[groovyd]

first of all you clearly have not done much embedded design.  all mcu contain dedicated peripheral support in hardware for busses like i2c/s and yes they are able to manage uninterrupted output by leveraging other technologies like hardware dma to the peripheral buffer which functions like a fifo on the mcu side and maintaining various dedicated clocks and busses to feed these peripherals.  if that fifo gets exhausted it means your firmware is completely wrong and needs work not just a little but a fundamentally unsound architecture.  further then the mcu-side hardware peripheral is the dac-side input buffering, with a full fifo and re-clocking of the data to the actual convertor usually using very precise jitter controlled clocks. Again if this were exhausted you will have and hear serious issues that would send any developer back to the drawing board.  If you are that close to your deadlines to be pushed over the edge by some unrelated user feature change then you don't really have a product.  All of these products run real time os internally (in this case embedded linux) which guarantees hard scheduling deadlines and priority based interrupt response.  i have worked on embedded system design my entire life and on safety of life systems and codecs and such, none of this is magic it is born through proven design.
 
if the fr graph is identical which i suspect to be true and is the reason i bring this up then all of these concerns about sound changes are a waste of time to discuss.  regarding better documenting the changes it isn't black and white like you are describing, there are shades of grey where they could spend an extra hour on every release just summarizing the changes in terms anyone can understand and as they truly apply to the details of the quality of the product, things like jitter as you mention if they are indeed related to a change or not then we should like to know that.  it isn't rocket scientist vs the dummy here it is passioned firmware developer reaching out to better explain changes to the passioned consumer who is indeed curious.  there is a way to communicate these changes between such people and they could do a much better job of it then 4 lines of features when we all know there were many many more internal changes involved.
 
i have worked aside these people, i know how they work.  they spend months every day changing tons of stuff and when it comes time to release the product no one ever takes the time to actually communicate the changes effectively.  management tends to try to keep the engineers quiet make believing their audience doesn't want to hear what they have done. management really has no clue in reality as to what the audience really wants and most engineers are in constant amazement as to how their management even got into the position they are in.

 

[Paul - iBasso]

  i need Paul's explanation about this. why they keep changing the sound? did anyone complained about the sound of 2.2?
i'm buying headphones and i refer to my dap's sound to decide.

TRANCE has stated it pretty well. We do not try to change the sound. Most manufactures go through this with FW changes. Very small changes to the coding has an effect and we constantly work towards having a FW that impacts the sound as little as possible. We do not shape the sound or use anything to purposely change it. We have our own SW team and are dedicated to get all of you the best functioning for a dap that we can. This means that every day they are working on the SW to resolve issues or to add features but it does take time. It is like some say, a cat and mouse game at times.

 

[windplr]

It's great that some people think the sound has changed (for better or worse), however I no longer take anyone else's experience with sound as anything except personal opinion.  There are just too many variables in the audio chain, each with the potential to alter what is processed by your brain. No two systems are the same, and even ears and your health can effect how you process the signal and "hear" the music.  Here are a few things that can change the sound:
 
1. The recording
2. The sample
3. DAC, rolloff, EQ, internal design and implementation
4. Amp, volumne, gain
5. cables
6. headphones
7. your hearing, the size and shape of your aural canal
 
There are many others.  No two systems are the same.  YMMV.  Just find what you like and go with it.
 
Cheers, Chris

 

[matinsoroudi]

Thanks for reply  Paul, i'll wait for another fw update then. 

 

[Paul - iBasso]

  You're not the only one around here prefer FW 2.2. I have DX90 for 6 months and enjoy the sound of every new FW released by Ibasso, except FW 2.2 -> 2.3. Sacrifice bass and full-body sound for better high and brighter sound is not my favorite choice.
 

Have you tried the slow roll-off function?

 

[AndrewH13]

^ The FR graph of each firmware would be identical. It would be pointless for ibasso to document software code changes that we don't even understand. You don't seem to think that software has an effect on jitter, well it does. The cpu can not send a completely uninterrupted I2S signal to the dac, it can only aim to send the most efficient I2S stream, firmware plays a part in this and influences jitter, every dap is the same in this regard. Some dap's use various methods to try and reduce this cpu generated jitter by using a separate cpld for I2S generation or an SRC, this can reduce the jitter, but doesn't eliminate it, again it is reliant on how well the cpld firmware has been programmed to work in conjunction with the cpu. Bottom line is that firmware has an effect on jitter.

If firmware alone can change sound, then its fair to say different Daps into a Dac can have even more effect? I believe I hear differences between the DX90, X5, and an iPhone into the Hugo. Many believe they all sound the same as they should all be bit perfect and that jitter is negligible these days.

 

[meat01]

The next thing you will see is people saying that their firmware needs to burn in for 200 hours, because the code needs to settle in.

 

[mosshorn]

^^^^ LMAO

 

[proedros]

  The next thing you will see is people saying that their firmware needs to burn in for 200 hours, because the code needs to settle in.

 
genius post lol

 

[AndrewH13]

Brilliant!

 

[Highsmith]

TRANCE has stated it pretty well. We do not try to change the sound. Most manufactures go through this with FW changes. Very small changes to the coding has an effect and we constantly work towards having a FW that impacts the sound as little as possible. We do not shape the sound or use anything to purposely change it. We have our own SW team and are dedicated to get all of you the best functioning for a dap that we can. This means that every day they are working on the SW to resolve issues or to add features but it does take time. It is like some say, a cat and mouse game at times.

I believe the answer to this dilemma lies in the design of the digital low-pass filters,
used in the D/A converters and sample rate converters.

Why fiters on ADC ?
Sampling without filtering will include ALL signals, from the baseband that you want
to keep, along with the out-of-band stuff you DON'T want, all the way out to infinity.
This folds down ( aliases ) to the baseband, producing alias distortion, which sounds
a lot like ring modulation, especially obvious on instruments like trumpets that have
lots of high frequency harmonics. That's why an antialiasing filter is required every time
audio is sampled.

Why filters on DAC ?
The sampled audio stream which is played back contains the baseband and every image
of that baseband, all the way out to infinite frequency. That's why an anti-image filter is
required when going from sampled to continuous ( analog ).

Filters of lower quality or which are unoptimized, exhibit tradeoffs such as low calculation
resolution, higher distortion, ripple, ringing and potential for aliasing. The artifacts of ripple
are time-smearing of the audio, and possible short ( millisecond ) echoes.
Aliasing is a form of distortion which occurs if the filter does not have enough attenuation
in the stop band ( Passband > Transition Band (Roll-Off Band) > Stop Band ).
To avoid aliasing, we must use a very steep filter, or a gentle filter with a higher cutoff
frequency which requires a higher sample rate ( over 44.1 kHz ).

Be logical : the human ear cannot hear above 20 kHz, so any artifacts we're
hearing must be in the audible band.

Use the "slow Roll-Off filter of the DX90 !

Ringer or echo problems in a filtered system can be completely eliminated by
adding a gentle slope filter anywhere in the reproduction chain.
It seems counterintuitive that such a filter placed at the tail end of a chain can repair
previous issues, but Dr. Craven, who has the mathematical proficiency to prove this, so his
paper ought to have a profound effect on how converters and digital systems should be designed.
His discovery explain why some digital audio systems sound better than others.
Craven's discovery alone is justification for using 96 kHz sample rate or UPSAMPLING
TO THAT RATE FOR REPRODUCTION. Another word about upsampling :
Studies about the objective characteristics of conversions at 192 kHz showed that at
those extreme rates, distortion increases and conversion accuracy decreases.
( Craven, Peter (2003) AES 114th Convention Preprint 5822 )

Cheers Highsmith

 

[akira281]

Although I only had the DX90 for only a little over a week, I would have to agree with the others in this thread that going from fw 2.2 to 2.3 the sound has gotten aggressively brighter. Running my Ety4S directly out of the DX90 or through an external amp, is now impossible. What was a very detailed, listen-able experience has become a fingers-on-a blackboard nightmare. My Bose QC20's are more forgiving and aren't effected as much. Unfortunately by going back to 2.2, which I will have have to, I loose the Mac USB DAC which is another reason I bought it.

 

[ReizeiMako]

 
Have you tried the slow roll-off function?

Unfortunately, Right now I'm using Rockbox which easier to create playlist from PC than stock Mango FW. Anyway how slow roll-off function affect the sound of 2.3 FW? Can it provide more bass/impact and make the sound less bright?

 

[matinsoroudi]

 
Unfortunately, Right now I'm using Rockbox which easier to create playlist from PC than stock Mango FW. Anyway how slow roll-off function affect the sound of 2.3 FW? Can it provide more bass/impact and make the sound less bright?

not really

 

[ReizeiMako]

not really

I guess so :l Anyway I need no new function provided by new FW (DAC for Mac, gapless fixed, etc.) so I don't mind using FW 2.2 + Rockbox.