The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Behavioral Models >> Ideal model of 0/90 degrees phase splitter
https://designers-guide.org/forum/YaBB.pl?num=1233062253

Message started by Helge Stenström on Jan 27^th, 2009, 5:17am

Title: Ideal model of 0/90 degrees phase splitter
Post by Helge Stenström on Jan 27^th, 2009, 5:17am

I'm doing a behavioral investigation of a concept involving mixers, and I need a function that splits a signal in 0 and 90 degree phase versions. There is such function in Cadence rfLib, but it is not ideal. The phase difference is always 90 degrees, but the amplitude varies with frequency. It's possible to do better than that with a polyphase network, but my question is:

Is it possible to model a phase splitter with constant phase difference and constant gain in both paths? Or at least the same gain in both paths regardless of the frequency? Or a frequency range that is wide and parametrized.

My wanted frequency range is about 500 to 3000 MHz.

Title: Re: Ideal model of 0/90 degrees phase splitter
Post by pancho_hideboo on Jan 27^th, 2009, 5:59am

Helge Stenström wrote on Jan 27^th, 2009, 5:17am:

There is such function in Cadence rfLib, but it is not ideal.

What model do you mean "rfLib/shifter" or "rfLib/shifter_splitter" ?

Helge Stenström wrote on Jan 27^th, 2009, 5:17am:

Is it possible to model a phase splitter with constant phase difference and constant gain in both paths?

If you don't care numerical costs ehausted in simulation, you can build model to approximate ideal phase shifter.

For example, there are three implementations.
(1) Use ideal Hilbert-Transformer by FIR realization.
(2) Use complex BPF of high orders enough for realizing wide band.
For example, use ideal integrator.
Of course, you can use multi cascaded poly phase filter for realizing complex BPF.
(3) multiply exp(-j*omega0) and filter this result with LPF or BPF and then multiply exp(+j*omega0), here you can get analytical signal of original signal.

Anyway it must be always big burden in simulation if you model wide band phase shifter.

Title: Re: Ideal model of 0/90 degrees phase splitter
Post by Helge Stenström on Jan 27^th, 2009, 6:11am

Quote:

What model do you mean "rfLib/shifter" or "rfLib/shifter_splitter" ?

Any of them; they have similar behavior, but shifter_splitter has about 6 dB higher output when the outputs are not loaded, compared with shifter.

To reformulate my question: how can a phase splitter be modelled, such as it does not have 20 dB/decade amplitude slope in a fairly large operating frequency range?

Title: Re: Ideal model of 0/90 degrees phase splitter
Post by pancho_hideboo on Jan 27^th, 2009, 6:16am

Helge Stenström wrote on Jan 27^th, 2009, 6:11am:

Quote:

What model do you mean "rfLib/shifter" or "rfLib/shifter_splitter" ?

Any of them; they have similar behavior, but shifter_splitter has about 6 dB higher output when the outputs are not loaded, compared with shifter.?

They are both simple phase shifter using LPF and HPF.

Helge Stenström wrote on Jan 27^th, 2009, 6:11am:

To reformulate my question: how can a phase splitter be modelled, such as it does not have 20 dB/decade amplitude slope in a fairly large operating frequency range?

See my answer.
http://www.designers-guide.org/Forum/YaBB.pl?num=1233062253/1#1

Title: Re: Ideal model of 0/90 degrees phase splitter
Post by Andrew Beckett on Jan 27^th, 2009, 12:34pm

Helge,

In case Pancho Hideboo's answer wasn't clear, I will say that it's not possible to produce a model with constant phase shift and constant gain over all frequencies, in the time domain. You can however come up with something that will do this over a reasonably wide band, using the methods that Pancho Hideboo referenced.

Regards,

Andrew.

Title: Re: Ideal model of 0/90 degrees phase splitter
Post by shilpashetty on Oct 24^th, 2011, 9:24am

pancho_hideboo wrote on Jan 27^th, 2009, 5:59am:

Helge Stenström wrote on Jan 27^th, 2009, 5:17am:

There is such function in Cadence rfLib, but it is not ideal.

What model do you mean "rfLib/shifter" or "rfLib/shifter_splitter" ?

Helge Stenström wrote on Jan 27^th, 2009, 5:17am:

Is it possible to model a phase splitter with constant phase difference and constant gain in both paths?

Hello Pancho,
Can you please elaborate the options you cited to approximate the ideal phase shifter, or give a good link to understand it
regards

Title: Re: Ideal model of 0/90 degrees phase splitter
Post by weber8722 on Mar 6^th, 2012, 4:22am

Hi Helge,

spectre can include (via nport element) s-parameters into a simulation. You may define the s-parameter 2-port file in Touchstone format for such ideal phase shifter and run spectre as usual! Numerically it is challenging, because spectre will create a model on the fly, using different algorithms. I would try the rational fit with low order (say 5). Give it a try. :) :)

Start with not extremely wide-band phase-shifter and increase the BW just as much as needed.

Bye Stephan