How to improve almost any phaser

[Gurner]

360^o of shift is one full cycle delayed with respect to the dry signal. It may look the same on a 'scope when analysing sine waves, but with a harmonically complex guitar signal.........

All a 'harmonically complex' signal is...is one made up of a lot of sine waves...so the theory stands. (besides most of the harmonics very quickly drop out....like within a few hundred milliseconds -  and depending where you're playing on the fretboard, your often left with something akin to the fundamental ...which is in essence a sine wave)

....and how can opamps possibly perform the time travel back into the past that would be neccesary for a 360^o shifted signal (and therefore delayed by one full cycle) to occupy the same time period as the dry signal?

Crazy  I know.... I never said I had the explanation, just the concept down! Equally, Rod Elliot hasn't got the explanation, but it does happen (section 7.5)....

http://sound.westhost.com/articles/active-filters.htm#s75

Extract....

"Version "A" produces a lagging phase. That means that the output signal occurs after the input. For the values shown, the delay is about 155us with a 1.59kHz signal. Version "B" has a leading phase - the output signal occurs before the input. While this seems impossible, for a signal that lasts more than a few cycles it really does happen. In the second example, the output occurs 155us before the input"

In reality, this is just one of those polarity peculiarities that depends on how you view it (ie where you're referencing ...eg -5V  will look like +5V if your 'reference' level is -10V!) . Is a signal with a 350 degree lag any different to  a signal with a 10 degree lead wrt to the original signal ....no it's not, they're the same.

If you think a 361 degree phase shift is any different to a 1 degree phase shift, you need to do a bit more delving.

[slacker]

- the output signal occurs before the input.

No it doesn't, it just looks like it does.

[Gurner]

- the output signal occurs before the input.

No it doesn't, it just looks like it does.

ahem - That quote was Rod Elliot's quote not mine! (& Rod's one of the main men in my eyes)

but like I say, it's one of perspective - after one cycle, a 340° phase lag essentially can be equally viewed/termed as a 20° phase lead wrt to the ref (dry) signal ...but we're losing the main thread/point here...and that is, you can't get any more phase delay than 360°! (ie one complete cycle wrt any given frequency). And that once you're at 360°, you're back where you started.

What you do get by adding more & more (modulated) stages....is the ability to shift lower frequencies further in phase - & more phase resolution for higher frequencies, eg 1°, 1.01°, 1.02° vs say 1°, 2°, 3° that you'd get with less stages (but with static phase shift stages, all you're getting is a phase 'offset').

[slacker]

I completely agree Rod's a top man.

This is how I see it though, which I think is relevant to what Rick is saying. If you look say at one of the troughs in the wave form, then the output looks like it's leading the corresponding input, but in reality the input that created that output trough isn't that bit of the wave that it appears to be leading. The input is the preceding peak, the phase shift turns this in a trough, so it then looks like it's leading the input, but it's not really, the output always lags behind the bit of the signal that actually created it. Which is what Rick was saying, if you phase shift it enough the bit of the waveform that creates the output won't be the preceding peak it will be the one before it, or even the one before that.
If the signal is completely uniform then, like you said this is irrelevant, but I don't know enough to say whether this would necessarily be the case for a guitar signal.

[Gurner]

I completely agree Rod's a top man.

This is how I see it though, which I think is relevant to what Rick is saying. If you look say at one of the troughs in the wave form, then the output looks like it's leading the corresponding input, but in reality the input that created that output trough isn't that bit of the wave that it appears to be leading. The input is the preceding peak, the phase shift turns this in a trough, so it then looks like it's leading the input, but it's not really, the output always lags behind the bit of the signal that actually created it. Which is what Rick was saying, if you phase shift it enough the bit of the waveform that creates the output won't be the preceding peak it will be the one before it, or even the one before that.
If the signal is completely uniform then, like you said this is irrelevant, but I don't know enough to say whether this would necessarily be the case for a guitar signal.

I read that a couple of times but i didn't fully understand the point you're trying to get across (a topic like this is actually hard to phrase in language agreeable to all readers!).

Where I came into this was wondering what the point of a static phase shift stage was...a rx'ed a reply that the explanation was in the first post,  & the suggestion there is that because phase stages are cumulative, that somehow by adding stages you get more 'delay' - for example 500 degrees of phase delay being more 'delay' than say 140 degrees - it's not ...they're the same for any given frequency (remember, we're not talking about delay as in the traditional 'time delay' sense, but phase delay through a filter ...& the max time you can actually time delay a given frequency in this situation is one 'period').

And the most phase delay you can get for any frequency is 359.99999999 recurring ...once you get to 360, you're back at the zero point.

That's the bit I'm picking up  on...not necessarily the resulting peaks/troughs (which are obviously a function of adding in/out of phase signals)

[slacker]

By peaks and troughs I meant looking at a sine wave fed into an allpass stage, and then looking at the output compared to the input. I didn't think about the double meaning in relation to a phaser, sorry for the confusion. Don't know if that makes my post make any more sense

[merlinb]

But when the FET is driven hard on then the + input basically is shorted to ground?? I got this:
http://s81.photobucket.com/albums/j207/merlinblencowe/?action=view&current=CIMG5703.mp4
I'm not seeing things right, that is swinging from zero to 180 degrees?

Further to my earlier post I see my mistake. One stage can theoretically give you up to 180 degrees shift, but you can never quite get that in practice. After cascading two stages I see that I'm get just slightly less than 360 degrees shift, and hence not as many notches as I assumed.
http://s81.photobucket.com/albums/j207/merlinblencowe/?action=view&current=CIMG5808.mp4

[frequencycentral]

http://www.youtube.com/watch?v=gYF2h5ry3kY&feature=related

[Taylor]

Gurner, what you are talking about, 361 degrees shift being equivalent to 1 degree, is true only in the special case of an unchanging periodic wave, which is constant both in waveshape and pitch at all times.

This does not hold true for any real life instrument. Although it's true that any signal is made up of a bunch of sine waves, these sines are all changing amplitude and phase through time, so as long as a cycle is not identical to the preceding cycle, you will have a result quite different from 0 degrees shift.

It's quite possible to simulate phasers in real time with Falstad's analog filter app:

http://www.falstad.com/afilter/

I will put together something that looks like 0 degrees phase shift on the scope, but which offers obvious phase shift in a frequency plot, so we can see what's happening.

Edit: here goes. The first link shows a 6-stage phaser on the scope. Input and output look to be the same, in the same phase.

http://tinyurl.com/2bv9lud

Now, we look at that in the frequency analyzer. Go to http://www.falstad.com/afilter/ and paste in this code (unfortunately the filter app does not allow direct code linking).

Code Select Expand

$ 1 5.0E-6 5 1 5.0 50
% 0 28853.998118144256
O 848 352 960 352 0
a 128 128 208 128 0 15.0 -15.0 1000000.0
r 208 64 128 64 0 10000.0
r 128 64 48 64 0 10000.0
w 208 64 208 128 0
w 128 112 128 64 0
c 128 144 48 144 0 5.0E-8 1.7514964460852651
w 48 64 48 144 0
r 128 144 128 224 0 50000.0
g 128 224 128 240 0
w 208 64 240 64 0
w 400 64 432 64 0
g 320 224 320 240 0
r 320 144 320 224 0 50000.0
c 320 144 240 144 0 5.0E-8 -4.232526694508851
w 320 112 320 64 0
w 400 64 400 128 0
r 320 64 240 64 0 10000.0
r 400 64 320 64 0 10000.0
a 320 128 400 128 0 15.0 -15.0 1000000.0
w 240 64 240 144 0
w 624 64 624 144 0
a 704 128 784 128 0 15.0 -15.0 1000000.0
r 784 64 704 64 0 10000.0
r 704 64 624 64 0 10000.0
w 784 64 784 128 0
w 704 112 704 64 0
c 704 144 624 144 0 5.0E-8 2.5647735075791105
r 704 144 704 224 0 50000.0
g 704 224 704 240 0
w 784 64 816 64 0
w 592 64 624 64 0
g 512 224 512 240 0
r 512 144 512 224 0 50000.0
c 512 144 432 144 0 5.0E-8 1.9216010425797787
w 512 112 512 64 0
w 592 64 592 128 0
r 512 64 432 64 0 10000.0
r 592 64 512 64 0 10000.0
a 512 128 592 128 0 15.0 -15.0 1000000.0
w 432 64 432 144 0
w 432 384 432 464 0
a 512 448 592 448 0 15.0 -15.0 1000000.0
r 592 384 512 384 0 10000.0
r 512 384 432 384 0 10000.0
w 592 384 592 448 0
w 512 432 512 384 0
c 512 464 432 464 0 5.0E-8 2.1448116575278924
r 512 464 512 544 0 27800.0
g 512 544 512 560 0
w 592 384 624 384 0
w 240 384 240 464 0
a 320 448 400 448 0 15.0 -15.0 1000000.0
r 400 384 320 384 0 10000.0
r 320 384 240 384 0 10000.0
w 400 384 400 448 0
w 320 432 320 384 0
c 320 464 240 464 0 5.0E-8 -4.147259331992092
r 320 464 320 544 0 50000.0
g 320 544 320 560 0
w 400 384 432 384 0
w 208 384 240 384 0
w 816 64 816 288 0
w 816 288 208 288 0
w 208 288 208 384 0
170 48 144 32 192 2 20.0 4000.0 5.0 0.1
r 624 384 848 352 0 992.0
w 48 144 48 624 0
w 48 624 848 624 0
r 848 624 848 352 0 992.0
o 0 16 0 34 10.0 9.765625E-5 0 -1
o 41 16 0 35 5.0 0.003125 0 -1

It's quite clear that we have 3 notches.

[Gurner]

There must be something wrong with my browser, because I can't view those links.

But when I see you talking of notches, - it's only when you start adding/subtracting the phase delayed signal back with the original dry signal do your get notching - that wasn't my point.

To be clear, I was picking up on an earlier statement about allpass stages being cumulative & therefore by adding more in, that you get in excess of 360 degrees phase shift - you don't (& can't!).

What you do get when adding in more stages is greater resolution  -  & more phase shift for lower frequencies

[puretube]

High time to re-read your good ole Dome, Bode, & Nyquist... 

[Gurner]

Not entirely sure we're all talking about the same thing! 

And re the video linked top above... (http://www.youtube.com/watch?v=gYF2h5ry3kY&feature=related )   ....at 6m46s he actually says the same thing I've being saying all along - ie that once you get past 360 degrees, you're starting at zero again  . And referring to something as having a 720 degrees phase shift, well, it's just a convenient way of saying a particular frequency has completed a full cycle of phase shift twice.

[puretube]

But when the FET is driven hard on then the + input basically is shorted to ground?? I got this:
http://s81.photobucket.com/albums/j207/merlinblencowe/?action=view&current=CIMG5703.mp4
I'm not seeing things right, that is swinging from zero to 180 degrees?

Further to my earlier post I see my mistake. One stage can theoretically give you up to 180 degrees shift, but you can never quite get that in practice. After cascading two stages I see that I'm get just slightly less than 360 degrees shift, and hence not as many notches as I assumed.
http://s81.photobucket.com/albums/j207/merlinblencowe/?action=view&current=CIMG5808.mp4

Once I find the time, I`ll link to a tube- (valve-) circuit that goes beyond 180° in one stage...