Improved envelope detector for monitoring peak or average rf envelope power

[AB2EZ]

Here is the url for the schematic of a peak detector that I am using in conjunction with an rf envelope detector to measure my peak envelope power. This schematic is very similar to many peak detector designs that are already published on the web... except that it uses a general purpose npn transistor to enable the second op amp to drive much more current into the 0.1 uF capacitor... and thus allows the peak detector to follow short-lived peaks. The 1N4148 diode protects the transistor from excessive emitter-to-base voltage.

http://mysite.verizon.net/sdp2/id22.html


 
The overall gain of the peak detector is unity. The peak-average switch allows me to measure the average of the detected envelope, rather than the peak of the detected envelope. This is important for AM applications, when one wants to know one's carrier level.
 
The envelope detector that I am using is the sensor of my PowerMaster power meter... which I have modified so that it will follow the modulation of the envelope of the modulated rf signal it is monitoring. [A standard PowerMaster sensor has too much capacitance across its rf envelope detector to track the modulation of the envelope of the monitored rf signal].

[k4kyv]

Better still, use a calibrated thermocouple rf ammeter working into a purely resistive (non-reactive) load.

[AB2EZ]

Don

Are there thermocouple-based rf current sensors available that are fast enough to respond to the modulated envelope of the rf current (e.g., voice peaks on an rf signal modulated with 0-5 kHz audio)?

The circuit shown is a simple analog circuit for capturing and "holding" the peaks of a voltage waveform (coming from whatever sensor is responding to the rf envelope). The decay time of the "held" peak of the incoming voltage waveform is 0.1 second... which is long enough for this circuit to be used to feed the sampler at the input of the PowerMaster; or for use between a Bird sensor and a Bird meter.

It is very similar to the conversion kits that are sold to convert Bird 43 meters into peak reading meters... except for the extra transistor providing current gain to quickly charge the capacitor that stores the peak value. I have modeled this circuit using SWCADIII, and I believe that using a transistor that is capable of delivering high peak current to charge the capacitor makes a significant improvement in the accuracy with which the circuit can follow, and hold, the peak of the incoming waveform... particularly when the desired decay time of the peak detector is three or more orders of magnitude larger than the duration of the peaks one is trying to capture and "hold".



Stu

[k4kyv]

I don't even try to measure voice peaks with a meter.  I use a thermocouple rf ammeter to indicate what the mean (or average) power would be if the load impedance were precisely known, thus telling me how much power I am really running. 

My monitor scope in the envelope pattern mode tells me precisely what the voice waveform looks like, particularly that the signal is clean and that there is no flat-topping at the positive modulation peaks, nor cutting off at the base line on negative modulation peaks.

I'm not sure why, but monitor scopes seem to have fallen into disfavour with amateur radio operators. To me, operating a phone transmitter without a monitor scope is comparable to driving at night with no headlights.

[steve_qix]

I don't even try to measure voice peaks with a meter.  I use a thermocouple rf ammeter to indicate what the mean (or average) power would be if the load impedance were precisely known, thus telling me how much power I am really running. 

My monitor scope in the envelope pattern mode tells me precisely what the voice waveform looks like, particularly that the signal is clean and that there is no flat-topping at the positive modulation peaks, nor cutting off at the base line on negative modulation peaks.

I'm not sure why, but monitor scopes seem to have fallen into disfavour with amateur radio operators. To me, operating a phone transmitter without a monitor scope is comparable to driving at night with no headlights.

Hi Don,

I find an oscilloscope very useful for diagnosing problems, etc. but for actual monitoring (unless one is using a storage scope), the peaks go by too fast to show a quantitative measurement.  The scope is definitely useful for giving you an "average" peak indication, but if I want to know the real numbers, a modulation monitor is the only way to go.  And, of course, you get a perfect detector and a full fidelity headphone output as a bonus, letting you hear problems.  At least in my experience, if I can see a problem on a monitor scope (flat topping, etc.), the problem is already pretty serious.  I can hear problems in the headpone monitor WAY before I can see them.

But, I absolutely agree - operating an AM transmitter without a monitor IS like driving at night without headlights !!  I feel just about powerless - WHAT IS GOING ON OVER THE AIR ??

Regards,

Steve

[Bacon, WA3WDR]

Scopes are not so important to PSK mode, CW, SSB and FM operators because on PSK modes, there is a spectral display and splatter is visible.  On CW, if it doesn't splatter with key clicks, or really bounce around with power supply filter resonances, etc, it's fine.  With SSB, peak flattening is almost inaudible, and splatter is simply blamed on the receiver.  With FM, if you overdeviate much, you cut out badly.  If you don't deviate enough, you have low volume.  It's so-so quality with peak clipping followed by low-pass filtering mandated by the FCC, and people are satisfied.

But AM.... well, we care on AM.

[KB2WIG]

    " But AM.... well, we care on AM. "

The kiinder,genteeeler mode...... klc

[k4kyv]

I have trouble monitoring my own signal with a receiver, even with the antenna disconnected and everything well shielded.  A certain amount of crud always seems to get through.  A simple xtal detector running directly into an audio amplifier with headphones is a much better approach.  Also, the selectivity filter in the receiver will alter the audio.  I sometimes use the receiver in narrow selectivity to serve as a manually scanning spectrum analyser, but again the trick is to get rid of the crud.

It is difficult to tell when using voice modulation, but if I test with an audio sinewave, and it looks good on the scope, I have always found the distortion level to be satisfactory over the air.   Perhaps a distortion meter would indicate otherwise, but I have always found audible distortion and/or splatter to be visible on the scope pattern.  Of course, with voice modulation you would need a dual trace scope and superimpose the input signal to the rf envelope pattern, since it is impossible to visually determine what is distortion and what is the natural waveform of a complex signal like the human voice.

One problem when testing the audio at modulation levels above 100% positive, using heavy metal, is the risk of blowing the modulation transformer during heavily overmodulated negative peaks.  A diode in series with an external wirewound loading resistor equivalent to the modulating impedance will protect the modulation transformer.  Connect the cathode of the diode to the modulated +HV line, and ground the bottom end of the resistor.   As soon as the final amplifier tube plate(s) go negative, the diode conducts, and the loading resistor maintains a constant load on the transformer during the period of time the final is overmodulated in the negative direction.  The resistor need be only rated at a small fraction of the DC input power to the final, since its duty cycle will be very low, as it conducts only during negative peaks exceeding 100%.  Of course, the PIV of the diode will have to be the full modulated HV positive peak voltage, plus a reasonable safety margin.

This test should be very brief, since a modulator that easily supplies modulation voice peaks at 150% or higher, may run well in excess of the rated plate dissipation if you try to do the same thing with a sinewave.