<< Prev TUBE CAD JOURNAL Next >>

Tube Phono Preamps Yet we drive. Yet records sound quite good.
Several topologies & tricks And records and tubes, V&V, "valves and vinyl"
Part 1 of 2 as the Brits say, go together well. Some have
argued that the only reason tubes were
resurrected was to hide or paint over the
blemishes of CDs: had CDs sounded better, we
would have been happy to continue down the
solid-state path to perfect sound, forever.
Maybe. We'll see if SACD (or DVD-A) buries
the tube or (what is my guess) it only furthers
bringing the tube's virtues to the forefront.
What I find amazing is not that vinyl persists,
even twenty years after the introduction of the CD Topologies
("perfect sound forever"), but rather that it ever Not too long ago, all tube phono preamps
became popular in the first place. looked (topologically) the pretty much the same.
Imagine that records were never made and that Usually they held two cascading gain stages that
someone today broached the proposal that the were often followed by a unity-gain buffer (a
delicate nuances of a musical performance could be cathode follower), which always ended in a
reproduced by dragging a rock against a piece of feedback loop that actively realized the inverse
plastic. Madness. If nothing else, rocks are hard RIAA equalization curve. This is the topology
and plastic is soft, so shouldn't it be plastic found in the Audible Illusion Mini-Mite, the
dragging against rock? Berning P-1, most Conrad Johnson's early
Of course, the same might be said if cars did preamps, Dynaco PAS-3, Lux 3300, Marantz 7
not exist and someone proposes creating one-ton and C-22, all of the MFA preamps, the Precision
steel structures that could travel over one hundred Fidelity C-4, and numerous Audio Research
miles-per-hour, controlled by anyone over the age preamps, SP-6, etc. (One notable exception to
of 16, no matter how aged, infirmed, drunk, high, this scheme was the Leak Point One preamp,
or mentally unstable who could turn a key. What if which used one pentode-based gain stage and
he also proposes that they be placed on tracks of wrapped the equalizing feedback loop around
road 10 feet wide, what then would keep people this gain stage in a plate-follower arrangement!)
from crashing into each other? "Painted lines on
the road," he tells us. Absolute madness.
First, psychologists would explain that the stress
resulting from the constant fear of dying from an
accident would render any driver mentally crippled
after only a few hours spent driving, as obviously 12A U 7
driving a car would be a hundred times more
difficult than flying an airplane because of the
intimate contact with the ground. Second, complex 12A X 7 12A X 7

computer simulations would show that if only one

47k
driver in a hundred were slow to react by more
than a few milliseconds, the whole streaming mass
of cars would collide, creating a vast sheet-metal
graveyard. And last, the environmentalists would
point out that 70% of all land animals would die
within one year of the car’s introduction. Conventional active equalization preamp

< PREVIOUS Pg. 1 www.tubecad.com Copyright © 2001 GlassWare All Rights Reserved NEXT >
 << Prev TUBE CAD JOURNAL Next >>
This basic topology offered a simple way to Starting in the late 70s and early 80s, several
achieve feedback and equalization in one step. commercial preamps appeared that used the
The topology's limitation was that the feedback passive equalization approach: the Counterpoint
network itself constituted a severe load at high 5.1, the NYAL NCP-1, and several very high-
frequencies and that the maximum gain that was end Conrad Johnson preamps.
realizable from any two cascaded triodes limited Then the CD came out and the need for a
the amount of available feedback. (The better tube phono preamp topology seemed to
equalization capacitors decrease in impedance disappear. (Yet in the 80s and the 90s we saw
with increasing frequency and thus they load some of the best tube phono preamps being
down the output; more gain stages endanger the made: the MFA MC Reference and the Audio
fragile feedback stability.) Was this topology the research SP-11, for example.)
end of the line for tube phono stages?
Back in the 70s, I remember the parade of Why Equalize?
boring schematics, the same conventional two- If implementing an RIAA equalization curve
gain stage active equalization topology, differing is such a hassle, why not just record and
only in component values, a relentless striving to playback a flat signal free of equalization? It
squeeze greater performance from basically the could be done, but the LP would have to be
same circuit. Then in the late 70s, the great renamed the VSP for “very short play.” To allow
French audiophile and tube fancier Jean Hiraga greater playback times (bass signals must be
designed a preamp with zero global feedback attenuated to conserve groove width) and to
loops and with a passively equalized output improve the high-frequency signal-to-noise ratio
signal. H.L. Eisenenson and friends at Audio (high frequency signals must be accentuated to
Directions in San Diego, California then overwhelm the ticks and pops of a record’s
mirrored his efforts. But passive equalization surface), the sound recorded onto a phono album
was not anything new, as anyone who had read a must follow a special equalization curve. The
RCA tube manual had seen the RCA- lows are greatly attenuated and the highs are
recommended circuit that used a passive greatly boosted. At playback the inverse of the
equalization in between two 7025-based-ground- recording EQ curve must be employed to return
cathode amplifiers. And many cheap tube-based the signal to flat by boosting the lows and
stereo consoles used passive equalization. But it cutting the highs.
did seem fresh to those who had painted Since the need for equalization is not going to
themselves in the old topological corner. go away, we must decide how to implement it:
actively or passively or a blend of both. Each
+300V approach has its adherents and distracters.

150k 100k
.1µF .22µF
976k
+150V +150V
106k Preamp Equalization Playback
Record Equalization
.001µF

1M
12A X 7 12A X 7
.003µF

47k 1M
1.1k 533

10 100 1k 10k

Passive equalization preamp

< PREVIOUS Pg. 2 www.tubecad.com Copyright © 2001 GlassWare All Rights Reserved NEXT >
 << Prev TUBE CAD JOURNAL Next >>
Active Equalization Now +60 dB equals a gain of 1000. This
Active RIAA equalization means feedback amount of gain could be had from cascading two
equalization: the frequency response is tailored +30 dB (gain of 31.6) grounded-cathode stages,
to fit that of the RIAA curve by varying the which is easily achieved with two 12AX7s or
amounts of feedback returned to the input. The even two 12AT7 triodes, but not by two 12AU7s
advantage of using feedback is consistency. or two 6DJ8s, as their amplification factors are
Each channel will track the other to a very great too low.
degree in spite of aging parts or circuit wiring If we increase the desired gain to +80 dB (a
dissimilarities, as the feedback tends to iron gain of 10,000), then even two 12AX7s will not
everything out. The disadvantage is that, do, as we never realize the full potential gain
because it is active, the circuit can more readily implied by a triode's mu in a grounded-cathode
suffer from input voltage overloads and the amplifier, except when it is loaded by a constant
preamp must have voltage gain far in excess of current source and works into no other load
nominal +40 dB usually specified, as the impedance. One work-around is to use a cathode
feedback uses the excess gain to force the output follower after each grounded-cathode amplifier,
to conform to the desired curve. Another but this approach was seldom if ever taken.
problem is potential instability, as each coupling So here is the dilemma: we do not want to
capacitor and gain stage add some phase shift. exceed two gain stages and yet we want more
Since the bass frequencies must be amplified gain. One solution has been to use at least one
+20 dB higher than the 1-kHz center frequency, cascode stage in the mix. The cascode circuit has
the +20 dB must be added to the +40 dB of gain, the very desirable attribute of realizing a gain in
yielding +60 dB of total gain. On top of this +60 excess of the mu of the triode used. (This plan
dB an additional 20 to 30 dB of gain might be was beautifully implemented in Audio
added to feed the feedback mechanism. Thus, Research's SP-10 and MFA's MC Reference
we need more gain, but we cannot risk adding preamp.) Pentode-based circuits and hybrid
more gain stages, as each coupling capacitor and circuits can also realize a much larger gain than
Miller effect capacitance adds some phase shift, can be developed by the triodes alone.
which can reverse the phase of the fed back Still another solution can be gleaned from
signal, creating an oscillator, not an amplifier. some solid-state phono preamps: forgo the
In the absence of this extra gain, the varying single-all-encompassing-equalizing feedback
amount of frequency dependent feedback can loop around the entire preamp stage. In other
result in looser bass reproduction because of words, split the equalization curve into its sub-
smaller amount of feedback at low frequencies curves and use two gain stages, each with its
and possibly a pinched, compressed high own equalizing feedback loop. For example, the
frequency playback due to excessive feedback fist stage can produce the 50-Hz to 500-Hz part
ratios at high frequencies since a 20-kHz signal of the RIAA equalization curve, while the
is attenuated by -40 dB relative to a 20-Hz second stage can yield the 2122-Hz low-pass
signal. In other words, a preamp with only +60 function of the curve. This arrangement greatly
dB of open-loop gain will have zero feedback at unburdens each amplifier stage and allows for
20-Hz and 40 dB of feedback at 20-kHz. the realization of a much greater total gain.

60dB Open loop gain

40dB Feedback

20dB in
Closed loop gain

< PREVIOUS Pg. 3 www.tubecad.com Copyright © 2001 GlassWare All Rights Reserved NEXT >
 << Prev TUBE CAD JOURNAL Next >>

10k 10k 10k

300

6922 6922
200 20k

6922 6922 6922 6922

47k 66 1M 1M 200 10k 1M

1k 9k 375pF

.3537µF 200k

One possible implementation of the this two The second stage is an inverting amplifier
section approach is shown above. Here the first made up of a grounded-cathode amplifier
stage is a fairly straightforward two triode cascading into a cathode follower. The feedback
cascaded amplifier. The first triode(s) inverts the loop consists of only the 200k resistor and its
input at its plate, which in turn is inverted again shunting capacitor. At low frequencies, the input
by the second triode back to normal phase. The resistor (20k) and feedback resistor define the
non-inverted signal is then given to a cathode gain of the stage based on the ratio of their
follower which buffers the second triode from values. At high frequencies, the shunting
the added load imposed by the following stage capacitor's declining impedance shortens that
and the feedback loop. The output from the ratio, which decreases the gain. At an infinitely
cathode follower is then returned to the first high frequency, the capacitor's impedance
triode's cathode via the two resistor feedback becomes effectively zero and the gain falls to
loop. zero, not unity (1), but zero output. This is an
At low frequencies, all of the resistance improvement over the conventional single
within the feedback loop is used to voltage equalization feedback loop applied across a non-
divide the output signal deeply as it returns to inverting amplifier, as the output should
the first triode's cathode. Thus the gain at these continue to fall with increase frequency, not go
low frequencies is great, as a positive signal flat once unity gain is reached, which effectively
applied to the first tube's cathode subtracts from results in a high-frequency boost.
the total gain of the amplifier; thus the less (To overcome this departure from the RIAA
signal returned, the greater the gain. But at curve two approaches have found favor: do
higher frequencies, the feedback loop capacitor nothing, as the record itself has its own high
shunts away one of the feedback resistor's frequency limitation, its own falling off with
resistance and now the voltage division only ever higher frequency function, which even if it
partially voltage divides the output signal, thus did once have infinite frequency response,
greatly reducing the output at these frequencies, playing it once would scrape the highs off its
as more positive signal is given to the first stages surface; and add a fourth pole to the equalization
cathode. This creates two gain plateaus: one network to help the output follow the RIAA
below 50-Hz and one above 500-Hz, the latter curve beyond the unity gain point, usually this
being down –20 decibels relative to the former. takes the form of a simple passive-RC-low-pass
This is the first half of the inverse RIAA filter added to the preamp's output.)
equalization curve.

< PREVIOUS Pg. 4 www.tubecad.com Copyright © 2001 GlassWare All Rights Reserved NEXT >
 << Prev TUBE CAD JOURNAL Next >>

SE Amp CAD
in

Other topological variations are certainly

possible. An input amplifier with no equalization
could feed an inverting plate-follower amplifier
with all of the equalization, which would
accurately follow the RIAA beyond unity gain.

in

An inverting plate-follower amplifier with

only the 2122 Hz part of the equalization might
make a good first stage of a two stage MC
phono preamp, as moving coils cartridges work
well with low impedance shunting impedances,
which this first stage could easily present. The
second stage would actively finish implementing
the 50 to 500 Hz part of the equalization curve.
Successful design and analysis of a
One complaint might be that any inverting
single-ended amplifier output stage
plate-follower amplifier will invert the phase at
requires an accurate model of the tube's
its output, which if it does cascade into another
plate curves. SE Amp CAD is a tube
inverting stage, will result in phase inversion at
audio design program that has a library of
the output. This is a non-issue, as the leads to the
30 tubes and over 100 output
phono cartridge only need to be reversed in
transformers and SE Amp CAD knows
phasing to set the output straight. The cartridge
how these tubes really curve in a singled-
does not know how it is being hooked up to a
ended amplifier.
preamp; its coils do not “know” what phase
configuration is “correct.”
In fact, there is one distinct advantage to Windows 9x / Me / NT / 2000
having the preamp invert the phase at its output:
it lessens the chance of the output signal re- For more information, please visit our
circulating back into the input and causing Web site or write us at:
oscillation. The danger any high gain non-
inverting amplifier faces is that its output signal GlassWare
is so much greater than its input signal that only PO Box 231
a small fraction of its output being fed back to Fenton, MI 48430 USA
its positive input can lead to wild oscillation. On
the other hand, an inverting amplifier cannot www.glass-ware.com
oscillate under the same conditions

< PREVIOUS Pg. 5 www.tubecad.com Copyright © 2001 GlassWare All Rights Reserved NEXT >
 << Prev TUBE CAD JOURNAL Next >>

Tube CAD
Reality Check Preamp
Before moving on, I recommend building a
reality check preamp. If several hundred dollars
worth of expensive parts and ten hours of
frustration and work cannot beat the sonics from
a twenty dollar IC-based preamp, then we need
to go no further. The schematic below shows a
preamp I built back in the 80s. It offers 40 dB of
gain and is very quiet. I was able to stuff all of
the parts and four 9-volt batteries (dual mono
power supplies) in a small aluminum box that
sat underneath my turntable. The input leads
were hard wired in place and only a foot in
length. I used a four-pole rotary switch to turn
on the unit. Battery life was easily 20-50 hours,
which meant weeks of listening; Costco sold a
brick of batteries (25) for under $10 back then,
which would last many years. (Yet friends who
owned $1000 phono cartridges that were only
good for 700 hours of use looked troubled by the
expense of replacing the batteries!)
.001µF

100 10k .003µF

78.7k
+9v
100k 976k
OP37
.05-.5µF +9v Tube CAD does the hard math for you.
-9v
AD711 This program covers 13 types of tube
47k
-9v
circuits, each one divided into four
variations: 52 circuits in all. Tube CAD
calculates the noteworthy results, such as
The circuit can be improved by adding gain, phase, output impedance, low
negative pull-down resistors to the IC outputs, frequency cutoff, PSRR, bias voltage, plate
but at the cost of less battery life or by using a and load resistor heat dissipations. Which
pair of batteries per Op-Amp, thereby doubling tube gives the most gain? Tube CAD's
the battery life. Now, do not get me wrong; this scenario comparison feature shows which
circuit is not the best phono preamp in the tube wins.
world, far from it. However, whatever tube-
based preamp we do build must decisively beat Windows 95/98/Me/NT/2000/XP
it or we are wasting our money and time.
For more information, please visit our Web site :
Next Time
The second half of this article will cover my
preferred equalization method: passive www.glass-ware.com
equalization and a few hybrid topologies.
//JRB

< PREVIOUS Pg. 6 www.tubecad.com Copyright © 2001 GlassWare All Rights Reserved NEXT >