# High frequency AM demodulation

I'm building an light communication device(pretty much lifi) using AM and am mostly done with every part except for demodulating the signal. The signal looks like with a 1.8432MHz carrier wave(serial communication at 115200Hz). The receiver consists of a photodiode, amplifiers, filters, and it requires an envelope detector or a circuit that outputs high when theres a AC signal present.

I've tried using a

1. A full and half bridge rectifier(with capacitor)

A lot of signal is suppressed with this can capacitor just shunts everything to ground.

1. Envelope detector (https://en.wikipedia.org/wiki/Envelope_detector#/media/File:Simple_envelope_detector.svg, basically peak detector):

It produces a lot of voltage spikes from capacitor discharging and does not really produce any recognizable demodulated wave

1. A diode and transistor

Transistor has quite a high breakdown voltage at 0.7V and the capacitor's response time is way too fast

1. A diode and capacitor

This circuit works for low signal frequencies but fails at higher signal frequencies

I'm also considering using a slow op-amp with response time faster than signal but slower than carrier wave, would it be reasonable? If so how should I build the circuit. Will there be any better options?

I have access to basic resistors, capacitors, some transistors and diodes, other components would have to be sourced.

Voltage graph after amplification and filtering:

Voltage graph with diode pointing away from ground:

Amplifier with filter:

Circuits tried:

• 1) include your actual circuit. Yes even when the circuit is "trivial". 2) make clear what (kind of) circuit supplies the unmodulated signal, include a schematic, the actual type of diode and capacitor value matter so show them. 3) There are many ways to do the "correct" thing in the "wrong" way so the fact that a circuit doesn't work does not mean it cannot work. You are confusing yourself with all these type of demodulators. Start with the simplest one (diode and capacitor) and do more effort to make it work. Jun 6, 2018 at 7:36
• What diodes are you using in your detector? Jun 6, 2018 at 9:30
• Schottky diode SD103A Jun 7, 2018 at 2:08
• Your 'scope photo shows a large signal, your final transistor looks to be over-driven into its non-linear region (clipped). A diode detector might accept such a signal to give sensible output. A much smaller signal won't reach a diode's threshold. You have a dynamic range problem, where the optical link amplitude must be adjusted carefully to get a detectable signal. Solutions: use an amplifier of type automatic-gain-control (AGC). RSSI type detectors have wide dynamic range, often found inside chips. Both approaches are not straightforward with discrete transistor circuits. Aug 23, 2020 at 14:03
• Don't forget you have the ability to bias things. So if the 0.7V forward voltage of the diode or transistor is causing a problem for you, don't let it be a problem, just add 0.7V onto your signal! (signal -> capacitor -> base, some volts -> resistor -> base) Aug 26, 2021 at 11:52

I have access to basic resistors, capacitors, some transistors and diodes, other components would have to be sourced.

If you don't use the correct diode you will have problems. Take the 1N4001 diode.

This one has a reverse recovery time of 2 us and given that your carrier is well over 1 MHz it just will not cut-the-mustard.

This one is specified at 30 us and will be useless.

So, check what diode you are using and if the specification in the data sheet does not show a figure then you can safely assume it will not work. Bridge rectifiers will not work and probably most of them will not specify reverse recovery time at all.

Alternatively, consider the 1N4148. These are really fast with a reverse recovery time of sub 10 ns and are ideal for what you want. Also consider the BAS16 if you are going surface mount. It is about 5 ns.

• I'm using schottky diode at the moment, can't remember which one, will have to check tomorrow Jun 6, 2018 at 10:35
• Please also check what capacitor you are using. Jun 6, 2018 at 10:49
• I'm using yellow ceramic capacitors Jun 6, 2018 at 13:19

Try this

simulate this circuit – Schematic created using CircuitLab

• Would this still work if the amplitude is around 50mV? Jun 9, 2018 at 9:41
• You might need to increase the ratio R1/R3 to create some asymmetry/ non-linearity in the charging and discharging of C2. Feb 26, 2019 at 2:05

I haven't done the math to recalculate capacitor values for your frequency range, but I found this circuit on the web which seems quite easy to implement. demodulator

Cp would need to be increased to 0.05uF and Cr will have to be adjusted for good detection of 115200 Hz, use a scope. If you end-up with too much gain, lower R2 value. D1 D2 should be germanium for better sensitivity. 1N34 is ideal.