-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Corrupt video files in Sentry Mode
- Thread starter EVnut
- Start date
Interesting finds after 2 weeks with this problem, after all troubleshooting failed (same on 2019.8.3):
1. Only happens when car not moving.
2. Only happen when camera facing the sun.
Never at night or while driving!
1. Only happens when car not moving.
2. Only happen when camera facing the sun.
Never at night or while driving!
michidragon
Member
Gonna throw my hat in the ring here too. Model S, got 2019.8.3 today, same deal: Corruption (stripes) on (usually) right repeater only, left repeater sometimes shows 0 byte files. Also happens only when car isn't moving. Seems to be fine when driving.
Sun/dark doesn't seem to affect it but I probably haven't let it get dark enough yet.
... I bet this has something to do with idle power, or low power modes to the cameras or APE/ECU. That'd both make sense for it happening more often when the car isn't moving (in lower power mode) and in different lighting situations. (may have more to do with temperature than sun.)
If the cameras are being run just at the edge of their power minimums when the car is in park, this may be causing dropped data. (I believe they're LVDS over coax??)
In any case, looks like I'm not alone at all; thought I was the only one in the world with bum repeaters. Wonder if this is fixable in an update. (if its power minimums, it might be.)
It never happened to me while driving. When parked, most of the files saved in the daytime (80+ degree) are either corrupt or having pixelated pictures; however, none of the files saved in the nighttime has any problems, which leads me to think if it has anything to do w/ the temperature. The cameras may not be able to work properly in a hot environment?
Model X. Just got 8.3. update with sentry mode. I'm getting the same problem. Noticed the following:
1. Happens only on side cameras.
2. Happens on either side but only on one side at a time. This leads me to think that it can be related to lighting condition.
3. Mostly happens when car stopped. The video may start just fine and then slowly gets more and more pixelated. Once the car starts moving all clears up and the video becomes normal.
4. In sentry mode the bottom part is corrupted and the top part is fine. Sometime the broken bottom part gets better while the video is playing. (Starts bad and finishes with just fine picture).
5. Sometimes corrupted video gets stuttering once pixelation starts.
6. Noticed just one single occasion of the same problem when the car was moving.
7. Watched some sentry videos from other people - the quality seems to be much better then mine. Clear and almost no pixelation.
Wondering that maybe the jumping cars on the main display at stops is the result of the same problem...
If it's a software problem then I would think it would happen to everybody. Maybe hardware is still different. Can it be MCU related? Mine is MCU2...
1. Happens only on side cameras.
2. Happens on either side but only on one side at a time. This leads me to think that it can be related to lighting condition.
3. Mostly happens when car stopped. The video may start just fine and then slowly gets more and more pixelated. Once the car starts moving all clears up and the video becomes normal.
4. In sentry mode the bottom part is corrupted and the top part is fine. Sometime the broken bottom part gets better while the video is playing. (Starts bad and finishes with just fine picture).
5. Sometimes corrupted video gets stuttering once pixelation starts.
6. Noticed just one single occasion of the same problem when the car was moving.
7. Watched some sentry videos from other people - the quality seems to be much better then mine. Clear and almost no pixelation.
Wondering that maybe the jumping cars on the main display at stops is the result of the same problem...
If it's a software problem then I would think it would happen to everybody. Maybe hardware is still different. Can it be MCU related? Mine is MCU2...
michidragon
Member
I really think it's power related, and *therefore* temperature related. The cameras use LVDS signaling, which has a very narrow margin for operation. (depending on the type of LVDS, its like 1.2v, 1.8v or 2.5v. This isn't really a 'bad choice' or anything, but they need to be able to get signals long distances, in electronics terms, without a whole lot of power draw. LVDS - low voltage differential signaling - is a good candidate for that. It just doesn't have a whole lot of room for error.) - Temperatures affect resistance of things like CMOS sensors, and therefore would affect small voltage drops like this.
I bet that what's going on is, when the car is stopped, the system is put into a lower power mode to decrease vampire drain. This may be at *just* the margin of some cameras at a given temperature. That low power mode may have been fine when Tesla tested it in Fremont, but parts (even the same ones) have variation, and if they're too close to the limit where a voltage sag is causing the cameras to drop data, it'll show up in some situations and not others. (Think of it like how processors are 'binned'; some LVDS 1.2v cameras may operate fine at 1.12v, others may go all the way down to 1.08v, and some may start having problems at 1.16v.)
A lot of people are reporting that it only happens 'in the sun', and 'not at night', so this really feels like it's temperature and voltage-drop related.
It's just a guess, but it seems to make sense to me as a good number of people, with Models S X and 3 are experiencing identical symptoms under the same circumstances.
The good side of it is, if it is this, then it certainly can be fixed in software by likely upping the current or voltage to the cameras (though the vampire drain may go up when Sentry is operating.)
I bet that what's going on is, when the car is stopped, the system is put into a lower power mode to decrease vampire drain. This may be at *just* the margin of some cameras at a given temperature. That low power mode may have been fine when Tesla tested it in Fremont, but parts (even the same ones) have variation, and if they're too close to the limit where a voltage sag is causing the cameras to drop data, it'll show up in some situations and not others. (Think of it like how processors are 'binned'; some LVDS 1.2v cameras may operate fine at 1.12v, others may go all the way down to 1.08v, and some may start having problems at 1.16v.)
A lot of people are reporting that it only happens 'in the sun', and 'not at night', so this really feels like it's temperature and voltage-drop related.
It's just a guess, but it seems to make sense to me as a good number of people, with Models S X and 3 are experiencing identical symptoms under the same circumstances.
The good side of it is, if it is this, then it certainly can be fixed in software by likely upping the current or voltage to the cameras (though the vampire drain may go up when Sentry is operating.)
Kira
Member
Tried out Sentry for the first time today while parked in a very quiet provincial park lot. Just spent a good chunk of my evening wading through dozens and dozens of 1 minute long video clips of - for the most part - absolutely nothing. 
Finally found a batch of clips of a family pulling up alongside the car and peering inside, pointing & waving at the cameras, etc. However, like others on this thread, a significant amount of the left repeater clips are corrupt and cannot be played.
Was on 5.15 (just updated tonight to 8.4).
Finally found a batch of clips of a family pulling up alongside the car and peering inside, pointing & waving at the cameras, etc. However, like others on this thread, a significant amount of the left repeater clips are corrupt and cannot be played.Was on 5.15 (just updated tonight to 8.4).
Last edited:
I’ve been out of the loop on Tesla updates so maybe you can help clarify for me. The update with the front camera I had used my 2GB flash drive and worked great.
So recently there has been an update with two side cameras and then Sentry mode?
Yes, though I’m not sure if the side cameras and Sentry were at the same time or quickly in succession.
MacGreiner
5YJ
Super_Popular
Well-Known Member
Red5StandingBy
Member
Tried out Sentry for the first time today while parked in a very quiet provincial park lot. Just spent a good chunk of my evening wading through dozens and dozens of 1 minute long video clips of - for the most part - absolutely nothing.
Was on 5.15 (just updated tonight to 8.4).
Let us know of 8.4 resolves any of the corruption issues.
Kira
Member
Thanks for the nudge - I forgot to pop back in and say that 8.4 seems to have resolved the issues with the left repeater files that wouldn't open at all. Also much less of the green half-screen issue on the clips.
The only problem now is that I may be addicted to watching Sentry videos.
Model X owner here. 2019.8.3. Definitely having Left Repeater green bands and 0byte files. Very annoying. I use a Sandisk Ultra Fit USB3.1 Flash Drive 130MB/s. I live in Sydney, Australia. The funny thing is the inconsistent time stamps. The file names are coded with the correct local time. But the file created / modified time is 11 hours behind, which is the Greenwich Mean Time. Go figure.........
srs5694
Active Member
I took delivery of my Model 3 on Saturday, but I'm still waiting on paperwork from Tesla so I can register the car, so the most I've driven it is back and forth a few feet in my own driveway. In testing the dash-cam features, I had the same problem initially, but it seems to have cleared up. In a thread I posted on the subject, somebody speculated it might have to do with quirky lighting conditions. That's as good an explanation as any I've heard at this point.
Teslas use Linux as an OS on at least some of their computers, and Linux, like other Unix-like OSes, uses UTC internally, and as time stamps on its files, at least on native filesystems. The FAT filesystem used by Tesla for dash-cam file storage, OTOH, normally uses local time for its time stamps. There are options to either do the conversion or not do it in Linux when mounting a FAT filesystem. It sounds like Tesla is not doing the conversion and you're reading the disk on a computer that is treating the time stamp as local time.
Also and FWIW, I've noticed that whenever I pull the USB drive and run a filesystem check on it, it's reported as having been incorrectly mounted, even when I explicitly turn off recording in the Tesla. Once or twice, actual filesystem damage has been noted. In other words, turning off recording does not properly unmount the disk. Doing this repeatedly is almost certain to cause serious filesystem damage, which could lead to lost recordings. I therefore strongly advise doing a filesystem check on the disk whenever it's pulled and read on a regular computer. The "dosfsck" command in Linux does this; or you can do it in a GUI tool like GParted. In Windows, "CHKDSK" will do it, as will various GUI tools. In macOS, use "fsck_msdos" or a GUI tool like Disk Utility.
I see the same thing on 5.15 also. I don't think it is just facing the fun, but anything bright even it is just a reflection off another car. And I have seen it on dashcam while the car is moving... but again with something bright in view of the side repeater camera.
Similar threads
