Another unusual waveshape...

[Paul Burridge]

Hi,

I'll avoid posting a screen shot of this one since so many people seem
to have difficulty finding them on abse or can't get the group at all
from their servers.

Imagine a two-state, high/low (on/off) wave form; i.e., negligible
rise/fall times. Amplitude 5V. A bit like a square wave, in fact,
except that the on and off period appears *different* between one
cycle and the next and the next and so on. There's quite a bit of
variation (anything from one whole time division to up to seven divs
and all times between the two). There does appear to be an eventual
series repetition, but it's only noticeable after about 30 cycles. I'm
not sure how I've ended up with this, but would it be rich in odd and
even order harmonics and suitable for generating them for frequency
multiplication purposes, d'you think?

Thanks,
P.

 

[Joerg]

Hi Paul,

Seems like a huge amount of frequency modulation. Or a pulse width
modulator? Where did it come from?

abse should be accessible from most servers. But one technique that is
easier for people is to upload it to a "scratch pad section" on you web
site, a sub domain which most hosting companies let you create. Then you
can use a fast FTP tool such as WISE-FTP and zoom it up there, and delete
it within seconds once you think it's no longer of interest. This allows
you to paste a link right here in your post. Just avoid doing it the cheap
way, via one of those "free" web space deals that swamp people with
pop-ups.

Regards, Joerg

 

[Walter Harley]

I'm
not sure how I've ended up with this, but would it be rich in odd and
even order harmonics and suitable for generating them for frequency
multiplication purposes, d'you think?

Seems to me that if you're not sure how you ended up with it, then it's not
suitable for anything, because you don't know when it'll change to something
else nor how to make it happen reliably.

One way to end up with a waveform like that is to pass random noise into a
comparator. Another is to use a shift register to create a pseudorandom
sequence generator, and then look at one bit of it.

 

[Robert Baer]

Hi,

I'll avoid posting a screen shot of this one since so many people seem
to have difficulty finding them on abse or can't get the group at all
from their servers.

Imagine a two-state, high/low (on/off) wave form; i.e., negligible
rise/fall times. Amplitude 5V. A bit like a square wave, in fact,
except that the on and off period appears *different* between one
cycle and the next and the next and so on. There's quite a bit of
variation (anything from one whole time division to up to seven divs
and all times between the two). There does appear to be an eventual
series repetition, but it's only noticeable after about 30 cycles. I'm
not sure how I've ended up with this, but would it be rich in odd and
even order harmonics and suitable for generating them for frequency
multiplication purposes, d'you think?

Thanks,
P.

Randumb duty cycle modulation would only alter the lower harmonic
spectrum; the rise anf fall times are the major source of frequency
multiples.

 

[N. Thornton]

Hi,

I'll avoid posting a screen shot of this one since so many people seem
to have difficulty finding them on abse or can't get the group at all
from their servers.

Imagine a two-state, high/low (on/off) wave form; i.e., negligible
rise/fall times. Amplitude 5V. A bit like a square wave, in fact,
except that the on and off period appears *different* between one
cycle and the next and the next and so on. There's quite a bit of
variation (anything from one whole time division to up to seven divs
and all times between the two). There does appear to be an eventual
series repetition, but it's only noticeable after about 30 cycles. I'm
not sure how I've ended up with this, but would it be rich in odd and
even order harmonics and suitable for generating them for frequency
multiplication purposes, d'you think?

Thanks,
P.

Think about it this way: say it has at least one 3f half cycle per 30
cycles. Therefore if you feed it to a high Q filter that rings at 3f,
it will duly give you 3f.

But only if the 3f is present in the original wave. To ring at 3f off
one half cycle per 30 cycles it needs to be reasonably high q, and a
high q filter is not going to work off 2.7f, so a fixed pattern is
going to be better than a random pattern that might not contain the
right f at times.

I wouldnt be surprised if someone critiques this 'visual' analysis
though, I wait and see.


Regards, NT

 

[Paul Burridge]

Seems to me that if you're not sure how you ended up with it, then it's not
suitable for anything, because you don't know when it'll change to something
else nor how to make it happen reliably.

One way to end up with a waveform like that is to pass random noise into a
comparator. Another is to use a shift register to create a pseudorandom
sequence generator, and then look at one bit of it.

That gives me an idea, Walter... How about I amplify the white noise
from a suitably noisy diode and feed it through a Schmidt or something
similar? What sort of harmonic richness would that give me?

 

[Walter Harley]

That gives me an idea, Walter... How about I amplify the white noise
from a suitably noisy diode and feed it through a Schmidt or something
similar? What sort of harmonic richness would that give me?

I'm sorry, but I've not been keeping track of what kind of problem you're
trying to solve. It sure sounds like you're trying to reinvent some sort of
wheel here, though.

If you are trying to start with a noise source in order to generate a signal
at a specific, adjustable frequency, then surely you must consider that the
way to do that is to filter the noise source; and that the task of making a
good tuneable filter is no different than the task of making a good tuneable
oscillator. And starting with broadband noise is not generally a good
approach to getting something quiet.

 

[YD]

Hi,

I'll avoid posting a screen shot of this one since so many people seem
to have difficulty finding them on abse or can't get the group at all
from their servers.

Imagine a two-state, high/low (on/off) wave form; i.e., negligible
rise/fall times. Amplitude 5V. A bit like a square wave, in fact,
except that the on and off period appears *different* between one
cycle and the next and the next and so on. There's quite a bit of
variation (anything from one whole time division to up to seven divs
and all times between the two). There does appear to be an eventual
series repetition, but it's only noticeable after about 30 cycles. I'm
not sure how I've ended up with this, but would it be rich in odd and
even order harmonics and suitable for generating them for frequency
multiplication purposes, d'you think?

Thanks,
P.

What's the circuit supposed to do? What's the frequency? What is it
made of? What's on the inputs?

You probably screwed up as usual and need your betters to bail you
out.

- YD.

 

[Paul Burridge]

I'm sorry, but I've not been keeping track of what kind of problem you're
trying to solve. It sure sounds like you're trying to reinvent some sort of
wheel here, though.

If you are trying to start with a noise source in order to generate a signal
at a specific, adjustable frequency, then surely you must consider that the
way to do that is to filter the noise source; and that the task of making a
good tuneable filter is no different than the task of making a good tuneable
oscillator. And starting with broadband noise is not generally a good
approach to getting something quiet.

I know. This is purely hypothetical. What sort of spectral range would
one get from this idea?

 

[Paul Burridge]

What's the circuit supposed to do? What's the frequency? What is it
made of? What's on the inputs?

You probably screwed up as usual and need your betters to bail you
out.

Damn! Forgot to tick the 'global' option in the Usenet filters list...

<Plonk!>
(again)

 

[Walter Harley]

[...]

I know. This is purely hypothetical. What sort of spectral range would
one get from this idea?

Presumably that depends on how "suitably noisy" your diode is, and how fast
your Schmitt trigger is.

Here's a project for you: how can you describe the behavior of a
(real-world) Schmitt trigger in the frequency domain, that is, in terms of
how it affects a signal's frequency spectrum?

 

[Paul Burridge]

That gives me an idea, Walter... How about I amplify the white noise
from a suitably noisy diode and feed it through a Schmidt or something
similar? What sort of harmonic richness would that give me?

[...]

I know. This is purely hypothetical. What sort of spectral range would
one get from this idea?

Presumably that depends on how "suitably noisy" your diode is, and how fast
your Schmitt trigger is.

Here's a project for you: how can you describe the behavior of a
(real-world) Schmitt trigger in the frequency domain, that is, in terms of
how it affects a signal's frequency spectrum?

To put it technically, "they generate loads of harmonics" - a perfect
one presumably an infinite comb of harmonics? Am I near the mark?