|Eduardo, share with us what transpired since you and Jim first decided to join forces.|
|To be perfectly honest, it was quite a lot harder than expected. Especially the finish. To get the level that we needed (exhales audibly). You realize that the materials, shape and general layout of our amplifiers are not what you're generally used to working with. Chromed brass. Automotive lacquers. The unusual assembly and internal two-story architecture of the chassis. You run into many little unexpected things which, to sort out, take patience and perserverance. (Chuckles.)|
|I remember calling Jim. For six months, I'd been in chrome and paint up to my neck - learning how to chrome brass, figuring out how to repeatably and flawlessly apply metallic lacquers to MDF. I was really upset. (Laughs.) I was really not feeling good. It seemed like a terrible waste of time. Now I'm happy that we've solved these problems. Mind you, it's still challenging. But now we know the rules of engagement. What to watch out for. How the whole process works. We still have to control it very tightly.
Also, we had to develop the whole TimbreLoc concept, from working-but-raw prototype to finished product. How to make it intuitive and easy to use. That entailed so many tiny things. Finding just the right value resistors for the ring of LEDs until they were very tightly calibrated to give the same read-outs from amp to amp; sourcing the proper controls with the right look and feel.
On the first prototypes, the action of those controls was too touchy. The LEDs responded too jumpy. So we had to find the right parts and implement them accordingly. It was a long trial-and-error process. Endless adjustments, tweaking, measuring. So the work was product development. Refinement. Defining how to best put things together, in what sequence. How to get the best results while shortening assembly times. By themselves, these were little things that took a long time. But add them all up and the difference is profound - like the difference between a perfect painting where every brushstroke seems deliberate -- none too few, none too many -- and one that's very competent but not quite it.
|Did you change the layout of the boards at all?|
|Very little. I did change the power supply from the prototype though. It grew a whole lot more quiet. That was important for Jim's >103dB efficient hornspeakers. It's a specific challenge peculiar to going with single-ended vacuum tube circuity and no feedback.|
|What kind of noise reduction are we talking about, Eduardo? Can you give us a figure that quantifies your advances in that arena?|
|(Without hesitation) 10dB. I can now measure 90 to 92dB as related to full power, unweighted! When people measure my amplifiers in upcoming reviews or tests, I expect at least 90dB weighted related to
1 watt. That is very good for a tube amplifier with no feedback, single-ended.
You see, when you add feedback, you automatically also reduce noise. With push-pull, you have inherent power supply ripple rejection that single-ended doesn't offer. In those two respects, single-ended is really the worst-case scenario. Add that we only use AC filaments. There are no DC filaments in this amplifier. That's quite remarkable (laughs merrily) but was achieved simply with the proper choice of configuration. As though we used directly-heated triodes, which we don't. You can use AC filaments if you mind the proper potential, attend to the details.
With the very unconventional topology of our twin series-connected amplifiers in each monoblock -- totally independent from each other including their power supplies -- we were able to achieve these unusual results. As it turned out, the entire circuit architecture really lent itself perfectly to this degree of fine-tuning. I have to confess that I didn't do it on purpose. I realized half-way into this project that it was really conducive to possessing a very low intrinsic noisefloor - if I managed to adjust certain variables just so.
(Shrugs.) Sometimes you must be lucky. You can't think of everything. You do account for new ideas and research, but certain other things are going to happen in the process. These things arise of their own accord. Perhaps like you sometimes write things that weren't in your head until you looked on the page.
|Sometimes? Eduardo, I write like that all of the time. In the dark like a bat trusting its sonar. I rely very much on my subconscious. That it has used its time to mull things over, digest them to tell the story spontaneously when I sit down on the keyboard. Afterwards, I do have to go back with my head. Of course. Tighten things up. Fix obvious errors. Check continuity of thoughts, the flow of the verbal cadence. Still, the initial and main process is very intuitive. I'd be a terrible danger designing electronics!|
|(Giggles.) For good and bad. When they occur, you have to deal with the bad things. The good things? You smile to yourself. You'd like to take credit but know you really can't. It's important to learn why - why did these good things happen? What rules do they follow? How to prompt them, maximize them, exploit them, regulate them. So that's how I work and look at it.|
|Why the brass chassiss?
Why not steel or aluminum?
|Well, I did try to fabricate the chassis from those two more common materials. Brass of course is far more expensive. Jim and I wanted to use the chromed look. Naturally, that ruled out aluminum. Still, I created a few trial pair chassis, of aluminum, steel and brass. It was very interesting. (Gets very animated.) I really noticed the difference in performance. It was not a slight difference, not a merely "different but so what?" kind of affair. It wasn't the kind of very small differential between one amplifier and the next, as is very common with cars off the assembly line. They all exhibit slightly different behavior, some of which can be compensated for with electronic feedback and error correction mechanisms.
With zero-feedback amplifiers, this potential for variability is very much a factor. So, to go off on a little tangent, each amplifier, before boxing, is tested for more than 15 different parameters. The graphs of distortion with frequency versus amplitude, at different points and five different power levels. The spectrum of distortion at several frequencies and power levels. Several verifications of voltages and impedances. You know, it takes about 4 hours to measure a pair of these. (Nearly sounds surprised himself.) I set aside half a day for each pair going out.
Going back to the steel vs. brass chassis, I was sure they'd measure very similar. In fact, there was nothing in the measurements that would have suggested any difference in performance, much less the very specific differences I clearly and demonstrably heard time and again. But don't ask me why! (Chuckles.) I couldn't quote a single convenient scientific reason why brass should sound better. It certainly does. I just don't know why. (Shrugs his shoulders and smiles like a contented Cheshire cat that just swallowed a mouse.)
|What did you hear that was different?|
Would you like to download the current image?