saw filter attenuation
I test one ECS-D480A SAW Filter , in 480Mhz insertion loss of signal is 20dB.This is the same as the value in datasheet.(www.ecsxtal.com)
But attenuation of signal in 450MHZ and 510MHZ is only 35dB(relative to input signal level)
,the desired value in datasheet is 50dB!
There is 15dB difference,what is the problem?how can I solve it?
is there any leakage or pcb problem?
I matched input and output to 50 ohm!
I would start by discussing your results with the manufacturer. Often this type of problem is caused by differences in the test setup.
The unfortunate truth is that it is sometimes also caused by enthusiastic marketing.
I like the approach of "Trust, but verify."
Have you used the specified terminating impedances, which includes a capacitor?
Data sheet: http://www.presto.co.uk/pdf/data-0082.pdf
Filter Application Note: http://www.presto.co.uk/pdf/apno-0017.pdf
do you have any PCB layout and matching network recommendation?
SAW produces 70dB isolation! this is too high ,how to do the pcb suitable layout?
Usually, the rejection degradation caused by so-called signal pick-up. It means that the signal have an alternative way to pass through other than the filter. It may be through the air or PCB.
In such cases you shoud consider carefully your PCB or the box in which your SAW filter has been mounted. Try to keep the input and output lines far. If the pick up would be through air, improvement of VSWR makes the condition better. If loss is not important put a T or Pi attenuator in both sides to improve the condition.
Based on experience , many times you will not be able to meet the specs in datasheet. so I strongly recommend that improve the condition by testing different cases and in design always set a margin from what datasheets are saying.
this problem is more complex, depend on you layout, at first y should check your circuit using the VNA,check the it is good match or not ,this can be done check the S11 and S22, then y should check the ground is good or not, then check the sapce isolation is goo or not , if above thing is good ,i think the result shall be agree with the datasheet!