How to filter a mixer's IF output?

[oac]

Hi all,

I'm building a receiver. The received signal is (low noise) amplified and has a frequency in the 2.45GHz band. This signal is mixed with a sample of the 2.45GHz generator. The resulting signal at the IF port has power around the frequencies 0Hz (DC) and 4.90GHz. The latter must be filtered away before IF amplification. The passband of the filter should be about 5kHz-1MHz.

I have two, possibly related problems.

1) I'm not sure how to make this filter. If I use a simple LC filter (theoretically lossless) it will reflect the 4.90GHz energy back to the mixer, and I'm afraid this will destroy the mixer linearity.

I'm aware of the existence of absorptive filters, but have no experience with their design. What do you recommend me to do?

2) How can I model the IF port of the mixer? The datasheet shows the IF output is matched to 50Ohms. I don't know how to interpret this at the low frequencies of the IF signal. Can i look upon the IF output port as a voltage source with internal resistance 50Ohm?

many thanks,
oac

 

[barry]

If you make a filter whose impedance matches your mixer, there shouldn't be any reflections.

 

[BigBoss]

You cannot realize 5kHz-1MHz filter with practical L-C components.Instead,I recommend you to use ceramic/piezo LP filter from-for instance-Murata or any manufacturer.
don't worry about 4.9GHz=2*fLO because this signal will be grounded in the filter.I wouldn't reflect.
If the datasheet has mentioned that IF output needs to be terminated by 50 Ohm, do it as specified. Output impedance will not impact so much the mixer's performance.
A good alternative..
https://www.linear.com/product/LTC1560-1

 

[vfone]

You can use at the output of the mixer a Diplexer calculated for the IF frequency, providing in this way to the mixer a good broadband load.

https://www.changpuak.ch/electronics/calc_16a.php

An absorptive low pass filter is using basically a low pass filter and a high pass filter, one on the through path, and the other terminated with a load resistance.
In this way the input sees a broadband return loss, on both, pass band and reject band.

 

[biff44]

yeah a diplexer...or you could just put a 3 dB pad in there IF you were sure there was a problem. BTW, the main effect I have seen from a bad port mismatch is to increase the levels of certain spurious products.