Separating IF frequencies of a mixer

[amal banerjee]

Could some electronics guru here please clarify the following
How to separate the IF frequency components at the output of,
e.g., a double balanced mixer?
The following is one possible scheme.
Feed the IF output into a 50:50 splitte, ideally not resistive.
Apply one splitter output to a narrow band bandpass filter
with a band center frequency? LO - RF and the other splitter
output is fed into another band pass filter with band center frequency LO + RF.

Are there any other more elegant ways ?

 

[Chris Jones]

On 24/05/2019 20:35, amal banerjee wrote:

Could some electronics guru here please clarify the following
How to separate the IF frequency components at the output of,
e.g., a double balanced mixer?
The following is one possible scheme.
Feed the IF output into a 50:50 splitte, ideally not resistive.
Apply one splitter output to a narrow band bandpass filter
with a band center frequency? LO - RF and the other splitter
output is fed into another band pass filter with band center frequency LO + RF.

Are there any other more elegant ways ?

I think you are discussing a transmitter, in which case I think you want
to look up the "phasing method" of generating SSB.

If it is for a receiver then you should look up "image reject mixers".

In either case, when this is implemented in analogue circuitry,
component matching will limit the amount of suppression that you can
get. For some situations, either a lot of trimming is needed (sometimes
this is even adaptive, during operation), or you might need to also use
some filtering at the same time, but with more relaxed filter specs than
you would need with a simpler architecture.

Often people do this stuff in DSP, e.g. in the analogue part, split the
rf signal and feed it to two mixers with quadrature LO, digitise both
the IF signals, and then do the image suppression in DSP.

 

[Tauno Voipio]

On 24.5.19 14:26, Chris Jones wrote:

On 24/05/2019 20:35, amal banerjee wrote:
Could some electronics guru here please clarify the following
How to separate the IF frequency components at the output of,
e.g., a double balanced mixer?
The following is one possible scheme.
Feed the IF output into a 50:50 splitte, ideally not resistive.
Apply one splitter output to a narrow band bandpass filter
with a band center frequency? LO - RF and the other splitter
output is fed into another band pass filter with band center frequency
LO + RF.

Are there any other more elegant ways ?


I think you are discussing a transmitter, in which case I think you want
to look up the "phasing method" of generating SSB.

If it is for a receiver then you should look up "image reject mixers".

In either case, when this is implemented in analogue circuitry,
component matching will limit the amount of suppression that you can
get. For some situations, either a lot of trimming is needed (sometimes
this is even adaptive, during operation), or you might need to also use
some filtering at the same time, but with more relaxed filter specs than
you would need with a simpler architecture.

Often people do this stuff in DSP, e.g. in the analogue part, split the
rf signal and feed it to two mixers with quadrature LO, digitise both
the IF signals, and then do the image suppression in DSP.

The 'phasing method' of SSB generation is the same thing as an
image rejection mixer (which is a more recent term).

An analog mixer is not an ideal multiplier resulting in just
two output frequency components. The mixers (balanced or not)
switch the signal with the injection oscillator frequency,
resulting in a multiplication with the sign of the injection.
This creates plenty of other undesired mixing products in
addition to the opposite sideband (image) signal.

An image rejection mixer needs two copies of the input signal
at 90 degree phase difference, and two copies of the injection
signal also at 90 degree phase difference. Mixing the signal pairs
in two mixers and combining the results will suppress the undesired
image response, provided there is good amplitude and phase balance
between the channels. The rejection does not usually apply to the
other undesired mixing products.

--

-TV

 

[Phil Hobbs]

On 5/24/19 7:26 AM, Chris Jones wrote:

On 24/05/2019 20:35, amal banerjee wrote:
Could some electronics guru here please clarify the following
How to separate the IF frequency components at the output of,
e.g., a double balanced mixer?
The following is one possible scheme.
Feed the IF output into a 50:50 splitte, ideally not resistive.
Apply one splitter output to a narrow band bandpass filter
with a band center frequency? LO - RF and the other splitter
output is fed into another band pass filter with band center frequency
LO + RF.

Are there any other more elegant ways ?


I think you are discussing a transmitter, in which case I think you want
to look up the "phasing method" of generating SSB.

If it is for a receiver then you should look up "image reject mixers".

In either case, when this is implemented in analogue circuitry,
component matching will limit the amount of suppression that you can
get. For some situations, either a lot of trimming is needed (sometimes
this is even adaptive, during operation), or you might need to also use
some filtering at the same time, but with more relaxed filter specs than
you would need with a simpler architecture.

Often people do this stuff in DSP, e.g. in the analogue part, split the
rf signal and feed it to two mixers with quadrature LO, digitise both
the IF signals, and then do the image suppression in DSP.

Another approach is to mix up to some high IF where the filtering is
easier on account of the sidebands being further apart. (For a SSB
transmitter, you want to use a low IF for the same reason. Filters get
easier to build as their fractional bandwidth gets up near 1/4 or
thereabouts. Very wide ones are usually best made using an LP and HP in
cascade.)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC / Hobbs ElectroOptics
Optics, Electro-optics, Photonics, Analog Electronics
Briarcliff Manor NY 10510

http://electrooptical.net
http://hobbs-eo.com

 

[John Larkin]

On Fri, 24 May 2019 15:18:49 +0300, Tauno Voipio
<tauno.voipio@notused.fi.invalid> wrote:

On 24.5.19 14:26, Chris Jones wrote:
On 24/05/2019 20:35, amal banerjee wrote:
Could some electronics guru here please clarify the following
How to separate the IF frequency components at the output of,
e.g., a double balanced mixer?
The following is one possible scheme.
Feed the IF output into a 50:50 splitte, ideally not resistive.
Apply one splitter output to a narrow band bandpass filter
with a band center frequency? LO - RF and the other splitter
output is fed into another band pass filter with band center frequency
LO + RF.

Are there any other more elegant ways ?


I think you are discussing a transmitter, in which case I think you want
to look up the "phasing method" of generating SSB.

If it is for a receiver then you should look up "image reject mixers".

In either case, when this is implemented in analogue circuitry,
component matching will limit the amount of suppression that you can
get. For some situations, either a lot of trimming is needed (sometimes
this is even adaptive, during operation), or you might need to also use
some filtering at the same time, but with more relaxed filter specs than
you would need with a simpler architecture.

Often people do this stuff in DSP, e.g. in the analogue part, split the
rf signal and feed it to two mixers with quadrature LO, digitise both
the IF signals, and then do the image suppression in DSP.


The 'phasing method' of SSB generation is the same thing as an
image rejection mixer (which is a more recent term).

An analog mixer is not an ideal multiplier resulting in just
two output frequency components. The mixers (balanced or not)
switch the signal with the injection oscillator frequency,
resulting in a multiplication with the sign of the injection.
This creates plenty of other undesired mixing products in
addition to the opposite sideband (image) signal.

An image rejection mixer needs two copies of the input signal
at 90 degree phase difference, and two copies of the injection
signal also at 90 degree phase difference. Mixing the signal pairs
in two mixers and combining the results will suppress the undesired
image response, provided there is good amplitude and phase balance
between the channels. The rejection does not usually apply to the
other undesired mixing products.

There are some seriously good analog multipliers around.

The "phasing method" is equivalent to vector rotation frequency
shifting. Imagine a little girl standing in the center of a carousel
that's spinning at N rpm. If she spins herself at M rpm relative to
the carousel, she's twirling at N+M relative to the kids sitting in
the dirt. Or N-M if she spins the other way.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics

 

[Tauno Voipio]

On 24.5.19 18:12, John Larkin wrote:

On Fri, 24 May 2019 15:18:49 +0300, Tauno Voipio
tauno.voipio@notused.fi.invalid> wrote:

On 24.5.19 14:26, Chris Jones wrote:
On 24/05/2019 20:35, amal banerjee wrote:
Could some electronics guru here please clarify the following
How to separate the IF frequency components at the output of,
e.g., a double balanced mixer?
The following is one possible scheme.
Feed the IF output into a 50:50 splitte, ideally not resistive.
Apply one splitter output to a narrow band bandpass filter
with a band center frequency? LO - RF and the other splitter
output is fed into another band pass filter with band center frequency
LO + RF.

Are there any other more elegant ways ?


I think you are discussing a transmitter, in which case I think you want
to look up the "phasing method" of generating SSB.

If it is for a receiver then you should look up "image reject mixers".

In either case, when this is implemented in analogue circuitry,
component matching will limit the amount of suppression that you can
get. For some situations, either a lot of trimming is needed (sometimes
this is even adaptive, during operation), or you might need to also use
some filtering at the same time, but with more relaxed filter specs than
you would need with a simpler architecture.

Often people do this stuff in DSP, e.g. in the analogue part, split the
rf signal and feed it to two mixers with quadrature LO, digitise both
the IF signals, and then do the image suppression in DSP.


The 'phasing method' of SSB generation is the same thing as an
image rejection mixer (which is a more recent term).

An analog mixer is not an ideal multiplier resulting in just
two output frequency components. The mixers (balanced or not)
switch the signal with the injection oscillator frequency,
resulting in a multiplication with the sign of the injection.
This creates plenty of other undesired mixing products in
addition to the opposite sideband (image) signal.

An image rejection mixer needs two copies of the input signal
at 90 degree phase difference, and two copies of the injection
signal also at 90 degree phase difference. Mixing the signal pairs
in two mixers and combining the results will suppress the undesired
image response, provided there is good amplitude and phase balance
between the channels. The rejection does not usually apply to the
other undesired mixing products.

There are some seriously good analog multipliers around.

Maybe the best analog multipliers used as mixers are the
Gilbert cells, but even they are usually well overdriven
from the injection side to maximize the signal port linearity.

The "phasing method" is equivalent to vector rotation frequency
shifting. Imagine a little girl standing in the center of a carousel
that's spinning at N rpm. If she spins herself at M rpm relative to
the carousel, she's twirling at N+M relative to the kids sitting in
the dirt. Or N-M if she spins the other way.

--

-TV

 

[Simon S Aysdie]

On Friday, May 24, 2019 at 3:35:27 AM UTC-7, amal banerjee wrote:

Could some electronics guru here please clarify the following
How to separate the IF frequency components at the output of,
e.g., a double balanced mixer?
The following is one possible scheme.
Feed the IF output into a 50:50 splitte, ideally not resistive.
Apply one splitter output to a narrow band bandpass filter
with a band center frequency? LO - RF and the other splitter
output is fed into another band pass filter with band center frequency LO + RF.

Are there any other more elegant ways ?

You can do that. I'm not sure why you need to pick off both LO+RF and LO-RF.. That is unusual. Usually one is dumped. An output scheme (on the IF port) that produces a broadband termination to the mixer is often desired.

┌─┐
RF─┤X├─LO
└┬┘
IF│LO±RF
┌────┴────┐
│Diplexer │
└┬───────┬┘
IF_LO IF_HI


You haven't specified separation, or much of anything, so it is hard to say much.