audio over USB v. optical

**mezz64** · 02-14-2007, 11:53 AM

Ok so I did some browsing through the data sheet for the DAC mentioned above. As far as implementation goes that would without a doubt be pretty much as easy as it gets. However as the OP mentioned the audio specifications are hardly of audiophile quality. Even so, some of that should be able to be fixed with output filters and of course op amps to boost voltage. The application example in the data sheet is fairly complete and a good first step would be to construct first that simple circuit, take some measurements and begin improving from there. I think we are going to end up in a position where the higher quality DAC's be them multi-channel or not are not going to be available with a built in usb interface, and finding a usb interface that simply recognizes multiple channels from multiple dacs, or even a single multichannel dac might be rather difficult, as a custom programmed solution might be necessary. The above dac is most definitely a step in the correct direction though.

**durwood** · 02-14-2007, 01:05 PM

I didn't get a chance to read through the datasheet. I did notice from the basic specs that the outptus were only 16 bit, which isn't a bad thing considering most Cd's are only 16 bit depth, but I would still like to see at least 24-bit. Also, I would like to know what type of digital to analog conversion the chip does. I need to look at the datasheet I guess, unless Mezz did you see it?

EDIT: Upon review of the spec sheet I am a little disappointed in the quality.

I don't think this will yield the quality we are looking for. SNR of 91dB and a dynamic range of 90 dB. So I continue on with my search...

**RedGTiVR6** · 02-14-2007, 01:34 PM

op amps have no bearing on output voltage though....

**mezz64** · 02-14-2007, 01:55 PM

Originally Posted by RedGTiVR6

op amps have no bearing on output voltage though....

That would entirely depend on the configuration, as a voltage buffer, no. But as a non-inverting configuration, yes they absolutely would. Referring to the attached diagram Voltage gain = (1+ RB/RA) where you would adjust RB and RA appropriately to get the desired gain. There is obviously a limit to this, as clipping of the signal will occur if too much gain is applied.

**durwood** · 02-14-2007, 02:02 PM

To avoid clipping, you need to increase the voltage supplied to the opamp to power it. So if you have a votage supply of +/-5V powering the opamp,and if it could output a signal at full swing, the max output voltage would be just a tad under 5V. If you can increase the suplly voltage powering the opamp then the outpust signal can be amplified further.

For example: So that is why when the X-Meridian has full swing output at +/-8V, the output can reach 5V as they suggest. 8x.707~5.6Vrms

So in theory, an opamp cable of full swing output (does not clip when it reachs the voltage rails powering it) is powerd by 12V and -12V, it could almost do 12V output max, with an aproximate RMS of 8V.

Edit: Fixed X-Meridan voltage statements to match their spec sheet.

**Felix509** · 02-14-2007, 02:23 PM

To be honest, I have been on this same quest for quite some time also, for my home and car.. Here is what I have found..

Jitter is the enemy, One way to get rid of it is to use 'bit perfect' over the USB, use a digital receiver chip(PCM2707) to convert USB to I2S Bus(also digital), Have a Buffer stage to eliminate clock error/jitter, then into a high quality DAC, finally, a voltage output stage with analog volume control..

I have already procured 3 USB to I2S boards and will be ordering an 8 channel DAC this week. I am deciding between a Burr Brown and a Wolfson eval board, not sure which, both would be very good..

Wolfson WM8768 (8channel DAC)
Burr Brown PCM4104(4 channel DAC)

One limiting factor is that the PCM270x family of USB audio digital receivers is max 16 bit/ 48KHz.. Unless you want to upsample, this should not be a problem as CDs are 16 bit 44.1 KHz.....

This can all be powered by a computer power supply, +/- 5, 3 , and 12v .

A bit of soldering and lots of time putting things together and it will be testing time..

All the Best,
Lars

**durwood** · 02-14-2007, 02:24 PM

Here's a good link explaining the basics of how opamps can work. It even has a few calculators.

http://www.bcae1.com/opamp.htm

There are other factors that limit output as well. Opamps can easily amplify voltage, but cannot drive low impedance loads. They just don't have the current capabilites by themselves, but combine it with a transistor, you can increase the output current.

**durwood** · 02-14-2007, 02:27 PM

Awesome Felix!

I would go with the Wolfson since it is 8 channels, unlesss you were going to use 2XBurr Brown ones. Keep us posted on this.

I agree with you that the way to go is bit-perfect and to eliminate Jitter. Upsampling should not be needed IMO. That's where Creative Labs runs into problems.

**Felix509** · 02-14-2007, 02:44 PM

The Burr Brown 4104 eval board has 2 chips, it is 8 channel also out of the box..

Upsampling and oversampling were all the rage far quite a while, 24/192 converters were everywhere, although CDs are only 16/44.1. It was said that there was more airyness(sp) or something, better depth, etc..

It seems that the last few years NOS(non oversampling) has become increasingly popular, inluding the DDDAC from Doede(linked above)

I have corresponded with Doede for a while about this subject, he is very much switched on on this subject.. I will be ordering one of his DDDAC 1543 Mk2 for my study/workshop soon, i think it will be a great success..

You can also order the USB to I2S moduled form Doede, based on the PCM2707 chip. the 2707 has a built in buffer already to help reduce the jitter, although Doede reclocks the I2S data between the PCM2707 and the 1543 DACs..

Laters!!
Lars

**Road_Dog** · 02-14-2007, 04:01 PM

Originally Posted by Felix509

The Burr Brown 4104 eval board has 2 chips, it is 8 channel also out of the box..

Upsampling and oversampling were all the rage far quite a while, 24/192 converters were everywhere, although CDs are only 16/44.1. It was said that there was more airyness(sp) or something, better depth, etc..

It seems that the last few years NOS(non oversampling) has become increasingly popular, inluding the DDDAC from Doede(linked above)

I have corresponded with Doede for a while about this subject, he is very much switched on on this subject.. I will be ordering one of his DDDAC 1543 Mk2 for my study/workshop soon, i think it will be a great success..

You can also order the USB to I2S moduled form Doede, based on the PCM2707 chip. the 2707 has a built in buffer already to help reduce the jitter, although Doede reclocks the I2S data between the PCM2707 and the 1543 DACs..

Laters!!
Lars

Lars

I was contemplating Doede's DDDAC 1543 after I found and posted his link. However, I was wondering if the PCM2707 is more than two channels. The tech sheet (http://www.chipcatalog.com/TI/PCM2707.htm) says stereo chip (I assume two channels) but up to 8 HIDs off of USB. Doesn't look to have greater channel input than two over USB.

I am going to continue looking for a 5.1 audio input over USB to I2S chip, unless you know of any? Or is it mythical?

Road_Dog