Extensive A/D Testing
Extensive A/D testing
This test took place in 2011 at Coast Recorders with the help of Jeffrey Ehrenberg at Vintage King, Michael Romanowski, and Paul Stubblebine. We decided to do a comprehensive shootout of the A/D capabilities of as many of the leading converters as we could gather together at one time.
The Analog source material for the tests was collected from the high resolution archives of The Tape Project. In all, the following converters were tested, some of which, as you can see, are of the high-end variety, and others not:
Apogee Rosetta 200
Burl Audio B2 Bomber
Digital Audio Denmark AX24
Lynx Aurora 16
Metric Halo ULN 8
Mytek Digital 8×192
Pacific Microsonics Model Two
Prism Dream AD-2
Universal Audio 2192
The source material was transferred from Paul Stubblebine’s one inch ATR two-track machine, featuring Tim deParavacini custom electronics, through MIT cables directly into each converter one at a time, with each converter acting as its own master clock. This source material is of extremely high quality, having been transferred directly to the one inch two track machine from the original half inch mix tapes. These mix tapes in turn were made directly from the analog multi-track master tapes so at no time was the source material in the digital domain.
The source material consists of three songs, a classical piece performed by the Minnesota Orchestra and recorded by Keith Johnson, a blues/rock piece performed by Dave Alvin, and a jazz performance by Patricia Barber.
We chose these particular songs because we wanted to see how the converters fared with handling the dynamic range of classical music, some material with guitars, bass, and drums, and some material featuring a big vocal sound with plenty of ambient information.
The files were captured, along with 1k tones for each converter, in a Sonic Solutions Soundblade system at 96kHz 24 bit resolution, using a Lynx AES 16 I/O card. Minor gain adjustments were then performed within the workstation using the tones as a guide so that all the levels of the resultant files are matched precicely.
Some comments about the test:
We transferred each set of three songs separately into each converter set to its internal clock. We did this because although we could have transferred the material into several units at once, this would have meant that one of them would have to be the master with the others clocked to it and this would give the master device an unfair advantage over the others since the optimal setting for these units is usually, according to the manufacturers anyway, internal. Therefore this seemed the most fair way to do it. In any case, if we had tested all the permutations of clocking possible we would have generated hundreds of files that no one would have the time to listen to!
Perhaps the decision that leaves our methodology the most open to criticism is that of putting the program straight across from the tape machine and adjusting the levels of the files in the workstation instead of adjusting the output of the tape machine, or using an analog gain reduction/boosting device in the chain to calibrate the inputs of the different converters. Obviously this is an important consideration because there are differences in the way that converters perform with different input levels.
In deciding to do it the way we did, our reasoning was that had we adjusted the output of the tape machine to make up for the differences in input level of the converters the character of the material would have also changed as it became removed from the sweet spot of the tape machine electronics. The same could also be said for an analog gain reduction/boosting device inserted in the chain.
In retrospect, perhaps had we had more time, and a very neutral source of analog gain adjustment readily to hand, I most likely would have calibrated the inputs, not so much because I feel the results we got are tainted, more to avoid the wrath and derision of those who inhabit the Gearslutz message boards where this is also posted!
In any case, all the files came out within a dB or so of each other with the exception of the Cranesong which, curiously, was quite a bit hotter than all the others so there may be some grounds for putting an asterisk against that particular result. In general I believe the results are a very good comparison and are very interesting to listen to, and you can drive yourself crazy too!
If anyone is interested in listening to these files that can be arranged. Please feel free to call or email me to discuss.