|
Digital Drive Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it. |
|
|
|
Digital Drive Upsamplers, DACs, jitter, shakes and analogue withdrawals, this is it. |
|
In Reply to: IME, it all makes a difference. posted by jusbe on October 26, 2017 at 06:50:33:
> > over an equivalent design using a Sabre DAC < <
That is a huge problem.
There are two main points that ESS seems to be chasing:
1) Ridding the algorithm used in the modulator (that reduces the 32-bit interpolated ("oversampled") audio data to the 6 bits of the actual DAC's output stage) of bugs that cause non-linearities not found in ladder DACs. Their original Hyperstream modulator was far more successful at this than pretty much anybody else's attempts, while their second-generation Hyperstream II does it just about perfectly. This should eliminate any potential for a generic sonic difference between delta-sigma and ladder DACs.
2) Offer a one chip solution for an entire digital product. The ESS chips include what used to be incorporated in what was called a DIR chip (Digital Interface Receiver that had a PLL to lock on to the incoming S/PDIF data), plus a microprocessor to automatically distinguish between S/PDIF, DSD, DoP, and compressed audio formats (eg, Dolby AC-3), plus what used to be an external digital filter chip, plus an ASRC (Asynchronous Sample Rate Converter) chip used as a jitter reduction tool, plus a digital volume control chip, plus the actual D/A conversion chip itself - all into one complex, highly configurable package. This is why it took years for manufacturers to adopt the ESS chips - it took that long for them to understand what it did and how to use it.
The input stage of an ESS DAC chip will not only lock on to any S/PDIF stream, but also automatically detect whether it is PCM or DSD or DSD-over-PCM (DoP), at any sample rate. It also has a "jitter eliminator" built in, which is just their term for an asynchronous sample-rate converter (ASRC), plus a 32-bit volume control, plus an 8x interpolation ("oversampling") filter comprising a 4x section feeding a 2x section. The original ESS only offered "Fast" and "Slow" rolloff options. Newer versions of this filter offer up to 7 options plus the ability to custom program with user-specified coefficients (oddly only at 24 bits, even though the filter actually operates with 32-bit filter coefficients when using the pre-programmed ones). Then there are a host of undocumented features that can be engaged or disengaged or programmed or altered.
The bottom line is that saying that a product uses a certain model of ESS chip is kind of like saying that I have a certain year of Mustang. Is it a convertible or a hardtop? Does it have a 6-cylinder engine or a small-block V-8 or a big block V-8? Does it have a manual or automatic transmission? Does it have the "sport" suspension option or the standard package? Does it have the vinyl upholstery or cloth or leather? Does it have idiot lights or the optional sport package with gauges? Does it have the stock steel wheels with cheap tires or aluminum alloy wheels with low-profile performance tires? Does it have a regular differential or the limited-slip option?
Just as every single one of these options will affect how the car performs and interfaces with the user, all of the various configurations available on the ESS chip will affect its sound quality - in many cases drastically. Just to give one example, the default configuration used by the vast majority of DAC makers using the ESS DAC chip engages the ASRC. While this allows for superb measured performance on the J-Test conducted by Stereophile magazine with very little effort, to my ears it sounds awful and amusical. If it were not possible to turn that feature off, I would never use the ESS DAC chip.
Bat as I said in the posting I linked, there is an irrational belief by most consumers that the DAC chip is one of the single largest factors that affects the sound. Unless you at least have the eval board and can play with all of these various settings to see what they do to the sound, you are (unfortunately and not to be denigrating) speaking out of ignorance.
Even with the eval board, you are listening to their choice of analog circuitry, which is just op-amps and then powered by whatever power supply you happen to hook up to it. It may be with the specific colorations of the specific op-amps used on the eval board, that some weird setting might sound "better" as it tends to minimize or complement the coloration of the op-amp.
It's not until you can put the chip into your own circuit with analog circuitry of known quality level, using power supplies of known quality levels that you can even begin to understand how these various configuration affect the sound of the DAC chip. This is pretty much only something that manufacturers can do. In the case of Ayre, about the only parts of the ESS chip that are used in (say) the QX-5 are the Hyperstream II modulator and the output stage current sources. That's it. We use our own S/PDIF receiver, our own digital filter, our own digital volume control (which works pretty much identically to theirs, which is an excellent implementation of a digital volume control), and our own just about everything else (including detection of PCM vs DSD vs DoP).
It's very, very easy for people to reach the incorrect conclusion when they only have access to limited information. It's not their fault, especially in the case of ESS, as they are unusual from the standpoint that they deliberately make access to their datasheets difficult, and even their datasheets are often incomplete. Ayre had a good relationship with ESS and when required can actually talk with the chip designer to figure out ways to get the chip to perform the way we want it to and not the way that other companies use.
Having built products where I've been in full control of all of the aspects of the design has led me to come up with the list I linked in the previous post. I strongly agree that everything has some impact on the sound quality - even things that shouldn't, such as what footers are used on the D/A converter chassis - but as noted previously, the actual DAC chip or the conversion technology (ladder vs delta-sigma) is fairly low on the list.
Hope this helps.
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- RE: IME, it all makes a difference. - Charles Hansen 10/26/1720:11:32 10/26/17 (0)
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
