SG9663DR update

This happened to me when I saw a lifted pickup...
I'm currently testing the h3ll out of a new 512GB card because a couple of weeks ago I happened to see one of the new mid-engine Corvettes on the road (in camo). When I got home to post the video it was missing. The camera was 'randomly' recording short files of 1:15 to 1:45 (red) duration then returning to 5:00 (green) like the camera was set for.

The time period from 6:35AM to 7:00AM was one continuous trip. Put the previously used 256GB card in the same camera and all is good. No abnormalities before or after using the 512GB card so I'm as certain as I can be that there's something 'wonky' with the 512 card.

Never got any indication from the camera that anything was wrong.


ShortFiles.jpg
 
I'm currently testing the h3ll out of a new 512GB card because a couple of weeks ago I happened to see one of the new mid-engine Corvettes on the road (in camo). When I got home to post the video it was missing. The camera was 'randomly' recording short files of 1:15 to 1:45 (red) duration then returning to 5:00 (green) like the camera was set for.

The time period from 6:35AM to 7:00AM was one continuous trip. Put the previously used 256GB card in the same camera and all is good. No abnormalities before or after using the 512GB card so I'm as certain as I can be that there's something 'wonky' with the 512 card.

Never got any indication from the camera that anything was wrong.


View attachment 47973
that normally indicates a card issue, did you notice any beeps at the time?
 
Would the central screen be able to show both camera streams at the same time, either as split-screen or picture-in-picture? This seems important to verify that both cameras are still recording.
yes

And would the system be able to only power one camera when in parking mode?
technically possible, unlikely to be something we would do though
 
@DT MI, which camera?
that normally indicates a card issue, did you notice any beeps at the time?
No doubt it's a card issue because the camera has been rock solid since day 1. Camera is SGX2Pro+ first put in service Sep. 24, '18 as an X2Pro and upgraded to a Pro+ on Nov. 7, '18. The card is a Samsung EVO Select 512GB ( https://www.amazon.com/gp/B07MKSGZM6 ).

At no time did I hear any beeps from the camera other than the one start up tone. The example posted above is my morning 'quite time' and there is no radio or music playing so if there were any tones from the camera I'm sure I'd have heard them.

A bit of background:

I used a 256GB Samsung EVO Select from Sep. '18 until May 13, '19 when I began using the 512GB card. I've checked all the video captured on the camera from Sep. '18 until May '19 and found nothing questionable.

Since May '19 the following has occurred:

...June 20 - there were 54 consecutive file pairs (A/B Front/Rear) that were 'shorted'.
...June 22 - one file pair
...July 6 - one file pair
...July 11 - two file pairs
...July 31 - 26 consecutive file pairs, of this group 5 pairs were corrupt and unreadable.
...Aug 1 - 7 file pairs (the example I posted above)

Other than the last 2 file pairs on Aug. 1 the shorted files are either 1:17 or 1:18 in length. In every case that I checked the elapsed time between last frame of the first pair to the first frame of the second is always 44 to 46 seconds.

On Aug 2 I went back to using the original 256GB Samsung EVO Select and no issues since.

I've checked the 512GB card with both h2tstw and Windows disk tools looking for bad sectors and it passed both. I also did a full format with the SD format program from SDA, again with no problems detected.

Given that no errors have been found on the card as well as the randomness and variances in the symptoms I'm reminded of this post by @reverend (https://dashcamtalk.com/forum/threa...00s-was-so-bad-in-dashcams.40054/#post-478964 ) since timing issues could be the cause.

The card is currently being 'bench tested' in a 'DC' just to see what might happen. It's been recording for about 50 hours now so some overwriting of older files should be happening. We'll see over the next couple of days what, if anything, may show up in that camera.
 
For several years now I've been running both a standard front facing dash cam along with a telephoto dash cam. Usually the telephoto works in concert with my normal dash cam as an "enhanced" auxiliary cam but on occasion one camera really captures what happens when the other one doesn't quite do the job. It's as if to some extent each camera works as a redundant back-up for the other. Since installing the telephoto, I've become rather addicted to having two cameras working together and such a set-up might have helped your situation.
What dashcam do you have with a telephoto lens? It'd be good for my highway patrol.
 
Last edited:
What do you think you'll see inside the car? A head facing down with a hoodie, shaded by the car's roof and pillars from outside lights, ie no face visible at all. Dumb idea.
Parking should record on all cameras.

Yeah, maybe I shouldn't be trying to shoehorn a security feature onto a dashcam, but I got the idea from some all-in-one products. The current cameras are designed for recording the exterior and are unsuitable for internal security purposes. The internal camera will need to illuminate the darkness, either by a motion-activated light or IR transmitters. The light may definitely get the person to look directly at the camera. I guess the only security feature of the DR model is that the main unit can be hidden or locked away. I wonder if anyone has tried to stick a wyze camera somewhere in the console but that would only be useful when near wifi.
 
The current cameras are designed for recording the exterior and are unsuitable for internal security purposes. The internal camera will need to illuminate the darkness, either by a motion-activated light or IR transmitters.
There is an IR version of the Viofo A129 which has IR emitters built in to the remote camera that records in IR, designed to be used facing into the car interior. I think it is feasible that SG could provide a colour + IR option for the DR remote cameras, as an alternative to the standard 2x colour remote cameras.
 
It may not be 'practical' but it's necessary. There have been many instances of card failure with no indication from or on the camera. The only way to absolutely insure that video is actually written to the card is to check the card. You don't want to wait until you need the video to find out it's not there.
I don't think this is acceptable because since the card can fail at any time, you would then need to check it every time. The firmware needs to have a way to indicate any card error. Or camera error. Basically, any error. Because that one time when you need the video is that one time you didn't check it.

Are the video being written to the card directly from the camera? Perhaps it should be written to a memory buffer so when it is written to the card, the two copies can be compared.

And it may be a good feature to have select videos be written to your phone wirelessly.
 
Last edited:
A bit of background:

I used a 256GB Samsung EVO Select from Sep. '18 until May 13, '19 when I began using the 512GB card. I've checked all the video captured on the camera from Sep. '18 until May '19 and found nothing questionable.

Since May '19 the following has occurred:

...June 20 - there were 54 consecutive file pairs (A/B Front/Rear) that were 'shorted'.
...June 22 - one file pair
...July 6 - one file pair
...July 11 - two file pairs
...July 31 - 26 consecutive file pairs, of this group 5 pairs were corrupt and unreadable.
...Aug 1 - 7 file pairs (the example I posted above)

Other than the last 2 file pairs on Aug. 1 the shorted files are either 1:17 or 1:18 in length. In every case that I checked the elapsed time between last frame of the first pair to the first frame of the second is always 44 to 46 seconds.

On Aug 2 I went back to using the original 256GB Samsung EVO Select and no issues since.

I've checked the 512GB card with both h2tstw and Windows disk tools looking for bad sectors and it passed both. I also did a full format with the SD format program from SDA, again with no problems detected.

Given that no errors have been found on the card as well as the randomness and variances in the symptoms I'm reminded of this post by @reverend (https://dashcamtalk.com/forum/threa...00s-was-so-bad-in-dashcams.40054/#post-478964 ) since timing issues could be the cause.

The card is currently being 'bench tested' in a 'DC' just to see what might happen. It's been recording for about 50 hours now so some overwriting of older files should be happening. We'll see over the next couple of days what, if anything, may show up in that camera.

Do you think that the problem is with the fact that you're using a Samsung EVO Select made by Samsung for Amazon, as opposed to using an EVO Plus original - tests show that the Samsung made for Amazon Select green version is both quicker and more expensive.

Or do you think the problem lies with using the larger capacity of 512 GB vs only 256 GB?

61gYHhe61lL._SL1500_.jpg

81EbtLEnNeL._SL1500_.jpg
 
Samsung has Endurance brand cards that have up to 5 years of warranty, but they are only up to 128GB in size. DR v2 should have multiple slots and a RAID 0 option...
 
I don't think this is acceptable because since the card can fail at any time, you would then need to check it every time. The firmware needs to have a way to indicate any card error. Or camera error. Basically, any error. Because that one time when you need the video is that one time you didn't check it....
In a perfect world your points are all valid, but there is no such thing as a perfect world.

In the real world the camera sends the data to the card with instructions to write it to memory and the card responds with a 'successful' status. If the data was not written the camera is ignorant of that.

Likewise a problem with the camera itself. If the error causes the camera to stop functioning how is it going to send a notification? The only solution is physically checking the card.

...Are the video being written to the card directly from the camera? Perhaps it should be written to a memory buffer so when it is written to the card, the two copies can be compared.
That would require cameras to have processors at least twice as fast as they currently are so that comparison of the previous segment could be done in real time while the current segment is being recorded. Likewise the internal memory buffers would have to be twice as large for the same reason.

While both of these could conceivably be done it would not be at the current price points. How much more would the average user be willing to pay for the convenience of not having to perform the common sense due diligence function of periodically checking the card?


...And it may be a good feature to have select videos be written to your phone wirelessly.

Which selected videos? 50%?... 10%?... user defined? Would also require the camera to have wireless communication capability (not free), and still faster processors to send the data - also not free.

Also, should the data sent to the phone be read back and compared to insure it got there correctly? All this will require still faster processors to get it done. There is no such thing as a free lunch or cheap, fast processors.

What if the phone is not functional? And there is still the data transfer rate which is generally inadequate for large volume data transfer.
 
Do you think that the problem is with the fact that you're using a Samsung EVO Select made by Samsung for Amazon, as opposed to using an EVO Plus original - tests show that the Samsung made for Amazon Select green version is both quicker and more expensive.

Or do you think the problem lies with using the larger capacity of 512 GB vs only 256 GB?
...
I have no way of knowing any of that - I could only speculate and I won't do that. Anything is possible including it might just be a random 'one off' issue with my specific card.
 
I think the camera SOC makers are just focusing on 1 memory card slot as you can get memory cards with plenty fast write speeds, but it would be nice if they would also focus on supporting even larger memory card sizes.
And i also think they will if dashcam makers keep pushing them and it is more easy than making support for 2 memory cards, cuz really the need is there for this market, for action cameras i am not so sure though the 4K resolution and high bitrates do eat up memory space fast, but in a action camera it is quick and easy to swap a card, and thats a luxury that are not there in dashcams.
 
In the real world the camera sends the data to the card with instructions to write it to memory and the card responds with a 'successful' status. If the data was not written the camera is ignorant of that.

Likewise a problem with the camera itself. If the error causes the camera to stop functioning how is it going to send a notification? The only solution is physically checking the card.

The central unit does need to periodically check how much space is left and clear out any videos that are not marked readonly, so there is activity outside of the cameras themselves. One of the central activities can be to check if the video files are ok. Encoded video files can have metadata stored in the beginning or end of the file and that metadata or lack of it can help determine, at a high level, if the original file was written correctly.

That would require cameras to have processors at least twice as fast as they currently are so that comparison of the previous segment could be done in real time while the current segment is being recorded. Likewise the internal memory buffers would have to be twice as large for the same reason.

While both of these could conceivably be done it would not be at the current price points. How much more would the average user be willing to pay for the convenience of not having to perform the common sense due diligence function of periodically checking the card?
Maybe I'm spoiled by my $20 wyze cameras that can write to the cloud and card simultaneously after the buffered video is checked for motion and human detection. And they are software upgradable with new features so the wyze's components are not being taxed to 100% of their ability. Your estimate of doubling this and that for new features would be true if the dashcam's components were being used at 100% their capacity or are using custom-made components that only satisfy a specific performance level. But it's also likely that the components are general purpose, were mass produced to work in multiple products and so more features or performance can be tweaked out of them.

Which selected videos? 50%?... 10%?... user defined? Would also require the camera to have wireless communication capability (not free), and still faster processors to send the data - also not free.

I was thinking only of videos that are being marked as readonly by the triggering of a user button or crash sensor.

What if the phone is not functional? And there is still the data transfer rate which is generally inadequate for large volume data transfer.

If only select videos are being sent to the phone, then the transfers can occur asynchronously.
 
Maybe I'm spoiled by my $20 wyze cameras that can write to the cloud and card simultaneously after the buffered video is checked for motion and human detection. And they are software upgradable with new features so the wyze's components are not being taxed to 100% of their ability. Your estimate of doubling this and that for new features would be true if the dashcam's components were being used at 100% their capacity or are using custom-made components that only satisfy a specific performance level. But it's also likely that the components are general purpose, were mass produced to work in multiple products and so more features or performance can be tweaked out of them.

You may be spoiled by your WyzeCam, "that can write to the cloud and card simultaneously after the buffered video is checked for motion and human detection", but the Wyzecam records at only 15 fps with much lower bit rates. It is not an apples to apples comparison.
 
...Maybe I'm spoiled by my $20 wyze cameras that can write to the cloud,,,
You may be spoiled by your WyzeCam, "that can write to the cloud and card simultaneously after the buffered video is checked for motion and human detection", but the Wyzecam records at only 15 fps with much lower bit rates. It is not an apples to apples comparison.
Plus the Wyze cam only transfers 12 seconds of video to the cloud and, if I remember correctly, will only do it a maximum of once every 5 minutes.
 
Plus the Wyze cam only transfers 12 seconds of video to the cloud and, if I remember correctly, will only do it a maximum of once every 5 minutes.

Yes, there's that too! Plus a few other factors as I understand.
 
Back
Top