Meridian Unplugged - Powered by MrTechGuy.co.uk Forums Discussion Meridian - Systems USB

I was curious about the USB audio input on the G61/G68 - whether it was asynchronous or adaptive. The only mention in the forum seems to be a similar question:


I wonder if there is any further information about this? I have been following the response to the Ayre QB-9, and am quite surprised - bearing in mind their high standards regarding SPDIF signal quality - that Meridian seem to have missed a trick regarding USB.

Of course, the very large amount of jitter inherent in an adaptive USB connection is no doubt dealt with by Meridian's famously strong jitter rejection, but asynchronous would be a much better start to the whole situation.

(Let's leave aside whether a USB input is valuable: in my particular case it is, extremely so.)

Nic

Hello Marbach,

As far as I know, the USB input is not an audio input but a substitute for the RS 232 input!
Gr. Paul.

Um there are two USB ports on the back: an audio input, and the service one you are talking about. They are nowhere near each other. Have a look at the back...

N

...but I see you have a G68ADV which if I read Duncan's website correctly doesn't have the USB audio input. I am referring to the current models of each.

N

The latest version of the G68ADV does.

Yes there is no mention anywhere of any details. typical
Basically the optical O1-SAT at the bottom left corner of the unit has been replaced by a USB input most likely at the end of 2007 (according to the date on the datasheets for G68 at the time).

The connection supports 16/48 (page 12 of current G68 Surround Controller PDF) so my guess now is that it is a cheap and quick adaptative implementation...

it would have been great to have a proper 24/96 implementation, may be it was a rushed job to get one in quick.

Doesn't seem like typical Meridian behaviour to do a rush job with a digital input, but I agree that's how it looks. And none of our friends here seem to either understand what we are talking about, or care... Ah well.

Nic

Here is the layout of that digital board from a G61R (Should be the same as in the newer G68), you can see the USB Audio input section shown in the Red box at the bottom right. If I remember correctly it is standard PCM270x chip that runs at 16bit and either 32, 44.1 or 48Khz.

Interesting. And depressing. Take a look at the comments on that chip here:

The problem with all of the chips in the Burr-Brown PCM270x series is that they have high levels of jitter. A fixed-frequency master audio clock is not employed. Instead a variable-frequency master audio clock is generated based on the timing of the incoming audio data. The computer sends packets of audio data at one millisecond intervals across the USB connection. The Burr-Brown chips have internal circuitry that measures the interval between the audio packets, and then uses this information to generate a new master audio clock that matches the rate at which the computer sends packets. This system, called “adaptive” USB mode, creates several opportunities where jitter will be added to the D/A converter`s master clock:
1) Any variable frequency clock will intrinsically have more jitter than an equivalent fixed-frequency clock. In the “adaptive” USB mode, the clock in the D/A converter must “adapt” to match the rate that the computer sends out audio packets. In the case of the Burr-Brown PCM270x parts, the master audio clock comes from a VCO and inherently exhibits high levels of jitter.
2) The computer cannot send audio packets at a perfectly fixed rate. Firstly, the computer`s internal clock is not designed to have low jitter. In a market segment where costs are literally shaved to the fractions of a cent, it would add too much the cost of a computer to use a high-performance, low-jitter clock. Secondly, the inside of a computer is filled with RF interference, making it impossible for even the best of clocks to maintain their spectral purity.
3) The Burr-Brown USB-enabled D/A converter chips update (change) the frequency of the master audio clock each time a packet of audio data is received. This happens once a millisecond so there will be a strong jitter component at 1 kHz, right in the middle of the audio band. Typical measurements of the PCM270x chips exhibit jitter levels in the thousands of picoseconds, orders of magnitude worse than a well-designed one-box player.

Finally, it should be noted that D/A converter boxes based on the PCM270x series parts are limited to a maximum sample rate of 48 kHz and a maximum word length of 16 bits. So while they can transfer CD-grade audio from one`s computer, they are incapable of playing higher resolution files.

That's from Ayre's QB-9 white paper here: [url=http://www.ayre.com/pdf/Ayre_USB_DAC_White_Paper.pdf][/url]

(The last point makes it clear that 48kHz is a physical limit and will not change with firmware.)

While I am not going to buy a QB-9 (preferring to plough more money into my Meridian set-up) the design seems to have been very thoroughly thought-through and is receiving very positive reviews.

I hope Meridian catch up with this. I would rather it didn't have a USB input than had one which was not up to their usual standards of leading the field.

Nic

I'm sure the output from the USB circuitry is routed through the FIFO buffers, so I wouldn't worry about that too much.

It was a weird thing to do though. I think M misunderstood market requirements.