WaveForm Generator
 

Dear All,
I want to synthesize a SINC wave with main lobe width = 100ns,
Is it possible to use a LC filter response or
a DDS or a similar approach mey be?
Can anyone help me?
Thanks

I'm sure there are some readers who know more about this than I do, but
I haven't seen any responses. So I'll post what I know.

A Sinc wave is a non-causal waveform. That means that it starts before
the excitation impulse. A perfect sinc wave starts at t = minus
infinity, reaches its peak at t = 0, and extends to t = plus infinity.
So only a truncated or windowed wave of some sort is possible. But even
a truncated one is non-causal. That means you can't make it from any
kind of passive components like an LC filter -- you need some sort of delay.

I think the only way you'll be able to generate even a truncated or
windowed wave is with a digital filter. I believe an IIR (infinite
impulse response) as opposed to FIR (finite impulse response) will be
necessary, but I'm not terribly well acquainted with digital filtering
techniques so will have to defer to others on that point. The filter
would look something like ones used in digital signal processing (DSP),
so you might look at references dealing with DSP for some clues.

Roy Lewallen, W7EL

wrote:
Dear All,
I want to synthesize a SINC wave with main lobe width = 100ns,
Is it possible to use a LC filter response or
a DDS or a similar approach mey be?
Can anyone help me?
Thanks

On 16 Dec 2004 04:01:34 -0800, wrote:

I want to synthesize a SINC wave with main lobe width = 100ns,
Is it possible to use a LC filter response or
a DDS or a similar approach mey be?

An address counter, a lookup table (ROM/RAM) and a DAC should be able
to generate any arbitrary waveform.

Since the SINC waveform is symmetric relative to T=0, you just need to
store half of the waveform and reverse the up/down address counter at
T=0.

Since you are going to need a sample frequency of several hundred MHz
to get a clean wave shape, it might be a good idea to use a wide
memory system (say 64 bit wide RAM for 8 bit DAC) and run the RAM
address counter at fs/8 and either using digital multiplexers to feed
a fast DAC at fs or use 8 DACs in parallel at fs/8 and use a fast
analog multiplexer.

An other approach is to note that sinc(x) is sin(x)/x or 1/x * sin(x)
and generate 1/x and sin(x) separately and use an analog multiplier to
multiply them together. You also would have to ensure continuity at
x=0. The multiplier accuracy and dynamic range can be a limiting
factor in that region. Generating 1/x can be done with lookup tables
or with analog multiplier divider chips (which also could generate the
sin(x)/x directly) and there are of course multiple ways to generate
sin(x).

A DDS chip with AM capability might also be usable by feeding the 1/x
as the modulating waveform.

Paul OH3LWR

wrote in message
ups.com...
Dear All,
I want to synthesize a SINC wave with main lobe width = 100ns,
Is it possible to use a LC filter response or
a DDS or a similar approach mey be?
Can anyone help me?
Thanks


Do you mean in the frequency domain? that is one pulse 100 ns wide, and
that is it, no filter needed.

Anything else, you need to be clearer on what you want.

It depends on what kind of accuracy you need.

You could do this analog wise by building a filter that has fast cutoff and
linear phase in the passband, and then
applying a very short pulse to it.

A bessel filter is easy to design and has somewhat this characteristic.
Better would be a phase compensated chebeshev design, but I don't think you
will find that in a table!

If you look at these filter's responses with an oscilloscope, they will look
very Sinc like.

Cliff



wrote in message
ups.com...
Dear All,
I want to synthesize a SINC wave with main lobe width = 100ns,
Is it possible to use a LC filter response or
a DDS or a similar approach mey be?
Can anyone help me?
Thanks