snapsy wrote:
It was worth a shot at least to check the EXIF temps, just in case Canon changed the location of the sensor used for the EXIF temp vs previous bodies. He seems to reach hasty conclusions based on observations without much technical knowledge or thought into the implications of those observations. I appreciate the fact he performs the experiments but he should exercise a bit more judgement and engineering thought before rattling off the headline and conclusion.
Well, when you already think you have the conclusion, you will find the factors to make the case, however inaccurate.
InnomnateViem wrote:
a good experiment would be to run the camera thru an application like APT. Apt writes the sensor temperature data as part of the image file name. one could write a script to set the exposure, ISO, aperture, WB, and record the time intervals it takes to run the script. one would be able to auto-document the temp ∆s for the duration of the script frame by frame real time. you could actually see if the Peltier Effect cools the camera down. you could even do long exposure testing. that would really heat things up.
just a thought.
You can achieve the same by using exiftool to rename the captured files based on the temperature. Here's a quick command line I just constructed to demonstrate - it renames the filename so that the temperature is prepended to the existing filename:
snapsy wrote:
Great experiment, thanks. Btw, you can also add internal temperature samples to your collection by taking a stills photos (before/after taking a video) and using exiftool to extract the temp. For example, this command line will print the temp of every photo in the mypics image. They have to be OOC images (so that the full EXIF is intact):
Any chance you can try the battery pull experiment I described in a link above? The results of your experiment imply it probably wont make a difference since the camera responded well to freezer cooling, but just in case there's some element to the camera keeping track of total recording time as part of its thermal algorithm. Pulling the battery while recording might corrupt the card so you might want to format it after putting the battery back in....Show more →
Btw, you can ignore the battery-pull experiment - I forgot that Canon bodies have a battery-door sensor that automatically does an orderly shutdown when the battery door is opened. The only away around it would be to find and hold that sensor in while the door is open when recording video - that way you can pull the battery without the orderly shutdown.
The CPU temp isn't useful, but the ambient (or close to it) temps are more useful to the end user. EXIF data is for the user to know when they took the photo, where they took (for GPS enabled bodies), temperature when it was taken, distance to subject, camera settings, etc.
in long exposure photography it is imperative that one knows the censor temp to mitigate image noise and dark current. the temp is part of the exif but you can't see it directly. it has to be extracted from the exif data.
this is basic common knowledge in the AP community.
I feel like maybe some people took some findings of mine to conflict with others.
Here is what my test showed:
The R5 is capable of regaining its entire record time after spending 25 minutes in my freezer with the battery and card ejected, body cap on.
The highest external temperatures measured were not even matching average body temperature.
The CFExpress card was well within it’s safe operating temperature, it was in fact 30f+ away from reaching the upper limits of the safe operating temp when the camera shut off.
Here is what my test proved:
Given sufficiently low internal temperatures, the R5 is capable of recording its full duration before “overheating”.
I also tend to think that rather than a software timer, the camera resets to a low temperature that must be reached in order to give back full capabilities (this is all my speculation). And by rapidly cooling the camera, it is possible to reset to full record time in much shorter than the two hours required at room temp.
I guess the next test would be to try and record from within the freezer but I believe that would be outside the operational safety limits of the camera. I could try the fridge tho...
snapsy, all i'm saying is there are tools/applications that already exist that can be used to test the camera automated in real time and record the test data.
Jesse Evans wrote:
I feel like maybe some people took some findings of mine to conflict with others.
Here is what my test showed:
The R5 is capable of regaining its entire record time after spending 25 minutes in my freezer with the battery and card ejected, body cap on.
The highest external temperatures measured were not even matching average body temperature.
The CFExpress card was well within it’s safe operating temperature, it was in fact 30f+ away from reaching the upper limits of the safe operating temp when the camera shut off.
Here is what my test proved:
Given sufficiently low internal temperatures, the R5 is capable of recording its full duration before “overheating”.
I also tend to think that rather than a software timer, the camera resets to a low temperature that must be reached in order to give back full capabilities (this is all my speculation). And by rapidly cooling the camera, it is possible to reset to full record time in much shorter than the two hours required at room temp.
I guess the next test would be to try and record from within the freezer but I believe that would be outside the operational safety limits of the camera. I could try the fridge tho......Show more →
That's a fair assessment. I think your test was very useful to demonstrate the camera isn't using a temperature-indiscriminate, time-only based algorithm, which some have speculated and concluded to mean was an intentional-cripple timer. My assertion all along was that the behavior some are attributing to outright crippling is really just the rough contours of the thermal management algorithm. There is likely some hysteresis built into the algorithm, which can look like temperature-agnosticism but is likely the logic trying to project thermal ramps before reaching critical temps. That hysteresis likely has hard cutoffs in both directions (rising and failing temps).
gregfountain wrote:
Hmmmm, I really don't care about video, and I think most Canon shooters were more interested in the "stills" features like IBIS, resolution and the focusing system than we were/are about the video features, but these issues still give me pause. I'm glad I held off until it's all resolved or the camera simply is deemed reliable for stills or not.
Add to that the issues with their website and reports of stolen data, they sure seem quiet about it all.
lets lend some perspective.... on this website.... most would probably care about stills over video.
but in the blogging, youtube, event shooting and commercial work..... many people are very interested in the R5/R6 video capabilities.
just the fact there is a thread on R5/R6 previews and reviews.... full of video's. Did you watch any?
How many videos do you watch about other items, world locations ... maybe politics, covid-19 on the news like cnn.com?
there is more video work going on every day... we all consume it in various forms. and it is growing.
TeamSpeed wrote:
I could see the peltier plate producing temps 40-50 deg below ambient right on that back panel, and that in turn would have to have some sort of effect on the air inside at least. It won't transfer well to the main board however, due to how "insulated" that is with all the other boards around it.
I bet it reduces the temps about 10%, maybe 20%, but will that be enough, especially for the bulk and extra power needed?
fyi ... true peltier coolers are terribly inefficient-electrically speaking. plus there needs to be a thermal path already in place. they are basically a pump.
Andrew J wrote:
That's funny he did an R mount lens tear down saying it was the best built lens ever? I don't recall any mention of it breaking.
Sometimes we do teardowns when things break, other times when we're ready to make our in-house repair outlines. The latter happens when 1) Aaron and I have time and 2) the item is available in quantities that allow us to pull one out of stock for a couple of days. With the R5, which currently has a long waiting list of people wanting it as soon as we get a copy in stock, it's going to be a month probably. Depends on how quickly we get our orders filled.
I really don't have any opinion on all the discussion otherwise, I'm not a software person. I would add, though, that my limited experience with throwing heat sink paste and runs over hot spots, has never been the panacea that some people think it's going to be. I'm not trying to say that first gen engineering (this is sort of first gen, with IBIS, a new sensor, and a CPU that hasn't been in a mirrorless body before) is the best at all. But I am saying there's a lot more to heat flow in an enclosed, small body with several heat sources (sensors and IBIS get hot and weren't even considered in that limited teardown) than what's being considered here.
Jesse Evans wrote:
I'm happy to give that a shot. Currently I have my EOS R5 is on its second round of overheating testing in 8k.
Here is what I did for my initial test:
Sat the camera on my desk. A peak design design tripod clip raising it off the surface. Left the display swung out to the left. All power saving features enabled. Display refresh set to power saving / evf would be 60hz but it was inactive anyway.
I just left the camera on the desk recording.
I used a Klein Tools IR10 to take both k type temperature probe readings as well as infrared readings in order to ensure the readings were pretty accurate.
Room Details:
75f ambient temperature with an overhead fan running which would be blowing air on the the R5.
Video settings:
8k / 23.98p / ALL-I
Card:
Sony Tough G CFExpress 512GB
Observations:
Camera started the test at 71f through both k-type probe and ir readings with a 0.95 emissivity rating.
Overheat warning came on at 17:00 - Recorded external temperature of 98.4f.
Video recorded for exactly 20:04 - Recorded external temperature of 98.0f
CFExpress card temperature upon ejection from the camera - Label / Top side: 134f - Bottom side: 92f.
K-Type probe readings from CFExpress slot ~30s after ejecting the card: 128f
CFExpress card temperatures after testing CFExpress slot temps ~30s after initial test - Label / Top side: 91f, Bottom side: 79f.
I failed to act fast enough to get a reasonable temperature from inside of the battery chamber.
I took measurements from around the entire camera body to find the hottest part while recording, the next highest temp area is between the grip and the lens mount at around 93.4f, this matched the left side of the rear panel.
On the left side of the camera the maximum temperature observed was 91f.
After the test, with the battery out and I placed the camera in the freezer after placing it in a ziploc bag which I pressed all of the air out of.
After 25 minutes I took the camera out to see how much time I had to work with on the camera. All tested internal and external surfaces were cooled to a temperature of 53-54f depending on location. The timer read 15:00 which it did during the first test as well.
On this second test, the camera achieved another record time of 20:10. Longer than the first test. And this was with 20 minutes in the freezer.
So, I actually think I have just provided some evidence *against* the idea that Canon is artificially causing a timer to tick down once the camera has overheated.
Edit to add: CFExpress card was only 119f upon being pulled out on the second test, and the interior of the CFExpress slot was ~113f....Show more →
Very nice test. I have one idea for an additional test, if you get the time and patience....
Once you get the overheating warning, take out the CFExpress card (and SD card you have one in). Keep the memory card door open and get a fan to blown some air into the slot to cool it down. Maybe try half an hour for a start to see what happens. I guess you could remove the battery too, but not sure that would matter much.
Jesse Evans wrote:
I feel like maybe some people took some findings of mine to conflict with others.
Here is what my test showed:
The R5 is capable of regaining its entire record time after spending 25 minutes in my freezer with the battery and card ejected, body cap on.
The highest external temperatures measured were not even matching average body temperature.
The CFExpress card was well within it’s safe operating temperature, it was in fact 30f+ away from reaching the upper limits of the safe operating temp when the camera shut off.
Here is what my test proved:
Given sufficiently low internal temperatures, the R5 is capable of recording its full duration before “overheating”.
I also tend to think that rather than a software timer, the camera resets to a low temperature that must be reached in order to give back full capabilities (this is all my speculation). And by rapidly cooling the camera, it is possible to reset to full record time in much shorter than the two hours required at room temp.
I guess the next test would be to try and record from within the freezer but I believe that would be outside the operational safety limits of the camera. I could try the fridge tho......Show more →
I have another idea to get a temperature measurement closer to the action - indirectly from the actual image sensor. Looking at the collection of R5 blackframes I have (doing statistical analysis on for reverse-engineering the R5's raw noise reduction), I noticed the stdev of noise is materially higher for the blackframes when the EXIF-reported temp is higher. For example:
That's about a 4% increase in noise for a 13C increase in temp.
The EXIF temp seems to top out at around 46C for both stills and video shooting, which may be because it's measuring an area of the PCB far removed from the image sensor/DIGIC (info). Based on the video-recording thermal shutdowns it seems likely the sensor and DIGIC temps continue to rise, and that continued increase should be reflected in stdev of noise off the image sensor.
The next time you do a recording time test could you take blackframes before, during, and at/after thermal shutdown and PM me the resulting raw files? The exposure parameters should be ISO 100 1/4000 f/8 EFCS (Electronic First Curtain Shutter)....with the lens cap on. Blackframes during the recovery period would be helpful as well. The goal is to associate the sensor temp to the available recording time the camera reports. Thanks!
you may be better off doing that test with a lens and lens cap attached to the camera. it would be more like real use conditions. some of the heat is synced into the lens.
Ziffl3 wrote:
lets lend some perspective.... on this website.... most would probably care about stills over video.
but in the blogging, youtube, event shooting and commercial work..... many people are very interested in the R5/R6 video capabilities.
just the fact there is a thread on R5/R6 previews and reviews.... full of video's. Did you watch any?
How many videos do you watch about other items, world locations ... maybe politics, covid-19 on the news like cnn.com?
there is more video work going on every day... we all consume it in various forms. and it is growing.
perspective provide....
I totally get your point. I was just pointing out what I believe to be a fact. But it does pose another question for me; how many content consumers are watching videos in 8K or even 4K? Does youtube offer those formats? I still have a 1080P television as I've not felt the need to upgrade (at least not yet) because my TV provider doesn't offer those high rez options yet. So yeah, I do watch videos, but I'm not actually seeing the benefit because even on my MBP, I don't have the screen resolution.
I was having a back-and-forth conversation with Andrew on his EOSHD site forums. He kept throwing insults at me, I kept sticking to the topic. He eventually grew frustrated and banned me. Here are the posts in case he decides to take them down as well - I kept them for both posterity and to help further the discussion/experiments along in this thread.
He obviously doesn't understand how the HDMI output works. Many cameras will output full sensor output to HDMI because it doesn't require alot of horsepower to do that. The horsepower needed is when you snap off frame after frame of a video feed and construct movie files. Debayering the data, applying picture styles, etc whoa all such hard things to do (just numeric processing on the data).
It's harder to run CLOG and take 4K or 8K snapshot files, run them to RAM and construct those frames into a movie file. That is what causes the processor and ram chips to go hot.
This is why the Atomos Ninja is around, many cameras that only do 1080 recording can actually output HDMI at a higher quality because the sensor has the capability, and the Ninja then does all the work constructing the movie files.
Also, the Chinese desconstructor aimed an IR gun at the processor to register what temp? Did he have the camera running while it was all opened up and measured the temps, or did he run the camera for a while, then took it all apart and then measured the chip?
I get frustrated just reading his responses. He seems like a total tool.
