View Single Post
  #10   Report Post  
Old August 29th 04, 09:43 AM
Ian Jackson
 
Posts: n/a
Default

In message , Reg Edwards
writes
A few turns on a teeny ferrite ring 1:2 transformer, centre-tapped
secondary, and a couple of diodes will double the frequency without any
power loss.



What Reg is describing here is undoubtedly the BEST and EASIEST and MOST
PREDICTABLE way to double the frequency of a signal. You can also use it
to obtain x2 and x6 etc, but with reducing efficiency. I have frequently
used it to get signals up to about 1000MHz.

It s exactly the came circuit as for a bi-phase rectifier on a mains
power supply, ie transformer with centre-tapped secondary and two
diodes.
The shape of the output waveform is the same as what you get from the
rectifier circuit, ie a series of half sinewaves (all +ve or -ve,
depending on which way you connect the diodes). This contains an
infinite series of even harmonics and no odds (including the
fundamental).

For the mathematical, the theoretical level of each harmonic can be
predicted from a knowledge of what Mr Fourier says about this shape of
waveform (from the coefficients associated with each frequency). See
below for several examples (just click on the examples at the bottom of
the page).
http://www.efunda.com/math/fourier_s...ier_series.cfm

In theory, there are no losses. But you are splitting a single signal
into a load of even harmonics, so the level of each must be less than
the original signal.

In practice, there ARE losses (in the diodes and the ferrite). From
memory, the x2 is about 8dB down, x4 is about 16dB down, x6 is about
24dB on the original signal. These ratios are essentially independent of
drive level, provided you have sufficient to overcome the knee of the
diodes. If you are after the x4 and x6, you can actually do things which
reduce the loss somewhat, and make the multiplication more efficient.
But I digress...

Finally, a big advantage of this circuit is that the odd harmonics are
well suppressed. They are not zero, but are typically 30 - 40dB less
than the adjacent evens (depending on the balance of the circuit). This
makes any filtering (if the application requires it) SO much easier.

Cheers,
Ian.

--