Diode Ring Mixer with LEDs

I wanted to use the Si5351 clock generator in combination with the LED ring mixer. The Si5351 together with an Arduino Nano as controller is easy to use and makes every frequency i want. Output level is about 13-14 dBm in 50 Ohm, that is 1 Volt. That is fine to switch on a regular diode, but not enough to enlighten an LED. To keep things simple, i did not want to use an extra amplifier, so i changed the l.o. transformer of the LED ring mixer. Normally the l.o. transformer has 1 primairy winding and 2 secondary windings. Both secondary windings have the same number of turns as the primairy one. Therefore each secondary winding provides the same voltage as provided to the primairy. To have the red LEDs lighten up, i needed about 2 V, so i doubled the number of turns on each secondary winding. The result you see on the picture. The 4 LEDs lighten up nicely and evenly. If there would be unbalance, you would see 2 LEDs more bright than the other pair.
About the LEDs: i am using an IF of about 8 MHz, so not a relatively low 455 kHz if. That means that these LEDs need to be fast. I checked a lot of datasheets for speed of response, and only a few manufacturers specify it. Avago (HP) does it, so i selected an LED with a short switch on / switch off time and a relatively small capacitance. Furthermore i selected an LED with the smallest Uf, so that means a RED LED. I chose the HLMP-EG. It has a speed of response of 20 ns, so on and off would be 40 ns. That means that it could handle frequencies up to 25 MHz, suits my needs. The LED capacitance in the off state is 40 pF, needs to be charged and decharged from the transformer. As i am stepping up the Voltage by 2, the impedance will go up by 4, so 4 x 50 = 200 Ohm. This impedance and capacitance (200 Ohm / 40 pF) is a low pass filter of about 20 MHz. This suits also our needs.
In practise, the LED mixer works very well, audio sounds well and it is simply nice to see all 4 LEDs lighten up to do the mixing job!

About the transformers: i used this time a multi aperture core, for a change. The BN-73-202, as it is widely available. Zi should be 5-10 times the imput impedance, so about 250-500 Ohm. I used 4 turns on the primairy side, and 2 times 8 turns on the secondary side. In this case, you can use this transformer also at 455 kHz as if, without changing anything. I used wire of 0.25mm, that is a nice size for this job. Not too thick that it doesn't do the number of turns, not too thin that it is uneasy to handle.
If you want you can also use the FT50-43, use the same number of turns for a frequency of 8-9 MHz, the standard if frequencies.
I first wound both secondary windings, and not bifilar, as it's simply uneasy. I took the middle of the wire and wound one turn clockwise for one winding, and the other end anti-clock wise one turn for the other winding. And so on, until you have 2 x 8 turns for both windings. Then the phase will automatically be correct. When you loose count for one winding, you see it directly as the ends of both wires have unequal length.
On top of the secondary winding you put the primairy of 4 turns, and you are done.
You can directly connect to the Si5351 output using a blocking capacitor, i used 10nF for 8 MHz, use a somewhat bigger value for a lower frequency.
You can check directly if you want that is works, by connecting the 4 LEDs to the l.o. transformer. When you switch on the BFO, the LEDs should go on and lighten evenly.
The transformer for the IF side, i used the same BN-73-202, and wound it with 3 windings of 4 turns, trifilar. Maybe i should have used the same windings as for the l.o. transformer. This is something for further investigation.