Andrew Reid

Yep absolute madness that it can be heated in 35C direct sunlight, almost scolding to touch and at same time not have any of the main temp sensors register high enough for a cut off, when doing HDMI 4K HQ 10bit. What next, can you go 20 mins of 8K with a hot CFexpress and camera fresh out the oven on a baking tray? Canon better have a good explanation in the draft stages by now. It will need to be about 20 pages long!

The only way it can possibly be really heat related is: 1. When 8K or 4K HQ are enabled (regardless of 10bit/8bit/compression type) the sensor fully sends it to the image processor and RAM, they all heat up in unison going full beans all at once, all 3 trying to melt one another, even in the menus, camera doesn't shut off the 8K feed in the menus, or in live-view, apart from maybe if you have Overheat Control turned on but I haven't really noticed that help much. 2. Maybe a margin is implemented in the thermal control system to take into account different speeds of card, some hotter than others, CFexpress temp is read by SMART, and added into the mix, once camera detects average temp on all 4 areas higher than a certain degrees (sensor, card, CPU, RAM), even if one of them crosses a boundary like the sensor as a result of the others running toasty, then maybe it will flash the overheat warning. 3. Maybe it tries to anticipate when this warning will occur by reducing maximum recording times in the limited modes, as you use the camera, whether this is artificial or accurate, I don't know. 4. I cannot in any way figure out the recovery times. Nothing about these make any sense. If the overheat trigger was caused by CFexpress card getting too hot, then removing that and putting a cold one in should reset the counter, right?

It does't add up does it! 1. CFexpress card throttling... Why would it even break into a sweat recording in 8bit 4K at low bitrates? Yet camera in 4K HQ 8bit quits long before the CFexpress card has had to fully utilise anything close to maximum write speeds. 2. Sensor overheating... I could appreciate this is sensitive to heat, it gets noisy if hot. But it doesn't cut off for external recording doing full pixel readout (8K) for 4K HQ so that's that theory gone and the Chinese thermometer tests show the surface area of the chip is not the main heat source in the camera. 3. DIGIC processor overheating. Well the EXIF temp correlates to camera activity in so much as I observed it go from 46C to 62C in 8K mode. If this temp reading is on a PCB far away from the image processor, it is indicative more of internal air temp and CPU could be 10-20C higher. I could understand if the CPU has a thermal run-away problem past 80C why it would stop. But why would it do so when you add 8bit 4K HQ H.264 to the mix? Pretty much any camera can do that unlimited. And we know the SD card and sensor are not the heat limited parts. So why would this compression tip it over the edge... And why does it need a 2 hour recovery time when it cools down to normal ambient temps pretty much instantly? 4. Menu overheating. Nothing about this makes any sense whatsoever. If the CPU has a fault and it cannot shut down all cores in the menus or something, maybe. I say unlikely. Surely the net processor usage downsampling 8K sensor data to 4K HQ for HDMI is higher than scrolling a list of text 5. Stills / Live-view = reduced runtimes even 0 mins of 8K. This is the other very artificial thing with no basis in physics. It just doesn't make sense that the CPU is happy to go all day in live view or 4K 10bit H.265 from a pixel binned sensor output, but combine this with the 8K sensor output and BOOM processor has had it in less than 15 mins. 6. Recovery times = too long. No basis in real thermal dynamics or temperatures. So it is all a big mystery and one I will be taking to the UK small claims court if Canon doesn't start cooperating or answering these questions.

Same limits with SD card puts a bullet through that theory Although I am sure CFexpress cards do get hotter and can thermally throttle all by themselves so it's another variable.

Are you looking at the EXIF temp data @mechanicalEYE during these hot sun shoots? Would be interesting to see what it says

No it doesn't at all. Just because you can freeze the temp sensor doesn't mean to say limitations are not artificial / cripple hammer territory. It could be tricked into pausing the timer below a certain temperature. But at normal operating temperatures NOT in a fridge, the cripple hammer runs as planned. Remember my test, read it again, and question why the camera reports 46C and can go for 15 mins 8K, and then reports 46C again and cannot do it again, even after waiting 30 minutes turned off. How is this based on real CPU temps?? Horshak does not have a camera.

If the 15 min 8K cripple clock is also based off temperature readings, it doesn't not make it artificial. Remember the EXIF temp at 46C results in two different maximum 8K run times depending on how long the live-view has been on for. In our tests it was 46C at the start of a 15 min clip and 62C at the end. And it has been observed back at 46C but suddenly camera refuses to roll a single frame. If you freeze the camera for 11 hours and trick it into going longer because a temperature sensor on the PCB takes longer to correlate with the internal temps being generated on the main CPU, not only is this dangerous and could brick the camera processor if it really is getting too hot, it doesn't prove that the timer and temp limits aren't at an artificially low cut off. But the recovery timer is still the most fake thing for me. As to see an inhibitor flagging just 5 mins of 8K after waiting over an hour and 30 mins is frankly bonkers and in no way can it be based on real physics.

He put it in the fridge for ELEVEN HOURS!! The CFexpress card doesn't make a difference for me. Sure they get much hotter than SD cards but I get the same run times in 8K to CFe as I do to SD card

Magic Lantern was able to read temps on previous Canon DSLRs. It gives us an idea of what's normal in stills mode for a digital camera. They also found the EFIC chip temperature correlates well with the temperatures in JPEG EXIF data, like we are able to see on EOS R5 https://sternenkarten.com/2016/04/24/temperaturanzeige-mit-magiclantern/ Internal camera temperature can mean a lot of things - sensor, processor, chassis air temps, or an average of a few different sensors. However generally I think it is giving a good impression of the internal temp around the main circuitboard. This is useful because we can measure the internal temp at the beginning of a cold start, before we hit record on 8K for 20 mins. Then when we turn the camera off, we can turn it back on again in 5 minutes and see if the temperature is back to what it was before the 20 mins. If it is back to what it was, then logically there is no reason why the camera cannot go again for another 20 min run in 8K. So the next test I am going to do is to put the camera in a fridge... Sample internal temp during over a period of 100 stills over about 30 mins, and if this ends up reducing 8K recording times at internal camera temps far below what the ambient room temperature was when we could record the first 20 min 8K in, then we know there is something fishy. Or... if after 1 hour an overheat warning is triggered in the fridge, with the camera just sat in the menus idle, then switched into 8K video mode, we know for certain the timer is an artificial limitation. The WIFI app is great for triggering the camera behind the fridge door 🙂 I am just making sure I don't f*** the camera.

One possible solution, will be to use a headless HDMI adapter and plug it into the HDMI port. These are basically a little plug that makes the camera think it has a HDMI recorder attached, then have a toggle in the @BTM_PixCDA TEK temp sensor monitor app that turns off the LCD image. It'll be interesting to see if this allows the camera to record internally for longer, but long term I see the only real fix being on Canon's side. It's up to them to step up to the table really.

Canon could have released a EOS R5 to compete against Sony A7R IV... for stills. It has the slightly soft 4K pixel binned mode like the Sony does, but in 10bit. That's not too shabby. Then, release an EOS R5 C, with all the video bells and whistles enabled and properly functional. For some strange reason they decided to go full beans with the 8K RAW and 4k 120p only to not let us use them! Marketing tick boxes only! Why bother engineering these features if they are not going to be usable by the intended customers? And how in any way would a mirrorless camera cannibalise sales of a $15,000 full frame C500?

Erm. How about they just give us what is advertised? And tell the truth.

The recording time charts don't take into account the time spent in live-view or menus. So at some point during the first 40 minutes of the day, you will be looking at a 5 min limit in 4HQ instead... and then 0 mins. Unless you work like this: Shoot a 30 min continuous shot. Power down camera for 2 hours. Come back and shoot another 30 mins. Power down for 2 hours. Nobody works like this.

Anyone spot the problem with the charts like this?

Absolutely. Nobody would care if it wasn't for the fact that an incredible camera has been kneecapped.

Very much agree, it's a new tactic from Canon based on usability rather than limiting specs. They've realised limiting specs hurt sales and brand reputation. So it's the same thing in different clothing now. Make the high-end features unusable. Apparently $4000 is not enough for Canon and they want the full $15,000 from you if you want high-end specs AND usability. And this is from a company who wants us all to rush out and buy $2K RF lenses for our crippled $4K bodies?! They can fuck off.

+1 I've already got rid of numerous accounts to save our eyeballs and sanity, but apparently even more need to go.

Basically the timer kicks in whenever 8K or 4K HQ is enabled in the menus and there is a live-view feed on the LCD or EVF (even if it is hidden behind a menu overlay). That's probably just how they implemented the cripple clock in firmware. When the screen is off it probably disables the mechanism they are using to calculate run times... either by error, or on purpose to satisfy Atomos. That the scorching hot ambient temps and black alloy casing absorbing so much external heat don't impact the timer, has to tell you something!

How can it "overheat" just switched on in the Wifi menu, but do 4K HQ via HDMI just fine? The processing loads surely are not higher in the Wifi menu compared to outputting 8K to the DIGIC X image processor, processing it, resampling it to 4K 10bit and sending it out to an HDMI recorder? So either the HDMI mode bypasses the cripple clock accidentally (a bug in the cripple hammer). or Atomos persuaded them to turn the timer off? Nothing else makes sense.

They permit long recording via HDMI for same reason they allow long recording in 4K pixel binned. It is probably pixel binned 4K via HDMI. Yet to test myself. Anyway, if the camera allows itself to get very warm from external recording, it would shut down. Ask yourself why it suddenly throws a wobbly when you insert an SD card? It is doing all that work for HDMI recording, casing mega hot to the touch in direct sun, then suddenly you ask of it to do a basic 4K H.265 compression that it does anyway in the normal continuous 4K non-HQ mode and all overheating hell breaks lose?!? C'mon. The artificial nature of the timers is a huge part of what is going on. It could even be all patent related. 8K RAW for long recording times may result in RED knocking on door for more money. The majority of the processing is in the first 4 stages of the image processing pipeline and happen regardless of whether you use an external recorder or not. Pipeline is probably something like this: Sensor rolling shutter Onboard sensor A/D conversion Debayering of 8K RAW sensor data / Downsampling to 4K Conversion from RGB to YUV Compression to H.264 or H.265 So just switching off the dedicated hardware H.265 encoder to 4K does not magically reduce heat. That encoder is doing 4K 10bit 422 in the non-HQ mode as well. So logically deduce what is going on in the different modes... With HDMI recorder: 8K sensor readout (or maybe not)... debayer... resampling... conversion to YUV... RUNS FOR HOURS! Without HDMI recorder: Exact same process. But add the H.265 compression that happens in the non-heat restricted mode anyway... Suddenly, camera is screaming OVERHEATING!!? C'mon.

So to clarify... Are you sure HDMI recording is actually 4K HQ oversampled from 8K sensor readout? Or is it pixel binned? Please do the frame side by side quality check. HDMI recording only goes for long if you deactivate live-view? Or does it go for long runs with the mirrored display too? HDMI goes for long without a card in the slot? What about with one? Please try to use an SD card for testing, as this shuts people up from claiming it's the hot CFexpress that's the issue.

Shutting down when the card door is opened is long standing Canon camera practice. The 5D2 does it. The 5D3 does it. It is to prevent data loss, if you pull card out of slot whilst accidentally recording to it without noticing. Horshack: "Also, the fact the camera allows near-unlimited recording over HDMI argues against intentional crippling since external recorders are common for professional video use" If anything this proves that the camera sensor isn't the main heat source, that it can go for ages, and that the processor can downsample the 8K sensor readout to 4K HQ on the fly, and put it out over HDMI for hours.

It doesn't need it. You should just be able to turn it off and the extremely thin slice of silicon & metal will cool down by itself from whatever it couldn't handle... Supposably 90C... To ambient temp or a comfortable for any CPU 45C... In less than a minute, if not even quicker, we're talking a few seconds to go from 90 to 60. Anyone have a PC? Open the CPU temp monitor and see how fast the temps fall once you stop rendering video in an NLE. And they fall even faster if you pull the power completely.

Ah the good old EOSHD is not legitimate campaign. I wouldn't be surprised if 90% of it was bot generated. DPReview still up their tricks hey? Yeah, probably a reason they haven't run a clickbait article based on our findings yet with a tiny little 'Source' link at the bottom.

Great good for you! Do us a favour and read what I find when I use it. I feel you have your blinkers on. We don't all just turn the camera on for a few seconds at a time. Do us a favour and realise that the camera overheats in the menus or just doing stills in live-view. Overheats as far as a complete shut down once you go over into video mode. But yes... Video compression blah blah blah, data intensive blah blah.