June 1958 Radio-Electronics
Wax nostalgic about and learn from the history of early electronics. See articles
published 1929 - 1948. All copyrights hereby acknowledged.
The old adage about a picture being worth a thousand words is
still true today, even in the Information Age in which we live.
A lot of people, especially those new to the field of electronics,
struggle with the concept of decibels as applied to power and
voltage (and to a lesser degree current). A plethora of computer,
browser, and phone app programs are available to make individual,
specific conversions, but what has been learned about the fundamental
relationship? A nomograph is still one of the best tools both
for teaching and performing conversions. This article that discusses
properly matching impedances of amplification stages includes
a nice nomograph.
Know Your Levels
By Norman H. Crowhurst
Both low - and high-level amplification have built-in complications.
Discover what they are and how to beat them.
Some things about designing or using amplifiers you can find
in any textbook. But for some unexplainable reason other items
of information that ought to be easy to find seem to get left
out. For example, operating levels at various points in an audio
The amplifier user needs to know this so he can put the right
items of equipment together and get the best performance out
of the whole system. The amplifier designer needs this knowledge
to select the right components for his amplifier.
A closely related piece of information - that of impedance
matching - gets discussed in every second article on amplifiers.
So it is common knowledge that a 50-ohm microphone must be connected
to a 50-ohm amplifier input. A number of articles have shown
how to make resistance pads to match one impedance to another.
I was called in recently on a case which aptly illustrates
the lack of this knowledge. My job was to make a microphone-amplifier-loudspeaker
combination work. The man who called me had paid careful attention
to matching. The microphone was 50-ohms and the amplifier had
a 600-ohm input, so he used a line transformer from microphone
to amplifier. However, he was unable to get any output from
Fig. 1 - Cascode input circuit for preamplifier
input stages where noise must be minimized. Resistors and capacitors
should be high-stability low-noise components.
Examination showed that the amplifier was designed to operate
from an input level of 0.5 to 1 volt at 600 ohms. The microphone
was one of the higher-sensitivity dynamic types which gives
about 3-mv output across 50 ohms for normal speech. The matching
transformer from 50 to 600 ohms stepped this voltage up about
3.5 times, delivering a little more than 10 mv to the amplifier.
But 10 mv is not enough for an amplifier that needs an input
of 0.5 volt (500 millivolts).
"You need more amplification," I told my caller.
"That's easy," he replied, "I have another of these amplifiers
here on the shelf, and you know how to make a matching pad so
I can work the 16-ohm output of one into the 600-ohm input of
I explained that what he needed was not another power amplifier
but a preamplifier for working at low level. This he did not
seem to understand. He thought amplification was amplification,
and between the two amplifiers there should be enough of it.
So I explained briefly why this arrangement would not work.
But how many of us have had to find this out the hard way
- by trying it - simply because there was no one on hand to
tell us what would happen? Fortunately this job did not prove
difficult, because he did have a comparatively high-sensitivity
microphone and it wasn't too hard to find a preamp that would
work successfully with this power amplifier.
If he had been trying to use a mike with a much lower output
level, there would have been bigger problems in finding a satisfactory
preamplifier. So let's start at the input end and see what it
takes to make a good amplifier that will handle signals at all
When we set out to build an amplifier for amplification at
low levels, from insensitive microphones and pick-ups, particularly
the ribbon type, we have to be very careful when selecting components.
The first tube gets a maximum signal of only a few millivolts
at its grid. It is expected to make these signals audible at
the output. This is getting down to the level of tube hiss and
the hum generated in a good many tubes.
The input circuit must be carefully shielded to avoid hum
pickup. This part of the story, though, has been well discussed
elsewhere from time to time.
Electronics voltage and power level nomograph.
Tube hiss is due to plate current, which consists of electrons
flowing from cathode to plate. Each electron transit is a separate
event, so the plate current is made up of a random sequence
of separate charges passing from cathode to plate. The average
rate of transit determines the measured current. When amplifying
low-level signals, changes in plate current due to the applied
audio grid voltage are not much more than the fluctuation in
rate of arrival of electrons at the plate due to the random
nature of their departure from the cathode. Therefore, tube
hiss is apt to be almost as loud as the audio signal we want
Ways have to be found to minimize tube hiss. The noise a
tube generates due to these effects is obviously proportional
to the total current flowing - the proportion of fluctuation
in electrons arriving at the plate is proportional to the average
total number arriving. The noise voltage they develop at the
plate is also proportional to the fluctuation in the velocity
at which they arrive. The fluctuation in velocity is proportional
to the actual velocity.
Therefore, halving the plate current will approximately halve
the noise output of a tube and halving the plate voltage will
also approximately halve the output. But halving the plate current
or plate voltage does not necessarily halve the tube's gain.
Over a wide range of variation in plate current and voltage,
a tube's amplification does not vary by too much. Operating
the tube with low plate voltage and current gives almost the
same amplification as a higher plate voltage and current, but
considerably reduces noise introduced by the tube.
From the standpoint of noise, two things are required of
an input stage: (a) it must have minimum noise itself; (b) it
should have as much gain as possible, to lift the signal as
far as possible above noise. Pentodes have good gain, but generate
more noise than triodes. About the best compromise seems to
be the use of a twin triode in the cascode circuit shown in
Another factor in operating tubes at low levels is their
microphonic characteristics. A tube's plate current is determined
by the electric field produced by the grid and plate at the
cathode. Fluctuations in this field vary the plate current and
this constitutes signal. If one of the electrodes, particularly
the grid, is subject to any vibration, the field controlling
the plate current is altered.
As the variations we are concerned with are very small, it
takes very little vibration to produce an audible output. This
characteristic is known as microphonics. To prevent microphonics,
special attention has to be paid to providing a rigid tube construction
so the electrodes are not free to vibrate.
Another way in which a tube can inject a spurious signal
is by hum, especially if ac is supplied to its heater. Tubes
designed for use at low levels have the heater specially constructed
to radiate a minimum ac field.
In short, a tube specially designed for operation in audio
amplifiers at low levels has to have the following special characteristics:
It should be a hi-mu tube which can be used with low plate voltage
Its construction should prevent microphonic vibration.
The heater must be made so it doe not inject hum into these
One tube to which careful attention has been paid is the
Usually, feedback is not applied over the first stage of
an amplifier designed for low level inputs. This is because
a feedback loop over a low-level input reduces the input voltage
seen by the tube to a fraction of the input voltage provided
by the microphone or pickup. This aggravates the noise problems
we have been discussing rather than helping reduce them.
However, in these days of low distortion, when some designers
are working to get figures in the region of 0.1%, feedback is
used to reduce the distortion factor of all the other stages
to a minimum, so the input stage is sometimes the limiting factor
When this is first discovered, it comes somewhat as a surprise.
The classic treatment of distortion tells us it is due to curvature
of tube characteristics and, if we use the tubes well inside
of their maximum rating, the distortion will be low. Surely,
by using only a few millivolts of the tube's characteristic,
distortion should be negligible.
This you would expect. But it does not work this way in practice.
You would expect a similar thing of high-quality transformers
because their distortion is also due to nonlinearity. However,
distortion measurements on the magnetizing current of high-quality
transformer cores, suitable for input transformer use, show
that reduction in level reaches an ultimate minimum in distortion
- about 1% of the magnetizing current.
Of course, the distortion is likely to show up only at low
frequencies because the magnetizing current itself is a small
fraction of the currents that the transformer handles as signal
currents at frequencies above the low frequency cutoff. This
means that such a transformer at low levels produces a distortion
of maybe 0.5% in the region of 50 cycles, dropping to about
1/20th of this, or .025%, at 1,000 cycles. This residual value
of distortion is due to a hysteresis effect in the magnetic
It appears, although little work has been published on this,
that tubes exhibit a small "hysteresis" effect , which is normally
so miniscule as no to be worth considering. But, in comparison
with the minute fluctuation of current we have to handle in
low-level input stages, this hysteresis effect in the tube current
characteristic is enough to represent harmonic distortion of
about 0.1 or 0.2%, according to tube type.
This kind of distortion does not disappear - like the transformer
magnetizing-current distortion - at higher frequencies, because
the tube current fluctuations are the same whatever frequency
the tube is amplifying.
So, in selecting a tube for low-level amplification, a number
of factors have to be considered. Fortunately, tube manufacturers
as well as amplifier designers are working on this, so we are
able to call on their assistance to provide the best tubes for
the purpose. What the amplifier user has to realize, however,
is that you cannot just plug in, say, a 12AX7 as a replacement
for a 12AY7 because it happens to give a similar gain in the
same electrical circuit. It may even give a little more amplification,
but it will probably introduce other troubles which have been
taken care of in the design of the 12AY7.
Other components that need special attention in low-level
input circuits are the resistors and capacitors. The capacitors
must be low-leakage types. Their insulation resistance must
be better than that required for normal amplifier applications,
because any leakage through this insulation will cause increased
Resistors should be made of a highly stable material that
will not give appreciably more than the basic minimum of noise
due to thermal agitation, which is a characteristic of all resistors.
The big problem in input transformers for low-level stages
is shielding against hum pickup. Where the input level is very
low, a triple-shielded transformer, giving a reduction in hum
pickup of as much as 90 db, may be necessary.
Now we can see why a power amplifier would not do as a substitute
for a preamp to provide the initial stages of amplification.
A power amplifier's input transformer is intended to handle
a normal signal of 0.5 to 1 volt. In trying to work at a level
of 10 mv, it will probably pick up excessive hum because there
is no necessity for superfine shielding in this application.
Further, the input tube of a power amplifier, while quite
satisfactory for its purpose, would probably prove to be quite
noisy and microphonic as well as introducing its own quota of
hum when used for amplifying at low levels.
From low-level operation we will turn to high-level operation
because that is where the next group of problems arises. There
seem to be no particularly tough problems in providing amplification
at intermediate levels because then we are well away from the
extremes which cause problems either low- or high-level operation.
As you know, the big problem in high-level amplification
is distortion. Here the curvature of the tube is the controlling
factor. The problem of output stage design has been discussed
many times, but there are other relatively high-level circuits
where you must be careful in selecting tubes to give the best
For example, pursuing the comparison made between the 12AX7
and 12AY7 for low-level work, when we compare the data, we find
that the 12AY7 gives a larger harmonic distortion than the 12AX7
for the same output level if the signal voltage on the grid
is more than about 0.1 volt.
Similar comparisons can be made between other tubes such
as the 12AU7 and its nearest counterpart in the octal range,
the 6SN7-GT. The 6SN7 will give both a little more output and
a little less distortion than the 12AU7, operating as a single-ended
tube. But at levels less than one-third of the maximum handling
capacity, the 6SN7 gives considerable less distortion than the
12AU7. However, there is a consoling factor for the latter tube
in that most of its distortion is second harmonic, and the pairs
in the same envelope are usually pretty well matched.
Nomographs Available on RF Cafe:
Voltage and Power Level Nomograph
- Voltage, Current, Resistance,
and Power Nomograph
- Earth Curvature Nomograph
- Coil Design Nomograph
Coil Inductance Nomograph
- Antenna Gain Nomograph
Posted July 2, 2014