[nraymond]

My understanding is that a DAC frequency response is a combination of the sample rate of the incoming signal, the digital filter being used by the DAC and the analog circuitry that outputs the sound (line level analog if a pure DAC, and if it's a DAC/amp combo it will be amplified by additional circuitry to drive headphones or speakers). So the sample rate affects frequency response, the digital filter affects frequency response and the performance of the analog stage affects frequency response. The analog stage, especially if includes an amplifier, can have a host of other effects on frequency response depending on the output impedance of the amplifier, the impedance of the drivers getting the signal (which can vary depending on frequency, a common aspect of dynamic drivers), and how much power the amplifier has and the volume being played on the drivers in question. You might say that all these factors together give the DAC/amp it's "character". Have I got this right?

Background of why I'm asking, for those interested:
In the Takstar Pro 82/GM200 Review, impressions and discussion thread that BenF has been curating since 2013, someone asked for specific advice on DAC/amps he has been recommending to use with those headphones, and BenF suggested the BTR3K for "lively yet detailed sound", and the P10 + EL-D01 for "euphonic sound". Someone else said, "...please post the FR for each to verify a difference." BenF responded, "You can't see the difference in FR.
Amps usually have have flat FR (except at the very ends), and DACs don't even have FR." To which I responded, "DACs do have a FR" and I provided a link to a review where frequency response was one of the tests. BenF responded by saying a review I had linked to had bad measurements, claimed the reviewer was biased, and said tube amps measure badly and implied they are actually better than they measure and that is why people spend a lot of money on them (implying that measurements aren't worthwhile). I went back to my original point, and then he said, "Frequency response is a result of a transfer function between input and output. In an ideal amp, the output should follow the shape of the input signal exactly, if there is a deviation - we can measure it in dBs. DACs don't have a transfer function, because their input is digital and output is analog - you can't measure a difference between a digital "1" or "0" and an analog signal (at least not in dBs)." Then he suggested I post to this forum instead of replying to anything he's written in "Takstar Pro 82/GM200 Review, impressions and discussion thread", so here I am.

 

[sander99]

If someone says a DAC doesn't have a frequency response then my conclusion is that this person is not the best source of information. Of course a DAC has a frequency response. And usually it will be ruler flat over the audible range.

(The 1's and 0's represent an audio signal that obviously can contain different frequencies. You can measure for different frequencies with the same digital amplitude what the resulting analog (voltage) output amplitude is. And the difference between the resulting output amplitude for different frequencies can be expressed in dB. You define the result at one frequency (for example at 1 kHz) as 0 dB, and express the values for the other frequencies relative to that.)

 

[castleofargh]

A DAC can have some small variations in FR within the audible range(usually it's well below 1dB between lowest and highest audible amplitude). But some DAC with some filter will roll off the upper treble in a way that can be audible.
We always have to make the distinction between what can and should consistently be done, and what some designer ends up making for whatever reason(personal beliefs, money justifications, or just being bad at his job).
I feel like seeking some audio personality in DACs is a massive mistake. The majority of existing DACs have stable high fidelity audio outputs. Amps already can show much more variations because they have to drive vastly different loads at sometimes vastly different voltages. bringing up the possibility for designers to focus on doing a great job withing a certain use range, and more or less drop the ball for other uses. Also headphone amps can still have fairly significant differences in impedance outputs like you mentioned, so that could affect the final response and overall damping ratio of a particular headphone. Some variations one some headphones could be audible.
And of course if you're looking for big sound differences, without using DSPs, just get another headphone and voila! Very different sound almost guaranteed.

Under such circumstances, Why would someone seek change by changing DACs? Why go for what's least likely to bring out a desired amount of change? I'm sure those DACs with a different sound do exist(and maybe in more ways than just FR differences), but I'm not entirely sure that they should exist or that it's to our benefit to search for them. Beyond that, it's just my opinion and someone else can do whatever he likes with his free time and hard earned money.



About digital to analog not having a usable reference, as explained above, we totally do have such reference. In fact all modern measurement devices convert the analog signal to digital before "analyzing" it and telling us all about analog variables. That's how easily the math goes from one to the other(and how accurate digital sampling actually is).

About reviews and feedback from people not being reliable. Well, they're human. Each time they say something, we have a range of possible explanations for it, sound heard merely being one of them.
Humans lie many times a day and they are wrong about many things every single day. Almost no feedback comes from a properly controlled experiment, so it would be truly unreasonable to just pretend that everybody's always correct.

 

[bigshot]

In theory a DAC/amp should produce a ruler flat response. The converter is designed from the ground up to do that, and it isn't rocket science to create a transparent analog stage.

In practice, most DAC/amps do produce a ruler flat response. The ones that don't are the exception, not the rule. Just make sure your impedance is matched properly and avoid obsolete DACs like non-oversampling ones and you shouldn't have any problems finding one that is audibly perfect.

By the way, price isn't a determiner of quality here. You can find cheap DAC/amps that sound perfect, and expensive ones that don't oversample and muffle the highs. No need to spend a lot.

 

[nraymond]

Thanks for confirming my understanding. The only pure DAC in the original discussion that led to my post was the ELE EL-D01, which was a very simple and inexpensive USB DAC when it was available for purchase, based around the Texas Instruments PCM2704. The Little Bear P10 is a MOSFET Class A tube multi-hybrid headphone amplifier, and the Fiio BTR3k is a combination of a battery powered USB DAC, Bluetooth receiver, and headphone amplifier, based on the CSR8675 BT chip and dual AK4377A DAC chips. It's so strange to me that someone would completely discourage anyone from wondering about the comparative measured performance of these devices, because I'm pretty sure they're going to measure differently (especially due to the differences in amplification), and those measurements could inform a potential buyer as to as to why and how they sound a certain way, and what they might expect when pairing those devices with different headphones. I think it's a definite missed opportunity for people in our head-fi.org community to get a better understanding of audio when someone is basically banning any discussion about measurements other than what one person has to say from a forum thread.

 

[bigshot]

Tube amps can be audibly transparent, but there really isn't much reason for them to be. If you want clean signal, you'd use a solid state amp, if you want colored, you'd use tubes (or better yet, a tube amp emulating DSP).

That said, there's no accounting for the ignorance of audiophile "experts".

 

[T 1000]

My knowledge of Dac is very poor. Can you explain to me, if the Dac chip does not need to be "modernized", why is it being done? Is it a deception in order to make a profit?

 

[bigshot]

Well, DACs have different features, different interfaces and different branding. You buy a DAC for those things.

 

[T 1000]

Ok, it is clear to me what the DAC is for, but it is stated here that once I buy a DAC, there is no need to ever change it again, because the one I have is enough, and a better one will not bring me any qualitative change?
If I understood correctly what @castleofargh presented

 

[Raketen]

Based on your account and without reading any of the links- it seems there was a category error/misunderstanding in the discussion (or perhaps I am just introducing my own error here )

Your frame of reference was global and comparative- the "input" being the ideal reconstruction of an analog signal as represented by the source file, the "output" being the reconstruction in actuality at different steps of the analog signal chain (dac output>amp output>transducer output). That is to say, the accuracy of the reconstruction.

Your interlocutor's perspective was local- the literal steps of the signal chain, 011010 output -> 010101 input; analog ~ output -> analog ~ input. That is to say, the accuracy of transmission at a particular point relative to itself, which isn't relevant to the discussion of "synergy", I don't think (at least not in the context of measurably transparent equipment).

Since they are most often visually represented with just the dB@frequency graphs, maybe that leads to some misunderstandings of what frequency response measurements are capable of capturing and what comparing measurements can illustrate? If anything is going to be able to illustrate "synergy" it is probably FR measurement comparisons of some form or other (even if perhaps we don't yet know everything there is to know about them).

 

[T 1000]

Good morning (here), thank you

 

[bigshot]

If the DAC is audibly transparent, it can handle a higher data rate which will extend frequencies and lower the noise floor, but you won’t hear the improvement with human ears. Transparent is perfect for human ears, and DACs that aren’t transparent aren’t common.

The Nyquist theory states that frequencies are perfectly reconstructed up to half the sampling rate. So CD quality of 44.1 will perfectly reproduce all frequencies up to just beyond the upper limit of human hearing. And 16 bit provides a noise floor with dithering of 90dB, which means you would have to turn the volume up to deafening levels to hear the noise floor.

16/44.1 was designed to be audibly perfect. And almost any DAC you might buy can perfectly convert 16/44.1 to an analog signal you can play on headphones.

DACs aren’t the area of your system that you should be looking to for sonic improvements. Better to choose one that has the features and price that works for you.

 

[T 1000]

I think that's the part that confuses me. The transparency you mention here is very important to my ears, but it requires a good implementation of the components inside the DAC, then an amplifier that will give each piece of information the right dose of energy...,am I inferring something wrong?
However, let's stay only on DAC
Thank you for your patience

 

[T 1000]

Let me simplify the question
If we add MSB Selekt DAC II to the system of 50K, and then remove it from the system and add Denafrips Ares II, can we not hear the difference
Of course, reference headphones or speakers are implied

 

[bigshot]

A DAC converts both frequency and amplitude. An amp just increases the power from line level to an amount that can push the transducers.

I’m not familiar with those specific brands of DACs, but it is the job of a DAC to convert a digital file to an analog signal that is exactly what is intended. A DAC should add no coloration at all. It is calibrated to produce a signal that is up to digital specs- which is transparent by design. If it does have a sound that is different than other DACs, it is defective, either by manufacturing or design.

So to answer your question, if those two DACs produce a line level signal that sounds different, then one or both of them are defective and are not performing to spec.