Adding an Envelope Detector to convert an RF input to a DC output

I connected the ADC Pi module to the GPIO header and stacked the DDS module on top. PCB support spacers were used to hold the modules together and to take the strain off the GPIO pins.

And here's how it looks from the other side...

All that was left to do was to use a simple envelope detector to convert the RF coming out on the circuit under test into a DC voltage that could be measured by the ADC Pi module - time for a little bit of theory...

A simple envelope detector is shown above. It consists of only three components - a diode, a capacitor and a resistor. An RF signal (Vi) is fed into the detector and a DC signal (Vo) comes out the other end. The RF input (blue line) and DC output (red line) signals are plotted below.

When I tried using a simple envelope detector I discovered that I was getting little or no signal out of the envelope detector. So much for the theory! I measured the peak RF voltage coming out of the circuit under test I found it to be around 0.2Volts (200mV) - this was the cause of the problem! Time for some more theory...

With reference to the graph above, a diode needs a certain level of forward voltage "Vd" to "turn it on". For a Silicon diode this voltage is about 0.7 Volts and for a Germanium this voltage is around 0.25 Volts (250 mV), which is around the same level as (or greater than) the peak RF signal voltage I was trying to detect. Therefore the RF signal I was trying to detect and measure was insufficient to drive the simple envelope detector.

The solution was to use a buffer amplifier with a high input impedance and a biased detector stage to convert of the RF signals to a DC voltage. A design by Rodney Byne for the front end of a Wideband RF Millivoltmeter was adopted and modified for use with the RPi Wobbulator. The circuit diagram is shown below:

The circuit uses a JFET transistor in “source follower” configuration, which has a gain of less than 1, but has a very high input impedance, therefore minimising the loading on the circuit under test. Diodes D1 and D2 are both Schottky type 1N5711 which have good RF bandwidth. With no RF input, VR1 was adjusted for a quiescent bias voltage of about 0.25 V (250mV). This seemed to be a good compromise between quiescent power consumption and linear performance when dealing with small RF signals.

The Buffer Amp / Detector was constructed on a prototype board and housed in a diecast aluminium box. A PP3 battery was used to power the circuit. The completed Buffer Amp / Detector is shown below.

The Buffer Amplifier / Detector was tested at various frequencies and at various input levels, and the DC output (minus the quiescent bias voltage) was found to be approximately equal to the peak voltage of the RF signal applied to the input. In other words, when a 0.2 Volt (200mV) peak RF signal was applied to the input of the Buffer Amplifier / Detector, the DC output from the Buffer Amplifier / Detector was approximately 0.2 Volts (200mV), plus the the quiescent bias voltage. This is well within the measurement capabilities of the ADC Pi module. Happy days!

So that about wraps it up for the hardware side of the Raspberry Pi Wobbulator. In my next post I'll discuss the development of the Raspberry Pi Wobbulator software.