SAW FILTER + DUPLEXER + ISOLATION!!!

[DDavid]

Hi guys,

I got some problem: I have duplexer for CDMA (UL 824MHz - 849MHz & DL 869MHz - 894MHz). My duplexer has about 90dB isolation is not enough for me so I decided to put Saw filter after Duplexer to increase isolation. Important to indicate that after Saw filter I have LNA.

So I did it, and I was expect for isolation at least about 120dB (Saw filter => 40 dB [EPCOS]).

But I was disappoint Saw filter as if no exist I got only 90dB from duplexer that all!!!

W H Y???????????

SAW filter according to the datasheet said that Zs=Zl=50Ω. So Saw filter suppose to be macthed. I check with Network I have above 10dB Return Loss. Dulpexer have much more and so the LNA.
I can't understand and truly don't know what to do!

Tnx a lot
Regards
David

 

[biff44]

Answer this yourself. If the diplexter gave you 90 dB, and the saw filter gave you an additional 90 dB isolation, would you have expected 180 dB of isolation? Why not?

 

[DDavid]

No,
I expect 120dB not 180dB. I am not close to noise...
I am really do know what the problem!

 

[biff44]

If you had one filter with 90 dB of measured isolation, and another filter with 90 dB of isolation, would you get 180 dB of isolation if you put them together in series? Use your noggin.

 

[DDavid]

But Saw filter give me 42dB

 

[RFDave]

You must be using a cavity filter for this, I'm not aware of a surface mount duplexer that provides 90 dB of isolation.

Another hint. How are you measuring this?

Dave

 

[biff44]

Alright, maybe there is a language problem here, so let us explain. A surface mount filter, unless you do some pretty special mounting/board design techniques, is good for around 55 dB of isolation. That is because the signal will leak around the filter via other paths (thru the air, ground loops due to disconinuities at the launch, etc). So you can put twenty filters in series on a board, and you would not get 20 times the isolation, you would get somewhere in the 55 to 70 dB isolation, beacuse the signal is simply going around all those filters.

If you are using connectorized filters, you still have to look at what the EMC leakage on the connector is giving you. Signals can easily leak out of an SMA connector, and wherever you have two of them around λ/2 apart, the signals will sneak right over the intermediary parts quite easily. 90 to 100 dB is the most you should expect.

Finally, your filter's isolation performance was designed to see an out of band 50 ohm termination. When you put a second filter in series, the out of band termination is no longer 50 ohms, and the isolation can suffer severely.

 

[RFDave]

Another potential problem is that to measure 120 dB of rejection, you need ~130 dB of measurement dynamic range from your instrumentation, apart from any fixturing errors. Again, not something that is impossible, but something that needs to have some attention paid. I don't thing it's possible to measure 180 dB of rejection, I don't think there is a test setup available with that type of dynamic range.


Dave

 

[DDavid]

First of all thanks to you replying...

In my system I have to kinds of duplexer cavity duplexer and ceramic.
True ceramic have max isolation 50dB & Cavity dup about 85-90dB.
But the behave with saw filter pretty similar of both duplexers [of course with cavity dup I have 40 dB more].

Biff44 you sad that i don't have termination 50 ohm between filters. What are you suggest? I think to put between dup and saw filter BGA427 it's amplifier from infineon [to improve matching]. Do I must improve matching if I will put another Saw filter [second (DUP + SAW +SAW)]??? Saw filter according to data sheet suppose to be match to 50 ohm [input/output]

I attached two scenario that I checked and no so understandable...

Tnx a Lot
David

 

[RFDave]

SAW filter's only provide a 50 ohm match in the band that they are supposed to be filtering. Since you seem to not be worried about power, try this

Input--(Cavity Filter)--(6 dB Atten)--(12 dB Gain Stage)--(6 dB Atten)--(SAW Filter)-Out

Where the blocks are seperated by (). THis will provide a decent 50 Ohm match between stages. The 6 dB atten stages provide a decent 50 ohm match, and will provide both filters with a good 50 ohm out of band match. Another possibility is to use an isolator between the two stages.

I still don't think you will be able to actually measure the 120 dB of attenuation that you are planning, though. What test equipment are you planning on using?

Dave

 

[DDavid]

So what you say that I will improve my systems if I will put amp between the saw and dup.

If I will build:

In [Dup cavity or ceramic] [AMP G=20dB] [SAW] [Att=20dB] [SAW] Out

How you think this chain will work???

According your question how I measure this, next chain will explain:

SG [F=880MHz @ P=15dBm] [DUP(in DL-out UL)] [AMP G=20dB] [SAW] [Att=20dB] [SAW] Spectrum Analyzer.

 

[RFDave]

Nope, you want to have the attenuators at the input and output of the filters to provide a wideband 50 ohm match so that the filters will give you the filter rejection that you are looking for.

What is the noise floor of the spectrum analyzer you plan to use? Assuming 90 db of insertion loss for the cavity filter, and 42 dB for the SAW Filters, I calculate you'll get -174 db of attenuation. With +15 dBm in, you'll need to measure -160 dBm . If you terminate the input of the spectrum analyzer with a 50 Ohm termination, what sort of displayed noise floor do you get?

Is there a particular reason you need this type of isolation? As mentioned above, this level of attenuation is hard to acheive, and can be destroyed by coupling through the air.

Dave