Owners Manual

7) DIGITAL OPTION: Some SLAM!s will have this converter option and some will not and have just a blank panel fitted. The 'Option'
is two parts, and one is this panel with jacks, switches, and its 3 circuit boards. It is intended to be something the average user can install
and similar to inserting a PCI card in your computer and attaching a few ribbon cables to it. The second part is a 5"x6" board that contains
the converter chips, clocks, PLL, two very high speed SHARC DSPs, and a micro controller. These boards are static sensitive and one must
be grounded if handling them. Like, don't be shuffling your feet on the carpet as you go to pick it up.
8) AES EBU INPUT: The standard digital input that feeds the D to A converter (DAC). It accepts data or sample rates of 44.1K, 48K, 88.2K
and 96K. The sample rate is asyncronously up-converted to 192K and jitter is removed in the process. A dedicated high speed SHARC DSP
chip running proprietary code and 40 bit math is used for that. The result is near zero jitter, and audibly less 'time smear'.
Note 1- The DAC outputs are the 1/4" phone jacks (5) and/or can be passed through the tubes, limiters and iron as needed via the SOURCE
switch on the front panel.
Note 2- We have used simple XLR to RCA adapters to use SPDIF outputs to feed this jack and it seems to work fine but there are true SPDIF
to AES converter/adaptors that would be the officially recommended method.
9) WORD CLOCK INPUT: Regular BNC jack that accepts a master word clock or Super Clock. In complex workstation installations
often a distributed master word clock is used to guarantee stable and accurate timing across the variety of digital gear in use. With most
DACs, this usually helps improve the sound.The most common report is an improvement in imaging. The biggest reason is that it offers
a better alternative than AES lines for carrying the clock component, which can be considered an analog signal, and most converters have
less than perfect clock recovery or jitter removal circuits (PLLs) (being very polite here). The other reason is that converter chips have
what is called 'fixed' coeffecient FIR filters that depend on highly accurate clocks and stand crystal oscillator clocks are not usually that
accurate. Because the Anagram converter is almost immune to jitter, and uses 'adaptive FIR filters' and not the internal (free) chip filters,
the usual audible benefits of the word clock may not apply to this converter. However, probably other components in your system like word
clock and this converter retains compatibility (and convenience) by the inclusion of this input.
10) AES EBU OUTPUT: This is the A to D converter's (ADC) output. As with the DAC, the actual sample rate is 192K which is down-
converted to your choice of data rate from 44.1K to 96K. A second SHARC DSP is used here.
11) DAC FILTER: In developing this converter, our research (and Bob Katz) suggested that often the biggest audible differences between
converters depended on the designer's choice of filters, so we gave you the choice. The toggle switch provides 3 different filter frequencies,
20K, 40K and 80K passive analog circuits. The 20K and 80K are 3 pole 18dB per octave, and the 40K is 2 pole 12dB per octave, each based
on the Manley Massive Passive filters. A good starting point is 80K in a great system and 20K is less harsh / brittle and perhaps warm.
12) ADC FILTER: Similar ideas as the DAC filters except the 40K setting has a 1dB bump or boost at 20K for 'air' and a unique feature
and is otherwise the steepest of the 3 settings. Note with both 80K filters that because the maximum data rate is 96K and Mr Nyquist's
theorem, the maximum true bandwidth is about 45K. If the only concern with filters was the final bandwidth, the 80K would be pointless
but because of a variety of factors there does seem to be subtle audible nuances between all 3 filters even at 44.1K data rates (20K BW).
13) SAMPLE RATE: The actual sample rate is always 192K and this knob is really a "DATA RATE" control. The seven settings are:
44.1K, 48K, 88.2K, 96K, AES EBU IN RATE (locks to the DAC input AES stream), WORD CLOCK, and SUPER CLOCK, (locks to
the clock rate input to the BNC connector). Why no 192K selection? There is no official standard for 192 connections and the defacto method
is 2 XLRs and obviously we didn't have room on the panel, besides we provide the audible benefits of 192K without having to use that high
of a data rate. The converter is 'ready' for 192 if it becomes more common and connection to 192 systems becomes more feasable.
Why no WORD CLOCK output? Experts agree that the best A/D clock is its internal crystal, and this ADC is always in that mode even
if you lock to the word clock input or AES. As described above, part of the reason better word clock generators sound better than cheaper
ones is due to the absolute accuracy (48K= 48,000.0000 Hz) and you are better off with a dedicated box with oven controlled crystal accuracy
than the built-in clock of this converter, and if desired, the better boxes will lock to the AES output of this converter and allow proper
distribution required. In other words, if you are going to do it - do it right, or don't do it. Always listen, and don't blindly trust word clocks.
14) NOISE SHAPE & 15) DITHER: The ADC samples at 24 bit resolution, but often we need 16 (or 20) bit data. Simply chopping off
the 8 bits results in audible artifacts and a loss of low level resolution. Modern converters add a small (almost inaudible) amount of noise
(random numbers from +1to -1) called dither which effectively smooths the data and removes artifacts (trading audible distortion for a tiny
amount of noise). In other words, the signal better appoximates the original analog. Noise Shaping is used to describe two different
techniques that also help. The first is a method where the dither is pushed to the near ultrasonic and most of the noise energy is focused
where we are least likely to hear it so we still retain the benefit of dither but don't hear the noise. The second method uses the difference
between the 24 bit data and the 16 bit data (those 8 bits) which can now be considered an error signal, and does the same thing - pushes
the energy to the extreme high part of the spectrum, which also seems to increase resolution. Some companies refer to this as 'apparent
resolution' and claim, for example, 19 bit resolution on 16 bit data. The biggest difference can be heard on reverb tails and the end of fades
and there may be personal preferences involved so it is worth evaluating for yourself. None of this applies with 24 bit word lengths.
16) WORD LENGTH: This toggle sets the ADC and AES EBU OUT word length for 24 bit, 20 bit or 16 bit data. Most recording today
seems to happen at the highest resolution of 24 bits, but mastering to CD still requires 16 bit data and some MPEG codecs prefer proper
16 bit data rather than 'raw' 24 bit. The NOISE SHAPE and DITHER described above are only active if 20 bit or 16 bit is selected.
8