How to decrease the interference between 2 paths in duplexer

I desigened a duplexer(2 passband filters), but the interference(in Rx path, in Tx frequency band it generate a sharp peak and in Tx path in Rx frequency band it generate a sharp peak ) between Tx path and Rx path is much big, I use the ADS to simulae, also got that interference, so how to decrease the interference between them?

The attachment is the filter simulation using ADS!

Welcome everyone contact me
Email:liangdian_78@163.com

first, make a real world filter to see how well you can make this isolation....

and only loss at 0.8 dB on 6 and 7 pole filters smells a (big) mechanical cavity filters soloution...

simulate is a easiest part, and mechanical skill and mechanical workplace is important and lot of hour on network analyser also needed... and rework, rework, rework ........ rework...

Possible you hunting simulator ghost now but in real word you have bigger problem to implemente it and reducer this problem to minior/non exist problem.


Attenuate around -115 - -130 dB to opposite band in one pice filter i really hard to make in real world depend of mechanic soloution and material select and how well you make screening between RX- and TX-side filter structure - you possibly need use lot of screws to hunting big current loops on wrong place (prefer on opposide filter structure) ....

in practical you can make around 70 - 90 dB isolations in mechanical filters for acceptable prices, not so much more...


---

looks on diagram and you marks called 'peak' is for me zero points, and oriented on falling edge of opposite filter shape (3 -9 dB) (you find this 'zero points' also on Chebyshev-II filter structures and cauer-filters)

I thinking opposite filters edge (around 3-9 dB ) make a impedances so low, in right phase and compare to filter output in junktion point and draw power from common output. Here draw in very low level around -120 dB and transporting to input side on opposite filter with 3- 10 dB loss.

for test this, measure between RX-port and RX-port with common port terminated and see how much power leaking TX and RX side - and if possible send in picture on this meeting.


Remember - filter is reactive impedances transformer and always finding impedances point in small spots to matches between filter (complex conjugate points?).

And my question, is 130 dB loss in opposide side not enough for you???
- and you have strong carrier/power on this point ???

and practical implement, you cannot calculate more than 70 -90 dB isolation in your design if not want a very expensive filters...

---

simulator is usefull - but cannot replace real world work ie simulator have not good enough modell of real worlds all type of parameters - this has only a real world itself - special in filter design...

(escuse for my bad english - is a not native english)


/xxargs

Toxargs
Thanks for your reply
I have made the really sample,also found that problem,so that interference cant be elemilated