Sunday, October 17, 2010

Another day, another twofold (6dB) reduction in noise

Actually this doesn't happen often, that I am able to improve my audio system by a factor of two in one day.  If it did, my system would already be astronomically good, two to the power of over ten thousand. But it is often possible to find improvements that can be made which are real, provably real because they are also measureable, and having an ability to do audio measurements can help find them.

The above picture shows a Harrison Labs 3dB attenuator like the one I am now using now on the output of my Behringer DCX 2496 digital crossover that feeds my midrange amplifier.  I'm using a similar 6dB attenuator on Behringer output that feeds my supertweeter amplifer, where I have achieved more than 6dB in total measureable noise reduction (also including some smaller noise reductions from other factors I'll discuss in future posts).  [Correction: In the following post, I give measurements showing that possibly the entire noise reduction came from attenuation, not from cord changes.]  I think the 3dB may be the optimal limit in improvement for the midrange amp, but I'm planing  to try a 12dB attenuator, when I can get one, on the supertweeter amplifier.

For some time, I have noticed a whiney high frequency noise from the supertweeters (like the right one shown above) when I put my ear right up to them.  I had figured the problem was ground looping (to be discussed in future posts) but ultimately found to my surprise and horror that the source of the problem was that the Behringer DCX is rather noisy for a hifi component.  (And this is after having spent months doing bench tests on the Behringer, which convinced me that it was fairly quiet.)   Since the Behringer both has excess noise and excess output capability for most hifi components, it makes sense to attenuate it for them.

I don't really notice any noise coming from my midrange speakers, the Acoustats.  But anyway I put a 3dB attenuator on the input to the midrange amplifier, so there is a 3dB noise reduction through the midrange as well, even if requires a listening session to hear, not just a quick listen.  As well as being less immediately obvious, the benefit in the midrange is uncertain.  In addition to being objectively quieter, there might also be a slight rise in distortion, which is much more noticeable in the midrange.  I sometimes think I hear a slight hardening of the sound, but it is superceded by a tremendous improvement in depth.  But because it may already be a mixed blessing, I'm not certain I'll precede to using a 6dB attenuator for the midrange once another one becomes available when I replace the tweeter attenuator with a 12dB one.  3dB attenuation in the midrange may be the best compromise option.

Actually, I have argued against using attenuators with the Behringer in a DIYAudio forum.  But that was when I was using a Krell FPB 300 as my midrange amplifier.  No attenuator is desireable for that combination, because the Krell can use the extended high output of the Behringer, and also because the Krell is less sensitive and already suppresses the excess noise of the Behringer.

But while I was making these arguments, I was overlooking the need for attenuation of my supertweeter amplifier.  Attenuation was clearly called for there.  And now that I am using a different midrange amplifier, a Parasound HCA-1500A, some attenuation is useful there also.

Under my loading conditions, the Krell requires 2.6dB more signal input to output the same level as the Parasound.  This means the Krell, relatively speaking, suppresses the noise of it's source by 2.6 dB compared to any THX approved amplifier.  So long as the preamp or source component can provide enough level (and the Krell requires about 5-8V input max) this can be an overall benefit.

Now why doesn't everyone play this game, make their amplifiers less sensitive to make better sounding systems (and sell more amplifiers)?  The answer is standards.  Currently, there is a defacto standard for audio amplifier sensitivity.  It is a proprietary standard, THX.  THX determines the "correct" level of amplifier sensitivity.  They have chose a comparatively high sensitivity level that tends to exaggerate the noise level of preamplifiers.  But anyone like Parasound who wants the THX seal of approval needs to use the THX sensitivity level (can't remember what it is offhand).  John Curl has complained about this, though his amplifiers do often (like my 1500A) have a volume pot on the back.  If I wanted to, I could lower the sensitivity using the volume pot.  I simply hate volume pots because of noise and other factors and will always use a fixed attenuator instead.

So anyway, the context of my old system was the Behringer crossover (with max output around 10V) and Krell amplifier (max input similar to that).  In that context, it makes no sense to attenuate the output of the Behringer, because if you do, you are limiting how much of the peak power of the Krell you can access.  And that peak power is a lot higher then the specified power of 300 watts, more like 550W into 8 ohms (and more into lower impedances).  So therefore don't calculate the max level from your preamp based on the amplifier power spec.  Based on that calculation, the Krell would need only about 4 volts to reach max output.  But if the preamp runs out of steam at 4V it will never access the ultimate peak power of the Krell.  It needs more like 8 volts for that.

In this context, not only was attenuation not good, it was not needed, because of the 2.6dB noise suppression from lower sensitivity.  Thus the Behringer and Krell are a good match without any attenuation.  Attenuating the Behringer in this context either reduces peak power or increases distortion.  In addition to being rather noisy, the Behringer gets slightly more distorted at the higher voltage levels above 4V output.

But the Behringer and Parasound are not a good match.  The Behringer puts out way more voltage than the Parasound will ever use, so that extra voltage is wasted.  And being more sensitive, the Parasound exposes the noise of Behringer much more loudly.  In this context, some level of attenuation is desireable, possibly 3-9dB.  I'm using 3dB now, and 6dB on my supertweeter amplifier, an Acurus A250, which has similar sensitivity.  The supertweeter generally only needs a few watts, so it's unnecessary to access the upper power ranges of the A250, so using lots of attenuation is not a problem.

Why use the noisy/distorted Behringer?  Because it is a miracle box, letting me build a fantastically complex speaker system that could not have worked 20 years ago.  It was designed to cheap semi professional standards, not audiophile standards, but the capabilities of the digital processing it offers is without peer (at least under $5K).  I plan to get my Behringer modified with new audiophile grade circuitry from Europe soon.  That will make it an audiophile grade unit.

I actually blame the THX standards somewhat also.  Given current technologies, the sensitivity and output levels of the Behringer and Krell work out better.  There is no sensible reason why amplifiers need to reach maximum output level with less than 2 volts input.  But historically, lower levels like that were common. In fact, for many years, amplifier sensitivies were around 1 volt.  Why?  I don't know, maybe just because it's an obvious number.

OK this is getting too long and I need to get to yard work before I post pictures, but there is more to come, and other tricks helped get lower noise levels also.

No comments:

Post a Comment