Crosstalk measurements for DACs?

[bachatero]

Why isn't stereo crosstalk measured for DACs? It looks like crosstalk might be a great test suite addition, considering how fine circuit board traces/components can get, and so it's possible that crosstalk on cheap devices can result in an audible result. Or maybe not? It's hard to know because it's almost never measured here.

 

[fpitas]

It might be interesting. I'll note that recordings seldom have better than 40dB channel isolation, so it's an easy bar to meet.

 

[bachatero]

I'll note that recordings seldom have better than 40dB channel isolation, so it's an easy bar to meet.

True, but this might actually be harder to meet than SINAD. Consider some instrument like a guitar or violin. The fundamental frequency is only a small part of the final product, so say it's something like 5 or 10 dB SINAD because of all the harmonics. That makes a small difference in THD due to the signal chain less audible than it would be if you were just listening to a single test tone with 0 harmonics due to how our hearing is logarithmic.

Therefore, assuming a stereo separation of 40 dB, the bar to reach for audibility is a lot lower than it is for SINAD which lends credence to the idea that stereo crosstalk may be an audible imperfection of DACs, and therefore should be in the standard test suite. However, this is all assuming you've got 40 dB of separation, so if it's lower than that then crosstalk won't matter as much.

 

[DVDdoug]

I remember the analog days when 20dB was good enough (and about as good as you can get) from a phono cartridge. That was better than we can hear under any "normal conditions". Your brain tells you the sound is coming from the louder side.

They'd publish the specs for everything else, but it never concerned me because everything else was way-better and always good enough.

 

[RayDunzl]

My DAC spec:

Nothing to see here...

 

[fpitas]

considering how fine circuit board traces/components can get

It's a matter of impedance, and whether you put ground between sensitive nodes.

 

[MAB]

Why isn't stereo crosstalk measured for DACs? It looks like crosstalk might be a great test suite addition, considering how fine circuit board traces/components can get, and so it's possible that crosstalk on cheap devices can result in an audible result. Or maybe not? It's hard to know because it's almost never measured here.

I stopped paying attention to crosstalk specs many decades ago when vinyl and cassette tapes were the source. I pay even less attention to it now. Can you hear crosstalk in any device you have ever had?

 

[RayDunzl]

Can you hear crosstalk in any device you have ever had?

Yes! My brain!

Well, sort of.

Play 440 and 445Hz through left and right speakers, and hear the 5Hz beat frequency in the air, as expected

Play 440 and 445Hz through left and right headphones. Left ear gets 440Hz (only) right ear gets 445Hz (only).

No beat in either ear.

Hear the beat manufactured in your head, like built-in crosstalk,

I was surprised by that result.

 

[MAB]

Yes! My brain!

Well, sort of.

Play 440 and 445Hz through left and right speakers, and hear the 5Hz beat frequency in the air, as expected

Play 440 and 445Hz through left and right headphones. Left ear gets 440Hz (only) right ear gets 445Hz (only).

No beat in either ear.

Hear the beat manufactured in your head, like built-in crosstalk,

I was surprised by that result.

It is surprising. Thanks for reminding.

 

[DonR]

I stopped paying attention to crosstalk specs many decades ago when vinyl and cassette tapes were the source. I pay even less attention to it now. Can you hear crosstalk in any device you have ever had?

FM radio.