John Broskie's Guide to Tube Circuit Analysis & Design

28 July 2006

I had hoped to post at least once from the East Coast, but never got the chance. Most of my computer time was wasted trying to access my email from a succession of phone numbers (dialup! I’ve been so spoiled by DSL.) In addition, the “Nothing Ships Until I Get Back” Sale was a big hit – so big, in fact, that I was kept busy with the orders for the whole week since I got back. (Thanks for the interest and congratulations. As I put together the packages, I kept thinking about how many new Aikido fanciers were about to be created.) So while a number of topics have been swirling around in my head, this is my first chance to flesh one out. I’ve chosen the one that tickles me most: Einstein’s amplifier.

Don’t be confused. A year and a half ago, I labeled an amplifier topology the “Einstein amplifier.” I chose that name partly because the amplifier shifted the usual frames of reference, creating a single-ended amplifier that looked like it was standing on its head; but mainly because the design was born of an Einstein mental algorithm, wherein instead of fighting your biggest problem, you make what had been the problem the first principle of your new system. But now I’m not talking about that “Einstein amplifier.” What I’m talking about here is Albert Einstein’s actual power amplifier.

The real Einstein amplifier

Click on image to see original size

Einstein's amplifier
No, Albert didn't design this amplifier; he enjoyed it. Albert Einstein was a dedicated, even ardent, music lover and he played a mean violin as well. Music was easily as important to him as math and physics. (Einstein also loved to sail.) Einstein had excellent taste; for example, he was a huge Mozart and Bach aficionado and, like all right-thinking music lovers, he couldn't tolerate Wagner*:

"I admire Wagner’s inventiveness, but I see his lack of architectural structure as decadence. Moreover, to me
  his musical personality is indescribably offensive so that for the most part I can listen to him only with disgust."
                                                                           —Albert Einstein quoted from Einstein and his Love of Music

So, what sort of audio system did Albert listen to at home? The partial answer comes from reader Allen:


   I've been reading your journal for almost a year now and have found it to be a fantastic source of information.
I haven't yet built any tube projects myself, but once I gain a deeper understanding of how the circuits work (I don't like to just blindly copy things), my desire is to build a nice amplifier to accompany my ever growing collection of LPs. (Just over 1600 at last count.)
   In the recent issue of "Make" magazine (Vol. 6), they had a back-page article about a home-built tube amplifier based on a one-off design done as a 70th birthday present for Albert Einstein.
   To summarize the article, it was designed and built by Jack Rosenberg in 1949. Rosenberg was an electrical engineer working on the earliest digital computer projects at Princeton, as well an audiophile. It was known amongst his colleagues that he built his own home amplifier and it was judged to be far superior to commercially available models.
    Erwin Panofsky, an art historian at Princeton and a friend of Einstein's, heard about Rosenberg's amplifier and asked Rosenberg if he would build one for "The Professor." He agreed, and along with Oppenheimer personally delivered the amplifier to Einstein at his home.
    I'm presuming that up until the recent home-built project, the design was not built again as the schematic was not published (it remained in Rosenberg's personal papers until recently). Also it wasn't printed in the magazine. But after several requests on the magazine's online blog, a JPG of the hand drawn schematic was made available.
   I've been studying it, but am afraid my knowledge of audio circuits is still limited as I am unable to figure out how it works. The audio signal itself is injected into the circuit at a very late stage and only worked on by a single dual triode tube. Also at that triode, the signal is applied to the plate on one side, and the cathode on the other, with the output signal drawn off one grid of the triode. Yet it appears there are multiple tubes amplifying the power before it is (directly) applied to the signal.
   I would appreciate any pointers as to how to understand better this circuit. And, perhaps it could be used as fodder for a future Tube Cad Journal entry.

Thank you.

--Allan B.


The topology
The Einstein amplifier is not as complicated as the schematic above suggests. First, 40% of the schematic details the preamp circuitry, which few would wish to copy today. The remaining portion (devoted to the power amplifier) is interesting, but not difficult to follow. This amplifier is a slight variation on the famous Williamson amplifier, consisting of a grounded-cathode amplifier feeding a split-load phase splitter, which then offers its balanced output signal to a simple differential amplifier, which drives the directly-heated triodes in a push-pull output stage. No feedback loop closes around the amplifier.


The only twist is found in the addition of the RC (the 1M resistor and 0.1µF capacitor) filter at the bottom of the split-load phase splitter. Not a good idea. My guess is that it was added in an atempt to re-balances the phase-splitter's output, which was thrown off because of the 1.8k cathode resistor (the 1M resistor is effectively in parallel with the bottom 47k resistor). A better ploy would be to ommit the 470k resistor and the 0.1µF capacitor, while moving the bottom output to below the cathode resistor, as shown below.

I like the 6SN7s and the DHT push-pull output stage (the 6B4G is a 2A3 type tube with an octal base and a 6.3 filament) and the absence of a global feedback loop and the use of octal tubes; but nothing really fresh or inspired stands out. Not that this is bad amplifier design, it just doesn't appear that novel today, maybe it was more so in 1949. Although I would be surprized if it was. In fact, below are some other designs amplifier designs circa 1950, which illustrate how similar amplifiers were being designed at that time:

The above amplifier is from Audio Engineering (1949) and the amplifier below is from High Fidelity No 48, a book from the Gernsback Library (1953).


The first amplifier used a feedback loop, the second didn't.

Nonetheless, I am sure that with quality iron, good tubes, and good coupling capacitors, the Einstein amplifier would sound quite good today. Therefore, I wouldn’t tweak/modify/improve/fix anything on the amplifier or leave anything out; instead I would build it straight off the schematic. Why? Because that way you would have an excellent conversation piece. Einstein’s tube power amplifier, what could be snazzier? Definitely, be sure to pick up a copy of Make magazine (Vol. 06) and look at the beautifully built amplifier on page 192. Mercy.


Make magazine
Speaking of Make magazine ( Prior to getting the e-mail from Allan, I had never seen it; now, I own the first six issues and I plan on subscribing soon. I recommend that you do the same. Why?

“Now we've got geek DIY (do it yourself) porn. Just as would-be Emerils pore over lushly illustrated cookbooks with recipes involving hard-to-find morels and complicated instructions for roux, Tom Swift wanna-bes are devouring MAKE.”
                                                                                                                                                                  — Steven Levy, Newsweek

Make is the best magazine ever published aside from Sound Practices.
                                                                           — Jeremy Epstein

“DIY porn” was an apt description. Make magazine is devoted to existing mechanical and electronic hardware hacking and to new DIY designs. It is a slick effort, unlike Sound Practices and most underground magazines; it is sumptulously laid out, with high-quality color photos and professional illustrations, and almost no ads. No, it's not a cheesy, ad-ridden, weak-kneed magazine that will end up in the recycle box an hour after receiving it; nor is it cheap at $14.99 a copy, but it is worth it. Really, it belongs on your bookshelf next toissues of Audio Technology, MJ and Morgan Jones's books. Its motto is “Technology on your time.” and the magazine sells T-shirts that read, “If You Can't Open It, You Don't Own It” and “Void your warranty, violate a user agreement, fry a circuit, blow a fuse, poke an eye out...” However, it is not an audio magazine. First of all, its editor and writers are not over 60 (nor is anyone obviously anal-retentive, twitchy, sycophantic...). Many issues of Make may come before an audio-related, let alone a tube-related, article shows up again. Follow this link to see the current issues table of contents.

Aikido-inspired amplifier for Einstein
If I were asked to build an amplifier for Einstein—what a great honor it would be (the only person alive today who would justify a similar request and prove such an honor is physicist Craver Mead)—it would look something like this:

It's Aikido time! Three audio Aikido techniques allow this amplifier to sidestep power supply problems (only two are shown in the above schematic). The input stage consists of a grounded-cathode amplifier with a symmetrical loading of the same tube and cathode resistor value. This stage then cascades into the split-load phase splitter. Because the first stage evenly divides the power supply noise, the phase splitter passes this noise equally between its two output phases, which means that the following differential amplifier will reject most of the noise because of its inherent CMRR. Additionally, the first two stages work in opposite AC signal current phase, so that the simple RC filter at the B+ has much less work to do, as it does not see the audio signal. Second, the differential amplifier finds an active cathode load in the bottom triode and its cathode resistor. This triode effectively equals a medium-sized cathode resistor. But as the triode’s grid samples some of the B+ noise, the power supply noise at the differential amplifier’s outputs can be nulled. The last Aikido technique is to introduce a small amount of B+ noise to the cathodes of the output stage, via an additional capacitor, which will null the noise at their plates.

For output tubes, choose 2A3s or 6B4s or 300Bs. I would use a rich class-A idle current and take the sweet sound over the lost watts.


Next time
I know that I promised balanced-output DACs and tube stages sometime ago.





*Actually, Einstein was swimming against the current here. Wagner was once universally held to be the apothiosis of musical development. Here is an excerpt from Modern Composers of Europe, writen in 1904 by Arthur Elson:

"In the latter part of the nineteenth century the musical world was overshadowed by one of the few great geniuses that the art has produced, —Richard Wagner. It seemed almost as if he summed up in his works the whole range of musical possibilities—almost as though he had reached the Ultima Thule, beyond which no further progress was possible. He towered above his fellows like a veritable giant, and he stood as an apparently unapproachable exponent of the school which he himself had founded and forced upon the public".

In other words, since Wagner had reached the top, it was amazing that anyone else should even try to compose after him. Sort of reminds me of all those Noble Prize winners in literature whose names no one can remember today.           Copyright © 1999-2006 GlassWare           All Rights Reserved