Topic: When hardware adds dither...what happens?

[Nick's Picks]

Hi all....

i've been thinking about hardware that adds dither upon playback. Lots of CD players, DVD decks, DACs and digital processors do it these days. They also typicaly upsample from 44.1 to 96k or 192k, making it hard to understand what you are hearing, or what "cause" is responcible for the effect.

I got to thinking about this as I just recieved (and am very impressed with) my DIP upsampler. It takes any signal, but lets use redbook for the example...any way, it takes redbook and adds dither so that the output is 24bit. In the manual, it says that these extra 8bits are just zeros. As the name suggets, it also upsamples to 48 or 96kHz.

So I started wondering about other pieces of hardware I've had or heard that do this same thing.
When you have a 16bit source, and pass it through a chip that adds dither, is it also just padding the extra bits with zeros? Or are there any products out there that actualy do something to the data here...like throw in a 16bit DAC stage and then re-clock at 24bit from that analog signal ?
Just a thought.
I wonder what the Bel Canto DAC2 does in this regard..as well as other high-end products. I think i'll start emailing some of these companies and see what they provide for an explanation. Fluff and glitter, most likely.

One thing for certain, I'm a fan of hardware dither/upsample. I dont know what it is exactly that is happening (other than the upsample) that makes such a difference, but a difference there be...none the less.

back to the DIP upsampler. I have not done any real test with it yet, but what I hear out of it is an immediate improvement. This could be another short money "must have" for anyone listening to digital sources...especialy CD/DVDs out of a less than ideal transport..such as myself. The differenec in my 24/96 DVD discs is quite "in your face".

[dmonterisi]

there was a thread over at Oade where Doug made some comments regarding certain playback gear that did a nice job of upsampling and increasing bit depth...i'll see if i can track it down.

[MattD]

When you have a 16bit source, and pass it through a chip that adds dither, is it also just padding the extra bits with zeros?

Dither is random noise added to the lowest bit (bits?). Padding with zeros is not dithering. However, any calculations that are done at a higher bit-depth than the original data will benefit from the extra space. I would assume that's why your unit does this - so the resampling is using the extra bits for calculation purposes.

One thing for certain, I'm a fan of hardware dither/upsample. I dont know what it is exactly that is happening (other than the upsample) that makes such a difference, but a difference there be...none the less.

You could prove it to yourself by recording the output of this box (unless it's analog output only). Then you'd have a pretty good before/after comparison of what is going on.

I'm hesitant to bust out the math because Damon might moderate me. 

[dmonterisi]

only if the math offends me.  go ahead and roll the dice...