GPS receivers Front-End Design Problems

I am designing a GPS L1(PWR=-160dBW) band Direct RF sampling receiver and the front-end:
(passive antenna + LNA(21dB) + BGA1(30dB) + jumper + BPF(L1(fc=1575.42MHz,BW=15MHZ) + BGA2 (30dB) + BGA3(30dB) + VGA(-13.5 to +18 Gain Control in 0.5 dB Steps) ).
all amplifiers are supplied with a +5 V power with decoupling capacitors and a test point after each component to view the signal.

The problems of the printed circuit:

• noise peaks through the RF line that gets amplified even though we have no input signal after each amplifier seen through the test points (+5V power is ON), no noise when +5V is OFF even when there is an input noise.
• The BPF filters the noise peaks generated by BGA1, but the Nosie peaks still get regenerated by BGA2 .
• One issue is that each amplifier amplifies with different GAIN values even though they are the same BGA, also I am sure I did not reach 1dB compression point(10dB).

The test point output using Network analyzer to see S-parameters ( the port1= input , port2= test points) S21 outputs are :

Port2= test point after LNA:



Port2= test point after BGA1:



Port2= test point after BPF:

• Port2= test point after BGA2:

any suggestions on the matter would really be helpful.

Post photos of both sides of the PCB?

At 1.6 GHz with 160 dB of gain makes it really difficult to isolate the high power stages from the input. Any conductor will radiate some power, and with that much gain, a tiny bit of leakage will be louder than the signals from the GPS satellites that are 20,000 km away.

I would have expected a superhetrodyne receiver with two frequency shifts so that any unintended radiation from the sections where the signal is more powerful will be well away from the GPS frequencies.

I've had GPS active antennas oscillate with 20 - 30 dB of gain when the output of the antenna amplifier radiates and the antenna picks it up. I really don't think that it's possible to isolate against 100 dBs of gain.

Diver300 said:

At 1.6 GHz with 160 dB of gain makes it really difficult to isolate the high power stages from the input. Any conductor will radiate some power, and with that much gain, a tiny bit of leakage will be louder than the signals from the GPS satellites that are 20,000 km away.

I would have expected a superhetrodyne receiver with two frequency shifts so that any unintended radiation from the sections where the signal is more powerful will be well away from the GPS frequencies.

I've had GPS active antennas oscillate with 20 - 30 dB of gain when the output of the antenna amplifier radiates and the antenna picks it up. I really don't think that it's possible to isolate against 100 dBs of gain.

Click to expand...

thank you for replying,
I can not work with super heterodyne because I want to use the Direct RF sampling architecture.
I realized that the amount of amplification is a lot therefore I removed the VGA and BGA3 to just amplify with only LNA+BGA1+BGA2 where it's output is actually the screen shots I provided in thread,

when it comes to the PCB I designed 4-layer PCB (1stlayer=microstrip (50ohm matching),2ndlayer= GND,3rd layer= PWR(+5V,+3V),4thlayer=GND).

I think the circuit contains enough decoupling caps between VCC and GND.

The second test:

The input is an antenna for GPS L1 signal(L1=-160dBW) received with modified PCB(no VGA and BGA3) and these are the outputs I see in the signal using spectrum analyzer:

• test point after LNA:

• test point after BGA1:

• test point after BPF:

• test point after BGA2:

one issue I realized that each amplifier does not provide the same amount of gain even though they are the same.

The PCB layout for such a system is incredibly critical!

Until you show that, there is no point trying to guess answers, because it could not be anything other than guesses.

rjenkinsgb said:

Post photos of both sides of the PCB?

Click to expand...

rjenkinsgb said:

The PCB layout for such a system is incredibly critical!

Until you show that, there is no point trying to guess answers, because it could not be anything other than guesses.

Click to expand...

This is the layout of the circuit currently I am not using any splitCombs Instead just short circuited the RF line to BPF1 line to just use the L1 band . the screen shots I provided are taken from the PCB that has no BGA3 and VGA.
The PCB is 4-layer PCB (1stLayer=microstrip (50ohm matching),2ndlayer= GND,3rd layer= PWR(+5V,+3V),4thlayer=GND).
CS = is coaxial switch for test point.

Are you sure of the strip impedance? It looks very narrow - but I don't know the dielectric type or thickness.

Some splitters/combiners only work properly if all ports have a matched load, so no open circuits.

Have you tried adding screening walls or boxes to divide the various gain sections and reduce unwanted coupling?

rjenkinsgb said:

Are you sure of the strip impedance? It looks very narrow - but I don't know the dielectric type or thickness.

Some splitters/combiners only work properly if all ports have a matched load, so no open circuits.

Have you tried adding screening walls or boxes to divide the various gain sections and reduce unwanted coupling?

Click to expand...

yes the strip impedance is actually matching.
Currently in the printed PCB of the design that I am using I did not connect the splitters/combiners I just short circuited the RF line from Mid RF line to top RF line to have less components when I am testing the L1 band.
What could be the reason that the same amplifiers are amplifying the signal with different Gains ?
As for the screen walls and boxes I did not try that.