Gear / Technical Help > Recording Media

SD 7.0 and SD Express - New SD Card Spec and Interface

<< < (2/3) > >>

voltronic:
I would not put too much stock in those speed ratings.  They matter, but not quite as much as you might think.  Remember that the class and speed ratings are based on burst read / writes as you see in video / photo cameras and other typical file transfer applications.  Audio recording doesn't necessarily work that way.  From what I understand (there is very little out there on this topic), audio recorders write to the media in a sustained stream, more or less.  That's not something SD card manufacturers measure or publish, which is why you see audio recorder manufacturers having to test cards themselves.  In other words, it's not enough to trust the numbers on the card package.  You see some very high-speed rated cards failing to work in these devices, because the manufacturers aren't testing them in this way.  For instance, I have a very old Transcend 16G card that is quite "slow" by today's flash media standards, but handles 4 channels of 24/96 with zero hiccups.

Standalone audio recorders should have a memory buffer before writing to the flash media, but if that buffer memory is small and/or the card can't keep up, you run the risk of buffer underrun errors which cause dropouts in your recording.  This was the central theme of the drama with the Tascam DR-70D and it being "picky" about SD cards.  The theory a few of us had was that this particular recorder has a small memory buffer, which would make the sustained-write performance of the SD card even more important that usual.  There were members that complained that their very expensive new cards were failing (while my old & cheap Transcend ran like a clock) and even after what I laid out above was explained by the company rep, they were convinced that the speed rating on the card packaging was the be-all-end-all of how it would perform in an audio application.  Clearly, we know that isn't the case.

jerryfreak:
correct, which is why im only concerned about sustained sequential write speeds in both the tests i did and the website i cited.. its all that is relevant for our application

voltronic:

--- Quote from: jerryfreak on July 21, 2019, 06:23:16 AM ---correct, which is why im only concerned about sustained sequential write speeds in both the tests i did and the website i cited.. its all that is relevant for our application

--- End quote ---

Ah, I somehow missed the word "sequential" in your post.

jerryfreak:
My Mixpre 6 is giving me ‘sd card is slow’ error and stopping recording of eight 24/192 channels on several of the cards above so clearly there are other factors, unless my recorder is defective. Still researching the best way to qualify cards. Even though I will prob never record 8 tracks at that high bitrate it’s the best I can do for a ‘torture test’ on the device itself

jerryfreak:

--- Quote from: voltronic on July 20, 2019, 09:16:15 PM ---Remember that the class and speed ratings are based on burst read / writes as you see in video / photo cameras and other typical file transfer applications. 

--- End quote ---

actually many of the ratings such as class 4,6,10, U1, U3, V30. V60, V90 are actually based on sequential write speed, though im still puzzled how 4.5MB/sec fails to write to a card that is rated at 30 MB/sec and tests over 40 MB/sec.... at some point i have to blame the device implementation

One thing i came across yesterday, though i cant remember where, is a statement that cards are faster when empty because they somehow use free space for caching. I started testing that 128GB sandisk in the mixpre with the aforementioned 4.5MB/sec write, it was flawless until the card hit 60-70% full after 4.5 hours of testing. The Anker USB battery i was using dropped from 3 to 2 lights around the exact same time (out of 4), so im wondering if its kicking out less power and making the mixpre perform not as well. still investigating

even fresh (quick) formatting didnt fix the card's slow speed issues, i tried a slow low level format in my computer and will then do a format in the device itself before resuming testing

Navigation

[0] Message Index

[#] Next page

[*] Previous page