The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> mixer NF comparison
https://designers-guide.org/forum/YaBB.pl?num=1185235714

Message started by fz2101 on Jul 23^rd, 2007, 5:08pm

Title: mixer NF comparison
Post by fz2101 on Jul 23^rd, 2007, 5:08pm

Hello everyone,

Since the DSB NF of a mixer is usually 3-dB less than the SSB NF for a mixer, then the NF of a direct-conversion receiver mixer should always be 3-dB less than that of a heterodyne receiver mixer (of course, excluding the rare case of DSB modulation).

If this is true, how come I have not seen this point being brought up as a selling point for zero-IF architecture?

Thanks,

frank

Title: Re: mixer NF comparison
Post by aaron_do on Jul 23^rd, 2007, 6:23pm

I don't think its so straight forward. Here's the way i understand it.

1) Both homodyne and heterodyne receivers require image rejection.

2) for homodyne, the image is itself. However, it is mirrored at 0 Hz on the frequency axis. To get the full information signal, you must use a Quadrature mixer.

3) The advantage of homodyne is that the image is the same frequency, therefore it has the same power. example if you receive -50 dBm, both the signal and its "image" are -50 dBm.

4) in a heterodyne system, the image could be much larger depending on the inteferrer at the image frequency. For example, if Pint (interferrer power) at 5 Mhz offset is +30 dB with respect to signal, and you chose an IF of 2.5 MHz, then your image rejection requirements are roughly 30 dB + SNRmin (required SNR at the ADC)

5) image rejection in heterodyne systems can also afford some image noise cancellation. If you use a polyphase filter, its image rejection will also reject image noise.

Please correct me if i'm wrong...

cheers,
Aaron

Title: Re: mixer NF comparison
Post by didac on Jul 23^rd, 2007, 11:15pm

Hi,
I agree with Aaron in his explanation. I just wanted to add that direct conversion receivers have a drawback that complicates a little bit their realization: since you downconvert to 0Hz any DC offset will cause distortion to your signal, also IIP2 will convert to DC furthermore LO can leak to RF port and then will be downconverted also to DC. This issues caused that the biasings of following stages can be difficult(it's necessary to compensate the "instantaneous" DC offset caused by the information signal and the other effects while preserving the bias voltage necessary). This problem can be solved of course(for example deciding to notch the DC,at the cost of losing energy of the signal) but I think is one of the reasons why people usually think that direct conversion topology benefit on NF don't pay for the extra work in solving this issues.
Hope it helps,

Title: Re: mixer NF comparison
Post by aaron_do on Jul 24^th, 2007, 1:54am

BTW direct conversion also has another selling point which is the filtering. A simple LPF can be used for channel selection, which is able to filter both positive and negative offset frequencies. For a low-IF, you need a complex BPF.

Aaron

Title: Re: mixer NF comparison
Post by fz2101 on Jul 24^th, 2007, 6:12am

Please correct me if I'm wrong, thanks!

From what I understand, homodyne does NOT require image rejection, and that is one of its biggest selling points.

It is true that interferer power at the image frequency can be much larger than the desired signal, however this is
an issue with the linearity of the mixer or image rejection, not noise figure. Of course, in the grand scheme of things (dynamic
range), we need to take care of both.

The quadrature mixer comment is important, although I don't think only homodyne needs quadrature, since
heterodyne eventually has to be downconverted to 0Hz as well (so there is always that folding of negative frequency portion
onto positive frequency portion, and vice versa). Since NF deals with signal to noise ratio, if we have two paths working in parallel (''I and ''Q''),
and we finally find sqrt(I^2+Q^2), while the noise add in power (are the noise correlated?), then the NF should
not change, right?

Thanks,

frank

Title: Re: mixer NF comparison
Post by didac on Jul 24^th, 2007, 11:25pm

Hi,
Frank I think that you are correct. The homodyne don't need image rejection(as Aaron said the image is the desired signal itself). For the use of a quadrature mixer in a direct conversion topology is compulsory(if you don't have a quadrature mixer you can find that your LO is orthogonal sometimes with the desired signal losing all the energy), but for a heterodyne topology it's possible but not compulsory(it all depends on how much power budget you can put on the ADC,I/Q is a low-pass signal non-zero IF is a bandpass signal so we find Nyquist here).

About the noise I think you are right except that in my opinion part of the noise is correlated. Let me explain my thoughts(any comments as always will be appreciated):
1) You pick noise from the antenna
2)You use an LNA that amplifies noise and adds noise itself
3)You feed this noise to the two mixers.
From this point all noise internally generated by the I/Q branches I think can be considered uncorrelated, but the noise at the input of both I/Q branches share a common origin so I think that at the output of the two branches there will be two noise components, one uncorrelated and the other one with a correlation coeficient.

Title: Re: mixer NF comparison
Post by aaron_do on Jul 24^th, 2007, 11:58pm

thanks for clearing up my mistake about the image rejection :)

I agree with didac...the correlated portion is before the mixer, and since the LNA is too, you don't necessarily get a 3 dB rise in NF.

Sometimes you do not need analog image rejection, you simply supply I and Q to the ADC and image rejection is done digitally. In that case, i think you will see a 3 dB rise in NF before ADC. I'm not sure how digital image rejection affects this though. Anybody know?

cheers,
aaron

Title: Re: mixer NF comparison
Post by James Bond on Jul 25^th, 2007, 1:10pm

hi,all,

I am confused with NF in mixer. When talking about the NF in a mixer, at which frequency man should look, LO frequency or IF frequency to get NF ?
Thank you very much.
James