gsm sensitivity

During my GSM handset design, I find some channels(1 to 25) have bad Rx Sensitivity about -97dBm, but others are normal (-107dBm).
What can I do to improve the sensitivity?

Thanks for help.

if in same band others channel is under -107dBm of sensitifity,
please check
1. receive path & track in PCB stack?
2. Is controll signal cross talk?
3. tracevier & LNA power supply
4. I/Q DC offset

Can you check perfect cover when you measure NF?
I have meet same problem for other system.

And you can check LNA input S11 in smith chart.Is it a big circle or small. If big ,it means cover a wide range of NF.

Try to find harmonic relationships between the frequencies used in your design. I once spent weeks finding the reason for a few bad RX channels and it turned out that 9 x one LO minus 11 x another LO was just our 2nd IF frequency...

Bert

Yeah,

Check TCXO higher harmonics, It will be fallen into Rx in band freuqency.
proper layout is very crucial to get good sensitivity over all channels.

First of all in EGSM (Extended GSM) first channels are not 1 to 25. These are the first channels in PGSM (Primary GSM).
- EGSM channels are 975 to 1024 (925MHz to 935MHz)
- PGSM channels 1 to 25 means 935MHz to 940MHz.
I do not think that is any manufacturer of mobiles in these days that are designing just for PGSM band. This was happened more than 8 years ago.
If you are talking about first channels in EGSM (925MHz to 930MHz), from my experience one of the most issue that gave me a low Conducted Sensitivity was a bad match of the RX SAW filters and high NF of the LNA at that channels.
If you are talking about bad Radiated Sensitivity on that channels, things are more complicated.
regards

Thanks for your replies:)

Yeah, my handset have three frequency bands(GSM900,DCS1800,PCS1900).Of course,the GSM900 works under the EGSM900.

In Conducted Sensitivity Mode(Test Port direct to the instrument), the sensitivity are all good(880 Mhz to 915Mhz of Tx Band).
But in Radiated Sensitivity Mode(put into the chamber), the 1 to 10 Channels are specially bad(about -95 dB). others are all normal.
DCS and PCS are also normal.

RF TRx of the handset is single chip but PA circuit. And it is two displays,but in the test,we have eliminated them.

So, I guess the circuit should be no problem. The problem may be EMI?
But I don't know where it is.

Thanks for your kind discuss.


Best Regards

So it happened what I believed, is about Radiated Sensitivity.
One of the reason of radiated desense could be the digital noise that is going to the front-end through antenna path. Are a lot of factors that can make this problem, staring with digital noise from LCD and finish with antenna behavior of some transmission lines which carry digital signals.
Why is happened on the first channels of GSM 900? Because this are the closest channels from the harmonics of the digital signals (usually 10MHz up to 200MHz). DCS1800 and PCS1900 are way far from these frequencies.
regards

I agree vfone' s idea.

I ever encountered the same question. The LCD is the max digital noise source, you may add BEAD to reduce EMI.

Cheers

Thanks for vfone . But how to find the harmonics or digital noise , or which ways to improve the radiated densense for these first channels of GSM900?