Simulating a Very Low IF (VLIF) receiver

Hi,

I'm using Matlab to simulate a complex mixer. It basically performs a
multiplication between the incoming complex signal "y" and the phasor
exp(jw).
It basically performs 2 operations:
1. Move the input signal from IF to 0
2. Remove the interferer at the image frequency
What I would expect more is a 3 dB gain in SNR, like the equivalent analog
image reject mixer.
Since the analog part of the VLIF receiver has a noise figure 3 dB worse 
than the direct conversion receiver, I think I should recover the gap in
the digital receiver. Is my assumption correct?
Thanks

Alberto

alberto.fuggetta wrote:

> Hi,
> 
> I'm using Matlab to simulate a complex mixer. It basically performs a
> multiplication between the incoming complex signal "y" and the phasor
> exp(jw).
> It basically performs 2 operations:
> 1. Move the input signal from IF to 0
> 2. Remove the interferer at the image frequency
> What I would expect more is a 3 dB gain in SNR, like the equivalent analog
> image reject mixer.
> Since the analog part of the VLIF receiver has a noise figure 3 dB worse 
> than the direct conversion receiver, I think I should recover the gap in
> the digital receiver. Is my assumption correct?

The IF is filtered already. There is no point in suppression of the 
conversion images. In the other words, no gain.

Vladimir Vassilevsky
DSP and Mixed Signal Design Consultant
http://www.abvolt.com

On 09/08/2010 09:00 AM, Vladimir Vassilevsky wrote:
>
>
> alberto.fuggetta wrote:
>
>> Hi,
>>
>> I'm using Matlab to simulate a complex mixer. It basically performs a
>> multiplication between the incoming complex signal "y" and the phasor
>> exp(jw).
>> It basically performs 2 operations:
>> 1. Move the input signal from IF to 0
>> 2. Remove the interferer at the image frequency
>> What I would expect more is a 3 dB gain in SNR, like the equivalent
>> analog
>> image reject mixer.
>> Since the analog part of the VLIF receiver has a noise figure 3 dB
>> worse than the direct conversion receiver, I think I should recover
>> the gap in
>> the digital receiver. Is my assumption correct?
>
> The IF is filtered already. There is no point in suppression of the
> conversion images. In the other words, no gain.

Depending on the gain distribution of your receiver there can be.  If 
your IF processing chain adds a significant amount of broadband noise 
then that noise, plus its image, will appear at the output of a regular 
mixer.  If you're trying to build a receiver with the lowest possible 
noise figure, then this cannot be ignored.

An image reject mixer from IF to baseband will only see the signal (and 
noise) at the IF frequency, and improve your overall noise performance.

But, there are other ways to skin the cat.  Probably the three easiest 
ways to do this are to use the image-reject mixer, to make sure that you 
have enough gain going into your IF filter to prevent the excess noise 
from being a problem, or to put another filter at the IF output, after 
all the amplification in the IF.

Each has its advantages and disadvantages.

-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html

On Wed, 08 Sep 2010 10:55:03 -0500, "alberto.fuggetta"
<alberto.fuggetta@n_o_s_p_a_m.gmail.com> wrote:

>Hi,
>
>I'm using Matlab to simulate a complex mixer. It basically performs a
>multiplication between the incoming complex signal "y" and the phasor
>exp(jw).
>It basically performs 2 operations:
>1. Move the input signal from IF to 0
>2. Remove the interferer at the image frequency
>What I would expect more is a 3 dB gain in SNR, like the equivalent analog
>image reject mixer.
>Since the analog part of the VLIF receiver has a noise figure 3 dB worse 
>than the direct conversion receiver, I think I should recover the gap in
>the digital receiver. Is my assumption correct?
>Thanks
>
>Alberto


If the incoming signal is already complex as you describe, then there
may be no image and the translation to 0 is merely a frequency shift
of the entire spectrum.  

But that'd be pretty unusual to have the incoming IF signal 'y' to
already be complex-valued in a real system.   You may want to make 'y'
real-valued, as it would be in a typical system.   If you are careful
about selection of sample rate and IF frequencies you can simulate
images with aliases (or just create the images if you so desire).


Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com

On Thu, 09 Sep 2010 07:31:04 -0500, "alberto.fuggetta"
<alberto.fuggetta@n_o_s_p_a_m.gmail.com> wrote:

>Let me add some detail more.
>Let's suppose I have a direct conversion receiver: it could be modelled as
>IQ demodulator with a certain Gain (G_1) and Noise Figure (NF_1). The
>frequency of the local oscillator is equal to that one of the input RF
>signal (f_LO=f_RF)
>Then, the I and Q outputs of the analog receiver are sampled by the ADC
>providing a complex digital signal.

With you so far.

>If I change my LO freq so that f_LO = f_RF - f_IF, with f_IF < 1 MHz, I end
>up with a very low IF receiver having G_2=G_1 and NF_2 = NF_1 + 3 dB.
>The increase in noise figure is due to the noise folding at the image
>frequency that cannot be filtered by an analog filter since the IF freq is
>too low.
>As consequence the SNR at the output of the VLIF receiver is 3 dB worse
>than the direct conversion receiver.

Are you saying you're tuning the direct conversion receiver to 1MHz or
are you saying this is a traditional het receiver with an RF and IF
mix?   Is only the last IF mix complex, or the first one?  Neither?
Both?

And you seem to also be saying that for your system you can't make an
IF filter at 1MHz, which I can understand if the signal has any
appreciable bandwidth.

1MHz is certainly low enough that you could sample that IF directly
and then mix to baseband digitially.    Is that what you're
describing?

>The sampled signal can be expressed in the form r=conj[y*exp(j*w_if)],
>where y is the complex envelope of the transmitted signal.
>In order to properly recover the information I have to perform the
>operation
>
>conj(r)*exp(-j*w_if) = conj(conj[y*exp(j*w_if)])*exp(-j*w_if) =
>
>= y*exp(j*w_if)*exp(-j*w_if) = y.

>This is done passing the sampled signal through a digital complex mixer.
>It can be shown (and simulated) that any interfer at the image frequency
>(i.e. -2*f_IF) is fully suppressed.

You lost me there.   You seem to be mixing the input signal by the IF
LO and its inverse so that you wind up with the input signal again?

>What I also expect is that I get rid of the thermal image noise, so that
>the VLIF receiver has the same performance of the direct conversion one.

I haven't seen a way to interpret your descriptions that would suggest
this, but I suspect I still don't understand what you're trying to do.


>I hope I better stated my problem.
>
>Thanks
>
>Alberto


Eric Jacobsen
Minister of Algorithms
Abineau Communications
http://www.abineau.com

Hi Eric,

I was not good enough to clearly explain my system.
The VLIF receiver is a Motorola patent.
Please have a look at the following link

http://www.wipo.int/pctdb/images4/PCT-PAGES/2000/492000/00074252/00074252.pdf

I tried to answer your questions too. Let me know if I can give you more
info.

Thanks

Alberto

>On Thu, 09 Sep 2010 07:31:04 -0500, "alberto.fuggetta"
><alberto.fuggetta@n_o_s_p_a_m.gmail.com> wrote:
>
>>Let me add some detail more.
>>Let's suppose I have a direct conversion receiver: it could be modelled
as
>>IQ demodulator with a certain Gain (G_1) and Noise Figure (NF_1). The
>>frequency of the local oscillator is equal to that one of the input RF
>>signal (f_LO=f_RF)
>>Then, the I and Q outputs of the analog receiver are sampled by the ADC
>>providing a complex digital signal.
>
>With you so far.
>
>>If I change my LO freq so that f_LO = f_RF - f_IF, with f_IF < 1 MHz, I
end
>>up with a very low IF receiver having G_2=G_1 and NF_2 = NF_1 + 3 dB.
>>The increase in noise figure is due to the noise folding at the image
>>frequency that cannot be filtered by an analog filter since the IF freq
is
>>too low.
>>As consequence the SNR at the output of the VLIF receiver is 3 dB worse
>>than the direct conversion receiver.
>
>Are you saying you're tuning the direct conversion receiver to 1MHz or
>are you saying this is a traditional het receiver with an RF and IF
>mix?   Is only the last IF mix complex, or the first one?  Neither?
>Both?

The ADC input is complex since it's the output of a IQ demodulator.

>
>And you seem to also be saying that for your system you can't make an
>IF filter at 1MHz, which I can understand if the signal has any
>appreciable bandwidth.
>
>1MHz is certainly low enough that you could sample that IF directly
>and then mix to baseband digitially.    Is that what you're
>describing?
>

Yes, it is.
>>The sampled signal can be expressed in the form r=conj[y*exp(j*w_if)],
>>where y is the complex envelope of the transmitted signal.
>>In order to properly recover the information I have to perform the
>>operation
>>
>>conj(r)*exp(-j*w_if) = conj(conj[y*exp(j*w_if)])*exp(-j*w_if) =
>>
>>= y*exp(j*w_if)*exp(-j*w_if) = y.
>
>>This is done passing the sampled signal through a digital complex mixer.
>>It can be shown (and simulated) that any interfer at the image frequency
>>(i.e. -2*f_IF) is fully suppressed.
>
>You lost me there.   You seem to be mixing the input signal by the IF
>LO and its inverse so that you wind up with the input signal again?

Exactly. Please note, r is the sampled signal and y is the complex envelope
of the transmitted information.

>
>>What I also expect is that I get rid of the thermal image noise, so that
>>the VLIF receiver has the same performance of the direct conversion one.
>
>I haven't seen a way to interpret your descriptions that would suggest
>this, but I suspect I still don't understand what you're trying to do.
>
>
>>I hope I better stated my problem.
>>
>>Thanks
>>
>>Alberto
>
>
>Eric Jacobsen
>Minister of Algorithms
>Abineau Communications
>http://www.abineau.com
>

On 09/08/2010 08:55 AM, alberto.fuggetta wrote:
> Hi,
>
> I'm using Matlab to simulate a complex mixer. It basically performs a
> multiplication between the incoming complex signal "y" and the phasor
> exp(jw).
> It basically performs 2 operations:
> 1. Move the input signal from IF to 0
> 2. Remove the interferer at the image frequency
> What I would expect more is a 3 dB gain in SNR, like the equivalent analog
> image reject mixer.
> Since the analog part of the VLIF receiver has a noise figure 3 dB worse
> than the direct conversion receiver, I think I should recover the gap in
> the digital receiver. Is my assumption correct?

Why do you believe that the VLIF receiver has a noise figure 3dB worse 
than a DC receiver?

(I won't go into why Motorola thinks they should be able to patent an 
idea that's fall-down obvious, and had probably been done already by 
amateur radio operators in the early 1950's)

Here's your link, from your sub-thread with Eric:
http://www.wipo.int/pctdb/images4/PCT-PAGES/2000/492000/00074252/00074252.pdf


-- 

Tim Wescott
Wescott Design Services
http://www.wescottdesign.com

Do you need to implement control loops in software?
"Applied Control Theory for Embedded Systems" was written for you.
See details at http://www.wescottdesign.com/actfes/actfes.html