Image Frequency Rejection in double/triple conversion radios

[enrico]

Dear All,

Alternatively to image rejection mixers (IRM) or image filter, many wide band radios working in HF (0 to 30 MHz) band adopt the use of a 2nd or even 3rd conversion (double or triple heterodyne conversion) claimed by some folks to supress the image frequency and adjacent interferences.

May be someone could explain how a 2nd (or 3rd) conversion would help to reduce (or eliminate) the image frequency, based on the fact that the image frequency falling in the same frequency of the 1st IF cannot be further separated or supressed afterwards within the 2nd (or 3rd) IF frequency ?

I would really appreciate if someone could give me an insight about the relationship of the 2nd (or 3rd) IF with the suppression of the image frequency. In my opinion, the 2nd (or 3rd) conversion would not further reduce the image frequency, but just the fact that the filters used in lower frequencies (IF) would be less expensive and have high-Q leading good shape factor (e.g. Collins mechanical filter, xtal filters). Also having two or more conversions would help to divide the necessary gain in different frequencies to prevent oscillation, against too much gain in one stage/one frequency. Thus, in conclusion, the 2nd (and 3rd) conversion would then help to eliminate adjacent interferences, so selectivity is the reason for adopting double or triple conversions and in fact the image rejection is reduced solely in the 1st IF conversion and as far as it concerns, the far (high) is the LO frequency from the desired signal (RF), better is the rejection due to the fact the image will fall further away.


Thanks.



Please note the two possible cases (high or low side injection) :

Fimage=Fwanted + 2*IF (if FLO > Fwanted, high side injection)
or
Fimage=Fwanted - 2*IF (if FLO < Fwanted, low side injection)

 

[BigBoss]

Image frequency is well filtered by using low frequency sharp filters, that's why double and sometimes triple conversion is applied.For instance ceramic filters, SAW filters etc.

 

[biff44]

Yes. And as the IF center frequency usually goes down with IF2 and IF3, it is easier to make the filter because the % bandwidth gets smaller--you have steeper rejection at the bandedges.

But, you do need those filters. Lets say you had a 455 KHz third IF, and were trying to receive a 915 Mhz input frequency. If you did not have a bandpass filter at 455 KHz, you could have an image frequency at 914.09. That image could screw up your receiver noise figure, or allow a jammer signal to enter.

 

[chuckey]

many HF receivers use an upconvertor so the first IF frequency is 38 MHZ, so any image in the HF band falls in the range 38.5 -> 68 MHZ, and are rejected by the first IF filters. The frequencies in the range 8 -> 37.5 are then converted to a 1.4 MHZ IF, this will then cover all frequencies with a good crystal filter for SSB and CW.
Frank

 

[vfone]

In a multiple conversion receiver each mixer (or mixing process) would have its own image frequency, whatever is the 1st, the 2nd, or the 3rd in the chain.

An image filter placed in the front of a mixer will suppress the image signal AND the image noise.
Usually the 1st mixer use at its input (RF input) an image filter with bandwidth equal to the received band. This filter could be a band-pass in most of the cases, but this could be low-pass or high-pass (high-side or low-side injection) if the distance is high between RF and 1st IF frequency.

The following IF mixers (2nd, 3rd,..), they use the IF filters in their front as image filters. For example the 2nd IF filter is the image filter of the 2nd mixer.

One thing that sometimes RF Systems engineers miss in their system noise calculation, is that if the 2nd (or 3rd) IF filter is not exactly at the input of the following mixer (e.g. has an amplifier after it), this cannot be consider anymore an image filter for that mixer, and another few dB's of image noise contribution should be added to the chain.

 

[enrico]

Dear Biff, Vfone and Chuckey,

Thank you so much for having shared your view.

Biff your example made to clearly understand the importance of a 3rd IF (in this example). Vfone thanks for detailed info, your statement about each mixer has its own image is also important, therefore it would be interesting to obtain the whole frequency plan. Maybe someone of you know a software for it, in the past WhatIF from Genesys, now Agilent was a tool. Microwave Office also had it, but I can't remember the name (the advantage here is the fact you can observe critical areas like spurs and image but already considering in case you choose to have a filter in the input of the mixer). Chuckey you also clearly pointed out in the example the case of IF > RF in the 1st conversion citing numbers.