Ayre Acoustics QA-9 USB A/D converter Page 2
"So now we are starting to get somewhereperfect transient response, combined with zero-feedback, fully discrete, fully analog circuitry."
Hansen had mentioned that the QA-9's response was flat to around 1Hz.
"Another thing that we wanted to avoid that is ubiquitous with ADC chips is the brick-wall high-pass filter that is intended to keep DC out of your system. Now, it is good to eliminate DC for a variety of valid reasons, but adding a brick-wall filter will create ringing no matter whether it is a low-pass or a high-pass filter. The Arda ADC chip carries its own similar DC filter, but we chose instead to implement ours from 'scratch' in the Field Programmable Gate Array (FPGA) instead.
"If you tried the conventional approach of using an FIR, you would need several bazillion taps to make a filter that operated that low in frequency. So instead, what people normally do is use an Infinite Impulse Response (IIR) filter, where the output signal is recirculated through the filter to get the desired response. This has the side effect of turning the filter into a minimum-phase type, which would normally be good, but in this case one does not want to introduce phase shift in the bass range. Also, I do not like to use IIRs, as the inevitable errors accumulate as the signal recirculates through the filter. The goal is that the signal is smaller than one LSB by the time the error is equal to one LSB, but it doesn't always work as planned.
"At any rate, we had to come up with a different technique to avoid these problems, so we just did the same thing at the low end that we did at the high end. There is a moving average that samples a few thousand data points and averages them together. Ideally they will all average to zero, but if they do not (ie, there is a DC offset), then the module slowly starts adding or subtracting numbers so that the average comes back to zero. By doing this slowly, it avoids any clicks or pops. By using a moving average, it is (once again) transient perfect, with no ringing, pre-ringing, overshoot, or pre-echo. Essentially what we have done is exactly emulate the way an analog servo works and implement it in the digital domain."
What about the Measure filter setting?
"If you are in a situation where you need totally flat frequency response to the highest possible frequencies (eg, when recording bats or dolphins), the Measure position allows for that. For those filters, we use the moving average to bring the sample rate down from 256Fs to either 8Fs or 4Fs, and then use an FIR low-pass filter to do the final division by two. While these are minimum-phase, slow-rolloff types, they still will introduce a cycle or two of ringing, while the Listen filters are absolutely perfect in regard to transient response."
The QA-9 has a single pair of XLR input jacks. I used RCA-to-XLR adapter plugs to connect the single-ended output of the Liberty phono preamp that Michael Fremer reviewed last May. Reducing the input level by seven clicks gave a maximum level of 3dBFS with all the LPs I played using this preamp and my Linn Arkiv B cartridge. Otherwise, setup was as simple as plugging a USB cable into my Mac mini or MacBook Pro and selecting the QA-9 as the default input device. (A couple of times, I did need to restart the computer for it to recognize the QA-9.) A rear-panel DIP switch selects the Measure or Listen filters, turns the Pro version's Word Clock and DSD outputs on or off, and sets the AES/EBU sample rate. When the QA-9 is connected to a host computer via USB, its sample rate can be set with the chosen recording software or with the system control panel.
My review sample was of the Pro version of the QA-9, which includes both Word Clock and DSD outputs and can transmit DSD data to the host computer using the DSD-over-USB 2.0 protocol developed by a team of independent engineerssee "Method for transferring DSD Audio over PCM Frames Version 1.1," by Andreas Koch (Playback Designs), Andy McHarg (dCS), Gordon Rankin (Wavelength), and Michal Jurewicz (Mytech). Charlie Hansen tells me that v.1.2 of the DSD-over-PCM standard for USB is in development, and that Ayre will offer, he hopes, a free update when that standard is finalized.
However, as I do not currently have a DSD-capable DACour review sample of the dCS Debussy is awaiting an upgradeI didn't experiment with the QA-9's DSD capabilities. That will have to wait for a Follow-Up. However, my dCS 904 and Metric Halo A/D converters all successfully locked to the QA-9's Word Clock output, as did the Logitech Transporter DAC. (A word-clock output is essential for pro-audio use, as it enables other converters to be synchronized to the QA-9, to allow multitrack recording. However, as the QA-9 lacks a word-clock input, those other converters can't be QA-9s.)
I had made only a few recordings with the first sample of the QA-9 when I ran some preliminary measurements. To my dismay, even though all 24 bits in its data output were active and the sound quality of the files it made was far from disappointing, the noise floor was higher than I'd expected (though still below the LPs' groove noise). At 48 and 96kHz, the converter appeared to be operating with just under 18 bits of resolution; at 192kHz, it was actually equivalent to only 16 bits. Concerned that I had a faulty sample, I contacted Ayre about this problem, who asked me to send it back to them to be checked out. (I wasn't giving Ayre special treatment; my policy is to let manufacturers know if we have a review sample that is broken or doesn't appear to be working correctly.)
It turned out that the first 20 QA-9s had suffered from both a firmware problem and what Ayre referred to as "subtle assembly errors." To their credit, Ayre recalled all 20 QA-9s from customers and updated them free of charge. At the end of June, they sent me a sample that was representative of current, corrected production. All of my auditioning comments and the published measurements refer to this revised sample.
An odd subtitle, given that the QA-9's output is a stream of digital data. The quality of those data can be assessed only by using a D/A converter, which will introduce its own sonic signature. I examined the QA-9's sonic performance in two ways: 1) by ripping LPs to 24-bit/192kHz-sampled AIFF files via USB, using Ayre's recommended Vinyl Studio app ($29.95 for Mac OSX and Windows XP, Vista, 7); and 2) feeding the QA-9's AES/EBU output to the D/A processors I had on hand, using a 192kHz sample rate when possible.