Personal View site logo
Make sure to join PV on Telegram or Facebook! Perfect to keep up with community on your smartphone.
Please, support PV!
It allows to keep PV going, with more focus towards AI, but keeping be one of the few truly independent places.
Canon EOS R5 - Overheating video failure
  • 250 Replies sorted by
  • So, according to your post, today only a top player in the smartphone world would be able to design decent electronics by bringing the level of miniaturization of a smartphone into the electronics of a camera.

    Only Samsung comes to mind...

  • @LongJohnSilver

    So, according to your post, today only a top player in the smartphone world would be able to design decent electronics by bringing the level of miniaturization of a smartphone into the electronics of a camera.

    Not only top player, must have foundry, but he also must have big amount of engineers.

    Things that we had in digital cameras market are:

    1. Secret cartel between Samsung and Sony that killed both big Samsung sensors and Samsung camera LSI (best at the moment, that Sony matched only 5 years after). In return Samsung got access to most Sony mobile sensors technology. Another consequence is that Sony hiked prices of large sensors, their FF sensor is now around 4x of cost it will be if Samsung had been around. If not this cartel we'll have $599 new mirroless FF cameras with 48Mp sensors (without build in EVF).
    2. Destruction of most LSI engineers teams at Panasonic, Sony and other companies, some remains still exist but usually they only help some third party teams. Top management got huge bonuses as it allowed to hike numbers for one year by slashing expenses (and selling factories).
    3. Inability to agree and design complex modern LSI for all players, ala Qualcomm 865. Camera LSI must be even more complex. Modern camera require 8-16 powerful ARM cores, 2 low power ARM cores, NN accelerator, DSP, big video encoding and decoding block, multiple PCIe/NVMe controllers, support for DDR4L dual or four channel memory.
    4. Economy of scale is now playing against all players, as sales going down at staggering pace. In 2020 cameras sales can not compete to any electronic device, so bad it all had been.
  • Well if Samsung LSI and imaging know-how took a hit it would explain the Samsung S20 Ultra 108MP camera. None of the advanced image processing is present. The Xiaomi Mi 10 uses the same sensor, and the results are miles in advance of what Samsung achieved with their own sensor. Marketing has taken over the asylum. 100x "space zoom". Pure crap.

    I agree Vitaliy about short term thinking and the bonuses. Management have all incentives to make the worst decisions long-term and best ones for their bank accounts.

    Samsung's top management were so corrupt, jail came calling.

    The NX1 was just a collateral victim.

    Regarding economies of scale, it just needs smart thinking. Canon & Nikon have basically left the DSLR market for dead. Big opportunity for Pentax to provide high-end DSLR / optical viewfinder experience for those who don't want to go mirrorless. Yes it's a much smaller market, but Canon and Nikon have practically "gifted" their core business to Pentax. Make the DSLR in the classical sense, for artists, and leave the LSI rat race to the mirrorless camera market and smartphones.

    The camera market is now resigned to being a niche thing anyway.

    Canon, Samsung, Sony, Nikon top management will be richer and walk away from the burning wreckage.

    But there is still Sigma, Panasonic, Fujifilm, Pentax there to make the "thinking man's" camera in smaller quantities.

    Canon deserve to lose their reputation for reliability.

    Nikon deserve to lose their place in the sales chart.

    And as for Sony, I'm not sure what they deserve. Maybe some jailtime when the cartel is confirmed.

  • Well if Samsung LSI and imaging know-how took a hit it would explain the Samsung S20 Ultra 108MP camera. None of the advanced image processing is present. The Xiaomi Mi 10 uses the same sensor, and the results are miles in advance of what Samsung achieved with their own sensor. Marketing has taken over the asylum. 100x "space zoom". Pure crap.

    Samsung just have old time issues with software and image processing.

    They mostly originate from Korean management style, to be short - good software developers are incompative with such management and leave or never even come.

    Also it is best to never listen to forums and most big sites considering smartphones photo quality. It is too profitable piece and even big sites now live on selling their opinions.

    Samsung's top management were so corrupt,

    It is such wrong statement. As attack on Samsung top management was inspired by US after famous mostly fake story about exploding batteries that started right after Samsung broke agreement with Apple on announcement and release dates. Consequences had been severe, but not so bad as Huawei getting now.

    Samsung management made superb deal by forming cartel, as instead of anemic and barely alive cameras market they got hundreds of millions of sales for their new sensors.

    Regarding economies of scale, it just needs smart thinking. Canon & Nikon have basically left the DSLR market for dead. Big opportunity for Pentax to provide high-end DSLR / optical viewfinder experience for those who don't want to go mirrorless. Yes it's a much smaller market, but Canon and Nikon have practically "gifted" their core business to Pentax. Make the DSLR in the classical sense, for artists, and leave the LSI rat race to the mirrorless camera market and smartphones.

    Logic never touched this "analytics" :-)

    But there is still Sigma, Panasonic, Fujifilm, Pentax there to make the "thinking man's" camera in smaller quantities.

    LOL

    Nice list. Management of one company is preparing for closure similar to Olympus (sale and ripping). Management of another talked about shutting down cameras manufacturing in 2021 if it won't be better. Management of third planning cut of 2/3 of marketing and PR people (that had been already cut 3 times) and huge core engineers and developers cuts after December (will be silent). Managers of final one had pleasure of listening to lot of f--k words for making camera that top managers now do not know that to do with, as it brings only losses for each sale - best is to kill but they are afraid to loose face.

  • https://www.eoshd.com/8k/removing-internal-battery-resets-eos-r5-overheat-timer-are-canons-pants-now-completely-down/

    It is again lot of emotions. Keep it simpler, correct and use logic, Andrew.
    Instead of running in circles and shout strange things, you can make proper independent test, find legal firm who will see it as opportunity and make class action in US out of it. At least it is something real.

    As it seems like shutdown happens due to some thermals on LSI die. May be it had been little safe, may be. But trying to use FLIR here won't prove it (as you need to understand design of this part and limitations of this tool).

    Turning on is made by timers and not temperature, well, logical.

    Battery is used for realtime clock and it seems like it is powering dedicated memory where camera store things between short usage periods, of course timer is also stored in this memory. Camera can't use main battery here, as it can be removed and replaced at any moment.

  • The "emotions" are all part of the story. We are customers, not robots. If we put our trust into a company based on their reputation for good kit, and what's being sold to us (8K / 4K 120p pro camera) and it comes back very different to what's advertised you can't expect people not to have any feelings beyond a rational desire to maybe return it, or maybe to try and work-around it on a creative shoot.

    It is not just a rational process to think this through, and it is difficult emotionally to justify $4K for a professional 8K / 4K HQ camera this crippled and cynical.

    Difficult for people to be happy knowing what they know, or to believe Canon treats them like clowns. Just another number, to upsell to Cinema EOS product.

    Regarding battery for the realtime clock... Why does a genuine real-time temperature monitor even need to watch the time at all, let alone store it in non-volatile memory. A genuine real-time temperature monitor would read the actual CPU temp in real-time during recording, and various other critical temperatures, and give best estimates on where on the curve these are and how long you have left before recording ends. Shut the camera down based just on that info. No timers, no bullshit. And 1 hour recovery timer is the most fake of them all in my opinion.

    Indeed, what kind of CPU begins to thermal throttle at 64C anyway?

    As for proper independent tests... Already done and reported. Unless the Baidu guy is completely making stuff up and risking his opened-up $4000 camera a trip to the dustbin... It would be very odd.

    The API and Wifi app on my side by CDA-TEK also contains enough clues, when combined with the real-world behaviour of my camera.

    And your opinions are also valuable on the technical side.

    The PCB on top of the main processor and RAM, preventing thermal conduction with the alloy back...

    Absolutely terrible and probably on purpose.

    In my opinion:

    Canon had no intention in mitigating real heat for 2 hours of 8K recording.

    They had no intention in allowing firmware timer to allow more than 20 minutes.

    This is how tightly the thing is strangled.

    And how badly they want you to use a $15,000 camcorder instead.

  • If we put our trust into a company based on their reputation for good kit, and what's being sold to us (8K / 4K 120p pro camera) and it comes back very different to what's advertised you can't expect people not to have any feelings beyond a rational desire to maybe return it, or maybe to try and work-around it on a creative shoot.

    Such thing as "trust", "reputation" have nothing to do with any camera company, they are from different universe.

    It is not just a rational process to think this through, and it is difficult emotionally to justify $4K for a professional 8K / 4K HQ camera this crippled and cynical.

    Well, looking at your last expensive purchases I do not see where this drama comes from? :-) Seems pretty usual price for usual next fast buy fast sold camera.

    Regarding battery for the realtime clock... Why does a genuine real-time temperature monitor even need to watch the time at all, let alone store it in non-volatile memory. A genuine real-time temperature monitor would read the actual CPU temp in real-time during recording, and various other critical temperatures, and give best estimates on where on the curve these are and how long you have left before recording ends. Shut the camera down based just on that info. No timers, no bullshit. And 1 hour recovery timer is the most fake of them all in my opinion.

    Emotions and not reading that I wrote.

    In camera permanent settings and many other values are stored in special memory.
    I explained why you can't use DRAM. Normally it is also not usual NAND due to technical restrictions. Lenovo not long ago forgot about this basic understanding in their Thunderbolt controller and got lot of headache and tens of millions of expenses.

    Indeed, what kind of CPU begins to thermal throttle at 64C anyway?

    You do not know on die temperature, 64 is number you see on thermal image, on top of additional board. And even this can be inaccurate as you have no idea that is the emission coefficient of the board.

    The PCB on top of the main processor and RAM, preventing thermal conduction with the alloy back...

    You are rushing here. First you need to cut this addon board and see, as original strange idea could be to use board with lot of copper as kind of heat distributor. We don't know. Boards like GPU ones and MB i certain areas can be quite effective heat spreaders. Average motherboard PWM area dissipates around 60-70% of generated heat (despite nice heatspreaders on top that take other 30-40%).

    Canon had no intention in mitigating real heat for 2 hours of 8K recording. They had no intention in allowing firmware timer to allow more than 20 minutes. This is how tightly the thing is strangled. And how badly they want you to use a $15,000 camcorder instead.

    Reality can be 100x times simpler. Due to recent cuts they got few inexperienced engineers, after this due to COVID most ambassadors were in such state that they will prize and lick even shithole. Add to this that Canon top management never much cared about video side of things (specs they can like, but not real experience and real ork).

    And how badly they want you to use a $15,000 camcorder instead.

    From economical POV it didn't make sense now, as market of such camcorders literally stalled, situation is much worse than even in camera market that is in turn in worst state of any electronics stuff. Most of camcorders had been bough by small business or individual pros and core had been to use tax deductions, If you don't have weddings, clients and income it is hard to buy camcorder.

  • The EXIF temp gives us just a relative indication... Camera rises to 59C in the not-limited by heat mode (pixel binned 4K). Camera rises to 61C, max 64C in hot ambient temps, in the "limited by heat 4K HQ and 8K mode".

    So the relative indication shows that there is just 2C between a mode that can record forever until the battery dies and a mode that is so hot that it has to take a one hour break after 15 mins.

    The temp is so stable in live-view in fact, over the course of 1 hour, that it doesn't even go up 2 degrees... Maybe 1.

    And yet the difference in recording time in 4K HQ between the start of that 1 hour of stills shooting, and the end of it is 30 mins.

    It goes from 30 to zero.

    Apparently we now live through an era of the camera industry where an entirely new LSI and sensor is made purely for marketing purposes and not to be actually used by the people who fund the development and manufacturing... At tune of $4k each.

  • Don’t get me wrong. They should either remove this firmware limitation or someone at Magic Lantern should try. Counterpoint : it is not us funding their development by buying or not buying a $4000 luxury stills/video camera. Canon probably made more from printer technology and other IP last year and this year than all cameras combined. However, just consider, real business is not so simple to move and be nimble with such a large old beast of a quite old company. I’m sure there are people inside the organization trying to do all kinds of different things. What they have achieved here is in some ways a technical marvel, and yet some ways not. Magic lantern proves that with outside of the box thinking 1080p raw onboard was possible on a camera from the year 2008/2009. This doesn’t mean it was pragmatically possible to have that kind of feature set from the get go from Canon. It’s all the things you don’t see that prevent this stuff from happening the way cottage creatives might want it to happen. Let’s hope we can all survive current world events and get back into the privilege of being an artist, even if it means only shooting 8K raw a quarter hour at a time on a tiny $4000 USD camera.

  • Apparently we now live through an era of the camera industry where an entirely new LSI and sensor is made purely for marketing purposes and not to be actually used by the people who fund the development and manufacturing... At tune of $4k each.

    Read carefully that had been written.

    Andrew do you have hysteria? :-)

    Unlike the Mordor autocrat government or the management behind VW’s “defeat device” I do not think all people are that stupid. I do not think all people can be sold lazy excuses and be told to eat candy, to keep your blood sugar levels up and avoid metabolic disorders that end up in screaming agony then a coma in Omsk. The same people who think they can get away with poisoning the opposition, and glibly lying to the public, are the same people who will try to cover up the cynical market segmentation of the EOS R5 or the VW emissions scandal, by glibly lying to customers and discrediting the journalists who reveal the stories in the first place.

    https://www.eoshd.com/news/hey-squarespace-reconsider-your-sponsorship-of-tony-northrup-perhaps/

    What is the issue, Tony is businessmen, Andrew is also businessman, Canon management are even better business people. If money are at stake - nothing else matters, right?

    Btw, shallow journalism examples all other the place:

    It is not "Mordor autocrat government", this guy had beep proven scammer and stole lot of money but had been awarded softest sentence country ever saw. Guy collected big sponsors money, including from companies, while all other real opposition had zero chance to do this. Youtube channel had been promoted by largest marketing agencies who also work with government, and promoted by top official news. Long story short - government and popular "opposition" are all of the same - servants of the ruling class. Sometimes things happen if they decide that one of the other are no longer required or need to change location and lead new complain. Main official version of our opposition is that he had drug overdose. Regime actually did everything to not tell public actual version as it will mean that the guy will be done in public field. Plus he had upcoming trial in 3 days as he called old war veteran with obscene words (and if you get meaning - actually prized fascists, people he direct fought in fascist groups during recent Belarus events) and it could end badly both legally and publicly also.

    Management behind VW’s defeat device is actually all top VW management and it is not because they are some kind of criminals, it is because EU bureaucracy set such goals that are physically impossible to achieve, but same bureaucracy told them to keep producing cars. As soon as world started to have issues with certain kinds of oil used to make diesel (yes, it is complex thing requiring tons of info) this same bureaucrats started holy war under pretense of ecology and green energy.

  • Continuity of song and sad story

    And reusing idea that other people already made before Andrew.

    According to Horshack and Magic Lantern, most digital cameras have two types of internal memory:

    • RAM, which is volatile and fast memory. Volatile means it loses anything stored in it when power is removed.
    • NVRAM, which stands for non-volatile RAM. This can store data permanently when the battery is removed, but should not be written into regularly as it has a limited read/write lifespan.

    I tried to explain above about such values storage, but it seems like I failed to get to Andrew (NVRAM is general term, btw). So until he saw ready to use video with this text he can't understand.

    As Horshack and Magic Lantern understand it better than anyone as skilled software developers, cameras typically don’t update the information in NVRAM until they’re turned off in an orderly fashion. So if you adjust the aperture and ISO, the camera will not immediately remember these settings by writing them to NVRAM every time they change…. only to normal RAM. Then, these settings are lost if they are not written to permanent NVRAM memory upon an orderly shut down.

    As we will see below Canon has special advanced idiotic algorithm (as it can make real mess), but on many other cameras values are written right away (but can have RAM copy for general usage).

    The battery door on the EOS R5 has a small hidden switch which cleanly powers-off the camera in a controlled way, should the door is opened. This gives the camera prior warning that it may be about to lose power from a battery swap. This is also to prevent a user from suddenly removing power from the camera without it having the chance orderly save the video clip it is recording, or the state it is in, such as the most recent shooting settings like shutter speed, ISO and aperture.

    It is just mess and disaster.

    Using screw around battery and connector near it is "smart" idea. Use proper plastic or wooden part if you so want this mess.

    https://www.eoshd.com/news/canon-eos-r5-so-called-overheat-timer-defeated-by-a-single-screw-in-battery-door/

  • I have two questions, sorry, three.

    First, it is same famous EOHD magical screw that can be used to solve any problems of humanity?

    Second, where is fucking logic here? If you disable thermal shutdown by not allowing it to write and read NVRAM value it DOES NOT mean that LSI temperature is in manufacturer specified limits. It just mean that thermal shutdown routine gets wrong value as its input.

    Third, could you at least this time find same nice naked woman masturbating for us for this whole 50 minutes? As everyone is questioning for few years why you need even existing 15 minutes otherwise.

  • I think he is saying that if the thermal management were tied to the temperature inside the camera, the shooting limit would be reached once a certain temperature is reached. However, that is not what is happening here.

    Here, it seems to be a timer unrelated to internal temperature - as you can run all day long with this method, seemingly without regard to internal temperature inside camera.

    Doesn't mean thermal management is not required, but the way it is implemented is inelegant solution and certainly raises eyebrows.

  • @balldawg14

    I think he is saying that if the thermal management were tied to the temperature inside the camera, the shooting limit would be reached once a certain temperature is reached. However, that is not what is happening here.

    It seems like ML guys and Andrew confused lot of people. They don't have firmware and fully reversed thermal routines. It is clear that shutdown is directly related to temperature (some kind of accumulated counter can be also used as second option), but it seems like it is sensor on die that is not accessible to outside software.

    By effectively disabling proper values storage in NVRAM the routine that makes actual stopping and making some model unavailable can't see the proper values (that will be written by low level routines). Top routine can even get some kind of flags and not exact temperature.

    But it is not solution at all.

    If we had just timers Canon reaction from the start will be different.

    And we already had here photos of disassembled camera, where thermal design can be seen to be worse that everyone ever saw.

  • Is it possible that the disassembled camera of chinese guy would be a preproduction model and not be a fully and proper thermal protected camera?

  • @Knotbar

    If this had been the case you could see a lot of videos from Canon ambassadors and close reviewers that had attack on the Chinese guy.

  • Firmware Version 1.1.0 incorporates the following fixes and enhancements:

    1. Useful messaging is now displayed when [HDMI display: Camera+External monitor] and [Overheat control: on] settings are enabled.
    2. When using certain RF lenses for movie shooting, the in-lens image stabilization mechanism has been improved.Fixes a phenomenon in which the "Slow Synchro" setting screen is not accurately displayed, when the language is set to English.
    3. Fixes a typo displayed on the communication setting screen, when the language is set to Korean.
    4. Connectivity during FTP transmission has been improved.
    5. Fixes a phenomenon, in which the card access time may take longer, when using certain CFexpress cards.
    6. Temperature detection and shooting time control in video shooting have been improved. In addition, the total shooting time when the short-time recording and power-on/off are performed repeatedly at room temperature is improved.
    7. The phenomenon in which the movie recording time available is not correctly displayed when the Date/Time/Zone is not set has been corrected.

    The update that’s most interesting is the change in temperature detection. The camera will now check the three sensors more often, to give a more accurate real-time reading. There is some hope that there will be reduced cooldown times and longer record times because of this. However, Canon has not changed the temperature in which the camera will give you an overheating warning.

    The 29:59 limit on record time has not been changed. Longer recording periods didn’t see a significant improvement in record times or recovery times. Though there was some improvement.

    In 4K120P, the Canon EOS R5 allows for 7 minutes and 30 seconds of record time. The new firmware does not increase that time. However, when shooting short bursts. With the original firmware, Gordon was able to record 22 30 second clips with 30 seconds of recovery time between each clip before the camera overheated. With the new firmware, that number increased to 96.

    https://www.canon.co.uk/support/consumer_products/products/cameras/digital_slr/eos-r5.html?type=firmware

    So, I had been right about multiple sensors and that Andrew measured wrong one.

  • According to Canon, the EOS R5 uses three temp sensors—two internal, and one external; Firmware 1.1 forces the camera to check these sensors more frequently, and puts more weight on the ambient temperature sensor to improve cool down times under ideal conditions. This may not have a huge impact on single long clips, but it should help with both recovery time and with shooting multiple shorter clips (which is how most people shoot anyway).

    Again, I had been right and Canon did all they could (for now).

    To be more precise - they have issue with thermals of one LSI part (one of the two on die sensors), in EXIF they record external sensor.

    And to Andrew - it is time to stop all all this clickbait induced emotional suicides.

  • Andrew is in the nervous breakdown mode

    Who the fuck do these social media salesmen think they are? Doling out Canon’s PR line like Marie Antoinette – “Let them eat cake!”

    Marie Antoinette never told this, it is legend.

    If you wait 1 hour with the camera you just paid 4 grand for – Marie Canon will throw you 15 minutes of further run-time instead of 10 minutes, but only if you don’t do any stills first!

    English and physics are classes that Andrew missed in school. As Canon guys and I before them even WROTE to him (see post above) about things that happen with camera.

    The firmware file encryption has been changed to make it extra difficult for Magic Lantern to take a peek inside. So unless I can confirm that it’s possible to roll back to the old firmware if necessary, I won’t be touching it with a barge poll.

    Are Canon obliged now to share privately keys and information to ML members (because as years ago I asked few questions of origins of few encryption keys ML leaders went into very unkind mode and one year after this one their ex team member sent me info that Canon provides some information to very few key people in ML)? It is also really strange to see anything similar and very probable that Andrew again did not get something right here..