|   How to do it right - but why bother | AllanH
Nov 22, 2001 5:14 AM | | I really don't think that in any home system the solutions suggested below are worth doing, because theoretical arguments and my own experience suggest that cheap and cheerful interconnects and speaker wires are going to be just as audibly good as the most outrageously fussy solutions.
However.
The purpose of any bit of wire is to get a signal presented at one end to the other end with as little distortion, phase error , frequency response error etc as possible. And in a practical environment to reject as much of any interfering fields - RF , magnetic, electrostatic as possible.
For interconnects the best approach would be to use professional screened balanced twisted pair or quad, sourced and sunk in it's characteristic impedance, driven and loaded with differential drivers and receivers . 120-150 ohms is about the correct impedance.
For speaker cables each driver unit in the speaker should have it's own power amp very close by wired to it with low resistance wire - it could be inch thick iron or 12 gauge copper or anything else so long as it's low resistance. The crossover should be done at line level before the input to these amplifiers.
I'll wait for a few responses before expanding on my reasons for why I suggest the above, and also why I can't be bothered to do it myself.
Allan |
| A couple of points... | Monstrous Mike
Nov 22, 2001 7:16 AM | | I don't think it is necessary to use a twisted pair for interconnects. What does the twisting give? A little inductance to counterbalance the capacitance. I don't think this is needed at audio frequencies. A twisted pair, shielded cable costs a little more and is not needed.
And I'm not so sure about your reference to the 120-150 ohms. If this is characteristic impedence then this too is an unnecessary parameter to be concerned about at audio frequencies.
And as far as EMI (Electromagnetic Interference has replaced all terms regarding RF noise and interference) goes, again I don't think the levels can be induced unless the interconnect is extremely long. Using antenna theory, the first significant antenna geometry is a 1/4 wavelength tuned antenna. At 20kHz, a 1/4 wavelength antenna would be over two miles long. The main culprit in audio interconnects is the 60 Hz power signal getting into the line causing a hum. A coaxial cable with low capacitance is all one would ever need in a typical home stereo system. |
| A couple of points... | AllanH
Nov 22, 2001 9:32 AM | | Well I did say it was all rather unecessary.. but I would defend twisted screened pair as it's excellent at rejecting eg 60Hz magnetic fields if coupled with a differential driver and receiver ( generally high quality audio transformers). All professional microphones use this system. And it is about 120 ohms impedance . If you're sending mike level signals over cables a few hundred yards long in eg a football stadium with loads of thyristor controlled big lights, you'd bless the solidity of this approach too. But a waste of time at home, I agree. |
| A couple of answers | audioengr
Nov 26, 2001 12:55 PM | | Twisted pairs - very effective way of getting field cancellation. Often eliminates the need for shielding. Eliminates both emissions far-field and improves susceptability.
You are correct, characteristic impedance is not an important factor at audio frequencies, unless the particular transmission-system uses termination to Zo to eliminate resonance and reflections. Questionable benefits of this anyway.
EMI - I have measured significant differences in audible HF noise in unshielded versus shielded interconnects. Depends on your proximity to sources, such as radio antennas etc.. |
| re: How to do it right - but why bother | mtrycrafts
Nov 22, 2001 3:51 PM | | Your approach to amp each driver and electronic crossover before it has great merit:)
Barrie Haverluck has a 2 part article on this:
Optimizing your Audio System, part 1, Audio Electronics 3/98, pg 16-27; part 2 Audio Electronics, 5/98, pg 30-43.
Short of this, 12 ga is just dandy:) |
| re: How to do it right - but why bother | AllanH
Nov 23, 2001 3:52 AM | | Glad you like the approach - and , unlike all the fuss about fancy interconnects it DOES sound significantly different, and I would say better. Pro reference speakers used by eg the BBC generally use this style of design , and very nice they sound too.
The reasons I believe are as follows :
- Each amplifier is now only dealing with a limited frequency range, so any intermod problems it exhibits have that much less chance to happen. I don't think this is a huge problem area, but it's there.
- It also means that the output of each amp just sees one unit, which will have a fairly straightforward impedance. Crossovers often have highish-Q peaks and nulls in their characteristic which require an amp to have large reserves of over-current capacity to deal with impedance lows which crossovers often present. This avoids that problem.
- It also avoids the use of a high-power passive crossover. These are made up of power inductors and capacitors and perhaps resistors.
- The inductors can easily suffer from saturation if they're not big enough, causing distortion .
- The reversible electrolytic caps commonly used require one of the 2 back-to-back caps in the package to be forward biased half of the time during which period their impedance is highly unpredictable. They're bypassed by the reverse - biased cap, but I still think the overall effect is of a slightly non-linear impedance which would introduce distortion . Perhaps better speakers use large paper or ( haha ) silver mica caps, but I suspect not many do - they're far too bulky and expensive.
- The line-level crossover which would be required is running at very low power levels and much higher impedances, so whether done passively or actively can be of very high quality reasonably cheaply. It also can be easily made adjustable for crossover frequecies and different unit sensitivites fairly easily, so some correction for room characteristics can be attempted.
So overall I think for a multi-unit speaker this is a very worthwhile, though expensive, approach. It does means that a single manufacturer has to integrate the composite speaker/poweramps/line-level crossovers as a block in order to get it right, and though this happens in the professional field I haven't seen anything in the non-pro world. Manufactures tend to specialise in a particular area - only Quad ( I think ) try and do both jobs - and their speakers are magnificent though fussy to drive. |
| re: How to do it right - but why bother | mtrycrafts
Nov 23, 2001 6:43 PM | | You again made some good points but
", though expensive, " :)
A marked reson is eliminating the passive crossovers and each amp can be properly matched for the speaker load and sensitivity to achieve the overall levels deemed necessary.
You may want to find the cited article, not easy, but worthwhile. Your local library can help:) |
| Too simplistic | audioengr
Nov 26, 2001 12:47 PM | | I believe it would help you understand if you knew the following:
1) capacitance is the most important parameter to minimize in interconnects
2) inductance is the most important parameter to minimize in speaker cables.
3) dielectric absorption can cause signal dispersion in both interconnects and speaker cables
Once you take these into account, design of interconnects and speaker cables becomes more difficult and challenging. Simple ZIP-cord for instance, no matter how large the gauge, will not be as good as a well-designed low-inductance cable. See the analysis and measurements on both types of cables at:
http://www.empiricalaudio.com
under the FAQ page |
| Too simplistic | AllanH
Nov 28, 2001 9:13 AM | | All the effects you suggest are measureable - but I seriously doubt if they're audible in any practical home system. Not by me anyway. But then I never did claim the ears of a bat, and don't have the pockets of a Rockefeller...
I note that empiricalaudio use 3-ohm speakers and 16 foot cables
to compare ZIP cord with their wonderstuff. The differences are a small fraction of a dB at 20kHz. I can't hear ANYTHING at 20kHz, never mind 0.2dB differences... And most systems have shorter cables and higher impedances. Further, the CD makers and certainly the vinyl record makers put on next to nothing above 16kHz anyway , so all you get up there is a bit of hiss. Similar comments for interconnects.
I once had dealings with a well known English HiFi manufacturer who were very proud that their output stages were only 3dB down at 40MHz.. that's MEGS. So you cable freaks aren't as mad as some....
Allan.
ps do you have shares in this empiricalaudio? |
| |
