EMRFD Message Archive 1185
Message Date From Subject 1185 2007-11-24 09:11:49 bkopski AN AD8307 BASED POWER METER MODIFICATION FOR PEP MEASUREMENT
W7ZOI, and Bob Larkin, W7PUA, that has become very popular as
evidenced by many contemporary references to that work. This simple
power meter design is based on the Analog Devices AD8307 IC and
features operation over a wide dynamic range and a wide frequency
range.
I built mine shortly after the original QST appearance but added some
operational features I wanted. Wes encouraged me to publish my
variant and this presentation appeared in the May/June 2002 QEX. Not
long after that I incorporated an additional modification to mine
that added even more utility with very little effort and expense.
I've been using that latter version as presented in the Sept/Oct 2003
QEX for several years.
Very recently I discovered that these power meter versions did not
perform well in ssb applications wherein I was seeking to measure
PEP. I could do this with a spectrum analyzer but I wanted the
convenience of a power meter measurement also. After some
experimentation I was able to incorporate a simple circuit addition
that accomplished just that: my power meter now reads out as it did
before and in addition correctly displays PEP levels on the DVM
display. The latter is true for two tone testing or with a very
steady voice level. Previously, neither audio input would produce a
meaningful meter output.
The circuit addition is detailed in the photos and files section. It
requires the addition of a dual CMOS op amp, and one each transistor,
resistor and capacitor. This simple addenda circuit built on a
small piece of Radio Shack holeboard - is "electrically inserted" in
the earlier QEX circuit via 5 short wires as illustrated.
The circuit functions thus: Detected and buffered signal appearing at
(the original) D1 cathode is now passed to one new op amp section
which is configured as a peak hold circuit. This same signal value
also appears on holding capacitor Ch and at the input of the second
op amp configured as a buffer. It then continues on to (the
original) R12 and the DVM metering circuit as before. Basically, for
the DVM readout, the added circuit is "transparent" save for some
possible small offsets associated with the new op amp. In my case,
NO recalibration was required "luck of the draw" as far as op amp
offsets go. Otherwise, the most that would be required is a recal of
the DVM portion of the power meter readout.
Continuing, signal peaks such as associated with ssb audio are "held
and stretched" briefly by the Ch/Rh portion of the peak holding
circuit long enough for the DVM to properly display. Previously, the
metering did not respond to such signal peaks. In comparison with
the SA measurement, the revised power meter DVM display is "right on".
It is important to use the components AS SHOWN. Do not be tempted to
use casual substitutions. For example, the emitter diode of the
transistor functions much better (much lower leakage) than a classic
1N914 (or equiv). The very high input impedance / very low bias
current op amp is a MUST. And the capacitor must be low leakage as
well (like polystyrene or equiv). Basically, only the Rh-Ch
combination is to determine the held-value decay and those values
shown work very well so long as other potential leakage influences
are out of the picture.
If you have the QEX variant of the AD8307 power meter and have
interest in an ssb or similar application, I believe this circuit
modification / addenda will work very well for you as well.
Cordially and 73,
Bob K3NHI1188 2007-11-24 10:41:35 Gary Belcaster Re: AN AD8307 BASED POWER METER MODIFICATION FOR PEP MEASUREMENT
>Bob,
> The June 2002 QST presented an RF power meter design by Wes Hayward,
> W7ZOI, and Bob Larkin, W7PUA, that has become very popular as
> evidenced by many contemporary references to that work. This simple
> power meter design is based on the Analog Devices AD8307 IC and
> features operation over a wide dynamic range and a wide frequency
> range.
>
> I built mine shortly after the original QST appearance but added some
> operational features I wanted. Wes encouraged me to publish my
> variant and this presentation appeared in the May/June 2002 QEX. Not
> long after that I incorporated an additional modification to mine
> that added even more utility with very little effort and expense.
> I've been using that latter version as presented in the Sept/Oct 2003
> QEX for several years.
>
> Very recently I discovered that these power meter versions did not
> perform well in ssb applications wherein I was seeking to measure
> PEP. I could do this with a spectrum analyzer but I wanted the
> convenience of a power meter measurement also. After some
> experimentation I was able to incorporate a simple circuit addition
> that accomplished just that: my power meter now reads out as it did
> before and in addition correctly displays PEP levels on the DVM
> display. The latter is true for two tone testing or with a very
> steady voice level. Previously, neither audio input would produce a
> meaningful meter output.
>
> The circuit addition is detailed in the photos and files section. It
> requires the addition of a dual CMOS op amp, and one each transistor,
> resistor and capacitor. This simple addenda circuit – built on a
> small piece of Radio Shack holeboard - is "electrically inserted" in
> the earlier QEX circuit via 5 short wires as illustrated.
>
> The circuit functions thus: Detected and buffered signal appearing at
> (the original) D1 cathode is now passed to one new op amp section
> which is configured as a peak hold circuit. This same signal value
> also appears on holding capacitor Ch and at the input of the second
> op amp configured as a buffer. It then continues on to (the
> original) R12 and the DVM metering circuit as before. Basically, for
> the DVM readout, the added circuit is "transparent" – save for some
> possible small offsets associated with the new op amp. In my case,
> NO recalibration was required – "luck of the draw" as far as op amp
> offsets go. Otherwise, the most that would be required is a recal of
> the DVM portion of the power meter readout.
>
> Continuing, signal peaks such as associated with ssb audio are "held
> and stretched" briefly by the Ch/Rh portion of the peak holding
> circuit long enough for the DVM to properly display. Previously, the
> metering did not respond to such signal peaks. In comparison with
> the SA measurement, the revised power meter DVM display is "right on".
>
> It is important to use the components AS SHOWN. Do not be tempted to
> use casual substitutions. For example, the emitter diode of the
> transistor functions much better (much lower leakage) than a classic
> 1N914 (or equiv). The very high input impedance / very low bias
> current op amp is a MUST. And the capacitor must be low leakage as
> well (like polystyrene or equiv). Basically, only the Rh-Ch
> combination is to determine the held-value decay – and those values
> shown work very well – so long as other potential leakage influences
> are out of the picture.
>
> If you have the QEX variant of the AD8307 power meter and have
> interest in an ssb or similar application, I believe this circuit
> modification / addenda will work very well for you as well.
>
> Cordially and 73,
> Bob K3NHI
>
>
I really enjoyed your original article and have started assembling parts
to craft a meter here.
I accidently deleted the reference to your peak./hold circuit. Any
chance you could send that
again??
Gary
ka9rao1190 2007-11-24 17:21:46 kerrypwr Re: AN AD8307 BASED POWER METER MODIFICATION FOR PEP MEASUREMENT
I have opened an Album in the Photos section entitled AD8307 Power
Meter and included photos of my version.
The album is available for use by all members so post a photo of your
meter there; it might be a useful resource for future builders.
I intend to build another meter unit, just a basic LCD display in a
very small case, pocket-size really; that's one advantage of having a
separate measuring head.1208 2007-12-02 15:29:05 Wes Hayward Re: AN AD8307 BASED POWER METER MODIFICATION FOR PEP MEASUREMENT
This is in regard to the work that Bob Kopski, K3NHI, posted the
other day (including a figure in the files) having to do with his
8307 based power meter. Bob had corresponded with us before that
Group post, for he had observed that the power meter was not
indicating the right output.
There were two experiments that were done, here and there. K3NHI
built an audio two tone generator and applied it to one of his rigs
and then looked at the resulting SSB. He turned on tone #1 and
measured a power with the meter. He then turned that signal off and
turned on tone #2 and measured that power in the power meter. Then
he turned on both tones. This is the normal scheme we use for two
tone testing of a transmitter. But Bob saw no increase in his power
meter over what was seen with just one tone! We know from
fundamental considerations that the average power will increase by 3
dB because there is, after all, two times as many signals.
Moreover, the peak envelope power, PEP, will be 6 dB above each
single tone, for the voltage will be double that of one tone at some
points in time. This is what the math tells us and what we see
when studying things with an oscilloscope.
Bob Larkin, W7PUA, did the related RF experiment with a power meter
like that in Fig 7.13 of EMRFD. Two signal generators were attached
to the power meter through a combiner. Bob L observed again that
the power meter produced a response that was the same as seen with
just one generator. Bob then started playing with the amplitudes
of the two generator. Bob discovered that that our AD8307 based
power meter measures the power output of the stronger of the two,
even when one is just a bit larger than the other.
I repeated Bob Larkin's experiment. I tuned one generator through
the frequency of the other, but saw nothing in the meter.
Next, I looked at the two tone signal in my spectrum analyzer. See
EMRFD Fig 7.53. The same AD8307 type part is used as the SA log amp
as is used in the power meter. But now the circuit behaves
properly. That is, when I hooked the output from my hybrid to the
SA, I saw two tones. when I then tuned one generator to sit on top
of the other, bringing the tones together, the composite seen on
screen was a single indication that appeared to be 6 dB higher than
either of the individual tones. That is, the signal voltage
doubles.
Actually, it was more complicated than that. What I saw in the
analyzer was a band around what used to be the trace representing the
single tone. The top of this band is 6 dB higher than the original
tone. The lower band drops more.
So next, after a few well chosen words that can't be repeated here if
we want to avoid the Yahoo adult themes police, I again attached the
power meter and now examined the output with an oscilloscope.
Finally things start to make sense; physics does prevail. The
output from a single tone is just a steady DC. It is a steady DC for
two tones that are well removed from each other. As you move one
generator frequency to that of the other, you see a composite beat
signal at pin 1 of U3A in the power meter. The same thing was seen
at the log amp output, but it was a smaller signal. There is a
positive going response that exceeds an applied single tone by what
corresponds to 6 dB. There is also a negative going signal that
drops by a larger amount. However, these negative going signals
are narrow. The average, which is what the meter sees, is
essentially zero.
I noticed a subtle thing with my power meter. The tones really had
to be on top of each other to get a good scope response with the
original circuit. I removed capacitor C3 (0.1 uF) and replaced it
with 1000 pF. This works better. You now see the composite beat
signal when the two signals are within 50 kHz or so of each other.
The capacitor is a low pass video filter that is applied to the
logged response.
So Bob Kopski's solution, which is a simple, but well designed peak
detector that follows the log amp in his version of the meter, is an
excellent solution. It does nothing to compromise the response when
only a single tone is present, but it reads the peak envelope power
when two tones are present. The two tones should be close enough in
frequency to be within the bandwidth of the low pass filter related
to C3 in the original circuit. The K3NHI solution should work just
fine with the original circuit that Bob Larkin and I presented in
QST, repeated in Fig 7.13 of EMRFD. Just drive Bob Kopski's circuit
(TS912/2N3804/TS912) from pin 1 of U3A in Fig 7.13. Then drive R6
and the meter from the output of the Kopski circuit. If there is a
problem with taking the output of the TS912 op-amp to ground, a diode
can be inserted in series with the 9 volt battery to generate a small
negative power supply, just as Bob did in his original power meter of
March/April 2002 QEX.
No matter what is done, it is worthwhile to replace the 100 nF at C3
with the 1 nF part. The spectrum analyzer used a 10 nF in this
position and that worked fine.
The other thing that I have done in my own version of the meter is to
replace C10, which was a feedthrough capacitor to the front panel,
with a simple 0.1 uF and a BNC connector. It is then easy to run a
BNC cable to a DVM for those higher resolution measurements.
This modification would work well with the KA7EXM power meter.
Indeed, it would be worth examining for the spectrum analyzer that
K7TAU and I presented. The result would then be a simple trace
instead of a band.
It's any one's guess what is in the appliance power meter that is
reviewed in the Dec 07 QST, so we can't say how that would be
impacted.
As usual, great work