2

We are testing a video playing system that uses SD cards as the means of storage. The test is being conducted on 6 systems that are playing exactly the same content, all the players are the same, the location of the players is free from any type of interference, and the players are not being disturbed or changed.

We are seeing about 60% failure in the SD cards after about 48 hours of run time. Each player is playing a 1 minute long video and it loops over and over 24 hours a day. When the test initially starts all content runs as expected. Around the 48 hour mark the first 200 frames in the video will begin to develop pixelation or freezing issues.

The player has no writing ability however those first frames develop this issue. In the pixelated videos the content will continue to play, when the frames change to a 'new scene' the pixelation stops, you can take the card out and put it in another player and the same pixelation is present, so it is damaging the content on the card.

The question is how does this happen? According to the research we have done there should almost no playback issues with these cards, there are a limited number of write's that a card can withstand but I have not been able to find anything about a number of read's...

Any thoughts on what may be happening

8
  • How big is that 1 minute video, and how big is that card? You can copy the damaged file back to your laptop and compare it bit by bit with the original. That should make it clear if the file is changed. You played this file 6000 times in two days. In a year that will be 1.000.000 times. What does the manufacturer say about this?
    – SPRBRN
    Nov 12, 2015 at 14:04
  • Thanks for the reply... the video is about 20mb and the card is a 4 gig card. You know thats a good point... I have not tried to copy the file back to a computer to play it.. And yes we did the math as well.... we are running a new test right now where we stitched ten of the videos together to make a 10 minute long video rather then the 1 minute long video. The manufacturer rating is the question, I have not been able to find any information about a limitation on the number of reads, only writes. According to what i've found the reads should almost be unlimited?
    – Monergy
    Nov 12, 2015 at 14:10
  • Link to what you've found. This is information that makes your question more valuable, and it shows that you've done research. If you have more links or info about what you tried, let us know. So you could make that video 200 minutes long, and that should prolong the life of the card. It doesn't help find the solution to this question though. Try a USB stick if possible and see what happens then.
    – SPRBRN
    Nov 12, 2015 at 14:29
  • Thoughts 1) Really bad quality card 2) Heat 3) Filesystem writes.
    – qasdfdsaq
    Nov 12, 2015 at 14:41
  • searchsolidstatestorage.techtarget.com/podcast/… photo.stackexchange.com/questions/40489/… en.wikipedia.org/wiki/Flash_memory#Memory_wear Everything seams to revolve around the writes, nothing really indicates a limited number of reads... and based on some of these write stats we should be able to get many more reads then 6k...
    – Monergy
    Nov 12, 2015 at 14:41

1 Answer 1

4

This is not a fix, but I needed the space to give a sufficiently detailed explanation, please go easy on the down votes ;-)

I cannot give you any technical details on why this may happen, only to confirm that it does & you are not seeing anything truly unusual, though you do appear to be getting a very high initial fail-rate.

I've been doing a very similar task [playing adverts on in-store screens] for over 10 years now, with an install base of approx 2000 player/screens.
The workflow would appear to be the same - write once, read many.
In the early days, these were compact flash-based, which suffered very few card failures [though the screens themselves were a different story].
In the past 5 years, all have been SD-card based - with a phenomenal early fail-rate, giving precisely the symptoms you describe - jerky, halting video playback.

From our initial batch of SD cards we randomly selected a couple of hundred to send back to the factory in China. [I was not privy to that conversation so don't know the technical details] From that sample, it was estimated that approximately 15% had manufacturing defects sufficient to display slow read speeds after only a few thousand reads. The rest were deemed 'at spec'.

At worst, the failing cards will actually eventually crash the screens, requiring a power-cycle to recover. The 'bad but not terrible' cards will consistently give halting playback of varying severity, but not crash the players. If you watch the same video cycling on one of these, you will see the read halts are consistently at the same points in each cycle.
Empirically, they are not data-load dependant; it is just as likely to happen on a 10MB video as a 120MB video of the same length.

Our quick test on these, when we come to update the content seasonally, is just
'Format. If it works it goes back in the machine. If it fails it goes in the bin'
- very rough & ready but the customer will not pay to swap out cards unless they have totally failed.

The end 'result' is that after about 5 years, we are down to about 75% of the original card batch, all of which now just seem to keep working. Some still halt a little, but not enough to upset the customer.

My only, rather limp, conclusion is a pretty obvious "some keep going, some fail early"

Late note: If I were to start over at this task, I would use players that contain their own internal flash-based storage, to which you copy the video to be displayed just once, from a known-good SD card. We have a few hundred of these in use, & whilst the ones we have are horribly 'cheap & cheerful' the one thing they have in common is they don't suffer from playback fails.

4
  • 1
    On the contrary, this seems to be a great answer that covers what the OP needs to know over what the OP asked for. And yeah the batchtub curve and the SD cards are what should be looked at.
    – Journeyman Geek
    Nov 13, 2015 at 9:12
  • Thanks @JourneymanGeek This is the long-form answer which explains my rather cavalier attitude to the many 'my SD card/USB flash won't x,y,z' questions. If it won't format, bin it - not worth the effort once they start to fail.
    – Tetsujin
    Nov 13, 2015 at 9:21
  • Thank you Tetsujin that was a great answer.. I guess the next progression to the question is do you have a process other then 'crossing your fingers' that you send the card though to determine their stability? You mentioned you sent a batch back to china where 15% had manufacturing faults with them.. how did they determine that? We are sourcing these direct from China as well, the obvious advise most would give is to get a name brand card, that's difficult at the factory level. Have you found a way to 'qualify' good quality from bad?
    – Monergy
    Nov 13, 2015 at 12:58
  • 1
    It's all done above my pay-grade I'm afraid. These cards were sourced direct from the manufacturer, the cheapest possible; presumably with a correspondingly low QA. I imagine 'respectable branded' cards will suffer less, possibly only because they have been more thoroughly tested. I always get the impression you're buying the same thing with different degrees of 'pass' in the initial testing.
    – Tetsujin
    Nov 13, 2015 at 13:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .