Variable Square Wave to Sine Wave

 

Hiya folks.  I've been interested in trying to build myself a radio
receiver for various bands, primarily as a learning experience.  The
thing is, though, I've heard that it's kind of hard to build a stable
VFO above 50mhz.  One of the bands I'd eventually like to receive is 2
meters (144-148mhz).  Might be fun to receive television audio, as
well.  Not to mention, standard FM.

Since I'm more familiar with digital electronic components rather than
analog, I've seen that there are programmable frequency oscillator
ICs, which can be set via a microcontroller.  This sounds like an easy
solution for tuning, and would open the door to setting station
presets and the like as well.  The problem is, I'm fairly sure that
these would output a square wave.

I've read about how you can basically use a low-pass filter to tune
out harmonics of a square wave and get a sine wave output.  But from
the way I understand it, an RC filter would just filter out a specific
frequency, defeating the purpose of the variable oscillator.

So that's pretty much my question:  How does one turn a variable
square wave into a sine wave?

Or, alternatively, what are more reliable ways to produce higher
frequency sine waves to begin with?  Though I'm still interested in
the former question as well, for curiosity's sake.

Thanks!

- Jeff

 

On Mon, 8 Mar 2010 07:29:40 -0800 (PST), FyberOptic


You could use a DDS chip. These guys make quite a few:
<http://www.analog.com/en/rfif-components/direct-digital-synthesis-dds/products/index.html>
Or, a lower frequency DDS -- something that can be built with a
microcontroller. Analog has a note on the fundamentals of DDS
http://www.analog.com/static/imported-files/tutorials/MT-085.pdf
(no pun intended) or do a net search on "DDS phase accumulator" or
similar. Pair the lower freq DDS to a phase-locked loop (PLL) and you're
there (modulo tuning the loop and filtering the assorted artifacts).

Here's http://www.myplace.nu/avr/minidds/index.htm  an example of using a
microcontroller to make a DDS. I've built this one (back when the
AT90S2313 was new ;-) and it does work as advertised.

--
Rich Webb     Norfolk, VA

 

On Mon, 8 Mar 2010 07:29:40 -0800 (PST), FyberOptic


Back up a wee bit: How is the square wave being generated?  Perhaps
from a triangle-wave oscillator?
        
                                        ...Jim Thompson
--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
            
      The only thing bipartisan in this country is hypocrisy

 

On Mon, 08 Mar 2010 10:21:41 -0700, Jim Thompson


Just realized that you are a gmail/google poster (kill-filed).  If
you've already replied to the above question, please re-post... you're
now in my "exception" filter.
        
                                        ...Jim Thompson
--
| James E.Thompson, CTO                            |    mens     |
| Analog Innovations, Inc.                         |     et      |
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
| Phoenix, Arizona  85048    Skype: Contacts Only  |             |
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  |
| E-mail Icon at http://www.analog-innovations.com |    1962     |
            
      The only thing bipartisan in this country is hypocrisy

 

FyberOptic wrote:

A square wave is made up of a sum of sinusoids in a special relation. In
fact all periodic functions can be thought of as a simple sum of sinusoids.
The only thing different is the relative amplitudes. A square wave has only
odd harmonics(integral multiples of the fundamental frequency) while a saw
tooth has every harmonic.

The low pass filter reduces the amplitudes of the higher harmonics which
essentially emphasizes the lower harmonics. Note that if you have a
fundamental frequency of f then the next harmonic is at 2f, the 3rd at 3f,
etc... So an ideal low pass filter inbetween f and 2f of *any* periodic
signal will return a sinusoid of frequency f.  Between 2f and 3f one gets
the sum of first two harmonics.

If you place a notch filter on any of the harmonics with a high Q then you
can "pick" out that frequency if one exists and it will be
sinusoidal(ideally). So say you have a square wave at 10Mhz. This mean's it
has sinusoids at 10Mhz, 20Mhz, 30Mhz, 40Mhz, etc... If you want to get some
sinusoid at 1000Mhz then you simply create a notch filter at that position.
Unfortunately the harmonics amplitude decause as 1/k which means that
amplitude of that 1000Mhz sinusoid will be 1/100th of the original.

If we used a 1Mhz square wave then it will be not only harder to pick out
the 100th harmonic because the amplitude is 1/1000th of the original but
also because now the harmonics are only 1Mhz apart which means our notch
filter has to have a much higher Q. (and this analysis is not considering
external noise involved)


Your first question is, at least for a rough method, is to simply LP any
signal. Eventually you'll arive at the fundamental which, ideally, will be a
perfect sinusoid. The second question's answer is to notch filter a
"spectrally rich signal. But as I have explained, this gets harder and
harder to do. You won't be able to take a 1V 1Hz square wave and recover a
1Ghz sinusoid as the amplitude is 1pV. Not only that you have other
harmonics right next to it. Of course depending on your application it may
not be all that important. If you used a 100Mhz signal then the amplitude is
only 0.1V and the side harmonics are 100Mhz away.