Continue to Site

# 90 degrees phase difference cirquit needed for 1-30MHz

Status
Not open for further replies.

#### neazoi

##### Advanced Member level 6
What is the easiest solution in achieving two 90 degrees phase difference outputs out of an oscillator in 1-30MHz range?

Assuming you are looking for a direct conversion receiver local oscillator, you could try using a digital divider circuit similar to this:
Ensemble RX Builders' Notes
Peter

tony_lth and neazoi

Points: 2

### tony_lth

Points: 2
So, the ff devider seems to be the easier solution... or using the DIP package 74F74 for a fIN max of about 120MHz, enough for 30MHz out
The only problem is that you have to have an ultra wide range oscillator as this works on fIN=4 x fOUT.

Also I am considered afout the harmonics this can produce.. since the output wave is square. any thoughts?

You guessed right, this is for an I/Q DCR demodulator

What is the easiest solution in achieving two 90 degrees phase difference outputs out of an oscillator in 1-30MHz range?

Sinusoidal or squarewave?

neazoi

### neazoi

Points: 2
Sinusoidal or squarewave?

Sinusoidal is ideal for this application.

Sinusoidal is ideal for this application.

In this case, a simple solution consists in an oscillator based on two integrating devices (double-intergrator oscillator, DIO).
Here, you have two outputs at the same time - with a phase difference of 90 deg.
In principle, you have several options:
One inverting (Miller-) integrator together with
(a) another Miller integrator plus inverter (negative unity gain), or
(b) positive BTC integrator, or
(b) positive NIC integrator.

The favourite solution is (a) (however, with 3 opamps). Google for "integrator oscillator" and you will find the circuits.

caspirt and neazoi

Points: 2

### caspirt

Points: 2
Cascaded Multiple Polyphase Filters may be a solution.
They are based on Hilbert Transform...(non symmetric LP/HP form in frequency axis)
If you google it, you'll find lot of docs.

neazoi

### neazoi

Points: 2
What about phase shift oscillators?
And what about delay networks?

Will they be able to maintain good phase accuracy over 1-30MHz?

What about phase shift oscillators?
And what about delay networks?
The problem is that didn't give a full specification, so the vacancies have to be filled up first.

I assume, you would want:
- constant and equal amplitude of both outputs over the intended frequency range
- constant phase shift, as said

In this case, a double integrator oscillator is in fact the obvious (analog) solution. It will need two tuning elements (usually resistors) to keep the said properties.

neazoi

### neazoi

Points: 2
Thank you very much. Your assumptions are correct.

Searching google for "double integrator oscillator schematic" I was not able to find a practical schematic that could operate in the HF frequencies (at least not noticed).
Since this is a completely new topic for me I would appreciate if you could suggest me a few practcal schematics working on RF frequencies.

Unfortunately, I only implemented amplifiers or filters, as well as nonlinear functions, but no oscillator or integrator circuits with high speed (> GHz GBW) OPs. So I can't help you with the detail design. I'm just convinced, that they are feasible. The acceptable phase error should be specfied, however. If the I/Q signals are fed to mixers, I wonder if the suggested frequency divider and respective square wave signals won't be the more convenient solution?

neazoi

### neazoi

Points: 2
Hi neazoi,

as an alternative you can google also for "quadrature oscillator".

In addition, please find enclosed two documents.

LvW

#### Attachments

229 KB · Views: 147
• active_oscillators.pdf
1.2 MB · Views: 156
neazoi and tony_lth

Points: 2

### neazoi

Points: 2
Unfortunately, I only implemented amplifiers or filters, as well as nonlinear functions, but no oscillator or integrator circuits with high speed (> GHz GBW) OPs. So I can't help you with the detail design. I'm just convinced, that they are feasible. The acceptable phase error should be specfied, however. If the I/Q signals are fed to mixers, I wonder if the suggested frequency divider and respective square wave signals won't be the more convenient solution?

I have been notified that the best signal to drive a double balanced diode ring mixer is the sinusoidal, but I do not remember the reason why. Although I have seen divider chips driving mixers with square wave.
From what said in this thread the easiest solution in terms of complexity seems to me to be the divider method though... No "exotic" oscillators and these do not require super fast opamps. The only problem is to have a multi octave oscillator to cover the whole 1-30MHz when divided down. This means that the oscillator must operate on 4-120MHz! A DDS could be the solution to this, but if a DDS is to be used I would already build a circuit using a quad output DDS.

Another solution may be the Mini circuits PSCQ-ED11294A/1 which is a two way 90 degrees power splitter and it can work on a quite wide range, but I am not sure if it will maintain it's phase difference over this wide range.

Last edited:

I think you are going to have a lot more things to be concerned over in your receiver design than whether you drive the mixer with a sine wave or square wave. There are arguments for both sides. The mixer is going to generate harmonics whether it is driven by a sine or square, the diodes do a good job of squaring off the input voltage waveform.
You could try generating the quardature at a fixed frequency and mixing down with two matched downconverters. This eases the requirement on wide bandwidth for the quad hybrid and allows for it to be adjusted. The two downconverters will use wideband components ( minicircuits have plenty of mixeers & splitters to choose from) so it should be relatively easy to keep good phase match.
There will be a compromise as to how high a frequency you could use between minimising mixing spurs, keping the quadrature to what ever accuracy you decide is needed and maintaining good phase noise. Somewhere in he 100 to 200MHz region should be OK, but I've not checked it out.
Peter

neazoi

### neazoi

Points: 2
You could try generating the quardature at a fixed frequency and mixing down with two matched downconverters. This eases the requirement on wide bandwidth for the quad hybrid and allows for it to be adjusted.
Peter

One of the two downconverters will then need a 90 deg local oscillator, wouldn't it?

You could use a 3dB hybrid coupler.

neazoi

### neazoi

Points: 2
An option is to use a type-2 PLL (one integrator in the loop filter). The output of the VCO is inquadrature with the input over the whole lock range.
Possible design issues to consider: acquisition time, tracking if the frequency changes quickly, phase noise, wide-range VCO.
The last one can be solved with a frequency conversion, i.e., insead of a 1-30 MHz VCO, use for example a 51-80 MHz VCO and beat it with 50 MHz.
Regards

Z

neazoi

### neazoi

Points: 2
You could use a 3dB hybrid coupler.

I am thinking, if it is to use a hybrid coupler, why would I need the extra mixers? I could just use the hybrid coupler directly after the local oscillator to drive the first mixer.

---------- Post added at 07:48 ---------- Previous post was at 07:45 ----------

The last one can be solved with a frequency conversion, i.e., insead of a 1-30 MHz VCO, use for example a 51-80 MHz VCO and beat it with 50 MHz.
Z
Yes mixing with a higher frequency oscillator could produce the losc signal and filtering of the mixer products would be easier

One of the two downconverters will then need a 90 deg local oscillator, wouldn't it?
No, only one of the inputs needs to be in quadrature to simplify things this wants to be the fixed frequency, but it does not have to be the local oscillator. The variable frequency input is split by an in phase hybrid. As to which one is the local oscillator, if you can keep the level of variable frequency input constant then the fixed frequency would be best being the LO as it is easier to keep a fixed frequency from leaking out and becoming a problem in other parts of your system than it is a variable frequency.
Aimto keep the mixing frequencies as high as you can. You will need low pass filters on the output of each mixer and the faster the roll off you need to supress unwanted mixer products the more difficult it will be to keep the match between the filters near the transition band. In the example given by zorro you would have to supress the 50MHz adequately ( lots of setctions ) while still maintaining good phase match, better than 1 degree, between the filters at 30MHz. This is not impossible but I think will be tricky and take some set up time. The higher you can go the more room you will have in the transition band to give you more options with the filter design.
Those spurs will be present in the receiver mixers and depending on the harmonic repsonse of the mixer will cause some "birdies". This is where having the high level signal at a fixed frequency can help; you know what you have to filter out.

Peter

In the example given by zorro you would have to supress the 50MHz adequately ( lots of setctions ) while still maintaining good phase match, better than 1 degree, between the filters at 30MHz.
Not really. The phase characteristic of the post-mixer filter does not matter. The output of that filter will be in quadrature with the PLL input regardless of the phase at the input of the filter. The accuracy of the quadrature depends only of the phase detector.
Regards

Z

Status
Not open for further replies.