GPS LNA Input matching Problem

[pk2010]

I am trying to match this GPS LNA with a 50hm antenna.

LNA Datasheet:

https://datasheet.lcsc.com/szlcsc/1912111437_AWINIC-Shanghai-Awinic-Tech-AW5005DNRZ_C401025.pdf

Datasheet says that all we need is an inductor around 9.1nH to match it to 50ohm source.But practically when i measure input impedance with different inductance values i get following values:

9.1 nH => 10.5 + j48
7.5 nH => 10.9 + j24.8
6.8 nH => 9.4 + j11
5.1 nH => 8.8 + j8
4.3nH => 8.8 - j8

So basically with all these values I am nowhere near the 50 ohm target ( and real part stays around 10 ohm no matter what inductor i try only the imaginary part keep increasing with increasing inductance). When i use the suggested 9.1 nH then GPS reception is not good. The receiver shows crazy 1-2 satellite bars keep coming and disappearing.

I understand that Noise matching may be more important compared to power matching for a GPS LNA and maybe targeting 50ohm was never a goal.But if that is the case , datasheet doesnt tell what impedance i should target. I am very confused what should be my goal.

 

[BigBoss]

Does the manufacturer suggest to use an particular antenna ?? They might be used a particular one to achieve better result ..
Not all antennas have 50 Ohm Impedance, some of them may have different than this value to get better signal strength.

 

[pk2010]

Does the manufacturer suggest to use an particular antenna ?? They might be used a particular one to achieve better result ..
Not all antennas have 50 Ohm Impedance, some of them may have different than this value to get better signal strength.

No they have not suggested any particular antenna. But of course they have mentioned at several places in datasheet.

"input matched to 50 ohm using a high quality‐factor 9.1nH inductor."


My antenna is 50ohm that i am sure about. And this is a general purpose gps lna so ofcourse it is ezpected that user will place it in a 50 ohm ecosystem.

 

[FvM]

5 dB input return loss specified in the datasheet is neither near to 50 ohm, but the measured return loss is much worse, around 2 dB. Something is apparently wrong, either with the measurement or the LNA.

 

[vfone]

The similar LNA, BGA824 from Infineon, provides S parameters. Adding the recommended series inductor, the input impedance will be exactly 50 ohms.

https://www.infineon.com/cms/en/pro...gps-glonass-compass-lna/bga824n6/#!simulation

Could be something wrong with AW5005 LNA. Usually the dedicated GPS LNA's are matched for 50 ohms.

 

[BigBoss]

In such cases, calibration of VNA is very important because you measure impedance on Smith Chart.
As I see, the Input Impedance changes its polarity between 4.3nH and 5.1nH that means the Input Impedance has a capacitive part around this point.
You compensate this capacitive part by adding inductance.
-VNA must be very well calibrated with a high quality cables and calibration set with SOLTP capability.Small amount of cable length changes too much the measured impedance.Therefore Phase Calibration is very necessary.Because cables don't have pure 50 Ohm impedance.
-If a Transmission Line is used, the Characteristic Impedance should be accurate
-Inductor value and tolerance ?? Q Factor ?
-PCB and parasitic components around IC Input and Antenna
-Does the Antenna have 50 OHM, sure ??
-Signal Processing after the LNA.. Is it so good ??
-Electromagnetic Interferences in the measurement medium that worsen BER performance
-Have you ever simulated that portion of the PCB in a planar EM simulator?? Compare with measurement results.

 

[RCinFLA]

5 dB return loss on input to application test circuit in the spec may done to yield a better noise match for the device. Best power match is not necessarily the best noise performance. Problem when the optimum noise performance match and best power match get too far apart is it creates problem for terminating source stage like an antenna or an input filter as it can detune them.

Suggest you try different matches, one adjusted for best noise figure then adjusted for best power gain. Sometimes you just have to partial pad the input with resistance to compromise between gain and noise performance. It really gets complicated when device stability also enters into match. Stability is best handled by padding output tank.

If running output into a bandpass filter the amplifier output Q has to be low enough not to become involved in the bandpass filter response. If output Q is too high it effectively becomes another pole in the output filter design totally changing the filter design, with ramifications to passband ripple, skirt selectivity and filter loss.