On resolution enhancement
Remember that in extending the 16 bit data to 24, you've now got 256x the number of discrete levels. There is a small rounding error, but technically we could reduce the gain to 1/256th of the original signal and not lose any precision (although you'll have lost the dither beneath the 16 bit LSB).
The problem with operating at very low gains (realistically this would only happen when used as a pre-amp and running a low volume) is that although we've not lost much precision in the digital domain, you're now asking a DAC to operate with low analogue levels at the point where errors will form a greater proportion of the signal. (A 24 bit DAC will operate better with 16 bits at the MSB, than with the 16 bits at the LSB). To get a DAC to work well, you want to use all of the bits (and all the dynamic range). Increasing the gain makes sure that you use all the levels available, and adding dither to the LSBs allows you to have controlled, random noise rather than nasty noise correlated to the signal.
Increasing the gain allows you to use more of the dynamic range of the downstream DACs, I think the theory is that the gains from this outweight the loss of a very small amount of data in the rounding (which realistically will be within the noise floor set by the dither anyway).
The other benefit to dithering the 16 bit word to 24 is that it allows your DAC to operate more linearly, since you're not asking that it switch to perfect 16 bit levels (which it won't achieve), but if it can get close to the 24 bit levels with the same accuracy then errors are lost within the dither noise, rather than being correlated to the 16 bi levels.
Output bits - C(onsumer) or P(rofessional)?
This controls the formatting of the control data within SPDIF. 32 bits are transmitted per sample on AES/SPDIF. Of these 4 are preamble (for synchronisation), 24 are sample data (regardless of what size samples you actually send) and the remaining 4 - VUCP are Valid (you can output the data safely without blowing up speakers), User, Channel and Parity (error checking). If you add up the C bit for lots of samples you get frames of data, consumer and professional are just different formats for the frames. (Thanks ultima_gtr)
For 96/24-capable DSP speakers, set output bits to 24. For 18-bit MkI DSP speakers, you can set output bits to 22 as the original DSP speakers can take a 22-bit input!
Thanks to ultima_gtr for most of this information.
Using 518 with the new 565
If you have a Meridian 518 Digital Processor, this will de-jitter a DTS signal for optimum sound quality. However the resolution enhancement features of the 518 are not appropriate as DTS is a compressed digital stream, not standard digital audio. Hence your 518 should be switched to Bypass mode prior to listening to a DTS CD or LaserDisc. There is an upgrade to the 518 software, Version 1.2, which allows it to detect a DTS stream. In this case, the 518 automatically selects Bypass when DTS is detected. Version 1.2 adds four DTS modes to the 518 which can be assigned to sources in Config. The modes are:
DTSMU – 518 mutes on receipt of DTS (used when 518 is a preamp). DTSCD – detects DTS with a normal strategy and selects Bypass. DTSLD – detects DTS with a rapid strategy and selects Bypass. DTSMA – mastering processor mode ignores DTS. These can be selected according to Type as follows:
Type 0 – all sources DTSMU Type 1 – all sources DTSCD except DVD and LD which are DTSLD Type 2 – all sources DTSMA
v1.22 firmware vs. v1.21
From Richard, who wrote the code for the 518:
The differences were in default settings for the firmware. DVD defaults to Data = Y and allowing for the different word length for DTS on DVD vs DTS on LD. So unless all source options are used and you do not have an Laser Disc player you do not need to change the firmware.
In-depth Stereophile review: http://www.stereophile.com/digitalprocessors/367/index.html
A review of the 518 by Stacey Spears with some more info on ROM version changes: http://www.hometheaterhifi.com/volume_3_3/v3n3a.html