Superheterodyne receiver channel selectivity

[fasset]

Hi.
I am new to RF so I have quite simple question about topology of superheterodyne. I am facing with adjacent channel selectivity problem
.
The problem is that I can't obtain selectivity more than 40 dB. F.e. when wanted channel frequency signal had ampl. level 1 dBuV and adjacent channel frequency ampl. have more than 40 dBuV receiving process is broken. Both frequencies are provided to antenna circuit by sumator. Similar problem - 20 dBuV (wanted freq.) and 60 dBuV adjacent freq.

It is not a issue of IF filters stage - problem doesnt't occur when IF frequencies are directly modulated and provided to IF stage skipping first mixer and LNA stage. Then almost 85 db selectivity is obtained.

I have tested several mixers: MAX2335, ADL5530.

The question: Is there any topologies or f.e. limiters / attentuators at the RF antenna signal stage to limit unwanted signal without limiting wanted 40 db lower signal?

Thanks for any answer.

[T3sl4co1l]

Is this an image, or out-of-band signal?

If the former, carefully reconsider your band plan: place the VFO as far from the RF band and harmonics as possible.  The harmonics of both LO and RF are relevant, to avoid higher order mixing products.  This works when the received band is relatively narrow, for example you can receive FM BCB (88-108MHz, a 20MHz span) with a ~30MHz IF (LO 118-138MHz) which has an image at 148-168MHz (easily filtered) as well as good distance between 2nd and 3rd harmonics (giving good overload rejection).

Note that a traditional AM BCB receiver accomplishes this by actively tuning the front end: it's tunable over the full BCB, but the bandwidth actually received at any given instant is a small fraction, and this provides image rejection.  (With a 455kHz IF and a tunable range 530-1600kHz, the VFO range overlaps some of the overall RF band!)

Traditional FM BCB receivers can do this as well, allowing a 10.7MHz IF, slightly more than half the RF BW and just barely avoiding in-band images.

If the latter, just improve the filters..?  Maybe you need some shielding to help them out?

Tim

[Fraser]

Adjacent channel rejection is all about filter performance ! If a strong adjacent channel signal makes it through the Filters you get problems.

You have not specified your channel spacing or the quality of filters you are trying to use. Quality filters have an excellent frequency response and steep skirts. Lower quality filters leak unwanted signals into the following stages causing problems like you are experiencing. Carefully check the frequency response plots of your filters, especially the skirts !

If you are using an LNA, make sure that it is not suffering any overload in the presence of the adjacent channel signal, or any other signals within its passband. Intermodulation products should also be considered.

Superhet  receivers are relatively simple topology but their performance is often dictated by the quality of the RF amplifiers, Filters, oscillators and mixers. If any are wanting in the performance department, performance will suffer. Adjacent channel rejection was a real issue when I was designing Superhet receivers for 137MHz weather satellite receivers as in the UK we have very powerful pager transmitters on frequencies very close to the Weather satellite channels. Selectivity is created in the IF stages whilst some adjacent band ‘hardening’ may be achieved using good quality front end RF filters before any amplifiers and the first mixer. At input frequencies, steep skirts on the high frequencies are challenging. At 10.7MHz and 455KHz there are a vast array of high quality filters available. You get what you pay for with filters though ! Remember to protect the output of a high quality filter from its input with appropriate PCB layout and shielding. Otherwise you can experience RF bypassing the filter and destroying the stage performance.

Fraser

[thinkfat]

Intercept point of the RF mixer, might be relevant as well, no?

[iMo]

I think you are overloading the mixers. Those are monolithic ~1-4GHz chips with built-in LNAs.
40/60dBuV is 100uV/1mV at its inputs. I would expect you mess with ~1uV levels.

[fasset]

RF of receiver is tuned to 455MHz.
LO ref. for mixer is ADF4355 (set to 500MHz).

IF frequency: 45MHz, second IF: 455kHz. Filters are high quality and as I mentioned - when 45MHz and 45.025MHz (channel spacing 25kHz) are modulated and set directly to IF filter stage then 80dB channel selectivity is obtained so if I understand it correctly problem doesn't occur here.
Also I have done test with set 45MHz / 10 dBuV directly to IF and unwanted signal from 10MHz - to 1100MHz (with step 1MHz) / 90dBuV then signal degradation haven't been observed. So any harmonic and other than IF 45 Mhz products from mixer which potentially could degradate signal don't pass IF filters.

The problem occur when RF signal is provided to RF stage (455MHz and 455.025Mhz) then I have obtained only 40dB ACS.
@imo Sorry for mistake, not ADL5530 (which is RF amplifier) but ADL5350 mixer. I have tested ADL5350 instead of MAX2335 (which have built in LNA) on evm board and had same problem - of course without LNA sensitivy was worsen but 40db of ACS was still obtained (30dbuV of wanted signal and 70dbUv adjacent channel).
It seems like phase noise problem but ADF4355 LO and ref in. crystal oscillator have enough good parameters to obtain min. 70 ADS.

Thank you for your answers.

[duak]

Fasset, It makes sense that the LO could be the cause of the poor ACS.   I would not be surprised if any bit of noise on the power supplies, loop filters and other analog components could cause phase noise which causes a dithering of the LO's output signal.  This leads directly to a widening of the apparent channel width of the receiver.   There is information available on phase noise and selectivity; here is one example: http://www.appliedmicrowave.com/MPD_Oct-13.pdf

The ADF4355 is a complicated chip with many functions.  I expect that any probing will cause the problem to get worse because the probe will inject noise.  I would try injecting low level signals into the various power supplies to see if the ACS is made worse and so get an idea of which section is susceptible.  After that, layout, shielding and component selection will have to be examined more closely. 

[radiolistener]

in order to get good dynamic range, you're needs to use ultra low phase noise oscillator for LO and components with high dynamic range and low non linear distortions. Another issue is a filter with a good slopes, which is hard to achieve in analog circuit.

[fasset]

Ok, it is confirmed - phase noises cause low ACS.
With high quality generator instead of ADF4355 VCO 65 dB ACS obtained.
I will test other ICs VCO / syntheizers including discrete components solutions.
Thanks in advance if anyone could recommend topology of VCO with ultra low noise verified in practice.