[sdiy] that darned record module i keep obsessing about

[rsdio at audiobanshee.com]

On Aug 16, 2014, at 8:42 AM, Henry Birdseye wrote:
> I've been way overthinking this idea. How much does the viability go up if it's 2 analog 10v p-p signal going into level control, a place to plug a flash card and a big red button that starts and stop recording directly to the card. nothing else. no dsp, no programming, just addac and data transfer.
> 
> Sound better?

Good and bad.

The 10 Vpp dynamic range is an important feature. That's +13.2 dBu! Most pro recorders might not handle that without external padding, and consumer recorders certainly wouldn't handle it. So much for my idea of a passive panel that just adapts from modular cables to standard audio cables for the typical recorder. Instead, if you put some high quality analog circuitry in there, it would be a good feature.

The bad news is that you can't write directly to a Flash card - at least not in a way that you can read it anywhere else. Nobody that I know of supports reading audio from a raw, unformatted Flash card. Instead, everyone puts some kind of file system on the memory card. As soon as you get into a file system, you run into lots of problems, and you certainly need programming and a CPU to handle that. The problems include bandwidth, unpredictable delays when the Flash card decides to work around bad blocks, limitations of WAV file lengths, and the limitations of file systems that are compatible with a variety of systems. FAT32 is probably the most compatible format, but it has all kinds of limitations, inefficiencies, and is basically old technology. But if you choose a better file system, your Flash card might not work with everyone's computer. WAV is a good choice for audio, but it's a dated format. CAF is much better, but not supported everywhere. These are the main reasons I said your idea was more trouble than it's worth.

What I recommend is to take your 2 analog 10 V p-p signal processing, feed it into a 5 V p-p ADC chip, and then output the digital signal in AES3 format. Two major digital chips and some analog. Put a 3-way rotary switch on the face to select 48 kHz, 96 kHz, and 192 kHz. Might as well do only 24-bit, since AES3 transfers that much data all the time anyway. AES3 is also an old technology that has certain problems. I'm totally against AES3 for outboard DAC, because it doesn't allow the DAC to be the master clock without serious PLL and data manipulation that ends up not being bit-transparent. But for ADC, you get to be the master clock, and AES3 just makes it easy to send the audio to any standard recorder or computer with digital input. On that note, you'll want to design a circuit with really good digital audio clocks for best audio quality. Put an AES3 output on your face plate, maybe with SPDIF coaxial and optical for good measure, and call your module done.

Now that I think about it, I really like this idea. It's actually doable without lots of programming or dealing with problems that will ruin the fun. Since we live in a world where you can patent ideas, I'll just rush off and patent this now...

Brian Willoughby
Sound Consulting