Phase shifting a low noise sinewave

Have you looked at passive all-pass filters?

Single-side band transmitters use exactly this sort of filter to suppress one or other of the modulated side-bands, and at 5MHz you can make quite nice inductors as single-layer windings on nickel-zince ferrites.

Horowitz and (Win) Hill's "The Art of Electronics" (ISBN 0-521-37095-7) shows a wide-band phase-shifting RC network in section 5.16 on Quadrature Oscillators.

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

Hi, any ideas you may have on the following appreciated.

I have a 5 MHz sine wave oscillator which was specially built to have a low close-in phase noise (-115 dB/c @ 1 Hz offset). What I would like to do is phase shift the output +- 90 degrees, without totally destroying the high quality of the output waveform. Most of the circuits I've used introduce too much noise. I am beginning to think I need to steer clear of any phase shifters that have op amps in the design. Any ideas out there?

TIA

Do a google on "all pass" and "phase shift". I have even built them on-chip for, 1.6GHz and up, for making image-reject mixers.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

Gains of +1 and -1 with passives (RLC) in between, gives 90° over about an octave, with flat gain.

...Jim Thompson

--
|  James E.Thompson, P.E.                           |    mens     |
|  Analog Innovations, Inc.                         |     et      |
|  Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    |
|  Phoenix, Arizona            Voice:(480)460-2350  |             |
|  E-mail Address at Website     Fax:(480)460-2142  |  Brass Rat  |
|       http://www.analog-innovations.com           |    1962     |
             
I love to cook with wine.      Sometimes I even put it in the food.

low

too

Do a google search on "phase shifter" and "voltage control". There are commercial ones out there that cover your frequency, or you could roll your own with quadrature hybrids and varactors.

A lot of the descriptions/theory are for very high frequencies, but this still applies down at 5MHz, you just need to use an appropriate hybrid and varicaps.

Regards Ian

One useful method is a passive version of the usual op amp phase shifter--see H&H for that one. The op amp circuit makes the phase go from 0 to 180 degrees by using the fact that the noninverting input has a gain of +2, whereas the inverting one has a gain of -1. Thus at low frequencies the cap on the noninverting input is open circuit, leading to an overall gain of (+2)-1 = +1, whereas at high frequency, the cap is a short, leading to a gain of (+0)-1 = -1. The math works out so that the gain magnitude is always 1.

You can do the same thing with a passive network by using a gain of +1 and -0.5.

Cheers,

Phil Hobbs

A thought is to make a 20MHz squarewave oscillator instead and use flip-flops to make the 5MHz frequencies as squarewaves. This requires figuring out how to use squarewaves in place of your sinewaves.

Any form of two pole low pass filter will make a 90 degree phase lag. Unfortunately it also magnifies the resistor noise. You may be better off if you go for 2 shifted versions of the 5MHz. Shift one early by 45 degrees and the other late by 45 degrees. These would still be 2 pole circuits but their noise peaks would be away from the carrier, so you may be able to get the needed low noise performance.

You could put two copies of the oscillator in the circuit. One of them will have to be modified to added a varactor diode. You can then make a PLL that brings them into the 90 degree relationship. This assumes that the two oscillators naturally run at nearly exactly the same frequency. It may mean using a common even. The varactor is a way for noise to get into the second oscillator. If the oscillators are near exact, the varactor can be deleveraged enough.

Also, it may be posible to copy all but the feedback of the existing oscillator and figure out how to sneek a signal from it to drive the second one to run at 90 degrees.

--
--
kensmith@rahul.net   forging knowledge

[...] Tim's suggestion overlooks the strays in the inductor. I've modified it to get the phase closer. Modified: [...]

Why not? The LM741 makes a fairly good AM detector for RF above about

2MHz.

But seriously: This is the sort of case where a high frequency op-amp is likely to be more trouble that it is worth.

--
--
kensmith@rahul.net   forging knowledge

This is the best imo. Often it is thought a sine-wave is needed, but it really isn't. Even if a "sine-wave" is needed, a near zero noise LC filter can remove harmonics. The div-by-4 signals will have symmetry and be absent even harmomics, so the filter only needs to get the 3rd and above out. So phase error (off 90 degrees) would probably be quite low.