Thelonious Monk
Headphoneus Supremus
- Joined
- Dec 6, 2006
- Posts
- 2,483
- Likes
- 13
Quote:
that's one hell of a parentheses. i'll look deeper into Meridian products, that sounds like a very interesting process...
Quote:
hook a brother up, i'd like to know a few authors who write on audio engineering. i enjoy a good read on something that i half-way understand, like linux books.
Quote:
very informative again. i thought Wadia was the tip of the iceberg, didn't know EMM Labs had even MORE insane sampling rates.
Originally Posted by Welly Wu /img/forum/go_quote.gif It does not surprise me that this is occurring. Digital coaxial RCA is preferable over plastic or glass fiber optic TOSLink because the jacks are uniform are more secure, the materials are more durable and reliable, it has higher bandwidth, and it has less signal degradation over longer distances plus it is dirt cheap. The reason why so many audiophiles can hear distinct sound quality differences between these two different types of S/PDIF connections is because TOSLink has the potential for greater variances which I will explain later. However, it is the digital interface and processor(s) that most contribute to higher jitter in the S/PDIF interface. One box digital source components do not use interfacing S/PDIF connections so that reduces the potential aggregate jitter numbers and spectra variances compared to separate transport and DA/C solutions unless you are wealthy and can truly afford a proprietary solution to this conundrum (EMM Labs, Wadia, TEAC Esoteric, etc). For this technical reason alone, this is why I always search for one box digital sources (Meridian 808i. It uses a multi-speed DVD-ROM that can make multiple passes over scratchy CDs which increases error correction 100x, recovers that data in bit perfect sequential order, buffers it into three stages of which the first two are FIFO, shoots it over to both Meridian multi-bit oversampled delta sigma D/A converters and Meridian's Resolution Enhancement Technology that utilizes three 48bit/150MIP DSPs to upsample 16bit/44.1kHz -> 24bit/176.4Khz all the while a custom Meridian master oscillator type clock operating at 16.9344MHz keeps jitter down to 90 picoseconds with less than 0.1Hz spectra variances. That data goes off into two separate cards: an analog card with discrete traces for both unbalanced RCA and fully balanced AES/EBU outputs and a custom card for Meridian' High Resolution 24bit/88.2kHz digital output). |
that's one hell of a parentheses. i'll look deeper into Meridian products, that sounds like a very interesting process...
Quote:
Regarding the choice of cables, a boutique brand name 75/110 Ohm digital coaxial cable can cost upwards of $250 USD, but a plain jane video or XLR cable can be bought for less than $5 USD per foot of cable, yet both will give you virtually the same jitter measurements (all things being equal of course). TOSLink is the standard alternative, but it uses jacks that do not have a physical locking mechanism and plastic or glass fiber optic have wide variances in terms of aperature, durability, material purity and consistency, low bandwidth of about 6MHz, and virtually all lose a steep amount of data over longer distances. Furthermore, both glass and plastic TOSLink cables have low tensile strength which means that you can not bend them around tight corners without either breaking them or creating such a steep curve that you virtually guarantee a higher degree of signal loss over shorter distances. For the purposes of high end audio, this is simply unacceptable. AT&T ST glass fiber optic cable has a locking bayonet, the digital interface and processors are ready out of the box, high bandwidth of up to 150MHz, and data can travel very long distances in excess of several hundreds of feet with no signal loss. The only major downsides are cost and availability because they are not only expensive at several hundreds of dollars per foot, but the transport and processor must have the same interface boards which automatically doubles the cost of using this terrific cable (plus the other electronics). Furthermore, the terminations are delicate and fragile if handled improperly. In my opinion, a good performance to cost compromise is the AES/EBU (XLR) because it has the potential of up to 500MHz of available bandwidth, it is widely available in upscale consumer/professional/computer electronics, and the cables can be had for less than $100 USD for a pair which is reasonable in my humble opinion. It is also fully balanced and transmits at 5V rather than S/PDIF at 0.5V. The major drawback is that not all audio engineers and designers adhere to the 110 Ohm impedance, +/-20% specifications. Consequently, impedance mismatches cause reflections and increase jitter. Compromises are abound in audio. I hope this was helpful and not confusing to the discussion. For my birthday next month, I will go buy those three digital audio engineering books. |
hook a brother up, i'd like to know a few authors who write on audio engineering. i enjoy a good read on something that i half-way understand, like linux books.
Quote:
Originally Posted by Welly Wu /img/forum/go_quote.gif AT&T ST glass fiber optic, especially when utilized in multi-node systems, has a higher potential for the best sound quality in digital audio systems because the AES/EBU (XLR connection) have greater variances not only in terms of grounding problems and impedance mismatches among different components, but also the circuit design can vary from true differential balanced mode to conversion of an unbalanced signal to balanced XLR outputs through the implementation of cheap op-amps. Also, the extra bandwidth that both AT&T ST and AES/EBU offer is critical because companies like Wadia or EMM Labs have designed custom in house AD/C and DA/C convertors that go up as high as 5bit/5.6MHz or 24bit/2.822MHz. So, bandwidth does matter. Also, TOSLink is simply inferior to AT&T ST glass fiber optic; listen to the new Wadia 581i or 781i and compare the TOSLink versus AES/EBU to the BNC to the AT&T ST glass fiber optic connections for yourself. Sorry, I am at work so I can not refer to my audio engineering books now. |
very informative again. i thought Wadia was the tip of the iceberg, didn't know EMM Labs had even MORE insane sampling rates.