You are reading the older HTML site

Positive Feedback ISSUE 15
september/october 2004


Kennedy on Audio
by Kevin Kennedy


 After a hiatus of many years, PFO is quite pleased to welcome Kevin Kennedy back as a Senior Technical Editor. Kevin wrote for PF in its print form for a number of years, and his departure for full-time work in audio manufacturing and his own business pursuits was lamented by many of us who appreciated his excellent DIY projects and design commentary. It's wonderful to have Kevin back in the PFO community once more… our readers will certainly benefit from his insights, and his obvious love for the audio arts.

Back in the saddle again!

I am thrilled to be back writing for Positive Feedback Online after all these years. A very few of you may remember those early days when I was writing for PF. Some of those early articles may still be found on my website at I don't necessarily recommend you build those projects these days, but they may make interesting reading for those of you so inclined. Hopefully I have learned something (at least!) in the ensuing years...

Since I've been out of the flow here for quite a while, and many of you will not know much about me, let me tell you a little about myself, my system and things as I see them in general. I am given to rambling a bit so please indulge me—it has been about six years since I last wrote an article, and it probably shows!

I have spent roughly the last 15 years designing audio equipment professionally in one capacity or another. I spent most of the 1990s working for Bose Corporation, and while one may debate the philosophical merits of their product design approach, there is no question they have achieved a large measure of commercial success in the face of severe competition from the likes of Sony, Pioneer, and Harman. Perhaps as importantly, they employ a fairly large work force, pay competitive wages and treat their employees extremely well. On the other hand, I will say that a fair number of my colleagues there did not get the tube thing.

Amazingly, I learned a lot about engineering practices at Bose that are perfectly relevant to the design of good high-end audio electronics. I have always believed that good engineering practice is as important to the high-end as are imaginative, original design concepts. There a lot of observable phenomena that the "meter readers" reject because they cannot measure it. In some cases it may simply be that we have not figured out the relevant measurement or correlated audible effects to measurements we do make. Just because I don't understand the root cause of what I am observing does not mean it is not real. Sometimes I will admit to a pretty strong degree of skepticism towards what I see as the "mystical" approach to audio. This is typified more by some high-end audio dealers who have neither the grounding in engineering nor the sciences to understand the gear they are selling. In many cases they feel that they know better than the guy who designed the particular device in question.

I spent a year and half after leaving Bose learning that I am not a particularly good businessman, as I struggled to bootstrap Kennedy Tube Audio (KTA) into a full-time concern. I had some initial successes with products I OEM'ed in Hong Kong. This line included a phono stage, an SRPP-based line stage, and a 20Wrms per channel PP 300B stereo power amplifier. The most expensive product in this line topped out around $1500 with Valve Arts 300Bs. I also had a custom line which included a two chassis stereo 300B SE amplifier with Tamura amorphous core output transformers and custom made Tamura power transformers on copper chassis. These were extremely expensive to make, and although I got a hideous amount of money for the three I built I did not make any profit on them. Each amplifier required in excess of 60 hours of assembly time alone as they were completely hand wired. Sadly, the dealer withheld several additional orders because I could not package them cost effectively for shipping (MONEY) in a manner that fully satisfied him. This, more than anything, is what killed the business. When KTA finally died it was for a variety of other reasons as well, not the least of which were problems distributing the product in general, a serious lack of funding, and a very disillusioned owner. A word to the wise: unless you really want to lose everything you have, keep your love of audio as a hobby!

I ended up working for Fishman Transducers, a small musical instrument electronics manufacturer. While at Fishman I designed a lot of onboard and off board pre-amplifiers for guitars, with a preponderance of them being for bass instruments. I also spent two years developing the first two amplifier products in their current line of acoustical instrument amplifiers.

Preferences, lessons, and future directions

Despite a solid grounding in solid state audio design, I have a marked preference for tube audio, although I prefer a distinctly modern acoustical presentation. I am in a word a detail and imaging freak who really likes well reproduced lower registers. (Bass!) I want full spectrum sound! SLAM! I would rather hear a recording as recorded without sweetening in any form, but the same token I don't want my system to be cold and analytical-sounding either unless the source material is—and my experience is that most isn't. Consequently I haven't made any provisions in my designs for tone controls or other compensation schemes like loudness. I believe in simplicity and strive for flat, extended frequency response. My biggest constraint of course is the budget, particularly now.

I have learned the hard way that there are often differences in the sound of passive components, and that there is usually a reasonable explanation for it. I have run across certain brands of metal film resistors, for example, that sound harsh and strident, and upon investigation discovered that the resistors in question were generating harmonic distortion (mostly 3rd harmonic!) Many popular vintage and not-so-vintage film caps—particularly some Mylar types—sound dull and slow to me. Interestingly, many of these also have high DAs, and yet they remain popular even outside of tube guitar amplifiers where their colorations are justifiable.

I have some exciting new designs I will be sharing with you in this new series of "Kennedy on Audio" articles. My design philosophy has evolved since those early Positive Feedback days. I have, for example, eschewed PP amplifiers in general for DHT based SE amplifiers, which today of course is not even trendy—I won't be raising many eyebrows with that assertion, I am quite certain. I still believe in the merits of stiffly regulated supplies, but acknowledge this is not everyone's choice. Topologies are various and may include plate or grid chokes, transformer coupling, parafeed and conventional SE. My designs these days rarely utilize silicon devices in the signal path, and may even incorporate battery bias where useful.

There is far more DIY activity today than when I first became involved with tube audio, and a lot of this can be credited to the ubiquity of the Internet, which allows like-minded hobbyists to share their knowledge. The fairly universal availability of simulation software on the Internet, plus data sheets and models for obscure tube types, has made it much easier to build something original knowing that it will probably work. My preferred simulation software is the free student/demo version of MCAP8 which is a product of Spectrum Software. Of course, if you can afford to buy the full version so much the better!

This series of articles will be oriented more towards the technically literate individual interested in building it him or herself. One who perhaps like me does not have infinitely deep pockets—or perhaps who does!

I like to design things, and have since my teen years when I hacked together my first hi-fi from salvaged parts. Oddly, this presages where I am today, in that this very early work was all tube and SE as well! In the ensuing years I of course pursued (was perverted by??!!) solid state, and ended up with the usual suspects: AR speakers, Adcom amps, and a lot of rather more mediocre sounding commercial gear. I was never satisfied with the sound quality... Salvation came in the form of a pair of McIntosh MC-30s that needed some work. They were good enough that after repairing them I traded my Adcom power amplifier for them. I have since had several pairs of Maggies, big PP amplifiers to push them, and several CD players and a TT. Today, with the exception of the digital sources in my system, there is literally nothing present that was not available in 1970—or for that matter 1955….

Notes on my system and listening room (under construction)

At this point, I guess some description of my system and as yet unfinished listening space is in order.

My system is located in a modest sized basement room with roughly the following dimensions: 10 1/2 ‘ wide and between 19' long and 24' long depending on where you are relative to the bathroom walls. (There is a 5' L x 6' W bathroom at the back of the room) The room was built specifically with this use in mind. The construction is 2"x4" with 5/8" drywall over ancient concrete. The walls are fully insulated and framed on 16" centers against the existing concrete foundation walls. The floor is linoleum over concrete, and the 6 1/2' ceiling is ancient acoustical tile. No effort has yet been made to control reflections off the floor or wall or resonances in the room proper... and yet the acoustics in this room are about the best I have experienced in my home environment, very lively and intimate at the same time. I have 2 dedicated A/V circuits at 20A, another 15A with GFCI for my bench, and another for general lighting, dehumidifier, etc….

The system as noted is primarily tube-based and currently supports CD, SACD, and vinyl playback. In addition, I have two tube FM tuners should I feel inclined to listen to an FM broadcast, but they are not currently connected. My current system reflects a heavy DIY orientation, as designing and building hi-fi electronics is one of my major hobbies, and I also have a healthy desire to save some money for other important aspects of my life!

I will discuss each source, its relative merits and shortcomings, as well as the electronics, speakers…and perhaps even the cables (if I can remember what they are!)

Vinyl playback is handled by a Thorens TD-125 in a modified plinth. The arm is an SME 3009 MKII which is fitted with a Grado 8 MR cartridge, which seems a good overall match. I have replaced the decoupling bushings in the arm, but otherwise it is totally original. Arm cabling is the stock SME wiring, as is the interconnect cable. I plan to replace the electrolytic supply capacitors in the motor drive shortly, as this TT is now about 30 years old. In addition I plan to have a friend who is a fine cabinet maker build me a solid plinth for it. Vinyl playback is the best I have had to date, although much better systems exist at considerably higher prices. This combination is mechanically quiet, the bass is solid and well anchored, overall timbre sounds natural, the mids are warm and highs are fairly open and airy. Imaging and sound stage are good, too. Overall, it is quite musical, and good at resolving low level detail. Obviously it is not the ultimate by today's standard, but it is competitive and relatively inexpensive by comparison. The Thorens TD-124 and 125 can provide the basis for an excellent vinyl playback system with modern or vintage arms.

CD playback is handled by a first generation PS Audio Lambda Drive driving an Audio Alchemy data link (active cable), which in turn drives an Assemblage DAC-1 with an all tube analog stage. The Lambda Drive has been problematic at times, but all problems so far have been traceable to the same cable assembly which connects the drive mechanism to the main PCB. Two inconvenient outboard power supplies are also used with this setup—one powers the data link, the other the tube electronics in the DAC. This setup addresses many of the early complaints I had about CDs: bass is solid, well anchored, mids and treble are smooth, with little hint of harshness. Quite open and detailed. Sound stage is good, but shallower than I would like. Overall presentation might be considered somewhat two-dimensional at times. Tonal balance is very "right."

Unfortunately despite a low output impedance, the analog output of the DAC has proven to be very cable sensitive, performance is substantially degraded by most cables I have tried. I am using a set of XLO homebrew cables I made from parts purchased from Michael Percy some years ago. There are far better high-end cables like the Dr. Lewis that are more than substantially better here, but that are fairly unaffordable. Were I to do this again I might go with transformer coupling instead of the tubes, as I heard an Adcom changer with Sowter I/V transformers in it that at the least gave this much more expensive combo a good run for the money, and arguably might have been superior to it.

SACD playback is handled by a SONY SCD-XB770 UK version. I use this solely for SACD playback, as frankly the Lambda/Assemblage combo sounds better on Redbook CDs. This model is not sold in North America, or indeed in any 120V territory as far as I can tell, and I am currently running it on a fully isolated step-up transformer. This model occupies a medium price point niche that is not addressed here, that of an audio only, single disk player with some audiophile pretensions. Plans for this player include tapping the DSD output directly with custom LPF and tube based buffers using sub-mini tubes. I am also probably going to install a clock upgrade. I am looking at the LClock XO/ or Guido Tent's latest designs as possible options if I build one. Another possibility would be the Audiocom Superclock or Superclock 2.

Unfortunately I have a very limited selection of SACDs at the moment, but the few I have definitely sound more like analog than good digital, and have superior soundstage to my CD playback system. I am certain that tapping the DSD stream directly will result in a considerable performance improvement. I do detect a certain amount of grunge on some material and am not too impressed with the quality of the analog design. Parts quality is very good, lots of Elna, and Silmic caps. Power supply uses two hefty transformers and is all linear. Voltage regulation is independent for analog and digital sections. Seems like a good place to start for tweaking.

The Phono Stage is based on one of my recent sub-miniature tube designs, and has an outboard regulated power supply. It is designed primarily for MM and high output MC cartridges. I used a pair of 5744 in the front end of each channel because their transconductance is quite high, and the mu of 70 is not shabby either. Equalization is passive and based loosely on Stanley Lipshitz's single stage implementation. The output circuitry is based on a 6021 configured as an SRPP. The pre-amplifier is rather quiet, the noise floor being established by the 5744's. I managed to get the equalization into +/-0.25dB of the target. Gain is around 40dB, which with the line stage is just adequate under most circumstances. As was the case in the past, I still believe in voltage regulation and the power supply provides tightly regulated 300V @ 30mA and 12.6V @ 1A to the filaments.

The Line Stage is transformer coupled, uses Holco based step attenuators and a pair of 26 direct heated triodes in the circuit path. The supplies are all regulated, but unfortunately are located onboard. Transformer vibrations are not resulting in audible microphonics, but I can feel them in the envelope of the left channel tube. I am planning to add some mechanical isolation between the transformer and the chassis. There is not really enough clearance to decouple the socket from the chassis. I had a lot of hum issues early on, but with careful relocation of the output transformers and some wiring changes, almost all noise has been eliminated. Gain is around 4dB! This pre-amplifier features switchable phase at the outputs, as I still believe in the importance of absolute polarity.

The Power Amplifier is an SE design based on 45 DHTs and can be found on my website. The design has not changed at all since 1998, and has not failed once in that time either. Currently I am running UX245 globes and WWII vintage 6SL7. The driver stage is an SRPP which is simple, sounds good, and provides plenty of drive for the output tube. Unlike many SE amplifiers I run mine with fixed bias, as this avoids the use of a large, and probably crummy cathode bypass capacitor. The sound is quite pleasing, and the measured performance is typical for this type of amplifier. It is quiet; despite AC heating on the output tube filaments there is no audible hum. It will deliver full power from 30Hz - 30kHz, and achieves about 1.5Wrms at around 4% THD, which is predominantly 2nd harmonic. Damping factor is about 4, which is about the minimum the JBLs will tolerate for good controlled bass.

The Speakers are completely stock JBL Rhodes as noted above. They are 1960's vintage, and were built near the end of this series. I would guess they are about 40 years old now. I have not even messed with the x-overs as everything is working well. I plan eventually to build outboard x-overs for these, and in the future I may experiment with bi-amplification. The Rhodes cabinet is an early 1950's reflex design for the D130 woofer, and bass down to 25Hz is possible with this setup. The tweeters are the fast and airy 075 annular ring horn tweeter. Hearing these in comparison to my old Maggies and a pair of Silverlines I once had convinces me that in terms of performance, things have not moved nearly as far forward as many would have us believe. The drivers in these cabinets were technically extremely advanced in their day, and even by today's standards are rather high quality. These cabinets will win no awards for appearance even if mint—they are butt ugly—but the sound is another story. The only better sounding speaker system I have heard recently was Edgar horn-based.

The Cables are a hodge-podge of inexpensive commercial cables and veritable antiques. Interconnects range from an entry level XLO cable hand assembled from cable bought from Michael Percy years ago, several sets of pure silver solid conductor braided cables bought on eBay and made by a company that appears to be gone now, cables by Merritt de Jong, also long gone, and some very early FMS Litz type speaker cables that have acquitted themselves well against much newer cables. Eventually the plan is replace most of the interconnects with newer ones, but so far I have matched what appears to be the correct cable amongst the ones I have to each of the various components in my system.

The Music I like runs the gamut, from Techo artists like Crystal Method to Jazz, Patricia Barber and Dave Brubeck, to Telemann and Bach Concertos. I have several hundred LPs, which compared to many audiophiles is not very many, of course. I also have a considerably larger (but considerably less bulky) number of CDs, and a whopping three whole SACDs so far. As I mentioned I am not much of a fan of local FM; however, there are a couple of good college stations, particularly WERS, and several public broadcasters, as well as WCRB which is the local commercial (!?) classical station. Sound quality of the above mentioned stations is quite good, but as most material (with the exceptions of WGBH's and WCRB's live broadcasts) is CD-based, there is not too much one can say about the sound quality.

That gives you a rundown of what I have at this time. As far as the future goes, stay tuned! My next article will feature the 26-based direct heated triode pre-amplifier I mentioned, and how you can build one of several variants on the theme.

Until next time, happy tweaking!