I got intrigued by his simple circuit and decided to replicate his design.
It turns out that I have one surplus XTAL that would put me on 50.8 MHz and a canned oscillator that goes to 50.35 MHz. All simulations were done using my xtalfind Perl program.
The XTAL has too short leads to be used and it is not an overtone cut, so I decided to use the 25.175 MHz canned oscillator and try to double it with the EX-OR doubler described in the previous post.
The local oscillator signal has a period of T=40ns. I need to delay it of 1/4T, so 10ns. According to the 74HC86 datasheet this amount is right in the typical propagation delay range for 5V supply.
I've assembled dead-bug the circuit and done two tests:
- check the output at the oscilloscope (max BW 100 MHz, so quite at the limit)
- use my finger to check for the base frequency
- see the output spectrum on a spectrum analyzer
Test #1 is visual. The resulting waveform is more sinusoidal than it actually is because I've operated the instrument at the limit of its bandwidth. But the output period is half the input, so there's no 25 MHz output.
Test #2 confirm. Using my FT-817 receiver tuned on 25.175 MHz carrier I touched the canned oscillator output and noticed a signal increase. Then I touched the 74HC86 output and heard only an increase on 50.35 MHz.
Test #3. My spectrum analyzer starts at 45 MHz, so I have no mean to check what happens at the LO frequency. But I measured 5dBm at 50 MHz (through a 7dB attenuator pad) and about -6dBm and 100 MHz.
Curiously there was a rather strong product at 75 MHz that I could null adjusting the delay line.
So in theory I have a 12dBm signal (15mW) at 50.35 MHz. I could directly modulate the 74HC86 chip to produce AM, but the propagation delay is a function of Vcc too.
Now I can either generate AM or DSB-SC. Both need a linear amplifier chain, with DSB being more energy efficient. Since I have a digital signal I could probably drive a simple balanced mixer, or an NE602. In case I use a NE602 then it can be re-used as RX mixer too (for AM or SSB, not DSB).
If the canned oscillator could be pulled I could generate NBFM too, but I think this is not the case.