Tip #67 (December 2009)
Audio Signal Attenuators
By Dr. Jonathan Noble
This month, the podium is turned over to Jonathan Noble (see bio note at the end of the article) for an in depth look at signal attenuators. He is about as passionate a DIYer as you’ll find on this planet and appears to share my musical priorities, meaning that in his ears I trust. I’m certain that reading this review article will change your perception of the lowly resistor, traditionally cast at the bottom of the passive part totem pole.
You don’t have to be a science major to develop a feeling for what is going down inside audio circuits – like anything in life, experimentation, experience and good old common sense, goes remarkably far. If you think about it, it’s rather obvious why a volume control is going to be a very critical component in the audio chain. The point of an amplifier is to provide gain, without distortion, and in many cases to lower signal impedance as well. On paper it looks easy, but in practice there is a lot than can go wrong. Phono stage and line level stage amplification is especially critical, due to the low level and delicate nature of the signal. A volume control does the opposite of an amplifier, reducing gain, taking the signal down, and at times back to a level that is lower than where it entered. Now it seems to me that when you drop gain you have an issue, because there is every chance that you will also kill the vibrancy of the delicate signal. Attenuation is potentially ‘problematic’ because the signal enters a lose, rather than a gain, type of situation, and its not surprising therefore that many audiophiles claim (from experience) that the volume control is one of the most critical aspects of a line stage amplifier. In my case, I have preferred to engineer my line stage into my phono stage, with line level amplification occurring at the 3rd and last stage of my phono. I do it this way, because I am a minimalist at heart, and don’t want to introduce unnecessary gain stages, connectors and interlinks into the path, and there is no reason why one can’t achieve line level at the output of a carefully considered 3-stage phono design (footnote 1). In my case then, attenuation takes place a little earlier than is the norm, and after the second triode gain stage inside my own design/build Flamingo phonostage (footnote 2). CD input also enters here, just prior to the 3rd tube stage that acts as a line level amplifier, before entry into the power amplification stages. The comments made below regarding sound are with respect to this configuration.
As caveat, one of the things you loose with a conventional stepped attenuator switch is the ‘luxury’ of remote control. Personally, I don’t like remote controls. I grew up pre-remote. If I want to change the channel on the TV I prefer to walk across the lounge and push a button on the set, ditto for my turntable and amps – but I do appreciate that others feel different. If you must have a remote and a discrete attenuator, as opposed to a potentiometer, then there are various IC based, and relay based options out there. Yet this article will focus on the conventional, mechanical switch, so if the loss of a remote is not going to work for you, then you might wish to Google onto a different audio site! Another option might be to opt for a motorised Alps Blue Velvet potentiometer (i.e. ‘pot’ for short), which is the generic type used in the high end these days – and which sounds OK, but more importantly allows for remote control. And at this point, lets be clear, its not a given that a switched attenuator will sound better than a conventional volume control pot. The wiper blade on a pot runs across a resistive track, whereas a switch has to click in and out of place. When you think about, it’s not clear that a switch is best, because you nevertheless have a ‘dynamic’ contact point, and the click-in, click-out is a complex mechanical procedure, and the more you use a switch the more likely the contacts will scratch and wear, and maybe, just maybe a smooth sliding wiper blade is not so bad after all. I think we can conclude that a switched attenuator requires a high quality switch else one looses the advantage. The one I have right now is the Elma 24-way switch, which is ubiquitous for this type of application, and works very well. The Elma is a Swiss made, high precision switch, with gold contacts, and has a quality feel about it. But of course you can get better still – the infamous Shelco switch, for example, springs to mind.
There are at least 3 types of stepped attenuator – what I shall refer to as a series attenuator, a parallel attenuator, and a shunt attenuator (others may very well use different names). With the series type, the resistive steps are wired in series. If we have a 24-step design, then that means 24 resistors (actually 23 resistors are commonly used, as the last connects direct to ground) from the input side to ground (see sketch 1, where only 6 resistors are shown for convenience), and the stepped output is tapped anywhere in between. The advantage of a series attenuator is that you only have one switch contact in the path. The disadvantage is that all 24 resistors are in the signal path all the time. Normally one would observe that the disadvantage out weighs the advantage, but it’s not so simple, as this type has received a new lease of life with the likes of DACT (which is a highly regarded piece of kit), and the much better priced Goldpoint attenuator. DACT and Goldpoint both take advantage of surface mount technology (SMT) type resistors, which are very small in size, meaning the 24 resistors can be tightly packed in a manner that almost resembles a single resistor. What I mean to suggest is this. Imagine a single resistive tract cut into 24 small pieces, and then soldered together, with switch contact between. If you can implement this neatly – tightly spaced and clean – then you start to feel a whole lot better about the series attenuator approach. DACT and Goldpoint do exactly this, thanks to SMT. However, the main draw back from this approach is that you pretty much have to use standard SMT resistors, whereas with a conventional switch one can explore ‘boutique-designer-HiFi-grade’ resistors, such as Shinkoh, or what have you, which is more likely to give you the sound that you prefer, and as we shall see this is no small advantage.
The parallel type requires double the number of resistors, and there are always two contact points in the signal path, because in each position a new pair of resistors is connected into place (see sketch 2, where only twelve resistors are shown). So we now have 48 resistors in total, and two contact points at each position. There is a lot to like about this approach. Yet naturally, this type costs a lot more, and you do have two switch contacts in the path (instead of one) which is probably not an issue provided the switch is good.
Lastly, there is the shunt attenuator. With this type, the top, or series resistor remains the same at all volume positions, and the switch merely adjusts the other resistors that shunt to ground. So a 24-way switch requires 25 resistors (actually its 23, because the first position shorts to ground, and the last is usually left open, i.e. at infinity - see sketch 3, where only 6 resistors are shown). This is obviously a lot cheaper and easier to engineer than the parallel type, plus one only ever has two resistors and one switch contact in the signal path – which strikes me as being an elegant plan. One also has the advantage of being able to ‘tune’ the sound of this attenuator, by experimenting with different types of resistors in the series position. Since the series resistor is the one that is in signal path as such – I say ‘as such’ because from another point of view everything is in the signal path! – it is the one with the largest influence upon the sound, and it stays there across all 24 positions, means it’s worth ones while to explore exotic options here, Shinkoh, Vishay, Caddock, Charcroft, and so forth. So we have a complete winner then – err, no, not quite, never a free lunch with audio! The disadvantage of the shunt type is that the output impedances (Zout) is mostly higher than what you would have with a regular pot, or with the other types of attenuator mentioned above. The maximum change in impedance with a 100K shunt type, is from 0 to 100K, whereas with the other options the maximum change is from 0 to 25K. On paper this looks like a significant disadvantage, only an important counter point is that if we measure Zout across the more commonly used parts of the switch (in my case between 8 and 2 O’clock), it becomes clear that the Zout does not vary so much, and the variance is not so much higher than a pot, so in practice the shunt type works rather well. The shunt type is the kind that was regularly used in the classic Audio Note designs.
I used to think there is a clear hierarchy of importance when it comes electronic parts – arranging from most critical to the least: active devices (i.e. tubes), signal transformers, signal capacitors, power supply capacitors, and resistors last. Recent experience, however, has seriously challenged this assumption. If you don’t believe, I would recommend swapping between different brands of resistor in the anode path of a common resistive loaded triode stage. If you do this, you will quickly see that different types of resistor impart a very clear, and unique character to the sound. In fact, the change of a single resistor – lets say from a generic metal film to Riken Ohm carbon resistor – can result in a substantial change to tonal balance, which may be better or for worse depending on the particular circumstance. Plate loads are big offenders, because these resistors pull lots of current, and therefore add-in noise due to heat dissipation, and such like, and of course there is also the issue of stray inductance, and residual capacitance, which will work with the resistance to form a filter that can chop highs real nice. But in fact, all resistors, in all manner of positions, throughout a circuit, impart a sound, and one tends to have more resistors than anything else, so the effect tends to multiply. So, in actual fact I would have to say that resistors are every bit as evil/critical as coupling capacitors. The really nice bit about the parallel or shunt type of attenuator is the fact that you can select the resistors of your choice.
Enough of the techno babble – lets move to consider actual use and sound. I have read mixed reports from audiophiles about the effects of attenuators. The general assumption is yes, these are good, and especially so if you want a more detailed sound. Others, however, have reported cold, overly analytical and bass shy sound from attenuators. And a warning light flashes up here, because cold, analytical and bass shy are typical attributes of generic metal film type resistors, and if your attenuator experience is of this ilk, then I would have to ask as to what resistors were used? In any event, some find good quality pots (such as the TKD type) to offer a better, more musical compromise. I am an audiophile who treasures correct tonal balance above all else, and I am simply not happy to trade more detail for less tone. And it is for this reason that I never fully bought into the ‘hype’ about stepped attenuators – that is until recent experience, which has proved just how brilliant these devices can be.
I recently opted for a 24-step attenuator kit from HiFi Collective in the UK – who offer various options. I bought their Elma switch with Takman metal film resistor offering. At zero hour, when soldering this up, I suddenly remembered that I had a pair of 100K Shinkoh resistors to spare, so decided to wire in the Shinkoh’s in the series position. The switch started off as an Elma+Shinkoh+Takman combination, and would have stayed that way were it not for Nick at HiFI Collective who kindly mailed me a pair of Charcroft 100K resistors to evaluate in the series position, as part of my review. The superb sonic reputation of Shinkoh resistors speaks for itself, and I happen to adore what Shinkoh’s do. As for Takman, you can read about these here:
http://www.hificollective.co.uk/pdf/takamn_pdf.pdf. And look, its obviously sales pitch, but something about the seriousness of it all impresses me, plus audio chat forums are full of very positive comments regarding Takman, so I figured this was a good option to try. Having bought this switch, I can only say THUMBS UP FOR TAKMAN, as these resistors are seriously good, and I can recommend them without hesitation. As for Charcroft resistors, these are tipped off to be one of the new champions on the block, and are offered by HiFi Collective at an admittedly high prices – in the region of 24 UK pounds a piece, see here: http://www.hificollective.co.uk/components/car_resistors.html. And I have to say that initially I balked at the idea of spending 24UKP on a resistor, but Nick’s point was clear, “initially, I too struggled to believe, but when you think about it 30UKP, or more, is quite common in the context of audio grade coupling caps, so why not pay similar for a good resistor – I mean a really good resistor”, he said. So let’s agree to approach this matter with an open mind, and let’s get some focus here, as audiophiles will happily spend a small fortune on interlinks, speaker cables and power cords – 24UKP is a joke when placed in this exalted context. So the question is, does swapping out for a single Charcroft resistor perform on a level comparable to a serious interlink upgrade? – you bet it does, but more on that later (footnote 3).
The general feel and operation of the switch is excellent. The 24 steps feel just right, and I have noticed I only really need half of the volume settings on offer – and I never feel the need for an in-between position. I opted for two mono switches, which does not cost much more than a single stereo switch, and the cool bit is that with two monos’ it’s possible to adjust for balance, when required. Adjusting the one side up one notch with respect to the other, is just enough to adjust stereo balance, and in my view, this added freedom is definitely worth having.
OK, regarding sound, lets first look at some of the other options out there:
1) Regular carbon pots sound mediocre at best. This kind of volume control rolls off on top, and adds huge amounts of bass bloat. If one of these sounds right in your system, then I am confident to say it’s masking issues elsewhere – a bright tube stage for example. No, I am not opposed to mix and match, and in actual fact some degree of mix and match is all that can be achieved. But there are degrees, and in my experience, one should try to mix and match within a margin of neutrality. The standard type carbon pots are way off neutral, and therefore, are best avoided.
2) Noble/Alps Blue Velvet pots are OK, but not so great really. These are over priced in terms of what you get, and anyone serious about audio should want to move on and up from here (unless remote control is the ‘deal breaker’).
3) In my experience, Bourns plastic pots are a decent, cheap alternative to Noble/Alps, and this is what I was using prior to the new Elma switch (I also had a Blue Velvet some years back). One way to improve on the performance of a pot is to wire it in a shunt configuration, where typically a 50K resistor (i.e. half the value of the pot) is wired in series with the top of a100K pot. The result is close to that of a 50K pot, one where the wiper blade only features in the shunt-to-ground part of the circuit. I had my Bourns wired exactly this way, with a 50K Audio Note tantalum as series resistor, and this performed fairly OK. But, like many a polypropylene capacitor (but not all), the Bourns pot has a presence region somewhere in the upper midrange – a subtle kind of 'shout', that you won't notice until you upgrade. Highs are not bad, but ultimately lack fine detail, and spatial information is somewhat blurred, and bass is a little fat and lacks true punch down in the lows. In truth, everything is a bit grainy and blurred, yet pretty darn good for the price!
4) HiFi Collective Elma/Takman attenuator kit, which is what I bought. I am blown away. Smiley face! BUY WITH CONFIDENCE. This attenuator kit ROCKS!!! But I am sure there must be other good attenuator options out there.
5) Finally there is the autoformer, or TVC (transformer volume control) approach, as touted by some. Well I am sorry if this sounds cold, but I have my doubts. I tried a TVC in my system, one made by a company that is hugely celebrated by the fans, and I have to say it did not perform well. I drove this device from my line stage, which has a Zout of 1K, and which in my book ought to have been low enough, yet results were rather poor. In an A/B test, my carbon pot was a fair bit better. What’s better? Better is better, I mean better in every conceivable way. Possibly, I had needed to drive the TVC from a lower Z? Fine, but my friend who bought this device tells that he tried every conceivable way to drive it, including ultra low Z options, yet consistently sound was poor. I should think there must a way to get this thing to work, but personally I cant be bothered, and am happy to leave this device to those who think that putting mediocre interstage transformers EVERYWHERE is good for sound – poor confused souls!
So what can one expect from switching to an Elma and Takman style attenuator? Well it’s just a switch, not a complete system over haul. You can expect what can be expected from your existing set up only more of the same, as the system will now enjoy playing with a lot more audio information. I listen via a pair of Dick Olsher’s Platinum edition mk2 BassZilla loudspeakers, driven by a lovely (my own) SE10 triode amp on top (yes, that’s 1.5 watts for mids and highs), and a diy version of a Nelson Pass F5 on the bass (the F5 is fabulous, am so glad I built it). The BassZilla, Platinum mk2, is a brilliant/genius piece of kit. Being a dipole, this speaker sounds a lot like an Estat, only one that also has horn-like dynamics. From the Estat side, BassZilla has the transparency, low distortion, the seamless integration, the snap, and the 3D depth perspective that one might expect from a top quality panel. Only, couple this to the mojo/boogie/pedal-to-the-metal dynamic reach of a horn, and you get some idea of what Dick’s brilliant design has to offer – in actual fact, think Martin Logan, minus that ugly bass discombobularity yet plus hornophile dynamic slam. So in this context, one might expect the addition of a stepped attenuator to allow the Zilla to strut its stuff, with a significantly improved sense of timing and detail, better resolution of texture, more extended and more natural highs, more integrated and detailed highs, wider and especially deeper sense of depth perspective, more nuance, more delicacy, dynamic resolution dialed to 10, and bass with real guts, i.e. punch/feel-it-in-the-chest type bass SLAM! Well truth be told, that pretty much sums up what I hear. Of course all these things were there before, only with the attenuator I can honestly say the combined effect is DILALED TO MAX.
The improvement is quite shocking, to say the least. Frankly it’s amazing how much ambient detail is there on vinyl, and equally utterly incredible to hear just how much of it comes through on my vintage Garrad401/SME3012/SPU front end, especially now that the attenuator is in place. Listening to Bach harpsichord concertos the other night. A superb EMI pressing dated 1974, with performance conducted by Yehudi Menuhin. There is a glowing and bubbling effervescence to the sound. The harpsichord has remarkable presence and clarity, with space between and behind, and a seemingly endless layering of orchestral accompaniment that recesses back into the depths of the recorded venue. BassZilla peels back layer after layer into the sound stage. During harpsichord solos, I can literally hear the sense of a dark and uncanny space behind the instrument – incredible! – and a halo of ambiance all around the locality of the sound. I can reach out and touch the air that surrounds the harpsichord, and I am particularly struck by the delicacy of texturing as strings are plucked, well up and fade. The improvement brought is utterly breath taking. And please don’t get the impression that this device works best with classical program – I have listened from Art Blakey to Zappa, and have to report it sounds brilliant at every turn, plus you can play at low volume and not miss on the sense of life. With low level information now dialed to max my BassZilla can show off in response, and thanks in no small part to the new attenuator switch, which allows the life and pulse of the music to flow. I think you owe it to yourself – get one of these oh so bad devices, but be for warned, you may find yourself listening into the wee hours then struggling to focus as the day’s work wears on!
That’s how things were until the Charcrofts arrived from the UK. Packaged elegantly in a black fold-over pouch, these Charcrofts certainly do look the part (see picture above). I selected 4 favorite LP’s for audition, and played a portion of each in turn before swapping the Shinkohs for Charcrofts, and then a re-listen afterwards to compare – “Brandenberg Concertos” by J. S. Bach on Decca, “Concored” by the Modern Jazz Quartet on Prestige, “Life Cycle” by David Holland on ECM, and “Jazz at the Pawnshop” on Prophone which is not really a musical favorite but is incredibly well recorded and a fine acid test nevertheless. Lets be clear, I am changing but a single resistor in the audio chain. WOW! These Charcrofts are good – really very good. Listening to Bach, I immediately noted a deeper and more focused sound stage, and improvement to timing and pace. I especially hear a significant improvement to micro-dynamic contrasts, and from this a more lively and focused kind of sound, with the Charcrofts. The Charcrofts are beautifully clean and pure sounding, such that the Shinkohs seem a little veiled and slurred by comparison (but of course, Shinkohs are already very good). And, as is the case with the legendary Shinkoh resistor, there is no top lift or over etching in the highs from these Charcroft resistors. The Charcrofts seem to be every bit as refined, as sweet, as organic and naturally integrated as the Shinkohs, it is just that the Charcrofts provide a cleaner window. On Jazz at the Pawnshop, improvement to timing and tempo got my feet tapping, and I could hear deeper into the background – you know, all those clinking tea cups and chattering amongst the audience. I also noticed that the clarinet was significantly more focused. The BassZilla is a dipole. Dipoles tend to give a wet, ambient sound, and at times the sound of a solo instrument can hover in the air, lacking a certain specificity in terms of image locality – err, exactly as you would hear at the live venue! With the Shinkohs the clarinet was big, airy and a little vague, whereas with the Charcrofts the image was more focused, and locked in, whilst retained the big and airy, box less, natural spatialisation one expects from a good dipole. I could also follow the bass line with greater ease and could hear more texture and detail in the lows. The other two LP’s showcased the exact same type and degree of improvement, and I have say I am won over by the Charcrofts, so much so that I now wish to redesign my passive RIAA filter for use with a 100K series resistor, because I wish to replace my current Shinkohs with 100k Charcroft types (footnote 4). I think that comment says it all. Do the Charcrofts provide value for money – of course not (since when did the high-end ever offer ‘value for money’ as such)! Do they improve sound on a level that is worthwhile – you bet your bottom dollar they do.
(1) The best interlink is no interlink, and just calculate what that means – what could you afford for the price of a top of the line Nordost link. Besides, my phono to line interlink is better than Nordost, cos there isn’t one!
(2) Well actually, the attenuator is at the 3rd gain stage, if you count the MC step-up transformer at the front.
(3) Me thinks the average cable fetishists is missing out on half the fun!
(4) Some are spending a good 10 times this price on LCR type RIAA filters, because it is claimed that R’s don’t sound as good L’s – which is why I am now rather keen to try a Charcroft series resistor in my passive, composite, RC based RIAA filter.
Bio for Dr. J. A. Noble
Jonathan is a senior lecturer in theories and histories of architecture, at the University of the Witwatersrand, Johannesburg, South Africa. He holds a professional degree in architectural design (B.Arch from Wits), as well as two research degrees (March from Wits, and a PhD from University College London). Jonathan is currently completing a book on the history of pubic buildings in South Africa, post-Apartheid … hey that’s enough … its my time to speak! … what really matters here is my life in the world of music, which began in early childhood with visits to classical concerts, piano lessons, singing lessons, trumpet lessons, then later playing electric bass in a band – oh, and my wife completed a B.Mus, also from Wits, and is classically trained on the clarinet. And if that weren’t enough, my life in audio began in early teens with envious visits to audiophile acquaintances and audio shows, loudspeaker kits and turntable modifications. A used pair of Quad II power amp (push pull KT66), obtained at next to nothing, at age 14, was the turning point that got me into tube audio. These days I love to spend many a happy hour with soldering iron in hand building triode circuits, moding turntables, and with eyes shut listing to music. Audio always remains something of a passion (my wife will probably say an ‘obsession’), an escape, and a way of fulfilling an incurable love of music.