69 Mins of 8K from a cooled R5

[Jonathan Thill]

This is some awesome testing

R5 recording limits temp/sensor correlations: Canon EOS R Talk Forum: Digital Photography Review

Expert news, reviews and videos of the latest digital cameras, lenses, accessories, and phones. Get answers to your questions in our photography forums.

www.dpreview.com

 

[Kit Lens Jockey]

Very interesting that they were able to get an additional 15min of record time just by swapping in a new CF express card. But also it was on the camera that had been pre-cooled.

 

[Jonathan Thill]

Yeah the goal of the testing was to either prove or disprove what EOSHD was stating but it has really taken off from there. One test I think that would be interesting is do you need to cool the Camera to get the big jump or does cooling the CFexpress card extend things on its own.

 

[Stu_bert]

The CFE Cards (based on Delkin) have their own temp sensor on the card and will throttle I/O when they get too warm. So it could indeed be that there was enough headroom in the temp in the camera, but CFE card was too hot....

Good testing, the more that is done, the more it is highlighting the complexity... Thanks for sharing

 

[Mt Spokane Photography]

It would be interesting to swap out cards of the same exact type every few minutes as well as to swap to different brands. The cards get potentially extremely hot.

 

[Stu_bert]

It would be interesting to swap out cards of the same exact type every few minutes as well as to swap to different brands. The cards get potentially extremely hot.

They do.

Although they have no choice - that is their only method other than the device they are in cooling them. When they (based on Delkin site) hit a certain temp, they throttle by 30% and if it rises more they throttle by 50%. Without knowing the sustained writes, it is not clear whether the throttling can in effect stop the R5 from writing at the speed it needs to. And we don't get a more detailed message as to the cause (what part of the system).

I think the author mentioned that was a future test (multiple cards). He doesnt have an R5, he is relying on others. I don't have two CFE alas.... There was also mention that switching off before the overheat icon and swapping out cards, leaving it for a short while may also help extend the recording time.

They didnt cool the CFE Cards in the fridge

Hopefully this sort of testing might lead to some sort of cooling device - I don't think there is much chance that Canon can change how quick the heat dissipates, but lower consumption through different bitrates may slow down that buildup to an acceptable level.

For anyone shooting indoors, this may help offer a solution for them and offers significantly longer times (will leave others to comment about cooling a Canon from hot or warm in a Fridge)...

For outside shooting of stills n video, I think we will need something else...

 

[Kit Lens Jockey]

They do.

Although they have no choice - that is their only method other than the device they are in cooling them. When they (based on Delkin site) hit a certain temp, they throttle by 30% and if it rises more they throttle by 50%. Without knowing the sustained writes, it is not clear whether the throttling can in effect stop the R5 from writing at the speed it needs to. And we don't get a more detailed message as to the cause (what part of the system).

I think the author mentioned that was a future test (multiple cards). He doesnt have an R5, he is relying on others. I don't have two CFE alas.... There was also mention that switching off before the overheat icon and swapping out cards, leaving it for a short while may also help extend the recording time.

They didnt cool the CFE Cards in the fridge

Hopefully this sort of testing might lead to some sort of cooling device - I don't think there is much chance that Canon can change how quick the heat dissipates, but lower consumption through different bitrates may slow down that buildup to an acceptable level.

For anyone shooting indoors, this may help offer a solution for them and offers significantly longer times (will leave others to comment about cooling a Canon from hot or warm in a Fridge)...

For outside shooting of stills n video, I think we will need something else...

I was thinking it would be interesting if someone came up with some kind of CFexpress card extension, where it's a dummy CF express card that actually goes in the camera, and then it just runs a ribbon cable with all the proper connections to an external slot that you plug the CFexpress card into. So the card is out in free air instead of bottled up in the camera generating heat.

You would just need a way to fool the camera into thinking the memory card door is closed because the camera won't function with it open.

 

[dcm]

I was thinking it would be interesting if someone came up with some kind of CFexpress card extension, where it's a dummy CF express card that actually goes in the camera, and then it just runs a ribbon cable with all the proper connections to an external slot that you plug the CFexpress card into. So the card is out in free air instead of bottled up in the camera generating heat.

You would just need a way to fool the camera into thinking the memory card door is closed because the camera won't function with it open.

Placement matters. Note the proximity of the Digic X and the card slots. To move a lot of data very fast they have to be located very close together to minimize the length the data travels to meet timing specifications. That's likely why it is already all on a single card. Using a ribbon cable, connectors, etc. adds delay and potential signal degradation. It's probably not going to work reliably if at all. It would be similar to adding ribbon cables to the RAM in your PC/Mac and expecting everything to be the same. I know, I've done this kind of work in R&D over the years.

Or you need to use an interface defined for this type of external data transfer, like USB 3.1 where you would have to live with lower speeds to increase the cable length.

 

[SteveC]

Or you need to use an interface defined for this type of external data transfer, like USB 3.1 where you would have to live with lower speeds to increase the cable length.

And the lower speeds would negate the point of donkeying with these expensive incendiary devices cards in the first place. If they could live with lower speeds, the camera would take two SD cards (which for me would be an improvement, frankly).

 

[Jonathan Thill]

I just let my Camera record for 30mins (Camera went to sleep) 4KHQ 24P into my Ninja and the Camera is toasty. Not sure why people say it is not warm cause that feels pretty warm to me.

Oh and Camera was showing I could still record 2mins of 8K raw.

30mins of DNxHD is 165gigs...

 

[Kit Lens Jockey]

Placement matters. Note the proximity of the Digic X and the card slots. To move a lot of data very fast they have to be located very close together to minimize the length the data travels to meet timing specifications. That's likely why it is already all on a single card. Using a ribbon cable, connectors, etc. adds delay and potential signal degradation. It's probably not going to work reliably if at all. It would be similar to adding ribbon cables to the RAM in your PC/Mac and expecting everything to be the same. I know, I've done this kind of work in R&D over the years.

Or you need to use an interface defined for this type of external data transfer, like USB 3.1 where you would have to live with lower speeds to increase the cable length.

You're right, I didn't think about that.

 

[Chris.Chapterten]

I just let my Camera record for 30mins (Camera went to sleep) 4KHQ 24P into my Ninja and the Camera is toasty. Not sure why people say it is not warm cause that feels pretty warm to me.

Oh and Camera was showing I could still record 2mins of 8K raw.

30mins of DNxHD is 165gigs...

Interesting, was that with the cards removed from the camera?

 

[dcm]

You're right, I didn't think about that.

We often tend to forget about these things since they only matter in extreme performance cases. The only things sitting closer to the digic than the CFExpress card are the 4 RAM chips immediately surrounding the digic.

 

[Stu_bert]

I was thinking it would be interesting if someone came up with some kind of CFexpress card extension, where it's a dummy CF express card that actually goes in the camera, and then it just runs a ribbon cable with all the proper connections to an external slot that you plug the CFexpress card into. So the card is out in free air instead of bottled up in the camera generating heat.

You would just need a way to fool the camera into thinking the memory card door is closed because the camera won't function with it open.

Yep that would certainly help isolate that heat out - you would still get the PCI bus and whatever chip is controlling it, but yes - an extension would be interesting.

I doubt the door sensor is too complex. The battery door sensor certainly isnt - cause I have had to fool that when I lost it - ha ha.

Oh, well as I was writing that I decided to grab the nearest body. The 5DS is indeed very simple. At the top there is a small bit of plastic - push it to the right which is what the cover does, and voila.

The R5 is redesigned, but is pretty similar. There is a little raised ridge of plastic just to the left (if looking side on) of the CFE slot. If you now look at that ridge from the back of the body, there are 3 indents - two at the top, one at the bottom. The top and bottom I think are just there to help lock it. But the 3rd indent which is close to the top one has a small bit of plastic. Press that in with a small clip and the camera will power up.

I am guessing that blowing air over the CFE will make little difference - you are just pushing cool air in, and I don't know thermal physics and whether that would still push out hot air or not.There isn't much spare space in those slots - so maybe an extractor fan would be better to suck the warm air out.

Alas I am having compatibility issues with CFE at the moment, and my Sony is back with the shop for exchange. Reading over at FM, someone has info from Delkin that they will change their card firmware based on the R5 to improve compatibility. Whether that means other manufacturers are also, I don't know.

 

[Stu_bert]

Placement matters. Note the proximity of the Digic X and the card slots. To move a lot of data very fast they have to be located very close together to minimize the length the data travels to meet timing specifications. That's likely why it is already all on a single card. Using a ribbon cable, connectors, etc. adds delay and potential signal degradation. It's probably not going to work reliably if at all. It would be similar to adding ribbon cables to the RAM in your PC/Mac and expecting everything to be the same. I know, I've done this kind of work in R&D over the years.

Or you need to use an interface defined for this type of external data transfer, like USB 3.1 where you would have to live with lower speeds to increase the cable length.

You're right, but we have PCIe riser cards & ribbons for PCIe buses - typically used to move a Graphics card from being directly plugged in to maybe vertical due to constraints on size - so doesnt that suggest there can be flexibility there as CFE is using PCI? I've not heard (happy to be corrected), that this impacts Gfx performance.

General principle - as you stated - no argument.

Whether the PCIe chip has different constraints to the one in PCs - not a scoobie .

 

[Jonathan Thill]

Interesting, was that with the cards removed from the camera?

Nah, was not looking to push it hard. Just getting an idea of how it works with the Atomos.

The battery was also nice and warm.

 

[SteveC]

You're right, but we have PCIe riser cards & ribbons for PCIe buses - typically used to move a Graphics card from being directly plugged in to maybe vertical due to constraints on size - so doesnt that suggest there can be flexibility there as CFE is using PCI? I've not heard (happy to be corrected), that this impacts Gfx performance.

General principle - as you stated - no argument.

Whether the PCIe chip has different constraints to the one in PCs - not a scoobie .

I wonder how much of a hit it would take, if, say the card was made the same size and shape--except for its length; let it project outside the camera 50%. Put all the heat-generating stuff there, and insulation to keep the heat from travelling back into the part that's in the camera. Heck, even put radiating fins on the part outside of the camera!

Added path for signals to and from the card would be about one inch, which equates to roughly 1/12th of a nanosecond of propagation time.

 

[Stu_bert]

I wonder how much of a hit it would take, if, say the card was made the same size and shape--except for its length; let it project outside the camera 50%. Put all the heat-generating stuff there, and insulation to keep the heat from travelling back into the part that's in the camera. Heck, even put radiating fins on the part outside of the camera!

Added path for signals to and from the card would be about one inch, which equates to roughly 1/12th of a nanosecond of propagation time.

Well if that does turn out to be the main heat source, then it will be interesting to see how much people look to mod their r5 as you suggest.

Re performance hit - well I guess it depends if it drops it below the required threshold of 325MBps and whatever new modes Canon can provide which will lower that bitrate. But if we can extend graphics cards with riser and cables, then I would assume not significantly.

Whether the design was motivated purely by that signal impact (which may well explain why there is a stacked design and the digic is in close proximity to the cards and the slot, or whether that was intentional to balance against where the screen is located and balance the hot zones I guess we won’t know), but given the size constraints I think Canon wouldn’t have tested your scenario.

I guess the only challenge would be the sealing of the unit- not weather, just dust and moisture...

I looked again at the card slot on the r5 - don’t think removing the door would be easy therefore whether any unit could fit on the side replacing the door may well be serious enthusiasts only...

 

[Stu_bert]

I wanted to try and do a couple of tests based around the open door. I have a small 5" fan, and I can lower the appt via air con to 18 deg. I did consider mounting my R5 on my tripod so it sat really close to the aircon unit but decided against that

I am using UHS II cards, and with 8k30 ipb I got 20mins 10s. But I can sit that USB fan in front of the camera, behind the camera (screen open), into the side where the cards are, and the outside of the body certainly gets cool, but the cooldown time remains unchanged. Amusingly I have to put a block behind the fan else it manages to push itself off the table with the blowing so close to the Canon. Battery door is open, no battery, no card etc.

If you check the DP thread, swapping CFE made actually a -ve impact to the record times, which was interesting. I was going to try the 8k30 at 12 mins, and then swap the cards, but waiting 2 hours each time is annoying, and the active cooling of the body seems to make little change to that - leading to the conclusion of others, that the body is perhaps too good at letting external temp affect it, unless you leave it for extended periods at really cold temps (which I am personally not prepared to do).

I am thus leaning towards another conclusion of others I have read outside here, that once you trigger the thermal limit, whatever it is that requires the cooldown isnt perhaps something we are thinking about - the UHS cards are cool, the slots feel cool although I have no temp probe, the screen is cool, yet something insider not cooling down. Which leads me to currently conclude there is a slow dissipatation "component" inside and Canon knows how long this takes to dissipate the heat based on the properties of the body and this "component" takes 2 hours based on calculation to dissipate the heat naturally through the body before it is safe to use again.

Cooling in the fridge slows down the heat build up, perhaps, but once the thermal trigger happens, then the cooling of this component isnt something we can accelerate.

Maybe on Friday I can try 8 minute recordings - I have 3 UHS II 64GB cards - and maybe that is enough to allow me to test it 3 times to see if I can exceed 20 mins. I don't hold much hope, given my limited & non-scientific tests today.

 

[Jonathan Thill]

I wanted to try and do a couple of tests based around the open door. I have a small 5" fan, and I can lower the appt via air con to 18 deg. I did consider mounting my R5 on my tripod so it sat really close to the aircon unit but decided against that

I am using UHS II cards, and with 8k30 ipb I got 20mins 10s. But I can sit that USB fan in front of the camera, behind the camera (screen open), into the side where the cards are, and the outside of the body certainly gets cool, but the cooldown time remains unchanged. Amusingly I have to put a block behind the fan else it manages to push itself off the table with the blowing so close to the Canon. Battery door is open, no battery, no card etc.

If you check the DP thread, swapping CFE made actually a -ve impact to the record times, which was interesting. I was going to try the 8k30 at 12 mins, and then swap the cards, but waiting 2 hours each time is annoying, and the active cooling of the body seems to make little change to that - leading to the conclusion of others, that the body is perhaps too good at letting external temp affect it, unless you leave it for extended periods at really cold temps (which I am personally not prepared to do).

I am thus leaning towards another conclusion of others I have read outside here, that once you trigger the thermal limit, whatever it is that requires the cooldown isnt perhaps something we are thinking about - the UHS cards are cool, the slots feel cool although I have no temp probe, the screen is cool, yet something insider not cooling down. Which leads me to currently conclude there is a slow dissipatation "component" inside and Canon knows how long this takes to dissipate the heat based on the properties of the body and this "component" takes 2 hours based on calculation to dissipate the heat naturally through the body before it is safe to use again.

Cooling in the fridge slows down the heat build up, perhaps, but once the thermal trigger happens, then the cooling of this component isnt something we can accelerate.

Maybe on Friday I can try 8 minute recordings - I have 3 UHS II 64GB cards - and maybe that is enough to allow me to test it 3 times to see if I can exceed 20 mins. I don't hold much hope, given my limited & non-scientific tests today.

One of the guys doing a lot of the testing at DPreview restored full record times from overheated after 25mins in the freezer.