Low noise 90 deg phase shifter

Hello Izik,

It sounds like you are trying to design an all-pass filter structure that maintains a 90 degree quadrature between two outputs from 0.5 to 2.5 MHz? Not to say that couldn't be done using correctly implemented all-pass filter sections, but I have not seen an op amp version for that high of frequency range. The ones I have seen utilize passive LC filters.

I expect an op amp version is going to require very high gain-bandwidth (GBW) op amps so as to minimize the phase errors. Our Precision Amplifiers op amps have a maximum GBW of 50 MHz and usually less so I don't think we will be able to provide you a solution. You could try posting your question to the High-Speed Amplifiers e2e forum and see if they have any suggestions.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Thomas 

First I didn't find any forum dedicated to high speed amplifiers. Only amplifiers forum. Please send me the link. 

Second, in regard to my question, I have in my system an IQ mixer downconverter, and I would like sum both I and Q to create a SSB converter. In order to do that, I need to create 90 deg offset between the two signals in addition to initial 90 degrees generated by the IQ mixer. Following that I am adding both signals and have only one side band. 

Hi Izik,

do you have already a circuit which you can post? Then we can try to find a suited OPAmp and carry out a phase stability analysis.

Kao

HI 

This is just a standard all pass filter, I'm using two stages to improve the BW response. 

We noticed that the phase shifter adds a lot of noise to the system, and would like to reduce it. the stage before the phase shifter is OPA847 which is used as an LNA right after the IQ mixer IF output. 

Thanks

Hi Izik,

the noise can be reduced by minimizing the resistances R1, R2 and R4 and by taking an ultra low noise OPAmp. A phase lead capacitance across R2 can sometimes help to improve stability. But this depends on the used OPAmp and the rest of circuitry.

Izik said:

We noticed that the phase shifter adds a lot of noise to the system, and would like to reduce it.

Can you show a schematic of this circuit where you have noticed a lot of noise?

Kai

Hello Izik,

TI's Filter Design Tool can be used to synthesize All-pass filters that use the MFB topology. You can see the tool here:

https://www.ti.com/design-resources/design-tools-simulation/filter-designer.html

Below I show an example of a 2.5 MHz all-pass I put in a couple of values for demonstration purposes only. It shows a GBW of 217 MHz. Certainly you would have to adjust things as you need them.

As I expect the op amp GBW requirements are going to be well beyond that of our precision amplifiers it will require a high GBW op amp. The OPA847 is one that falls within the High-speed Amplifier product line. I am unable to access that e2e forum directly. About all I am able to do is move this discussion to their forum and I can move it if you like.

Regards, Thomas

Precision Amplifiers Applications Engineering

Hi Izik,

the all-pass filter you have shown can generate a phase of 90° only at a single frequency. For a 90° phase shift over an extended frequency band several of these all-pass filters (all with different corner frequencies) would have to be put in series.

Here's a simulation when using the 1.1GHz OPAmp OPA856 providing a corner frequency of about 1MHz:

izik_opa856.TSC

As Thomas already mentioned an ultra high bandwidth OPAmp would be needed. And even with this 1.1GHz OPAmp an ideal performance can only be achieved up to about 100MHz.

C2 is necessary for stability, increasing the phase margin to about 64°.

izik_opa856_1.TSC

And this would be the noise:

Kai

Hello 

Please see my schematic. 

How would you recommend to reduce the noise?  we use ADA4897 (not as in the drawing), and willing to replace it according to your recommendation. 

Thanks

Hi Izik,

wouldn't you need termination resistors at the outputs of transformers?

To decrease the noise I would decrease the 1k5 feedback resistors.

Kai

Hello Izik,

Have you determined that the noise is indeed being generated within the all-pass filters? If it is coming to them from the preceding I-Q stages, then bandwidth limiting the signals before they reach the all-pass filters would reduce the noise that passes through them. If it is being generated within the filter stages then the only options are to use the lowest noise op amps you can obtain, and minimize the circuit resistances to their minimum realistic values. That pretty much leaves the 1R5 k resistors that Kai mentioned.

I can move this post and make High-Speed Amplifiers the responsible organization. They should be able to assist you with selecting the lowest noise op amp from their product line that provides the high GBW needed for accurate phase response.

Regards, Thomas

Precision Amplifiers Applications Engineering