Improving sound quality by optimizing micro SD storage performance for free
Jun 20, 2017 at 9:02 PM Thread Starter Post #1 of 31

WindowsX

Member of the Trade: Fidelizer Audio
Joined
Apr 27, 2007
Posts
1,962
Likes
364
I'd like to share simple tips to improve micro SD for better sound quality, more reliable performance and save your DAP and micro SD's lifespan. Sounds too good to be true? Let's find out. :)

1. Format micro SD with exFAT filesystem with 128kb cluster size

upload_2017-6-21_7-48-31.png


2. Copy your music to micro SD card

3. Use DiskFresh to refresh data writing from physical drive directly without partition mapping

upload_2017-6-21_7-51-14.png


They're all free and and you can notice that device will becoming more stable with less playback glitches and also sounds great. Share this tips to your audiophile friends if it works for you. :)

Regards,
Keetakawee
 
Jun 21, 2017 at 2:30 AM Post #2 of 31
Sounds interesting.
 
Jun 21, 2017 at 4:24 PM Post #3 of 31
Sounds interesting.

Some people asked me about unstability issue with device so I recommended formatting his 200GB micro SD card with exFAT 128kb and that solved it. He also told me it sounds better so I figured I might as well sharing these tips. :)

Regards,
Keetakawee
 
Jan 5, 2021 at 10:35 AM Post #5 of 31
Thanks for the tip!

Does memory card speed make a difference? I also notice you use Quick Format (checked) rather than full format (uncheck) - wouldn't that, in theory, improve performance as well because full format checks for bad sectors of the disk?
 
Jan 5, 2021 at 11:10 AM Post #6 of 31
Thanks for the tip!

Does memory card speed make a difference? I also notice you use Quick Format (checked) rather than full format (uncheck) - wouldn't that, in theory, improve performance as well because full format checks for bad sectors of the disk?
Speed is seldom a problem with respect to music playback on most modern systems. It is mainly about access noise which can pollute the common lines, especially ground planes. Maybe latency/jitter to an extent but buffers can minimize these effects too. In fact the high speed ones can employ throttling and bursts which can help achieve rapid speeds but might increase other problems. There's also differences between hardware - SLC, MLC and TLC (coming soon - QLC). SLC is known to be the least complex if them all so possibly easiest to control noise from.
 
May 22, 2021 at 3:02 PM Post #8 of 31
Interesting. But why exactly 128 ? What is with the other options?
Larger allocation unit size generally speeds up the read speeds (especially for our case where music is a much larger file than the cluster size). There's also the aspect of the buffer size and transfer frame size (if it sends in frames that is) of the sd controller being in good sync with the card storage arrangment.

Do note that matching speed requirements is generally a trivial task but the deal here is to do it in a way where the overall access noise is least. I think op either meant 128kb or above, or if 128kb is the sweetspot. These things are anyway system specific (sd card and controller standards look very complicated), so it's worthy for you to try yourself and stick to what you find best.
 
Last edited:
May 22, 2021 at 3:35 PM Post #9 of 31
So the theory basicly is: The larger the cluster, the lower the noise, because of less accesses are made. Which probably means less electricity which means less disturbance. I guess that makes kinda sense.

I just recognized that the OP only had a 128MB card. So I guess 128 KiB was the maximum because the card was so old or something. Lol.
 
May 22, 2021 at 7:32 PM Post #10 of 31
There's a tiny difference. Larger cluster size means each access will have more noise but the total frequency of access will be low. Small cluster size means each access will have lower noise but the total frequency of access will be high. The deal would be to find a sweetspot which matches well with the whole system.

The noise levels mentioned is a vague rule of thumb as the buffer size, bus width, protocol etc will actually influence a non linear pattern (that is below a certain cluster size your noise per each access may not reduce as you reduce the cluster size).
 
May 22, 2021 at 10:38 PM Post #11 of 31
Last edited:
May 22, 2021 at 11:38 PM Post #12 of 31
Thanks for the links. The file structure, the playback library/code/instructions all make a difference when it comes to pc playback and I would assume it is quite similar when it comes to sd cards. One of my biggest confusions with sd cards though is that every single manufacturer seems to do their own variant of it, even within different batches of the same card. It's a miracle that a sd card from one company even works with another. Also there are multiple ways to access a sd card. There's different variations of sdio (choice between 1 bit serial and 4 bit parallel I think) and other protocols and as a standard also support SPI access (this is what I have seen used in most of the diy projects). Wonder what type of access happens in these daps!
 
Last edited:
May 22, 2021 at 11:44 PM Post #13 of 31
Thanks for the links. The file structure, the playback library/code/instructions all make a difference when it comes to pc playback and I would assume it is quite similar when it comes to sd cards. One of my biggest confusions with sd cards though is that every single manufacturer seems to do their own variant of controller etc even within different batches of the same card. It's a miracle that a sd card from one company even works with a other. Also there are multiple ways to access a sd card. There's different variations of sdio and other protocols but they also support SPI access. Wonder what type of access happens in these daps!
While fascinating, I failed my own double blind test. Even though there was some sort of placebo effect with the sighted test.

Perhaps measurable but not audible.

That and some players may be more or less sensitive to these noises. I read that Sony's class D amps are more succeptible to this noise while normal dacs have some sort of noise rejection built in.
 
May 22, 2021 at 11:57 PM Post #14 of 31
Yep. A lot of things depend on the whole system and the interactions. This is a hobby, and most of these changes are free. So it kinda becomes YMMV. The lack of change may or may not be because of perfect isolation (so many factors). Most noise rejection techniques will have their own tradeoffs
 
Last edited:
May 23, 2021 at 1:29 AM Post #15 of 31
Reminds me of the mSD card sound quality debate in a certain Sony forum. Ie Sandisk Extreme pro sounding the best and Lexar sounding the worst. How Sony daps have a special way of formatting that sounds better than using a standard windows format or using the formatter from the SD foundation.

https://www.head-fi.org/threads/new-sony-nw-zx500.914486/page-320#post-15850548
https://www.head-fi.org/threads/new-sony-nw-zx500.914486/page-332#post-15860873

Thanks. This is actually what I've searched for because I have to choose between 1 Terabyte Integral, Sandisk Extreme and Extreme Pro. Now I see why the Extreme Pro is like 200 bucks more expensive. lol.



I remember Sony wanted to release a audiophile SD Card for "premium sound": https://www.whathifi.com/news/sony-claims-high-end-sd-card-offers-premium-sound

The explanation how this works basicly is what we are talking about here.
 

Users who are viewing this thread

Back
Top