180 degree phase shifter

...

not

the

of

To trigger neurons or nerve fibers directly, I'd use a pulse.

Have Fun! Rich

Must this device do this with any combination of frequencies within this band, or only on one frequency at a time?

I have seen circuits that shift a fixed 90 degrees over a limited audio spectrum in a Single Sideband Transmitter. It was quite complex and required carefull control of tolerances on resistors and capacitors. To extend the range to a 1000:1 range, and then to add variable phase control would make the circuit even more complex. I doubt you will find what you are looking for without a lot of complexity.

-Robert Scott Ypsilanti, Michigan

The issue is that nothing simple will give you the phase shift throughout the range.

So long as you are dealing with one frequency at a time, it's easy. But if you need it to be broadband, and you do indeed state a broadband need, it gets difficult.

Most wideband phase shifters used a bunch of them in cascade, and maybe more important split the incoming signal into two channels. By careful math, the cascading phase shift circuits more or less work together, so throughout the range it is more or less a 90 degree phase shift. But we are usually talking a 10:1 range.

Since you seem to be talking about variable phase shift, I think the complication is infinitely worse.

If you are generating a signal, then you need to look at generating both phase signals right out of the box. Even if this is to shift a steady amplitude (and single frequency) signal that already exists, some sort of PLL may be the easiest solution.

Beyond that, conversion to other frequencies may be in the cards.

Better to step back and explain the need, because too often people go down the wrong path because they've jumped a step.

Michael

Anyone know of any (simple?)circuit that will do a unity gain variable degree phase shift between 0-180 degrees? Frequency range is 0.1 to 100Hz.

Could be. My natural inclination is to do it digitally ie A/D, hold and process, D/A Which is not difficult at the frequencies I want. At worst, I have 10mS to do all the maths.

Well, I don't anticipate using a very complex waveform so storing it, from one zero-crossing to another, and tracking its period should not be too difficult. I was hoping for an easy analogue solution though. The frequency would only drift rather slowly.

Well, the need is simple. I am applying a signal to the senses eg audio/visual/direct electrical. There are almost certainly delays associated with each mode of perception. If I want to eliminate those delays I have to apply different phases. Since I do not know those delays, and bearing in mind they may vary from person to person and according to time of application, a phase shifter is my first guess. I want the effect to hit the brains processing centres at the same instant, irrespective of nerve conduction rates.

degree

do

degree

not

the

of

Entrainment.

FFF Dirk

The Consensus:- The political party for the new millenium

Likely to slowly slide around this band, < 10Hz/min

I'm sure there is a simple signal processing algorithm for this in the frequency domain. If nothing else, you could take an FFT, convert the peak bin into phase/mag representation, increment the phase by your desired phase shift, then inverse transform.

Maybe ask in the DSP group. Is it comp.dsp, or comp.arch.dsp? I don't remember.

--Mac

Analog Devices sells a DDS chip - the AD9959 that has four outputs that can be programmed to have arbitrary phase differences (to 14-bits). The AD9958 has just two outputs.

These parts, and the single oputput AD9951, can be synchronised to other parts, each of which could be programmed with its own 14-bit phase shifts.

----------- Bill Sloman, Nijmegen

And you're not allowed to synthesize the 0.1 to 100Hz too ? In that case a quadrature DDS might do. They do have phase modulation registers.

Rene

If you are generating the signal can you generate a 2-phase sinewave?

Sin(A+B) = SinA.CosB + CosA.SinB

I've done this with a dual 4-quadrant multiplying DAC, (DAC8248), and a few opamps. Since a desktop computer was driving the system I just used the floating point to calculate the CosB/SinB phase-shift gain constants to send to the DACs. For a uP use a lookup table.

of

There _are_ a lot of complicated things going on in sensory transduction, let alone the "post-processing" that happens in higher centers. If you want _delay_, though, you don't want to simply shift each frequency by a single number of degrees (e.g. shift 100Hz by x degrees, shift 1kHz by x degrees).

In some cases (the auditory system comes to mind) specific stimuli can be generated (frequency sweeps) to make inner hair cells respond "synchronously" to incoming sound. Some interesting results occur with these stimuli.

If you want time delay, the amount of phase shift is linear with frequency. It is probably much easier to generate a real time delay with some sampled-data system (whether with some kind of bucket-brigade device, or ADC/digital delay/DAC system).

Perhaps it will turn out that you need what you've asked for, but at this point the odds of that don't seem high to me. I think you need to be a lot more explicit about what you're trying to do before chasing methods for doing what could be the wrong thing.

HTH -- -frank

"Dirk Bruere at Neopax" schrieb im Newsbeitrag news: snipped-for-privacy@individual.net...

Hi Dirk,

don't know if that works ...

Use a differentiator (OPAMP) to get a -90 degree shiftet signal and then do a linear combination of original and shifted signal. If one branch of the adder is capable of factors with both signs, you'll get 0-180 degrees.

The problems are (1) to get the differentiator to work over 4 decades, (2) to account for the magnitude response of the differentiator (you would have to use either a peak voltage detector, which is cumbersome at 0.1Hz, or square both original and differentiated signal, add them up and use that as the control voltage of an variable gain amplifier), and (3) to account for the non-unit magnitude after the (IQ-)addition.

Things get MUCH easier if you can put your hands on the signal generator, which, deep down in its bowels provides a quadratur signal already, I suppose.

Regards,

Paul

Use a Hilbert Transform. See