA CAD/CAM/FEA software (or whatever they use for 3D CAD!) to ensure that the heat transfers into the heat sink quickly, stays there and transfers only slowly out to the bottom of the camera.

V
Wow, thanks for the additional information, very interesting!

You should share with Canon - not cause I think they will retrofit it - just they might consider it for future products. Thanks again.
 
Upvote 0
what are you talking about here? MTBF significantly degrades with temperature increase.
My point is the camera is heated anyway then CPU should have the priority on heat dissipation over other components, except the sensor.
And MTBF degrades with temperature increase then Canon should improve the heat dissipation then? 2 hours completely turn off to back to cold start is just weird.
Please reminds that even your oven could not hole the thermal that good. You could try this at home, turn the oven to 60-70 Celsius (the temperature when recording 8K) for 30min, then turn it off and measure after 1 hour, 2 hours.
 
Upvote 0

koenkooi

CR Pro
Feb 25, 2015
3,543
4,082
The Netherlands
My point is the camera is heated anyway then CPU should have the priority on heat dissipation over other components, except the sensor.
And MTBF degrades with temperature increase then Canon should improve the heat dissipation then? 2 hours completely turn off to back to cold start is just weird.
Please reminds that even your oven could not hole the thermal that good. You could try this at home, turn the oven to 60-70 Celsius (the temperature when recording 8K) for 30min, then turn it off and measure after 1 hour, 2 hours.

When I turn off my oven it starts actively venting the hot air. That fan stays on for quite a while.
 
Upvote 0

zim

CR Pro
Oct 18, 2011
2,129
318
My point is the camera is heated anyway then CPU should have the priority on heat dissipation over other components, except the sensor.
And MTBF degrades with temperature increase then Canon should improve the heat dissipation then? 2 hours completely turn off to back to cold start is just weird.
Please reminds that even your oven could not hole the thermal that good. You could try this at home, turn the oven to 60-70 Celsius (the temperature when recording 8K) for 30min, then turn it off and measure after 1 hour, 2 hours.
I don't think my oven goes down that low :giggle:
 
Upvote 0
Wow, thanks for the additional information, very interesting!

You should share with Canon - not cause I think they will retrofit it - just they might consider it for future products. Thanks again.


Down below is an attached JEPG image that details an OPEN SOURCE hardware design under GPL-3 licence terms that is for a general purpose heat sink applicable to still cameras, video cameras, smartphones, tablets, standalone computers, embedded devices and peripherals with ALL them being of any form factor and design!

You can use multiple thermal conductor and/or thermal insulator materials which SHOULD BE MATCHED to the thermal radiation rates of the components, the heat transfer pipe/plate pathways and the desired amount of thermal storage capacity for the camera runtime required.

Again .... THIS IS AN OPEN SOURCE GPL-3 hardware design created by me on August 13, 2020 and is free for use under open source GL-3 licence terms!

OF COURSE YOU MUST DO YOUR OWN THERMAL ANALYSIS, SIMULATION AND TESTING before you make your products using this or derived design available for sale or use as this design is a mere unfinished testing and design prototype and has NO WARRANTY and NO expressed or implied fitness-for-use made by me HarryFilm! This means Use and/or Modify it at your Own Risk and Own liability! I ain't responsible in any form or manner for what happens!

V
 

Attachments

  • Open Source GPL-3 Heat Sink Design covering Stills-Video Cameras, Phones, Tablets, Computers i...jpg
    Open Source GPL-3 Heat Sink Design covering Stills-Video Cameras, Phones, Tablets, Computers i...jpg
    497.4 KB · Views: 120
Last edited:
Upvote 0

AlanF

Desperately seeking birds
CR Pro
Aug 16, 2012
12,298
22,371
Harry
I have completely misunderstood your design and its concepts. The LI-900 you propose to use is a a very efficient insulator of very low thermal capacity, so it isn't a heat sink, it just greatly slows down heat transfer without absorbing heat. If it surrounds copper, it will keep the copper hot, and if the copper is in contact with the heat producing components, it will keep them hot. So please explain what is going on so even I can understand it.
Thanks
 
Upvote 0
I know you signed up just a few days ago and this shit is all you post about. Hope Sony is paying you well.
Well, currently I do paid works with a Nikon Z6 and a D800, film cameras for hobby includes Canon F-1New, AE-1 which is FD, EOS-1 which is EF and a AF35M which is Canon PnS.
I have other cameras of different brands but I think you don't really care about those. I do camera repair both film and digital as side job, and of course have stripped down some Canons to their shutter blades and mirror box.
I used to have some Sony actually about 1 year ago and have Canon DSLR from time to time.
Is that enough for you to stop blindly giving people name tag ?
Is this a forum where people share their opinions or anything not-praising Canon is considered hostile?
 
  • Like
Reactions: 1 user
Upvote 0
I don't think my oven goes down that low :giggle:
Mine got 50 and 75 Celsius mode. Well I don't think it's that accurate but generally in that ballpark.

When I turn off my oven it starts actively venting the hot air. That fan stays on for quite a while.
Mine does not have fan but I forgot my pizza there for 1-2 hours, it's quite cold when come out.
 
  • Haha
Reactions: 1 user
Upvote 0
Yes, but once that air heats up, even the heat sink will heat up and the two will reach the same temperature. Once that does, it stays there. the cool down time will be very long. in the end, it doesn't buy you much. The delta T between that and the outside ambient temperature will slowly decrease over time, with or without a heat sink. Heat sinks are good for spiky heat - but sustained heat such as recording for minutes of video? it's not going to work. yes ideally, spreading the heat over the camera body is the ideal solution, however, I"m not convinced they are entirely legally allowed to do that.

On top of that, the cards are a major source of heat - this which we know because the damned camera records for hours without the cards in.

A camera is not going to run super hot. It simply can't .. again, something EOSHD can't comprehend, you're not going to run it much past 50C because of the thermal characteristics of a CMOS sensor and the fact that people have to hold these cameras in their hands for long durations of time.

If you have active cooling and/or are allowed to actively shunt as much heat as possible to the outside, then your design rules are different.

Also - be aware that the temperature reading that Andrew is using is what Canon calls the "EFIC temp" which is for speedlights and lens interface, and not the core CPU cores, or even sensor. There is a correlation, but that depends on body design. In other words, it's not like measuring core temp on a processor.
True that the CPU's core temp is different with what you can measure on its surface. There are some sensor inside the CPU package providing those info.
I'm just unsure the Temperature in the EXIF of JPEG file point to which part of the camera.
And Andrew keep complaining about CPU surround by RAMs while it's quite a general design not only in camera but in PC and phone, to maximize speed, minimize interference.
But, after seeing Sigma Fp, I think there are other way to work out the heat which Canon could apply for R5. Sigma Fp do RAW internal, fullframe 4K with much smaller body size compare to R5.
 
Upvote 0
Apr 25, 2011
2,502
1,878
My point is the camera is heated anyway then CPU should have the priority on heat dissipation over other components, except the sensor.
Where can the CPU dissipate the heat in a weathersealed body, except toward the sensor?

And MTBF degrades with temperature increase then Canon should improve the heat dissipation then? 2 hours completely turn off to back to cold start is just weird.
Maybe because it's not true? DPReview says that half an hour is enough.
 
Upvote 0

adigoks

EOS 750D
Jul 12, 2020
68
80
i have some speculation that i want to share

the other major source of heat could be from voltage regulator or VRM. this module is needed to regulate voltage & electric current to other component such as CPU, RAM, Storage, etc.

this part is gonna be hot all te time while the system is on. in PC you can find it on motherboard around the CPU socket and on Graphic Card .
those VRMs are usually covered by heatsink like this
asus-sabertooth-990fx-processor-socket-area.jpg


or in Graphic Card it looks like this
52eaa773_gtx480_pcb.jpeg


if not cooled properly VRM can become hotter than the other electronic parts like CPU,GPU or RAM.
EVGA-thermal-fix-before.jpg

after searching for canon tear down image. in canon R5 , one of the VRM is located in here


eos-r5-vrm.jpg

the specification about R5's VRM can be found here

i hope someone can confirm it . also i hope someone can make some minor customization to cover it with termal pad + thin aluminium/copper plate to see if its help or not.
 
  • Like
Reactions: 1 user
Upvote 0
Where can the CPU dissipate the heat in a weathersealed body, except toward the sensor?


Maybe because it's not true? DPReview says that half an hour is enough.
Weather sealed is different from heat sealed, think about your cooker, heat transmit quite well through there but it's quite weather-sealed if your lid is closed properly.
After 30min, you could record high quality again, but not up to the 25-30min that Canon state, you will get like 2-5min something after 30min rest (completely off).
 
Upvote 0
Apr 25, 2011
2,502
1,878
Weather sealed is different from heat sealed, think about your cooker, heat transmit quite well through there but it's quite weather-sealed if your lid is closed properly.
Not quite getting the analogy. Are you saying that a watersealed camera could conduct heat better if it were immersed in water instead of being handheld?

After 30min, you could record high quality again, but not up to the 25-30min that Canon state, you will get like 2-5min something after 30min rest (completely off).
Again, that's not what DPReview is saying.
 
Upvote 0
Not quite getting the analogy. Are you saying that a watersealed camera could conduct heat better if it were immersed in water instead of being handheld?


Again, that's not what DPReview is saying.
Well, a lot of other sources disagree with Dpreview then.
As for weather sealed and heat, what I mean is in 3 ways of heat transfer: conduction, convaction, radiation. Weathersealing reduce convaction to minimum, but it does not neccessary reduce conduction.
 
Upvote 0
Haha... I think accusers of canon regarding this overheating thing are going overboard, and are just forgetting to look at things in relative terms when I comes to purchasing. What is the number one rule when considering a product and it’s value?? Compare it to other products on the market.

1. The R5 can do everything the A7R4 and Z7 can do in terms of video without overheating, and more(albeit with the overheating and limitations).

2. Want better video specs and this want to compare it to the A7SIII?? Fine do so. But the A7SIII is the opposite of the R5z great for video but worse for stills (12MP).

So, besides complaining that the R5 isn’t a unicorn, what is the best option for you? A7R4, Z7, A7SIII, or R5???? Cause those are the options right?
 
Upvote 0