LED at transformer secondary (50 kHz) not lighting up. Why? Capacitance? Trr ?

[JeanF]

Hello,

I have a smart electricity meter that has a data output at 9600bps, on-off keyed on a 50 kHz carrier frequency. The carrier is a distorted sine, around 5 Vpp, available on a terminal block where the user is expected to connect twisted pair. This output is isolated from everything else by a signal transformer. There is a mains-rated cap in series with one of the legs at the secondary of the transformer, to limit the current to a non-destructive value in case someone accidentally connects mains to the terminal block. This cap is the only thing present at the secondary, besides the terminal block.

I was playing around with interface circuits in order to read the data stream with an Arduino or Raspberry etc., and I noticed something odd :

I placed an LED in series with a resistor across the two output terminals, and was expecting the LED to light up with varying brightness as bits are transmitted. (the meter is transmitting continuously). I thought that in this half-wave rectification scenario, the LED would let current flow during one half of each cycle and light up accordingly. But no... the LED doesn't light up. Instead, here's what I found:

- I'm not able to measure any current in the LED by probing the series resistor with my Analog Discovery (which has diff inputs)
- the LED emits a very brief and dim little flash of light when I connect it (with its resistor) in one way; then nothing, then again a brief flash if I manually swap the wires.
- when touching one of the legs of the LED with my fingers, there seems to be some sort of leakage and the LED lights up very dimly
- if I add another LED  in antiparallel with the first one, then the two LEDs light up brightly as bits are transmitted. Same thing with a 1N4148

How is that possible? I'm puzzled. I know it's a good idea to put another diode in reverse across the LED to clamp the reverse voltage and avoid damaging the LED, but I can't see how this second diode could be needed  for the first LED to light up in the 1st place... Is this "blocked-in-OFF-state" thing a standard feature for LEDs that I just didn't know?

Also,

- I used my Analog Discovery to output a 50 kHz sine into the same LED + resistor arrangement, this time the LED lights up as expected
- I also tried to record the actual waveform with my AD and then play it back : same thing, the LED lights up
- in the electricity meter, the secondary of this data output transformer is fully floating. So I repeated the two tests by adding another transformer (salvaged from another utility meter, so I know its suitable for 50 kHz) after the AD2 : same thing, the LED lights up again

That's not really a problem (I have a working demodulator, some of you may remember old posts of mine about that) but I'm just really curious. Thank you for any hints about what could be happening that prevents current from flowing into the LED!

[Benta]

The "secret" is described in your first paragraph.
The protective capacitor is the culprit. When you only do half-wave rectification, as your LED does, the cap will build up a DC bias voltage until your LED no longer lights up.
The anti-parallel diode (LED, 1N4148, whatever you have in your drawer) prevents this DC bias.

Your symptoms fit exactly to this behaviour.

[mikerj]

In simple terms you've added a half wave rectifier to the circuit (the LED) so converting the AC signal to DC, but capacitors block DC. Adding an anti-parallel LED or signal diode allows current to flow through the capacitor on both +ve and -ve cycles of the AC signal.

[David Hess]

Once the LED half-wave rectifies the signal, the series capacitor has enough DC bias to prevent the LED from turning on.  A resistor across the LED to discharge the capacitor would have the same effect as an anti-parallel diode.

[JeanF]

Thanks to all three of you. I get it now !

indeed I forgot/ignored the series capacitor when I tried to find the solution...  "well, caps pass AC so no problem, nothing to see here" ... D'oh !

I wrongly focused on the textbook-like schematic of the half wave rectifier with only a transformer winding, a diode and a resistor.

Oddly enough I remembered the capacitor when writing my first post, but that wasn't enough to put me on the right track...

indeed, when the cap has finished charging, the current in the loop drops to 0! Thanks again.

[TimFox]

There are two common ways to connect a single diode and single capacitor to make half-wave rectifiers:
The usual way, after a grounded-secondary transformer, the series diode develops the DC voltage across the capacitor to ground.
The other way, with series capacitor from source and diode to ground, works as well, developing the DC voltage across the diode. 
The important difference is that the DC level of the input contributes to the output in the first case, but only the variation of the input contributes to the output in the second case.