Channel Separation/ Isolation testing Part 3

Regarding this test, we first should revisit why i did it in the first place. The discussion in another thread about SQ and system design took a turn to talk about stage width, and in this discussion we covered topics covering speaker location , disperion patterns, vehicle acoustics, etc etc. This discussion included a big dialog concerning Dual mono amps (and dual-mono system architecture) and whether the channel separation afforded by such designs could be heard and/or thus "contribute" to stage width in an otherwise capable and tweaked SQ system.
Sure i am paraphrasing and dont quite recollect the particulars, but the thread link was posted by Senor Kevarino for your review if you are so inclined.
So it was suggested that testing could be used to prove/disprove the audibility of said channel separation, or "crosstalk", within the amplifier. Some people believe anything above a 30dB L to R channel separation spec on an amp is "acceptable" and we cant tell a difference. Others believe that virtually all modern amplifiers in controlled circumstances behave so similarly with respect to channel isolation that differences are inaudible. Still others think they can tell the difference between amps spec'd at 75dB ch sep vs. amps spec'd at 90dB+ ch sep. I belong to the last group , but the reasons for this are not on topic here.
What is on topic is the audibility of amplifier crosstalk and the measured decibel differences between a group of amps in a controlled experiment. I admit I had preconceived notions going into the test based on previous testing and tuning experiences, but I found some surprises that were somewhat shockingly unexpected. Let's first discuss the hearing accuity of a "trained" set of human ears.
Humans have remarkable hearing accuity. We can discern overall loudness (level) variations of approximately 0.5 to 1 decibel. We can distinguish anywhere from 1 to 3 decibel deviations at any given frequency within the audible (by us!) sound spectrum, and trained ears can easily detect small "dips" and "peaks" in a systems frequency response using reference material. Several sources of info on this are readily available both online and in print and several of us posess a set of the "trained ears"...but this is another topic altogether. Just wanted to mention this as a preface since this test collected results that are within a few decibels of each other, and there is the question of "audibility".
With regard to audibility, this is totally a listener-sensitive aspect....what you hear may be different from what I hear, and vice-versa. Some people would need 5 dB or more of a swing bfore being able to tell a difference. But, keep in mind the following...an increase of 3dB is a DOUBLING of perceived overall loudness...some of these results are DOUBLE the loudness level...and then some.
Let's read on.

The first shock i got was regarding a dual-mono amplifier, namely the DLS A3, as this one I mistakenly assumed the crosstalk would be inaudible based on channel isolation of the physical layout of the amp. YES i could hear it, even tho it was faint, it was still there. The only explaination is that this amp does NOT exhibit complete channel isolation. It is very close, but not total.

The second shock I got was regarding the Sinfoni and the exceptionally poor rt channel crosstalk into the left output. WTF? Is this a mfgr defect or a byproduct of this amp being a purported "bass amplifier" meant to mono-bridge, or what? I repeated this measurement 3 times, and got identical results so it wasnt user error. Hmm.

3rd shock was how clearly audible the worst-performing amps were in this experiment. I knew I'd hear it, but not this easily.

Another interesting note about the worst-testing amp in the majority of these sessions...it is a 4 channel amp. When I tested the FRONT to REAR channel isolation, I discovered that it was suddenly ON PAR with the dual mono amp, in essence, it performed as if it was a DM! WOW. Guess this is the reason there are two power supplies in the amp eh? So, when bridged to two channels, this amp suddenly was the best-testing of the bunch in a stereo config by a 0.1 dB crosstalk margin.

So lets look at these results for each channel all grouped together from best to worst>

1) Left channel w/balance to right
DLS 61.0
Arc 61.1
Sinfoni 61.2
JL Audio 62.4
JBL 63.1

2) Right channel w/balance to left
Arc 60.9
DLS 61.2
Sinfoni 61.5
JBL 61.7
JL 62.1

OK, so the head unit seemingly is doing a decent job muting the opposite channel, or is it doing a crappy job but the signal voltage drop of shifting the balance is making it seem like a little jump. I think the h/u balance control should NOT be used in this test b/c of the possibility of adding other variables. BUT if you are so inclined to ponder...the decibel margin between the best amp and worst amp on the left channel was 2.1dB, on the right channel 1.2dB. I think these small margins are the result of a dropping output voltage with balance shift in the head unit.

The IDEAL situation is to send out identical-level signals to both channels (essentially a MONO signal) and simply remove the input of the channel tested (see first part of thread). Here are those results, again from best to worst performer>

Rt channel w/o rt RCA input
DLS 60.6
Sinfoni 61.1
Arc 62.1
JL 63.6
JBL 66.1

Margin 5.5 decibels...almost 4X the loudness between first and last place!!!

Lt channel w/o lt RCA input
DLS 60.4
Arc 61.7
JL 64.3
Sinfoni 66.5
JBL 67.2

Margin 6.8 decibels!! That is OVER 4X the loudness between best and worst.

Keep in mind we can differentiate best-case-scenario volume diffs of 0.5 dB on overall loudness and you can see just how "audible" this can be. I plainly heard the crosstalk on all the amps, even faintly on the best amp. Ideally, we want these amps to measure ZERO crosstalk and be the same reading as the ambient quiet room level of 59.8dB. I guess the only way to completely isolate the channels and ELIMINATE crosstalk completely is to use two separate amps , each bridged, to drive the left and right speaker independantly of each other.
Some will argue that the poorer ch sep can actually help get a system to produce a center image, and it seems logical at first due to the amp exhibiting characteristics nearing MONO output (both channels the same), but this is virtually not the case , and the topic of other discussions.
Having already done extensive testing in a "SQ-capable" system with some of these amps, I know the differences heard in reference-level systems when a "dual-mono" amp is compared to a standard stereo amp cuz I hear em plainly(please see other threadz). To bring this full circle, the ideal system should not have any crosstalk whatsoever. It should be obvious that dual-mono amps are the closest single-chassis solution to achieving this. A true dual-mono system with separate left and right amplifiers is likely THE most ideal setup. Furthermore, I stand by my previous reports that, during extensive testing, I found the dual-mono A3 to provide a notably wider soundstage compared to other "stereo" amps, and we can now see that channel isolation likely has a BIG part in this phenomenon.