I just looked again at the Linear LTC2202 datasheet. I didn't notice before, but they have a suggested driver here. The converter is $50 and the driver is $4, not including some 1% high tolerance resistors. I could make it work... maybe the performance would be better than the other ICs, even if running at just 16x oversample.

Changed my mind on the ADC and ADCRAM. I really only need one ADC. Most likely I will go with the Maxim IC, and a 1Mx16 SRAM. As this board is really only meant to make sound effects for games, there's really no need for a large SDRAM for a large ADC sample. Still 6 sec's of recording should be possible with 1M at 165ksps.
Just did a quick general schematic to get a picture for the I/O's I will need.

To keep it simple the DACRAMs will only be written to when the Devboard CPU needs to change the data. 15 Address lines will be MUX'd with the Devboard CPU so the 3 DACRAMs will be mapped into the 65Org16 Devboard. The Devboard CPU can't read from these RAMs, it really has no need to. It will be able to see what the DAC's are outputting through the ADCRAM... Otherwise, the DACRAMs will be outputting their data to the I2S cores and those cores will feed the DACs through 9 lines. Lastly, 5-10 lines for various control. This makes for approx. 45 I/O pins on the FPGA for the DAC section.

For the ADC section, the common 16 bit data bus lines will be used to primarily read the data the ADC puts in there, through the FPGA. This ADCRAM can also be able to be written to for program, data, etc. Possibly a battery backup/watchdog circuit will be used. 20 address lines will be MUX'd with the Devboard CPU so this RAM will be mapped into the Devboard as well. 4-5 lines for various control makes for about 25 I/O pins.

So that is 70 I/O pins, which is cutting it too close to use a 100-pin FPGA. Time to use another 144-pin Spartan 6! Which is good, as I have become most familiar with it's pinouts. It will fit no problem into the current layout, and there are 2 PLLs which can generate exact frequencies.

I changed my mind again about the ADC. I think I will go with the high speed Linear Technologies 16-bit ADC. 6 sec's of recording time is not properly utilizing the ADCRAM. We need a higher A/D sampling speed when sampling a mix of 6 16-bit DACs. I don't know the formula's, I don't have time. But I do have a feeling of what to shoot for. An oversampling speed near 441kHz will yield 2.3s of recording time.
I think now it is time to copy what I have drawn on paper today into a general diagram on MSpaint, to keep my self on pace when I revisit this thread after a month or 2.

What exactly is the purpose of having multiple DACs and feeding the outputs back into an ADC ?

Why not a digital mixer on the FPGA, a single DAC, and no ADC ?

Sorry, that diagram is abit misleading, but I sort of rushed it at the end of the day last night. I copied it from my earlier scribblings to get an idea for FPGA I/O pins needed.

I would like 6 channels for stereo front, center, rear. Or, stereo front, stereo rear, front center, and sub.

The diag shows 6 inputs all from the DACs. That's not correct. 2 DACs will be going to the digital pot, along with an external stereo input.

I see, but what is the ADC for ?

Also, I think it would be easier to use a single RAM. Even with 6 channels, you'd still be able to play over 100 simultaneous waveforms per channel, in less than 50% of the bandwidth. You can use the remainder to update the wave tables.

Let's say you wanted to add a sample of a morphed voice that has been put through a vocoder on the PC. This is a great way to allow one to do that with the ADC and allow them to add anything else they have stored in their wavetables with the DACs.

Also, let's say one has a pure digital mix of 3 or more sin waves in different octaves. You can put the original through one channel, put another copy of the original a few degrees out of phase on another channel, another copy of one a few more degrees out of phase, and so on. Sample that, put it back into the DACs and start the process all over again.
Point being, sin waves are one thing but a complex sampled waveform like a morphed voice is another. Granted it will add noise the more times you do this, but it will be interesting to see just how much noise... Like 2 mirrors pointed at each other, slightly offset, and the sample is in the middle.

Also, with the digital pots, one could simulate attack/sustain/decay/release like the C64 6581 SID, which has been replicated by a Microchip or Atmega device. I will search for it...

Original datasheet Here...
Some links to 6581 projects Here...
There is one other site, dedicated to 6581 replication. I will continue looking...

In the mean time some cool stuff:
1

SWINSID, that's it!
1.
2.
3.

Originally, from bound:
1.
2.

I've been looking at some frequency generator IC's on my spare time the past couple weeks, more specifically the DDS (Direct Digital Synthesizer's) family from Analog Devices. The ones I've been looking at could be used for audio if they were run at slower speeds, but the output range can go up to 400MHz. Automatic/programmable sweep, programmable internal RAM for custom waveforms are but a few of the features of the AD9910. It is a 3.3V CMOS 100-pin QFP Device. This particular device can be programmed at 100MHz with a parallel i/f...

I'm still familiarizing myself with the datasheet, but it looks awesome so far.

Just to let everyone in on a secret, I actually bought all the IC's I need for this audio daughterboard (I have a fear of part non-availability, so when I had the money I bought a few here and there), not including the Analog Devices IC in the last post. I can picture using that one in a car project, one that involves alternate fuel, i.e. hydrogen, created from the splitting of H2O. I've done this easy enough using straight DC already, but when one can approach frequencies into the GHz, I theorize the splitting of +H and -HO becomes much more abundant with much less power in... But that's for another day, most likely in another forum...

The Linear Technology QFN packages worry me. The 50MHz 16-bit ADC LTC2202 is a 48-pin QFN with .5mm spacing. The LTC 6403-1 is a 16-pin QFN with .75mm spacing, it is a output amp that converts a single ended input into a differential output which is needed by this ADC.

OK, I'm going back to the software on the Devboard!...

O.T.: H2O is burned hydrogen though and has no energy in it; and to "unburn" it requires putting the energy back in, and that takes more energy than you'll get out of it.

Continued O.T.: Fascinating isn't it? H2O is burned Hydrogen (does it implode, explode, both?)... Tax free H2O that is... But, I get the flavor of your post Garth. Let me just say I am not one of these "free energy" people. There is no perpetual energy machine, not on planet Earth anyway, however there are machines bound by gravity that are closer to perpetual than others. I am in search of the one that is closer to the "impossible". Resonance and feedback loops are ideas I would like to pursue. In this regard what is the natural oscillation of the H and O atoms? I've read >13GHz. If you can just begin to approach this frequency, it would most likely be more efficient than brute force 2V DC was my only point.

EDIT: I've read today the natural resonance for a Hydrogen atom is 14.5800 MHz, but this is not a molecule of water... Interesting read here.

And, if you would've told someone ages ago that you could break a piece of glass by using a single frequency through a loudspeaker, they probably would have doubted you.

Back to On-Topic Digital Sound Synthesis.

There are no such things as coincidence which is why I post now... Recently I had brief 2 post rant, towards the bottom ofanother thread regarding the 2012 Olympics being held in London. This made some of us think of all that is good with Britain. Anyway, I had further diverged off-topic regarding British music, namely Pink Floyd.

For the D.S.S. board I aim for something similar to what I came across today. One of these devices was actually used by Pink Floyd! It's a digital audio building block made for the PC. Pure digital now though.

This DSS board will have an optical SPDIF output...

I plan to open this unit up and see what IC's they use. The one I've been using for over a year, but only yesterday successfully got the optical SPDIF output working with MP3 audio I have stored on an external USB HDD, is a nice piece of equipment for the money. I bought a separate piece of equipment that strips audio from an HDMI source into SPDIF, but that doesn't seem to work yet for some reason, which is what sort of delayed my enjoyment of optical streaming digital audio.

I haven't opened up this product yet, and probably won't.

I have found a nice family of Audio ICs made by Analog Devices that can take multiple SPDIF inputs and perform 3000+ operations per sample @172MHz. There is software for controlling the audio DSP...These IC's could be 8 channel SPDIF QFP IC's suitable for injecting programmable sound effects into the multichannel SPDIF output stage.

Chuck T mentioned some audio products in another thread. I've not had a chance to check them out yet, but thought I would post here for future reference.