RH 807 SE - HISTORY

It all started with a small SE amplifier based on EL84 output pentodes. The output tubes were connected in triode mode, the second grid being "strapped" together with the anode through a 200-ohm resistor. The amplifier was small but sounded quite well, especially since the power supply was ambitiously powerful (much more capacity than necessary) and all components were of proven quality. Still, 1.5 W leaves you wishing for more power, especially on medium efficiency speakers (88 dB).
One evening, after listening to some SE amplifiers during a visit to a friend who acts as a dealer for various local "small manufacturers", I set down to listen to some music on my own system. Sincerely, I felt that my pet amplifier was lacking power, but not finesse, in comparison with quite good examples of SE amplifiers using the more powerful DHT-s: 211, 845.
If the output tube was used in pentode mode, I could reach 5 W or even 6 W of power, in practice closing the gap in perceived loudness and dynamics on all medium and higher efficiency speakers. However, as all of those who have tried that already know, pentodes and tetrodes sound very, very strident and unpleasant: bad.
Of course, in times long gone, many approaches were developed to control the strident sound of pentodes and tetrodes, mostly overall feedback and ultra-linear transformer taps (the latter a form of feedback as well). But, my little "triode" amp had no overall feedback, and I am strongly opposed to the use (especially indiscriminate) of overall negative feedback: among other reasons, I feel that the "correction" signal is always late as compared to the original signal it corrects, as the loop is quite long.
At the time, I was thinking about some old approaches to lowering distortion, with shorter feedback loops. However, literature always considers triodes and distortion, and it might be possible that a different facet of the whole story was left unnoticed.
There are two main differences between triodes and pentodes: output impedance and efficiency. Triodes, especially the low mu versions, tend to have a low internal resistance (actually, impedance, as it is dynamic resistance we are talking about); low internal resistance yields a good damping factor and better performance, even with smallish output transformers. On the other hand, the efficiency of a SE triode output stage is seldom more than 15-20% - that is one of the reasons pentodes were invented, since the efficiency of a SE pentode output stage is usually closer to 45%. Pentodes, on the other hand, are characterized by high internal resistance, leading to poor damping in the low frequencies and exaggeration in the high frequencies: strident sound.
The RH design principle uses local feedback (the Rfb resistor in the schematics), but the loop is very short and efficient. Actually, the output tube acts like an I/V (current to voltage) converter, while the driver acts like a V/I converter. The purpose of this text is not to "uncover" everything that lies behind such topology, but merely to explain the basics to those interested in building an RH amplifier based on the published schematics. Yet, the main theoretical reason why those amplifiers sound good is the drastically lowered internal resistance of the output tube. One of the main factors determining the factor by which the internal resistance (output impedance...) will be lowered is the transconductance (Gm) of the output tube in question: that is why the EL84 behaves so good, as it has a decent transconductance.
There are always some limitations: in this case, the main limit to output power is not the B+ or power supply voltage, but the capability of the driver tube to push current into the output tube. In order to be a good driver tube for this type of circuit, its internal resistance should be as high as possible (an ideal current source has infinite internal resistance) - on the other hand, such tubes seldom are capable of drawing much current, therefore the driver clips before the output stage.
The development of component values took some time and lots of intellectual "sweat" as everything was first modeled (spice), than simulated, and simulated... until some form of methodology was built, enabling good results in designing RH amplifiers from scratch.
The mainstay of the RH series is a (simulated) 1% distortion at near maximum power: actually, those amplifiers never saw any distortion measurements, but distortions are impossible to hear until clipping - that's enough for me, as I could hear distortion when the above mentioned behemot DHT amplifiers were pushed hard, but not clipped. Of course, the maximum power is further defined by the available B+ voltage - the higher the voltage, the higher the output power.
Another important technical and economic promise of the RH design principle is the ability to use comparatively low priced parts in achieving outstanding results:
1. High quality pentodes and tetrodes are generally much cheaper than DHT's (especially NOS), as well as more readily available;
2. Due to it's low output impedance, the need for a large output transformer is relaxed (the tube has enough "bite" even for lower inductance primaries);
3. Furthermore, due to the higher efficiency of the pentode (tetrode) output tube, less current flows through the primary of the output transformer - which can therefore be either smaller (smaller gauge wire) or achieve higher inductance than an identical transformer used with triodes;
4. Since efficiency is higher, the output tubes need less idle current as compared to the output power, reducing power supply costs (smaller PS transformer, smaller chokes - or higher inductivity chokes for the same amount).
All these funds saving characteristics can be used to the audiophile's (DIYer's) benefit - either through initial savings that can allow him (or her) to construct the amplifier, or enabling purchase of higher quality parts on the same budget. The RH amplifier is so generous, that it will perform well even with second hand parts (i.e. SE transformers from old radio receivers - provided you can match a pair...) - but if constructed with premium parts, truly outstanding results are a certainty.
To cut a long story short, the little amplifier from the beginning of the story was rewired by adding two switches (triode mode/pentode mode - for fast comparison), a few resistors and some wire, in order to be used as the first RH amplifier - model with EL84. The first version used EBC81 triodes/double diodes (the anodes of the diodes and the shield are grounded) as drivers, and achieved outstanding sonic results. The EBC81's were replaced with an ECC81 (half per channel), with obvious changes in component values - representing the first "official" RH-84-SE. The driver valve change was dictated by the Miller capacitance limiting the high-frequency response of the amplifier in triode mode; with ECC81 as driver, the amplifier sounds almost identical in both modes, the differences being mere nuances and favoring the pentode mode in terms of power, low bass and transparency.
The RH-807-SE is a design derivative of the original RH amplifier, born to prove that it is possible to design a great amplifier with any given valve by sticking to the "winning" design methodology. Elvis Rakic, a very good friend of mine, was to experience something completely different and new with the amplifier I designed specifically for his needs (both of us are great fans of the 807 tetrode). You can read his comments filed as "construction" (and he did all the work on building this web-site, as well).
Although specifically designed for the 807 tetrode, the amplifier can actually be constructed using 6L6, 5881 or KT66 tetrode as direct replacements (of course, those valves do not have a plate cap and fit to commonly available octal sockets). Since their electrical characteristics are basically the same, all of them can be represented by one spice model (as used during simulations). Furthermore, the limiting conditions like maximum Ua and maximum Ug2 are at their lowest values with the 807 - and the component values and circuit specs (B+, etc.) are safe for all of the above-mentioned tetrodes. Actually, the B+ voltage can be raised from about 350V to about 500V if you will be using good quality 6L6s or KT66s, resulting in higher output power (or lower distortion at the same power levels). Keep in mind, though, that higher voltage means higher idle current operation, larger transformers and chokes, as well as higher voltage capacitors - the price difference might not be worth the extra power, since the amplifier is more than enough powerful for a thrilling listening experience. Operating your tubes at lower voltage and current consumption will undoubtedly result in extended tube life...
Please, keep on reading the remaining technical sections, before gripping your solders! There is plenty of information and advice waiting for you...

A.K.