Vacuum Tubes Part 2
Note: Part 1 can be found here.
For many years the bible for tube design was the RCA Receiving Tube Manual. It gathered into one convenient volume the data for hundreds of tubes and it was indispensable whether you were designing a tube circuit from scratch or modifying an existing one. These days, you can find quite a bit of information on the web and that's good... except that it's hard to “browse” the web the way you can flip through the pages of a book. Something is lost when you let others decide what you'll see on their website. Many worthy tubes get lost in the shuffle and even in cases where the information is online, the format of the web doesn't lend itself to page-flipping. For those who want to go back to the source, the edition to get is the RC-30. For browsing, there's nothing that beats a hardcopy, but you can download a copy and wallow in hundreds of pages of data and charts on tubes, tubes, and more tubes.
So why would you want to look for random tubes? Why not stick
with the tried and true? Because there are perfectly good tubes that you might
skip over if you didn't know they were there. For example, the 6FQ7/6CG7 was
originally designed to be used as an oscillator in black and white television
sets. That didn't stop Conrad Johnson from using it to design award-winning
audio circuitry. The 6DJ8 was designed for TVs too, but it went on to enjoy a
glorious second life as one of the go-to triodes of the tube renaissance.
But let us suppose for the moment that you're not necessarily
interested in re-purposing neglected tubes. What tubes would you want to
consider as candidates for, say, a preamp? It turns out that you can cover a lot
of ground with a half-dozen tubes:
Mu Plate voltage Plate
5W (7.5W total, both units operating)
The numbers in parentheses after the 12AX7 and 6DJ8 are
alternate versions of the same tubes that have better performance, particularly
regarding noise. Note that many tubes have so-called industrial or military
versions, sometimes more than one. The 12AX7 is also known as the 6681, for
instance... something that you might not realize unless you thumb through a tube
The second column, Mu, is also known as the amplification
factor. This is a theoretical number that places an upper limit on the gain you
can expect from the tube. It's a good number to have handy when you're designing
a circuit. For instance, if you're designing a tubed phono stage you're going to
want all the gain you can get. The clear winner for that application is the
12AX7. For a line stage, you might consider the 12AU7 or 6SN7. There's no
benefit to amplifying a signal beyond the level you need — you'll only end up
throwing away the gain with a voltage divider or pot. It's better to match the
tube's gain to your needs while you're still in the design stage.
Plate voltage is the maximum amount of voltage you should
apply to a tube. Yes, you will sometimes see tubes run beyond their ratings
(tubed guitar and bass amps are notorious for this), but for hi-fi use it's best
to stay within the limits.
Plate dissipation is the maximum amount of heat the tube's
plate can shed. If you exceed this value, the tube will not necessarily
immediately go belly-up, but it will degrade the tube's performance...and what's
the point of building a tube circuit if you're not going for performance?
Each tube has pins at the base. Which pins are grids, plates,
cathodes, or heaters varies. These are all twin triodes, but you'll need to
consult a data sheet to determine which pins are which.
The filament will require its own power supply, separate from
the main rail voltage. Never operate a tube without tying the filament supply to
ground (or sometimes the rail — more about this later) in some manner, or the
electron cloud surrounding the cathode will cause the filament supply to drift,
relative to the rail voltage, with the result that the cathode and the filament
can arc. Every tube has a maximum voltage that it can tolerate between the
cathode and the heater. For a DC filament supply, the easiest thing to do is to
connect one side of the heater supply to the audio circuit ground. DC is
generally better for a filament supply than AC in low level audio circuitry
because it won't introduce hum into the signal, however, it's not uncommon to
see AC; it's easier and cheaper. In that case, you don't tie one side of the AC
to ground, you use two resistors to create a voltage divider from one side of
the filament supply to the other, then connect the mid-point to ground. A few
hundred ohms will do; the exact value isn't critical. Just be sure to use
resistors with a high enough wattage rating.
In passing, I'd like to say that I've known a number of
aspiring tube circuit builders who assembled projects, only to have the cathode
arc to the heater. As a result, they decided that electronics was “too hard”
and gave up. The sad part being that if they had made a single connection
between the heater supply and ground the circuit would have worked perfectly.
Having come this far, it is time to start building a circuit. Assuming that you've selected a tube type that is appropriate for the circuit you want to build, how do you go about determining the values for the resistors and capacitors that surround the tube?
There are several possibilities. One is to take a circuit that
is known to be good and modify it to suit your purposes. It may not be elegant,
but it works, and you can learn a lot about how tubes behave simply by fiddling
about for a while and paying attention to what happens when you change one
resistor or another. Or you can take a chart of the tube's performance
characteristics, select an operating point, and solve the problem graphically.
If you have detailed enough information at hand, you can calculate the values
necessary to seduce the tube into doing what you want it to do.
Or — and this might seem too good to be true — you can
turn to the back of your trusty RC-30 and locate the chapter on Resistance
Coupled Amplifiers. In it, you will find tables giving pre-calculated values for
rail voltages, resistors, and capacitors for a wide selection of tubes.
Let us say you want to build a line stage using the 6SN7.
Reading from the table, we get:
Ebb = 300V (this one we get to choose)
Rp = 0.047 (this is in MegOhms, i.e. 47k)
Rg = 0.047 (also in MegOhms, hence 47k)
Rk = 1300 Ohm
C = 0.061 (this is in μF)
—Ebb—the rail voltage: Choose the highest voltage
available. In this case, that's 300V. High rail voltages not only give you
greater voltage swing, they also stretch out the curved lines that define the
performance of the tube, which translates as lower distortion, more or less for
free — not a bad thing. There are times when you might want to lower the rail,
like if you need more output current and find yourself approaching the power
dissipation limits of the tube. For a preamp, though, we're not going to be that
badly pressed, so 300V, it is.
—Rp—the plate resistor: A question of tradeoffs. The table
that covers the 6SN7 offers plate resistor values from 47k to 220k. Other
factors being equal, a higher value plate resistor means higher gain. I chose
the lowest plate resistor on the table for two reasons — we don't need the
higher gain and, more subtly, a larger plate resistor will limit your bandwidth.
Why? Because a tube has internal capacitances. They're small, to be sure, but
the plate resistor interacts with that capacitance to cause a high frequency
rolloff. And here we stand on the precipice of a controversy.
You'd think that the choice of a plate resistor's value would
be a relatively simple, innocent thing. Not necessarily. My view is to go with
the lowest plate resistance possible so as to have the widest bandwidth. The
opposing view is that you should always maximize gain, then burn off the extra
gain as negative feedback in order to extend the (now narrowed) bandwidth back
to where it was in the first place. From where I sit, it seems that a little bit
of negative feedback goes a long way and if by some chance you can get away with
no feedback, better still. Others feel that negative feedback is a gift from the
electronics gods and the more the better. You can find multitudes of websites
where people argue (sometimes quite nastily) back and forth about the matter.
It's part of the art of electronics, as opposed to the science. Seemingly
innocuous choices like this help define the house sound of all the big audio
companies. Just as a guideline, you'll find that many tube gear manufacturers
tend to use from 10 to 15 dB of negative feedback, but there are plenty of
examples of both lower and higher feedback ratios.
—Rg—the grid resistor: This might seem counter intuitive,
but this is not the grid resistor for the stage you're designing. It's the one
for the tube that follows. Huh?
The solution to the riddle is simple. Electronically speaking, the grid resistor
for the following stage forms a voltage divider with the plate resistor for the
current stage. The greater that resistance, the more signal appears at the grid
of the following stage. Unfortunately, the same caveat applies, in that a higher
resistance reacts with the interelectrode capacitance of the tube and the
bandwidth suffers. Again, your choice.
—Rk—the cathode resistor: The cathode resistor is in Ohms
and directly affects the bias. The way this works is actually pretty clever.
Remember, you want the grid to be negative, relative to the cathode. The current
through the cathode resistor sets up a voltage drop. The cathode sits on top of
that voltage, so to speak, while the grid remains at ground potential, which by
definition is 0V. So from the cathode's point of view, the grid is, in fact,
negative. Better still, the tube self-adjusts if the rail voltage changes.
—C—the DC coupling (a.k.a. blocking) cap: The coupling cap
goes between the stage you're designing and the one following. It keeps the high
voltage at the plate from screwing up the bias of the tube that follows. The
thing about the capacitor value in the table is that it's too low. It's set by
default for a -3dB point of around 100 Hz, which is waaaaay too high for a high
fidelity circuit. Fortunately for math-ophobes, it's a linear relationship —
to halve the cutoff point, double the value of the cap. To be on the safe side,
shoot for a cutoff point around 5 to 10 Hz. Eight times the value from the table
is 0.488μF which is not a standard value, but in this case having a little
extra helps, so a 0.51 or 0.56μF cap would be a good choice. For extra
elbow room, throw in a 1μF cap.
Assuming that you've got the RCA manual in front of you, you'll notice that there is another capacitor value in the table. The part is optional and for our purposes, it's best to leave it out. The capacitor is wired in parallel with the cathode's resistor and serves to shunt the audio signal to ground. It increases gain, but it also increases distortion. Tube guitar and bass amplifiers use bypass caps because gain and distortion are part of the game, but you don't want to go down that path for high fidelity use.
The DC blocking cap should be rated at or above the full rail
voltage of the circuit. The cathode and grid resistors can be 0.5W. The plate
resistor can be 0.5W for some tubes, like the 12AX7, which tend to run at pretty
low currents, but I'd suggest 1W or even 2W parts to be on the safe side,
particularly if you intend to experiment on the circuit. For the 6SN7, use a 1W
No, the 220 Ohm resistor in series with the grid isn't in the
table. It's there to reduce the chance of oscillation caused by radio
frequencies. In an earlier era, they weren't necessary and depending on where
you live, you may or may not have problems today, but it's safer to put it in.
The circuit as it stands will amplify, but it's not yet a
complete preamp. Next time we'll add an output stage and I'll give some hints