To name a capacitor after those little beasties that fill the hot summer night air with noisesome chirruping might seem counter-intuitive until one recalls the size of a cicada is uncannily similar to an audiophile high-voltage capacitor and one realises how that insect's sound allows its position to be pinpointed (by potential mates) and fills the air.

It transpires that testing capacitors is more difficult than testing vibration control or cables! My transformer coupled test rig did not work as it produced inconclusive results. Placing a CRCR network between line level transformers and using mono to exaggerate the differences did not work either as the brand/type of resistor made as much difference as the capacitors.

To add to the audiophile's rich tapestry of paranoia, and the reviewer's rich tapestry of experience, it transpires that the audible effects of capacitors vary from context to context. In some applications the audible contribution of a cheap generic capacitor is minimal so there is little justification beyond pure audiophile snobbery or lust, to populate hardware with fancy capacitors. That is NOT to say than some ancient polar electrolytic will suffice, just that the difference between an inexpensive metalised polyester capacitor and an expensive audiophile type is not noticeable. Read Part 1 to get some background before proceeding further. In a power amplifier power supply, big inexpensive capacitors are ideal for power amplifier smoothing and storage, bypassed and with further local bypass close to the output stage active devices, while input and driver stages are fed by smaller higher quality types (proprietary low ESR electrolytic or polyprop, not fancy pants low dielectric constant types). However, pre-amplifier power supplies (especially microphone or pick-up cartridge) do benefit from banks of smaller high quality capacitors rather than big individual capacitors (unless there is a stock of genuine Cerafine or Black Gate hidden under the bench) if the circuit is transparent enough. Otherwise spend the money on the coupling capacitors (in valve amplifiers) and bandwidth limiting capacitors.

The DC blocking interstage coupling capacitors are the most sensitive positions for upgrading, with just two exceptions. Exception NUMBER 1 is the RIAA network in a phone pre-amplifier. Passive 2 stage RIAA EQ is the best for sound quality when all else is equal and also the least likely to be upset by capacitor changes. Compound passive RIAA networks and 'active' networks (those that obtain a filter pole with feedback round the active device, be it valve or transistor) should NOT be altered unless you have access to an inverse RIAA network and endless time and patience. Exception NUMBER 2 is the capacitor bypassing the resistor in any feedback circuit. As the feedback loop is using a signal as a distortion cancelling control, any distortion imposed in the feedback loop is amplified by the gain of the amplifier. Hence the very best resistors and capacitors should be used here. It's also a good opportunity to try different levels of feedback before buying those components.

Entirely different problems present themselves to the loudspeaker passive crossover designer. Your old scribe's well known aversion to passive crossovers does not prevent testing of capacitors in this context. Cicada's biggest customers are high end loudspeaker manufacturers so their expertise in this area is established in the marketplace. Part 3 will cover a brief comparison in a passive crossover. Your old scribe so objects to passive crossovers that only one pair of any consequence are allowed under this roof at any one time.

Tom Fung, of Cicada has pedigree. He worked from 1994 for Aeroplax Ltd (distributing German lough Audioplan), then HiVi Research (well known for some outrageous bass drivers) from 2000 and then his CV, in 2003, includes distribution of Mundorf MCAP to the Hong Kong, China and Macau markets. From 2004 Tom has been a consultant to various audiophile companies and working with a Taiwanese company (established in 1980) producing OEM capacitors for European high-end audiophile brands, the decision to market them under their own brand became an obvious move. Hence, one pair of pre-production samples (in the photo above) have the name of a manufacturer deleted with correction fluid!

The factory is located in Taiwan and had exclusively specialised in manufacturing various high quality capacitor types for OEM use. They develop their own plastic film and electrolytic capacitors for industrial use as well as high-end audio manufacturers. Their reason to jump into the waters of branded audiophile component supply stems from their capacity [groan - another capacity pun] to manufacture pure copper and aluminum foil capacitors. While dielectric materials are as diverse as imagination allows, few manufacturers stray beyond metalised coatings or aluminium alloy foils. Hand made examples of paper in oil types lend themselves to this at high prices but bulk manufacture of plastic types with exotic conductors is more unusual. Equally rare among component manufacturers (outside the hand rolled on maidens thighs in mountain top monastery types) is a team comprising music lovers and audiophiles. Hence after some years making components to satisfy external designers' specifications, so applying their years of R&D for others, CICADA finally developed their unique design and construction to high quality aftermarket and DIY capacitors. The old saying "You cannot make a silk purse from a sow's ear" seems to have been applied as Cicada import 4-nines (99.99%) pure copper (4-nines purity is the equivalent of 24ct or 1000fine in gold assay terms) from Germany. This is cold rolled to a 0.005mm (5 microns) thin film, which, while similar to Hovland Musicap construction, is different from ordinary plastic capacitors. Indeed, many plastic capacitors deposit a metal film on the surfaces of the plastic dielectric, whereas the Cicada capacitors utilise sheet materials in the traditional way of waxed or oiled paper types. Cicada use specially designed machinery with four layers winding technology, which they claim perfectly winds the copper or aluminum foil together with the plastic. They reckon to tolerance to standard deviation less than 3%.

The choice of materials means that these are BIG capacitors, the copper foil examples being even bigger than their aluminium counterparts. They are similarly proportioned to variants on paper dielectrics, be they waxy or oily. Only the paper in oil test rivals are bigger than the smallest Cicada copper capacitors and it may be this mass and solidity that contributes as much to their neutrality as any more esoteric design decision.

Amplifiers included the reference hot-rodded Assemblage SET300B and the Canor TP106 VR+ with the B&C DE-400TN-8 compression drivers being used as a reviewing tool for their transparency despite their drooping response above 18kHz.
"What about extended HF effects?" demand plebs chorus, stage left.

Transparency proved to be more important, and results were checked with Focal T120TDX (prototype with aluminium phase plug). Most of the differences noted between the capacitors are in the domain of HF transparency and timing. It is interesting to recall the first serious capacitor test for audiophiles used the legendary Onken tweeter beloved of Japanese tube-o-philes. The capacitors were tested in series 'coupling' positions, parallel anode positions for local power supply bypass, and cathode bypass (bass bandwidth limiting) position.

In the series 'signal path' position (the whole amplifier is signal path, but readers should know what is commonly meant by 'signal path' in the garbled syntax of audio journal speech), immediately the Mundorf connection is apparent by the similarity to recent review items that were fully M-capped. The aluminium Cicada examples have the same etched clarity as the MCAP (a metalised polypropylene type), rather like the tinkly emphasis of many moving coil cartridges. Percussion stick noise is explicit and impressive in a hi-fi manner. They're similar to Infinicaps and Hovland Musicaps in this respect too. However, the Cicada aluminium foil types are generally better than any of those three plastic ranges. The etched quality of the former three is as though an outline of an image is etched onto crystal glass overlaid upon the same image printed opaque on photopaper. The Musicap and Infinicap add emphasis to the outline, or its musical equivalent, the envelope, but is never perfectly registered. The Cicada aluminium capacitors are much less pronounced in this effect, but it is still present, and they are an improvement on the Mundorf in this respect too.

It is much harder work comparing capacitors products than any other audio components. Retaining audio memory is a struggle when changing lead out connections on a jig separated from the circuit by 150mm or so. The Cicada aluminium capacitors do manage scale and depth as well as these plastic film competitors, if somewhat layered. They took an extraordinary time to run in, with a variety of signals but nowhere near the rated voltage of the capacitors. That capacitors have to be run in is probably a revelation to many; one hypothesis might suggest that the constituent layers settle into the least state of tension as the constant slight vibration of the signal disturbs the homeostasis of the original manufactured construction. Whatever the reason, some capacitors take an eternity to settle (Audio Note being equally time consuming) being unpredictably bright after switch-on and taking longer to warm up before they are run in. The physics is hinted at in those aforementioned long forgotten Wireless World articles.

In a power supply, it would need 40 of the large 10uF Cicadas to compete with the 200+200uF Cerafine capacitors, thus making NOS Cerafines look reasonable value! However, the Cicada might have a more effective role as a bypass to a similarly sized reasonable quality electrolytic like BHC. Indeed this turns out to be the case, but very different in character. This arrangement is reminiscent of a cascade of reducing values in parallel (using a 1/1000 rule instead of a 1/100 rule) they are also not justified on price compared with a combination of less expensive types, although a tiny difference can be heard. The cascade of 200uF BHC + 2uF polyprop + 200pF mica would cost much less and be very close to the 200uF BHC + 2uF Cicada while the latter would cost much more.

"So what?" chorus plebs, stage left, "What does this tell us?"

It tells us that sometimes a bank of less expensive capacitors, featuring excellence in the smaller values, is almost as good a a pairing of two top class capacitors (main and shunt) in a power supply. Hence, in the B+ supply of a valve amplifier, good commercial 200uF, bypassed with 2uF foils, bypassed by 0.01uF audiophile caps (or silver mica) will be almost indistinguishable from a 200uF legend, bypassed by a 0.1uF legend. The latter pairing costing many times the total cost of the former and therefore being a lesser engineering solution. The smarter engineering solution is local B+ bypass (anode to ground) by audiophile capacitors of the quality Cicada produce, rather than fancier smoothing caps from the rectifier.

The same becomes more true with solid state applications, where the high voltage ranges of Cicada capacitors are unecessary. Starting with a bank of 64,000uF 35V monsters for maximum smoothing, bypassed by a 10uF aluminium foil 200V Cicada and ending with the smallest copper 0.68uF local bypass across the transistors themselves might be more cost effective than banks of super low ESR (equivalent Series Resistance) types across the rectifier output.

Cathode bypass capacitors (where fitted) are almost as important in determining the voicing an amplifier as the interstage coupling capacitors. The need for fairly big values operating at modest voltages (16V capacitors will often be adequate) tends to conspire against film types on grounds of size and cost. The Cicada aluminium foil 10uF type were tested in such an application and did alter the sound compared to some 16V Black Gates. The Cicada Aluminium Foil types seemed more explicit and forward with previously unnoticed detail.

The trend throughout the electronic circuit tests is that the aluminium foil capacitors are very similar to the best of the competitive aluminium foil and film capacitors like Hovland or TRT Infinicap and sharing many of the qualities of AudioNote aluminium and paper in oil types. remarkably, the sound quality difference between AudioNote aluminium and AudioNote copper and paper in oil is almost identical to the change between Cicada aluminium and Cicada copper foil types. The sample size of just two manufacturers is far too small to draw any conclusions, and it may even be that the foil suppliers are the same rather than it being an inherent quality of these two metals.

The Cicada aluminium foil capacitors (visual analogy approach alert) emphasise detail like a mackie line developer (Acutol) with contrasty black & white film (Pan F), or excessive use of unsharp mask in photoshop. The Cicadas are much better than Solen fast caps, the old TRT Wondercaps (favourites in the late 80s), Multicaps and assorted Wimas; they are better than TRT Infinicaps (there were several versions of these, the test ones were the white 'Type d' and silver SETI ones), TRT Dynamicaps, and Hovland Musicaps. The improvement is in clarity, detail and high frequency timing. They are subtely different from aluminium in oil types like the AudioNote (which I believe is a rebranding exercise of another paper in oil) but have qualities that put them in a similar ball park. The Cicada aluminium are better than most plastic film (polyprop etc) competitors but characteristically very similar to them with their emphasis on detail. All these differences are tiny compared to pick-up cartridges or pre-amplifiers.

The Cicada copper foil capacitors differ from the Cicada aluminium foil capacitors in exactly the same way as the AudioNote copper capacitors differ from the AudioNote aluminium capacitors. The sound is more evenly balanced (less spotlighted than AudioNote silver too). Although the differences between the copper and aluminium types are subtle (scoring only slightly higher differences from each reference using the judging system in only a couple of parameters, but in A-B comparisons the copper wins every time.

The Cicada capacitors work very well but are unlikely to fit under the hood of most amplifiers or CD players. The Macbook Pro, on which this is being written, is thinner than the Cicada capacitors, so it's audio card cannot benefit from an upgrade. Big standalone separates hardware, especially of the valve persuasion, is the most likely beneficiary.

Conclusion

This author has no plans ever to try to evaluate capacitors again. To undertake this task properly is more time consuming than any other previous review activity. Most of what has been written on this subject elsewhere has been based on single experience of capacitor substitution in just one context. Only Ben Duncan and Martin Colloms in HiFi News and Bateman in Wireles World stand out as attempting academic rigour, which becomes easily explicable as soon as such an exercise is attempted. Inevitably visual analogies abound, because we are a species whose sentient data is dominated by the visual and therefore our language reflects this. Reflections may be audible or visible, but thess reflections will be metaphorical.

The Cicada Copper capacitors are definitely preferred over the aluminium versions in critical positions, but the aluminium are available in a wider range of values. They are all MUCH BIGGER than most competitors, and the copper are slightly bigger than the aluminium.

To summarise, in series applications (coupling caps), their character is somewhere between Hovland Musicap or Infinicaps (indeed, similar character to MCAPS), but with greater clarity, and the better types of paper in oil (old school Audio Note copper foil), regardless of circuit voltages (something of a surprise). The Mundorf MCAP connection is obvious in the aluminium foil Cicada voicing and the copper foil versions go that bit better. Frankly these differences are tiny in the great scheme of a whole audio system (they would have seemed bigger in a single A-B test against a singkle reference) but would affect a low component count project more than a complicated circuit. A single valve line stage would foreground these differences, while a sophisticated class AB bipolar transistor amplifier would mask them completely. Anyone considering building an SET or the Raven or an Allen Wright FVP project, might notice the difference.

In power supply bypass, cathode bypass or local PS decoupling they are not sufficiently better than some less expensive types, and while a proprietary electrolytic bypassed by a CIcada is not as good as a Cerafine or Black Gate alone, a shunt stack topped of with a little silver mica or Cicada aluminium serves well enough for most applications. The less expensive Aluminium foil types are indistinguishable from copper in power supply bypass postions and neither are much better than much cheaper types here.

You'll notice that there are some comparisons here between the Cicada capacitors with legendary capacitors that are no longer available new; to be compared with such company is high praise indeed, but also keeps some of the myths in perspective. Old capacitors are not works of art, they are just old capacitors, with all the risks attached to any old engineering component. However, there are also the risks in a market that behaves like the art market, that things are not always what they purport to be. Brand new reputable supplies will always be a safer bet than NOS (New Old Stock), but the lack of retail price information renders value judgements impossible.

[Go to Part II - tweaking passive crossovers]