Crosstalk/Crossfeed Questions
Apr 15, 2006 at 12:38 AM Thread Starter Post #1 of 42
Xerophase

Xerophase

New Head-Fier
Joined
Jan 26, 2003
Posts
31
Likes
0
I have a few questions about crosstalk and crossfeed and how these distortions/features affect the sound of an amp.

First, my understanding of crosstalk is that it is a negative attribute of an amplifier that reduces the soundstage. Crossfeed, on the other hand, is a feature added to amps to reduce soundstage and thus reduce fatigue for some individuals. With one considered negative, and one a feature, what is the real difference? Is the net result the same? Is it simply that crossfeed is something that can usually be defeated, where as crosstalk cannot? Do the effects actually sound different?

Second, is there a threshold for crosstalk past which you can really notice a reduction in soundstage? For instance, I was comparing the PIMETA and PPA/M3 crosstalk benchmarks:

http://www.tangentsoft.net/audio/pim...nch/32-ohm.htm

http://www.tangentsoft.net/audio/ppa.../10ma-bias.htm

Without truly understanding the impact, it would appear the PIMETA has significant crosstalk in the mid to high end of the spectrum. Is this the case, or am I just reading a disparity in numbers that doesn't really matter? If this was THD, are we talking 0.0002 vs 0.0003 or 0.0002 vs 0.3?

Also, when adding a source with its own crosstalk issues, are you compounding the problem? How close to mono are we getting here?
very_evil_smiley.gif


Finally, is there any "general standard" for crosstalk? IE, the average amplifier introduces X crosstalk with which to compare devices? In doing some searching, the only other crosstalk analysis I could find on mainstream hardware is this Anandtech review of an X-Fi sound card, which still appears to be better than the PIMETA.

http://www.anandtech.com/multimedia/...spx?i=2518&p=9

Thanks!
 
Apr 15, 2006 at 1:56 AM Post #4 of 42
Crosstalk is the characteristic that audio from one channel directly bleeds over to the other. Generally crosstalk at low frequencies occures because the power supply can't keep up with the current requirements and low frequency energy modulates the power supply which can be detected in the other channel; but at high frequencies capacitive and inductive coupling in and around all audio and power supply circuits can cause crosstalk problems. Crosstalk in the highs are tamed with good circuitboard layout, grounding and goundplane techniques, and sophisticated local power supply bypassing (which basically means power supply rail storage caps right at each chip or active stage), among myriad other "devil is in the details" engineering techniques.

Crossfeed, on the other hand, is a signal intentionally feed from one channel to another through a short time delay. The signal is intended to mimic the sound path from the left speaker to the right ear or the right speaker to the left ear. Sound from the left speaker reaches the left ear slightly befor the sound from the left speaker reaches the left ear. This inter-aural time difference is the main psychoacoustic cue your brain listens for to tell you how far left or right a sound source is positioned.

So the main difference between cross-talk and cross-feed is the cross-talk signal has no interaural time delay and is unintentional and not controlled. A cross-feed signal has a very specifically designed delay and frequency rersponce intended to model what we hear when we listen to speakers.
 
Apr 15, 2006 at 2:12 AM Post #5 of 42
Quote:

Originally Posted by Tyll Hertsens
So the main difference between cross-talk and cross-feed is the cross-talk signal has no interaural time delay and is unintentional and not controlled. A cross-feed signal has a very specifically designed delay and frequency rersponce intended to model what we hear when we listen to speakers.


Great, thank you very much for the response, Tyll! That definitely answers my first question.

Hopefully someone can still answer the questions related to measurements vs audible impact?
 
Apr 15, 2006 at 4:00 AM Post #6 of 42
Quote:

Originally Posted by Xerophase
Great, thank you very much for the response, Tyll! That definitely answers my first question. ...


You couldn't get a more authoritative response than one from Captain Crossfeed himself!

I'm afraid I don't have the technical knowledge to help you with your crosstalk questions - but, if you don't mind, I will take exception to the way you described crossfeed in your initial post as... Quote:

...a feature added to amps to reduce soundstage and thus reduce fatigue for some individuals.


For many folks, including myself, crossfeed contributes significantly to the enjoyment of headphone listening far beyone merely reducing fatigue. I posted my most extensive comments about the topic in this post.

If you're interested in the science, implementation, and effect of good crossfeed circuits have a look at the info on the HeadRoom and Meier-Audio websites.

Best,
Beau
 
Apr 15, 2006 at 12:01 PM Post #7 of 42
Quote:

Originally Posted by Beauregard
I will take exception to the way you described crossfeed in your initial post as... Quote:

Originally Posted by Xerophase
...a feature added to amps to reduce soundstage and thus reduce fatigue for some individuals.





Yes, about 12 hours ago I still considered crossfeed something "dirty" that I wasn't interested in
eek.gif
, but I think this line from Tyll sums up why I'm interested in it now:

Quote:

Originally Posted by Tyll
This inter-aural time difference is the main psychoacoustic cue your brain listens for to tell you how far left or right a sound source is positioned.


I'm entertained by listening to an electronic song where some noise moves back and forth across my head, but truly annoyed by a rock song in which the cymbals are only played in my left ear.

...what would be really interesting is a crossfeed circuit that let you to adjust the time delay and thus "how far away" you were from the speakers.

I'm still interested in the technical aspects of crosstalk if there are any takers?
smily_headphones1.gif
I've been doing some research on the decibel system and "adding sounds" and I'm starting to think there may not be that much of a difference, even in a crosstalk situation such as the PIMETA's?
 
Apr 15, 2006 at 4:46 PM Post #8 of 42
Apr 15, 2006 at 5:47 PM Post #9 of 42
At one time I had a fine HeadRoom amp that had a crossfeed switch, and now I have a Corda Aria amp that also has a crossfeed switch. IMHO I see absolutely no purpose for this function. Whenever I toggled the setting of these switches back and forth between crossfeed and no crossfeed, I never heard any clear difference in sound or soundstage. Although both of these amps provide wonderful performance, I don't see any purpose for this feature.
 
Apr 15, 2006 at 9:50 PM Post #10 of 42
I have never quite seen the point of crossfeed. I had a conventional blend control some years back and found it of no real use. Blending channels and putting in delay will simulate speakers but why go that route at all when you can have real 2-channel stereo with headphones?

Also why mess up the signals that other engineers have gone to such efforts to make separate and clean?

The business of adding two delayed signals basically means adding phantom channel signals to your phones. You are turning two channels into 4 by adding 2 spurious channels. That is of course what speakers do and what headphones don't.

Sure there are some recordings with excessive channel separation, mostly early stereo recordings, but there are a whole lot more with inadequate separation and any system which blends signals will make these worse.

These set-ups strike me as comparable to the proprietary circuits for enhancing or simulating stereo sound, e.g. my Sony tv has an SRS circuit. They do expand the stereo image at the cost of some fidelity of the sound.

The day I see Stax putting crossfeed in their amps is the day I will have to start taking it seriously.
 
Apr 15, 2006 at 11:16 PM Post #11 of 42
The idea that hearing dual mono reproduction through headphones of a stereo recording is anywhere near what some 'engineer' ever intended is not true, and simply perpetuated misunderstanding.

Dual Mono listening is just as inaccurate as crossfeed is, from a different sonic stand point. By not blending the channels in any way, the dual mono is an inaccurate reproduction of anything except binaural recordings. To say it's 'correct' because you enjoy the separation and clarity misses the point. It is NOT closer to the source. I've never heard anyone argue that you could possibly achieve the result inteded by this so called 'engineer' of binaural material played back on stereo loudspeakers...

Crossfeed is also not correct - its flaws lie in the fact that the delay and frequency responses are not exactly what would occur in air, they are not tuned to your individual head and can introduce distortion. But technically, if the crossfeed implimentation were perfect (impossible in the analogue domain, somewhat possible in the digital), you would be listening more accurately than without it.

You might not like it, but I fail to see how you can argue its pointlessness on any grounds other than taste. Many things in audio do come down to personal taste, however in this case there is significant objective basis to the principle and to write it off seems a little ignorant.
 
Apr 15, 2006 at 11:25 PM Post #12 of 42
The problem is that recording engineers are mixing for speakers, no headphones. All the panning and various effects are deigned to deliver a certain experience when listened to on speakers. When you listen on headphones the acoustic information is not being presented in the manned it was designed for. If you listen carefully for a long time you can hear verious things that don't happen in natural listening on headphones.

But, as you say, many records do not have much channel separation. Live classical recordings most commonly. These types of recordings do not gain much with crossfeed. On the other hand, studio recordings often have wide and/or significant image specificity. To my ears these recordings are made significantly more real and organic sounding by a well implemented crossfeed.

Quote:

These set-ups strike me as comparable to the proprietary circuits for enhancing or simulating stereo sound, e.g.


I'll have to admit that I find the SRS process particularly artificial sounding, but in the same breath I'll say that the Dolby Headphone process I listen to for movies is amazingly immersive. The point is: there's theory, and then there's does it sound better to you. I'll guaranty (and can show you piles of scientific investigation and experimentation by people working on utilizing the technique) that it's a legitimate approach to solving the problem of getting better acoustic imaging on headphones with material designed to be played back on speakers. I think you'll find the long time, well experienced listeners here at Head-Fi split somewhere down the middle on whether they like it or not. What you should take away from that is that it is a technique that has a large enough following that it should be considered legitimate worth checking into for personal preference.
 
Apr 16, 2006 at 5:19 AM Post #14 of 42
Quote:

Originally Posted by edstrelow
I have never quite seen the point of crossfeed. ...


TheSloth and Tyll have addressed your post quite nicely. The opinion you expressed is one of my pet peeves about what frequently seems to be the prevailing Head-Fi zeitgeist in regard to crossfeed - the effect of which was expressed by Xerophase's comment that he'd believed crossfeed to be something "dirty" and not of interest to him until he'd learned more about it.

If you'll allow me the conceit of quoting myself from the link in my last post: Quote:

... I'd bet a month's income that if, say, Sony introduced a new Qualia headphone with traditional single-diaphragm transducers for US$15,000 that somehow produced EXACTLY the same result as the HeadRoom processor, Head-Fi audiophiles would be liquidating their kids' college funds and camping out at the production plant in Japan...


A bit hyperbolic maybe, but I can easily see the rave reviews for such a headphone - "some HF suppression in the center and maybe a little loss of detail but you guys won't friggin' believe what these babies do for the headstage!!!"

Why should scientifically sound attempts to solve in electronics the genuine problem of dual mono stereo be somehow less legitimate - and less worthy of audition - than how different headphone designers attempt to deal with spatial cues within the constraints of the dual mono paradigm? And why should the outcome of an evaluation of crossfeed for an open-minded audiophile seeking improved sound reproduction be any different than the audition of any other factor in the playback chain? With the end result, of course, being a determination of preference.

Quote:

Originally Posted by edstrelow
... The day I see Stax putting crossfeed in their amps is the day I will have to start taking it seriously.


I'm curious, what makes this the determining factor for you?

Best,
Beau
 
Apr 16, 2006 at 6:08 AM Post #15 of 42
Quote:

Originally Posted by Tyll Hertsens
The problem is that recording engineers are mixing for speakers, no headphones. All the panning and various effects are deigned to deliver a certain experience when listened to on speakers. When you listen on headphones the acoustic information is not being presented in the manned it was designed for. If you listen carefully for a long time you can hear verious things that don't happen in natural listening on headphones.



I am more than a little bothered that a Headroom spokesperson would write such material without one iota of factual evidence to back this up. Headroom sells crossfeed as a feature, I would say gimmick, and this quote is nothing more than a sales pitch. Everytime I have seen recordings taking place I see lots of headphones. Sure the ultimate mix is determined by the target listener's equipment but these days that is as likely to be a headphone user as a speaker listener. I have a Beach Boys re-issue that points out that it was mixed for monaural AM car radios.
 
You must log in or register to reply here.

Users who are viewing this thread

Total: 2 (members: 0, guests: 2)
Back
Top