23 June 2002
Dave's system had lots of changes but the overall character of the sound was very similar to what I had heard some three years ago: sweet, detailed and musical (I guess one always gets what he looks for). As this is a very complexe and out-of-ordinary system, I asked Dave to write about it. He was very busy at the time, so the text has just arrived and here it is:
"My system is definitely a work in progress, and is constantly evolving (and expanding) as I zero in on the sound that pleases me. This evolution, like most is a slow process of making little steps and then assessing if it was a move in the right direction. The bulk of the changes from the last visit have involved minor improvements in power supply design, the switch to direct coupling in place of IT coupling, and moving to a 2 Way speaker system.
The easiest way to put the system into words is to just follow the music from the CD to the speaker.
I really haven't put much effort into my digital front end for some time, one of my first "high-end" purchases was the original giant killer the Audio Alchemy DDE 1.0 after hearing rumblings on how great it was with a few minor tweeks suggested by Arthur Loesch where you bypass the last gain stage and feed a Loesch linestage. This evolved into a unit where I replaced the cap output with first a transformer feeding an autoformer volume control and then a transformer volume control, with the end result being a very non-digital sound.
Attenuation is handled by a magnetic volume control, at this point a transformer is needed because it eliminates the need for the cap on the output of the DAC, if isolation is not needed I much prefer the bandwidth of the autoformer volume controls... in any event from my first few attempts at winding these units, it was apparent to me they were far superior to any resistive units, and 30 revisions later... I am near my final design.
At the output of the volume control the signal is split to feed two different amplifiers. The bulk of the music (150hz on up) is directly fed to a three stage direct coupled amp feeding some custom fiberglass fronthorns with lowther 5A's.
The input of the amp is a 6C45 with an 80% nickel plate choke direct coupled to a AV-8 with various different plate chokes available (80% nickel 49% nickel and amorphous) currently it is another 80% nickel choke due to the balance in sound it gives with the fronthorns. The choice of the choke core in this position has a suprising impact of the flavor of the final sound and depending on the surrounding tubes and iron I can "season to taste". The AV-8 is direct coupled to a mesh 2A3 loaded with some amorphous cored output transformers. An interesting thing I noted was when I was listening to the bigfuns fullrange and I used the amorphous outputs I found the best sound was with the 49% nickel plate chokes on the AV8. Switching to some nickel outputs wound by Michael Ulbrich, I found the amorphous plate chokes on the AV8 gave the sound that best fit my tastes.
The PS of the hf amp uses mercury vapor thyratron rectifiers feeding a CLCLC filter that has the following characteristics:
Small (approx 1 mfd) Oil cap input - amorphous cored filter choke - medium sized oil cap, nickel cored filter choke, and a final oil cap right at the tube. Each stage has its own supply and all three supplied are stacked. This topology uses no coupling caps, cathode resistors or bypass caps, and provides the best sound I have had to date.
The fronthorns are made of fiberglass and were made by Martin Seddon, using a custom expansion designed by Jean-Michel LeClech. It is a true 160hz front horn with a measured -3db point of 80hz and measures 40 inches across the mouth... so they are big. These horns run fullrange using the natural roloff as part of the crossover.
The other half of the signal leaving the volume control feeds the input of a digital X-over to fill in the bass below 150hz and boost the lower frequencies for bass extension. I was never unhappy with the bass from the bigfuns so it seemed a natural next step to use them as basshorns while I explore other options that may be better.
The bassamp was an experimental design for the most recent NY Noise, it uses an input trannie to split phase feeding a differential pair of 6C45's loaded with a 1:1 + 1:1 gapped nickel IT for each tube on a common core. The cathodes of the pairs share a gapped nickel choke which sets the bias. The outputs of the IT's feed the grids of a pair of 6C33's with an autoformer output. There is a negative supply in the cathodes of the 6C33's which has a CLCL filter with the last L biasing each pair of tubes.
The trick thing is the Output autoformer is a surplus toroid 1:1 isolation trannie with extra windings added for the autoformer secondary. I am by no means a "scope jockey", but this amp with an input trannie, IT, and surplus toroid autoformer output measures 25W from 40-70K with 70K being the -1db point -3db on the top is 150K, going the other direction the minus 1 db point is 25hz and -3 at 14hz... going below 40hz the max power decreases due to the limited turns on the surplus trannie I used, and in a revised design of this output I would probably use an amorphous core and design it for full power bandwidth and a slightly lower minus 1 db point. The 10K square wave looks suprisingly well given that most trannies have problems with them, and this amp has 3 trannies... The square wave is perfect to about 6K where a little leading edge ripple shows up by 10K, but it still looks respectable even at 20K. It works great for the bass, but when used fullrange, it sounds like what it is... a very clean uninvolving push pull amp, so sub 150hz is where it stays!
At one point I tried using the digital X-over full range dialing in the "perfect" X-over and adding a delay to the fronthorns to time align the drivers, but the damage done to the higher frequencies by the X-over was awful, the entire system was unlistenable. I also played briefly with the ultracurve, but again the sonic signature was not small and made the system very tiring to listen to... So neither will ever be allowed to get in the way of the music again.
I don't mind the digital X-over on the bass, since below 150hz, there isn't really much damage it can do... and it makes tweeking the points really easy... So I will keep it until I work on a few different basshorn ideas, but ultimately I want to build the X-overs into the amps and do away with any unneeded stuff in the signal path.
In general, I find it's best to follow two simple approaches:
KISS... "keep it simple stupid" the less stuff you have, the less chance you have to screw up the sound.
"The squeaky wheel gets the oil" that is to say, look for the major flaws in a system and address them first... I haven't really even thought of cable, and don't much care to... I look at cable swapping as a last frontier... If my system is so good that the last improvement possible is from trying interconnect after interconnect, then I think I might just enjoy the sound... but until that day, I still have lots of work to do with power supplies, circuits, magnetics, and speakers. Just think... at some point I have to dust off that Garrard 301, and design, build and tweek a phono stage that will sing with the rest of the system."