Softrock Lite 6.2
Adventures in Electronics and Radio
Elecraft K2 and K3 Transceivers
Receiver AGC Threshold
This page collects AGC test data I've taken on four
- Elecraft K2
- Drake R7
- Racal RA6790/GM
- Cubic R3030
The purpose of the collected data is to identify the
threshold signal level at which each receiver begins to apply AGC action and
also to assess the AGC's effectiveness at maintaining constant audio output with
changes in input signal level. Attack and release times were not measured in
As a quick overview, AGC--or automatic gain
control--automatically reduces the total gain of a radio receiver to maintain a
more-or-less constant audio output as the input signal varies in strength. If
AGC is applied to all signals, regardless of strength, however, the result is to
degrade reception of weak signals. Consequently, good design practice calls for
an "AGC threshold" below which signal strength no AGC action is applied--the
receiver runs at maximum gain. When the signal exceeds the threshold level, AGC
is applied and the output is leveled.
In an ideal receiver, once the threshold point is passed,
there will be little or no change in audio output with changes in signal level.
Below the threshold point, the receiver's audio output will be linear with
respect to the signal input, i.e., a 1 dB change in input level should result in
a 1 dB change in output audio.
And, of course, good design practice calls for an AGC
on/off switch, allowing the operator to disable the AGC when desired,
substituting instead manual RF and/or IF gain control instead.
All data was collected with the same automated setup, as
The receiver being tested is connected to a Panasonic VP8181A synthesized signal
generator, controlled via the GPIB interface with a Prologix GPIB-USB interface.
http://prologix.googlepages.com/. The test receiver's audio output is
measured with a Fluke 189 digital multimeter, with AC dB scale selected. The
Fluke's output is collected via the RS232 serial port. A custom program written
in Liberty Basic
http://www.libertybasic.com/ sets the VP8191A to the test frequency and
steps the output level from the entered starting point to the desired end point,
with user selectable steps (0.1 dB resolution). At each output level point, the
controlling program pauses 1 second to allow the receiver's AGC to stabilize and
then reads the Fluke 189 to determine the receiver's audio output level.
The collected data is written to a text file which I then
reformatted in Excel and plotted with Origin 7.5.
The test data was taken over the range -130 dBm to 0 dBm,
with either 0.5 or 1.0 dBm steps.
The absolute receiver audio output level is not the
important data in this analysis, as the audio level is arbitrary, related to the
audio gain setting in the K2 and related to the Line Level setting in the
RA6790/GM and to the line level output in the R7. Rather, the important thing to
observe is the change in audio level versus change in RF level.
Let's start with a Racal RA6790/GM. The 6790 is a 1970's design, synthesized,
microprocessor controlled receiver developed for military and commercial
applications and was manufactured between 1978 and around 1990. I find it to
have the best ergometrics of any receiver I've ever used, and to have
exceptional performance. The new price of an RA6790/GM was in the $5-6,000
range, in 1979 dollars, or well over $10,000 in 2007 dollars.
The AGC response looks textbook, in that below the threshold
the response is linear, dB for dB and above the threshold the audio output is,
for all practical purposes, unchanged over about a 110 dB range. If one were to
criticize the AGC response, a question could be raised as to whether the abrupt
transition at the threshold (set at -113 dBm according to the specifications)
might cause objectionable audio performance. I've never noticed it, but then
again I do not use this receiver for DX'ing. It's more of a piece of test
equipment in my basement shop than a working receiver.
I should also note that the RA6890/GM has an operational mode
that allows the user to set the threshold from a front panel control. This gives
the user three AGC modes:
- AGC off
- AGC on, with preset threshold
- AGC on, with user adjustable threshold
The RA6790/GM is designed, like many commercial and
military receivers, for optimum high signal performance and hence does not
include an RF amplifier ahead of the mixer.
The Drake R7 is of similar vintage to the RA6790/GM and
was intended to be a state-of-the-art amateur grade receiver to replace the R4C.
I bought mine new around 1978, if I recall correctly, and paid around $1400 for
the receiver and a complete set of auxiliary filters. The R7 has an optional
pre-amplifier stage, quite similar to the K2's design, as a matter of fact.
I collected test data from the R7 in four operational
- USB (2.8 KHz bandwidth) pre-amplifier off
- USB (2.8 KHz bandwidth) pre-amplifier on
- CW (500 Hz bandwidth) pre-amplifier off
- CW (500 Hz bandwidth) pre-amplifier on
In general, we see a similar shape to the RA6790/GM curve,
but with some important differences:
- The threshold is not a sharp break, but rather AGC
takes effect over a range of a few dB with a rounded threshold. Presumably,
this is a decision by the Drake engineers to smooth the transition between no
AGC and full AGC. An alternative explanation is that the AGC loop has less
- There is a small upward slope to the audio output,
around 1 dB for a 100 dB increase in RF level above the threshold. The
RA6790/GM has essentially no change over this range, but in truth the
difference between 1 dB and 0 dB change over this range is not perceptible to
- The curve most closely matching the RA6790/GM
test conditions (300 Hz bandwidth, no preamplifier) is the green curve. Drake
set the AGC threshold around -105 dBm for this operational mode, about 8 dB
above the threshold chosen by Racal's designers.
- The curves have some artifacts (bumps) that are due to
local RFI caused by the dehumidifier in my basement shop cycling on/off whilst
collecting the test data. This effect is most visible in the black curve, but
also can be seen in the blue curve as well.
The Elecraft K2 is a ham-band only transceiver of
relatively modern design and is in current production. Depending on options, its
kit price (it is available only as a kit) runs from $600 to $1200. It has an
switchable pre-amplifier. The AGC is applied only to one stage, the K2's 4915
KHz IF amplifier, a Motorola MC1350 integrated circuit. AGC is not applied to
the RF amplifier stage or the post-mixer amplifier. Motorola specifies the
MC1350's AGC range as 60 dB minimum, 68 dB typical.
I collected test data from the K2 in four operational
- USB (2.2 KHz bandwidth) pre-amplifier off
- USB (2.2 KHz bandwidth) pre-amplifier on
- CW (400 Hz bandwidth) pre-amplifier off
- CW (400 Hz bandwidth) pre-amplifier on
These curves differ from the Racal and Drake data in several
significant areas, and also from the Cubic R3030:
- There is no defined threshold. Rather, there is a wide
range over which AGC is slowly introduced.
- Even with strong signals (-60 dBm and more) the AGC
does not hold the output constant to the same degree as either Racal or Drake.
- Cubic's R3030 also implements a soft AGC curve, but
with significantly better control than the K2; 120 dB change in signal level
causes the R3030 to change audio output about 10 dB, whilst the same change in
the K2's input causes an audio output change of 25 to 40 dB, depending on mode
and pre-amplifier status.
In my opinion, the K2's AGC characteristics are
sufficiently different from a "normal" receiver to cause it to sound
different. When operating with AGC enabled, the Drake or Racal (and in fact
Drake and Racal are typical of other receivers I've examined) many, if not most,
signals are audio leveled to the same loudness, with only relatively weak
signals falling below the AGC threshold. The background noise is reduced
proportionally, e.g., a signal that's 20 dB above the AGC threshold results in
20 dB gain reduction; the audio beat note is just as loud in the headphones as a
signal right at the AGC threshold, but the background noise is reduced 20 dB.
With the K2, in contrast, many more
signals will either not be AGC leveled, or if AGC is effective, it is only
partially effective, in that it may, for example, translate a 6 dB change in RF
level into a 2 or 3 dB change in audio level. If the signal increases 6
dB, this gain action results in the audio level going up 3 dB and the background
noise going down 3 dB.
Thus, tuning across the band
will produce a different audio experience with a K2 than with a Drake R7, or a
Racal RA6790/GM or most any other amateur receivers or transceivers.
The Cubic R3030A is a dual receiver, with two complete
receivers packaged in one 19" rack panel configuration. All elements of each
receiver are modularized, with highly shielded plug-in modules. This another
multi-kilodollar receiver made for government and commercial facilities. It is
synthesized, with analog IF chains.
Cubic's R3030 has a "soft" AGC curve compared with the Racal
RA6790/GM or the Drake R7, but not nearly as soft as Elecraft's K2. As with the
K2, there is no distinct knee threshold; rather the AGC action is applied
gradually. The AGC is, however, reasonably effective, with 80 dB of RF input
increase (from -80 to 0 dBm) resulting in about 2 dB audio output increase.
All Four Receivers
The following graph is rather busy, as it shows all four
receivers for all test conditions.
As a final reminder, the absolute audio levels are
unimportant; rather the change in audio versus the change in RF signal level is
what should be examined.