"What's this?" Demand plebs, stage left, "More audiophile paranoia?
Another audio-thing to lie awake at night worrying about?"
"We're worrying about global warming; about Sunspots; about Carbon emissions; about the ozone layer; about a new ice age; about a Nuclear Winter; and now he wants us to worry about whether we can hear the friggin switches in our stereo systems!" screech the increasingly excitable plebs
"Well, dear reader," replieth humble scribe "Do you know if your hifi might sound better if you replace your selector switches with higher quality items?"
"What? As well as tinkering with cables, support tables and shelves, magic stones and happy feet?" enquire plebs in that quiet tone of voice that bodes of suppressed rage...
Well, dear reader, prompted by the experience of a poor switch connection in the ancient Concordant Excelsior, I ponder the possibility that some switches may have more positions than those of 'ON' or 'OFF'. I wonder whether the OFF-position may be slightly 'ON' with some switches, and whether the ON-position may be slightly 'OFF' with others. Indeed, I wonder, is 'ON' ever truly 'ON' and 'OFF' ever truly off. This implies that there is only ever 'nearly on' and 'nearly off'. Is that shortfall from 100% ever audible?
"What is he on about now?" scream plebs, stage left, now tested beyond
endurance by the ancient scribe's incomprehensible Zen-like ramblings
And as anyone who has ever played an old Fender Stratocaster will know, sometimes it is possible to connect two sources (in that case pickups) by holding the switch midway between detents. This resulted in a flourishing aftermarket in 5-way switches that fitted the cut-out for Fender's switchcraft 3-way.
Just like crosstalk between channels, there can be crosstalk between inputs. Those of you who choose to leave all your system components switched on 24/7 to reduce warm-up times will therefore have a certain amount of tuner output, cdp & dvd power supply noise, cartridge and phono-stage impedance noise, music server psu noise (etc) adding to the ever present background mush that is even already lurking below every music signal in the most purist of single-source systems.
So before we even debate switch audibility, there's an argument that we should switch-off every source or other component we're not actually listening to. Debates about mains noise suggest that these should even be unplugged from our specialist audio-mains-spurs as their on-off switch suppressor capacitors, soft-start circuits, and mains leads & plugs will inevitably affect the sound just as much as we believe that 1 metre of mains hook-up cord will affect the sound. It must; so if you subscribe to the mains plugs & leads affect the sound of their directly connected components, they must affect the sound of the others too as they're all linked at the mains sockets. So in test 1 I left every component plugged in, switched on, live and playing programme. In test 2 only the desired source is connected to the mains, but the other sources are still connected to the input sockets. In test 3 nothing but the desired source is even in the same room!
If we accept the problems that 'nearly on' and 'nearly off' propose, perhaps we have to consider their
nature. Is there any difference between one switch's nearly 'ON' and another's nearly 'ON'?
How does one eliminate differences in earthing (that is, ground or return) arrangements when trying to identify how well a switch disconnects? We are all familiar with amplifiers that reproduce a ghost signal of an unselected input if the other source is left on. This is often caused by earth-loops (ground-loops) in those amplifiers (or pre-amps) that only disconnect the live signal and cluster all the hook-up wires from the input sockets to the switch. Older amplifiers with tape monitor switches sometimes added an extra degree of isolation as they often broke signal and return even when the rotary input selector only switched between + terminals (or 'live' or 'hot').
The ancient Concordant Excelsior I tested has poor connections on most of the positions of the main rotary input selector. Only the Tuner position can reliably be made to connect. This does not matter so much with this pre-amplifier as it also has the bypass selector switch that takes the phono direct to the circuit. Thus line connections are made to the tuner sockets and the Bypass - Low position used to hear them, while cartridges are heard in the Bypass - High position, neatly bypassing the input selector rotary. However, I became intrigued by the intermittent connection of the cd input. My experience has been that switches either connect successfully, connect intermittently or not connect at all. Sometimes intermittent connection is accompanied by noise, usually of a crackly nature, such as accompanies wobbly 1/4" jacks on guitars and amps, and is a popular feature of a band's arrival on stage.
The Excelsior offered a new world of misconnection, due its being 20 years old and filled with dust, fluff and contact corrosion. Instead of noise, the Excelsior cd input is often plagued by distortion. Wobbling the selector shaft changes the nature of the distortion, as well as adding noise during the movement. This is not mild distortion, this is truly horrible. This is not the clean-clipping of fuzz-tone, nor the creamy overtones of an overloaded Vox AC30 input. This is nasty asymmetric non-linear grunge.
Out comes the hand-held scope and the probes connected across the selector switch while a cd plays tones. The distortion varies with time, with switch position, with signal amplitude and probably with barometric pressure or whether there's an R in the month. Some of the time it looks like partial half-wave rectification, like a single diode is passing the signal in parallel with a big resistor. Sometimes it looks like a kink in the zero-crossing point, like an industrial non-audio class B amplifier. Cleaning with delicate solvent spray makes no difference. Cleaning with the potent stuff used for auto-electrics also makes no difference. This switch is unserviceable.
This switch prompts me to ask "perhaps all switches might do this to some degree?"
Enter the DACT CT3 four pole 5 way switch.
"Why 4 pole?" demand plebs, stage left, "Surely we only have to break the live or positive connection, as
no current can flow in a broken circuit and earth or ground is common in most audio gear!"
"Because it is inevitable that the disconnection is not of infinite resistance so some residual current may flow, and this will be proportionately more similar to the desired connection if that too has any resistance, which of course IT MUST HAVE", asserts the old scribe.
I asked DACT to furnish me with a couple of 4 pole switches for this test, which arrived very promptly. This contrasts with the 2 years it has taken me to finish this review! Their customer service is obviously superior to that of your ancient scribe. I purchased cooking quality 4 pole selectors too. Then I realised that this test will involve soldering and resoldering numerous connections, thus negating any outcome validity, because every connection is another variable. Then I allowed this test to meander along for several months of tinkering before I established the best questions to ask when testing an audio switch. No one else seems to have bothered conducting switch audibility tests for audio, even though many manufacturers, from cottage-industries to global giants, have claimed advantages for various magic-switch configurations. Mercury contacts were a typical pseudo-scientific offering in the late '70s and early '80s, for example, and my Transcriptor Saturn turntable had a particularly noisy magnetically activated vacuum encapsulated reed switch for the mains connection, arcing with each make or break.
Then I have an idea. If all the switches, including the knackered old one from the Excelsior, are wired up to a harness of equal lengths from a common input node to the sector, and from the selector to a common output node...no...that will not work if any of the switches have an 'OFF' that is not 100% OFF.
Having also experienced prototype amplifiers become unstable when the length of a capacitor leadout wire is changed on the breadboard, I realise that the test-rig could be a minefield of possible errors that might lead to wrong conclusions. I decide that a common node on the selector side would be too risky, but on the selected side it might be OK. If it is not OK the default result will inevitable be that there is no difference as the less than 100% OFF switch will constantly mask any other changes as it is always a little ON. Thus, if the result of test cycle 1 is "no audible difference", it merely means "no audible difference in test cycle 1: proceed to test cycle 2".
On the source side there will be screw connectors to freshly stripped wire to enable quick changes. A torque screwdriver should satisfy the more obsessive reader that every effort is being made to minimise variables between samples.
Whatever the conclusions of test 1, test 2 will try each switch under test via one position of the spare DACT switch. Thus if all are roughly equal, the DACT switch will mask any differences and no one would be able to justify the extra cost of this switch beyond the £3 or 5€ demanded for a good generic 5-way. It's all a bit of a faff with such tiny solder tabs to play with. I use the same non-directional multistrand nickel hook-up wire for this as it is easy to resolder, being nickel, and has slightly higher resistance than copper so should swamp any resistance variations caused by slightly different volumes of solder on the pins.
Sadly the results are deafeningly predictable. An old generic switch, the worn out Concordant input selector, sounds worst. Doug Dunlop, the Concordant designer, believed that the usual input selector arrangement was inherently flawed. He designed the simpler 'bypass' switch and circuit for the designated principle source for each pre-amp: lp for the Excelsior and cd for the Exhilerant. The Exquisite had a more complicated arrangement designed to make the best of both, hence the much higher cost.
The old Excelsior switch sounds rough, like it is partially half-wave rectifying. A cheap 9v fuzz-box (old skool guitar effects pedal) sounds just like this with an almost exhausted battery and overdrive set to zero. So the sound is not like overdrive (fuzz) symmetrical clipping, but more like asymmetric clipping together with massive crossover distortion (a step in the waveform amplitude at the zero-crossing point).
All this from a mere switch? I am afraid so, dear reader. The generic (I will not name the brand as I imagine they all have similar performances) switch was obviously superior to the old one in the Concordant. You didn't need me to tell you that!
Comparing the DACT to the generic switch is more interesting. The DACT does sound superior. I really mean that the DACT adds to or removes from the signal less obviously than the generic switch. The difference is one of very slight (and I mean VERY SLIGHT) veiling of some of the subtler nuances at the highest frequencies, like the character of gentle stick-work at the centre of a ride cymbal, or a slight catch in the throat during a vocal performance. The difference is akin to replacing a commercial grade coupling cap in a pre-amplifier circuit with a well chosen audiophile cap. Not re-capping the whole pre-amp, just changing one.
More noticeable is a reduction in 'grain'. Again this is analagous to a capacitor upgrade, but this time more akin to replacing the power supply smoothing capacitors with higher quality items. The DACT CT3 selector switch clearly offers advantages both ways: less of what we don't want and more of what we do want.
So is it as much of a no-brainer as a dedicated audio spur or the nifty Imp valve tester I tried recently. An upgrade whose parts cost less than £50 (or 80€ YMMV) and whose installation will be very fiddly in some amplifiers needs careful cost-benefit analysis. If you are building your own project I would say that a component of this quality easily justifies its place in your circuit. You will be able to hook it up out of the box and drop it in as a sub-assembly.
If you have to replace the switch in an old high-end piece of equipment it is equally justified. The worn out switch was probably high quality to begin with and a more modern specialist audio switch like this one may raise the bar further. There have been a few comments in places like Naim forums suggesting that many anachrophiles have cottoned on to this upgrade already, but the most informative post on one of them is from our own Hartmut explaining how to extend the life of Naim's Lorlin switch at no expense.
If you are considering replacing a still serviceable switch on mid-price gear I would be less hasty in spending this much on the input selector. I suspect most 15 year old circuits would give a better bang-for-your-buck by replacing every electrolytic capacitor in the power supply and the circuit. If at that point you still think the design has potential for further improvement, the internal wiring and selector switch should be next.
Finally, I did try A-B testing (yes I know we normally eschew this procedure at TNT-audio as it exaggerates differences
at the expense of more considered evaluation, but that is precisely why I did it this time) of the DACT against another
MUCH MORE EXPENSIVE SWITCH that is not specifically an audio type (and not the above mentioned Shallco either), but does boast audio credentials like anti-arcing design,
break before make, much higher current capability than an audio switch ever demands, minimised mutual inductance... blah, blah, blah...
and the DACT specialist audio selector switch was indistinguishable from the high-spec switch. The DACT is about half the price of the other item, demonstrating that good engineering concentrates on the relevant
Conclusion 1 is that if we have vintage gear (anything over 10 years old) we must establish that the switches are in good order, if not, replace them now.
Conclusion 2 is that input selectors must break 'earth' (or 'negative' or 'ground') as well as 'plus' (or positive' or 'live' or 'hot') to avoid breakthrough, hum-loops (ground-loops) or distortion. This reduces or eliminates the need for shorting plugs.
Conclusion 3 is that even switches in as new spec may sound different!
Conclusion 4 is that a further test is needed to establish whether an audiophile switch is actually audibly different from another brand of audiophile switch.
I am not saying that DACT switches are the best in the world (and if they misquote this sentence I'll marmalise them) but I am saying that DACT selector switches are significantly and audibly better than generic switches advertised as being of high quality.
If you have high quality vintage gear they will be worth the cash and hassle. If you have high-end modern gear with unidentified flimsy switches, they will benefit too. If you are building your own project with any pretension to greatness there is no question that you should be using switches of the quality of DACT CT3.
© Copyright 2008 Mark Wheeler - www.tnt-audio.com