On the first date when I just broke open my D16 and listened to it, I was surprised by it performances, the liquidity and viscosity of it musical flows VS my 320Max. I couldn't stop thinking about the reason why it should have been this way.
Allow me to share it with what I thought it was
1/ We have an in house developed products VS a off the shelves products. The differences is stark! The In house products are developed so that every each of it footprints will match the criteria required for specific circumstances VS any off the shelves, which were generic and with multiple features that the DAP/DAC developers have to meet in order for it to perform. Just think about you cooling your own steaks the way you want VS a quick meal portion dropped at your door and all you need to do is to choose a way to heat it up. They are very different and needless to say, your will always be better for your taste.
In this way, Ibasso was able to extract and flex their own developments of their references and standard on what they think is the most suitable for their flagship devices. Just like foods, it is all subjective. The chefs enjoyed his dishes but doesn't mean all of us are.
Now, let's put subjectivities aside, the D16 was an in house developed products. So it power supplied to it pulse wave modulator and the algorithms with the Fourier values from hardware and so on are all satisfying Ibasso criteria
2/ If we took that D16 as stated having 16 elements, then thinking about ROHM chips with 6 elements each while sporting 4X in ultimate mode....we do have 24 elements inside the 320Max. Therefore, the elements are not just yet the decisive factor.
Then what can it be ?
I could not stop thinking about any reasons that could make senses of it. To then finally I think I have came up with a plausible untested reason.
It is the Fir filters together with the discrete built of the D16.
The discrete built is actually the FIR analog filter itself, while the DSP is being done by DSP chip. Other IC such as ROHM has it own algorithm to tackle the digital filter stages. It could have been a processing unit or it could have been a waffle silicon based arrays. The pros of the IC chip is that it tolerances for all transistors/capacitors/resistors on silicon is very very tight of which is 0.0004% or so. The pros of the discrete array from D16 is that it is progressive and continous because capacitors have 10-20% tolerances, the same thing for resistors, which is 0.5% I think ?
So why is the IC chip pro a con of it ? Well, because under digital processing techniques, we can only assume a certain values and not absolute value, it comes from quantization errors and so on. Everything is a close assumption. This is one of the reason why everything will effect the end performances...assumptions on top of assumptions...under some circumstances can completely altering the nature of some things. This is when oversampling are coming into play.
Some people prefer NOS because it has the liquidity of analog quality, not precisely, but similarly, though it has higher floor noises .
From this camp, we have some that enjoy OS because it has lower floor noises and yielding better fidelity, but lacking the liquidity of it analog counterpart...because aggressive filtering and trimming away errors, you inevitably altering the end result to be more sterile....this is what we know as digital signature ? Possibly ?
So then, when we learned to accept that. What makes D16 FIR discrete arrays to be this special ?
Take a look at this slide from IBasso
We see that it stated each DAC can be summed up into an output. We have 32 delays, by 16 elements. That means each sampling cycle is oversampled by a factor of 2x. The errors once being summed up this way will cancel itself out, similarly to how a balanced signals not only is more powerful but also in itself is a anti-interferences signals. This was the original reason why balanced signals were created for pro equipments.
So then, the errors of the discrete capacitors and resistors are being negated out. But why does it has such analog quality ? Because of it own architecture that has it own errors to become it strength. This is the reason why I recommend people not to swap out the orange capacitors or the opamps on the buffering stages in the D16. Doing so will completely change the performances, better or worse ? Unless you also have lab equipment to measures, and even so, I would refrain from doing it because you are ruining ibasso performances
So what is it errors that become it strengths ? The capacitors capability to hold charge and discharge is of a longer timing in tempo than the silicon based capacitors. While the silicon one is more precise, it is also too quickly discharged that the whole circuit needs to be moving in a faster tempo manner and or drawing more power...at this stage it already has negated most of the errors....
But!!! A huge But!!! What is the nature of musicality ? Yes, a rhythmically distortions...so distortion is an errors, and errors will cancel it own errors ... resulting in a more sterile signatures.
Let's stop for a moment here. Does that mean Either topology will finally be imparting it own signature onto the final signals ?
Yes!!! You got it!!! The reasons why each DAC has it own signatures from brands to brands. It is just the nature of it being an IC, it results in a more sterile signals where as utilizing discrete components, it additional errors will self cancel out reserving the best of both worlds...a lower floor noises while retaining the original errors from the musical distortions of the original signals.
From all of this together, it makes senses that PCM sounds better with the signature of Ibasso on D16 than DSD files...which were already decoded by another DAC and leaves not much rooms for the D16 to process it.
