Forums

How to improve SNR in Antenna Cirsuit Design

Started by Johnson Liuis March 10, 2005
I am looking into an existing GPS mini-size patch antenna design (1575MHz),
which includes an LC low-pass filter, an amplifier with RFI suppression
(BFP640), and a pi attenuator about 3 dB. The SNR values are generally 5dB
lower than normal. So I am wondering if I remove the pi attenuator, will the
SNR be able to get improved?

I also found the antenna ground was not big enough (about 26mm* 26mm), and I
suspect that it was the reason that caused bad SNR values. Any idea to
increase the SNR values without significantly increasing the Antenna ground?
Can I modify the LC low-pass filter and amplifier to improve the SNR as
well?

Thanks.

Johnson


On Thu, 10 Mar 2005 11:44:30 -0700, "Johnson Liuis"
<gpsabove@yahoo.com> wrote:

>I am looking into an existing GPS mini-size patch antenna design (1575MHz), >which includes an LC low-pass filter,
An LC filter made of lumped components at 1.5 GHz ????
>an amplifier with RFI suppression >(BFP640), and a pi attenuator about 3 dB.
Is the attenuator between the antenna and preamplier or after the preamplifier ? If the attenuator is in front of the amplifier, removing it would improve the SNR by 3 dB. The only sensible reason for putting a resistive attenuator in front of the preamplifier would be that the amplifier was unstable, when connected to a highly reactive antenna. Inserting a resistive attenuator would create a nearly 50 ohm resistive source impedance and the amplifier would be stable. However, this would be a very bad way to tame the amplifier. Paul
Sorry I made a mistake. The LC network is not a low-pass filter, actually it
is a bandpass/bandstop resonant filter around 1575MHz. I just calculated the
resonant frequency and to my surprise, I found the calculated frequency is
about 1.9GHz instead of 1.575GHz, so it seemed that the previous designer
took the parasitic capacitor and inductor into account. Could anybody please
let me know if it is reasonable to leave about 0.3GHz room for parasitic
capacitor and inductor?

BTW, the RF signal is first feed into the LC bandpass filter, then to a
Low-noise Amplifier, then to the Pi Attenuator.

Any idea?

John




Thanks for your quick reply, Paul,

I just made a mistake. The LC network is not a low-pass filter, actually it
is a bandpass/bandstop resonant filter around 1575MHz. I just calculated the
resonant frequency and to my surprise, I found the calculated frequency is
about 1.9GHz instead of 1.575GHz, so it seemed that the previous designer
took the parasitic capacitor and inductor into account. Could you please let
me know if it is reasonable to leave about 0.3GHz room for parasitic
capacitor and inductor?

BTW, the RF signal is first feed into the LC bandpass filter, then to a
Low-noise Amplifier, then to the Pi Attenuator.

If the attenuator is behind the LNA, can I still get some nenefits in
improving SNR?

You mentioned that it is a very bad way to tame the amplifier by adding Pi
attenuator, do you have any better way?

Thanks.

Johnson

"Paul Keinanen" <keinanen@sci.fi> wrote in message
news:ci8131hpobpsvvamr72in1nfvh4eiugape@4ax.com...
> On Thu, 10 Mar 2005 11:44:30 -0700, "Johnson Liuis" > <gpsabove@yahoo.com> wrote: > > >I am looking into an existing GPS mini-size patch antenna design
(1575MHz),
> >which includes an LC low-pass filter, > > An LC filter made of lumped components at 1.5 GHz ???? > > >an amplifier with RFI suppression > >(BFP640), and a pi attenuator about 3 dB. > > Is the attenuator between the antenna and preamplier or after the > preamplifier ? > > If the attenuator is in front of the amplifier, removing it would > improve the SNR by 3 dB. The only sensible reason for putting a > resistive attenuator in front of the preamplifier would be that the > amplifier was unstable, when connected to a highly reactive antenna. > Inserting a resistive attenuator would create a nearly 50 ohm > resistive source impedance and the amplifier would be stable. However, > this would be a very bad way to tame the amplifier. > > Paul >
On Thu, 10 Mar 2005 13:36:20 -0700, "Johnson Liuis"
<gpsabove@yahoo.com> wrote:


>I just made a mistake. The LC network is not a low-pass filter, actually it >is a bandpass/bandstop resonant filter around 1575MHz. I just calculated the >resonant frequency and to my surprise, I found the calculated frequency is >about 1.9GHz instead of 1.575GHz, so it seemed that the previous designer >took the parasitic capacitor and inductor into account. Could you please let >me know if it is reasonable to leave about 0.3GHz room for parasitic >capacitor and inductor?
I would not use discrete inductors and capacitors at such frequencies due to the low Q and spurious resonances, but rather make a bandpass filter of a 1/4 wavelength resonator grounded at the other end. This could be made of a stripline or microstrip construction. While this has spurious responses at 3f, 5f etc., the transistor gain has dropped quite a lot at these frequencies anyway. There are a quite a lot circuits published in amateur radio amateur literature for the 1.3 GHz (23 cm) and 2.4 GHz (13 cm) amateur bands, which should be easily adaptable to 1.5 GHz.
>BTW, the RF signal is first feed into the LC bandpass filter, then to a >Low-noise Amplifier, then to the Pi Attenuator. > >If the attenuator is behind the LNA, can I still get some nenefits in >improving SNR?
Only if the cable to the actual receiver is long and lossy or the actual receiver noise figure is _very_ bad. The preamplifier gain (including any post-amplifier attenuators) must be large enough, in order to "mask" the cable or receiver noise. With current semiconductors, you can get quite a lot of forward gain at these frequencies, so I do not think this is an issue.
>You mentioned that it is a very bad way to tame the amplifier by adding Pi >attenuator, do you have any better way?
It is a bad way to put an attenuator in front of a low noise preamplifier, since the attenuation is directly added to the noise figure and thus dropping the SNR by the amount of attenuation. If an amplifier has a lot of leakage from output to input (the s12 parameter, reverse transducer gain), the undesired wave travels from amplifier output to input and continues towards the antenna. The antenna is well matched, this wave radiates into space through the antenna. However, if there is a mismatch, the signal is reflected back and enters again the amplifier input port, it is again amplified and possibly causing oscillations, if the phase relations are favourable. When you put an attenuator between the mismatch and the amplifier the s12 wave is once attenuated going through the attenuator, it is then reflected from the mismatch, attenuated again in the attenuator before entering the amplifier input port and amplified. Due to the increased attenuation of this unwanted wave, the ability to sustain oscillation is greatly reduced. To avoid the need for this attenuator, look at the overall input matching and also look at s-parameters (s11, s21, s12 and s22) to find a more suitable active device. It should be noted that if some input filtering is used, there are certainly going to be mismatches at some frequencies (unless some diplexer system is used), the s-parameters should be studied at all frequencies at which the active device has forward gain and not just on the desired reception frequency. Now thinking about the attenuator between the preamplifier and the actual receiver may have something to do with a bad mismatch between the amplifier output and receiver input, which might also cause a strong reflection at the receiver input, which would travel through the preamplifier reverse gain to the input, reflected at some input mismatch and entering the preamplifier input, causing oscillation. The preamplifier output attenuator would also attenuate the wave reflected at receiver input and thus reducing the risk of oscillation. Provided that the preamplifier gain is large enough to mask the receiver input noise, this attenuator between preamplifier and receiver should not harm the noise total figure. However, since you experienced lower SNR than expected, either the preamplifier gain is too low (below 10-15 dB) or the preamplifier is very noisy in the first place. Paul
Thank you very much, Paul,

You mentioned that there are some circuits published in amateur radio 
amateur
literature for the 1.3 GHz (23 cm) and 2.4 GHz (13 cm) amateur bands. Could 
you please recommend a few websites or literatures to me?

Johnson


Thanks  Paul,.
You mentioned instead of using a discrete inductors and capacitors at such
frequencies
due to the low Q and spurious resonances, I'd better make a bandpass
filter of a 1/4 wavelength resonator grounded at the other end. This
could be made of a stripline or microstrip construction.Could you please l
et me know what is the Pros and Cons of this 1/4 wavelength resonator
compoared to adding a ceremic filter directly to the RF circuit before LNA?
Thanks in advance.
Johnson


BTW, 1/4 of the GPS wavelength is about 5cm. Does it mean that my antenna
circuit board has to be more than 5 cm in length to fit the 1/4 wavelength
resonator?
I am just an amateur about antenna design, sorry if my question is too
"naive".
Johnson


On Sun, 13 Mar 2005 21:43:16 -0700, "Johnson Liuis"
<gpsabove@yahoo.com> wrote:


>You mentioned that there are some circuits published in amateur radio >amateur >literature for the 1.3 GHz (23 cm) and 2.4 GHz (13 cm) amateur bands. Could >you please recommend a few websites or literatures to me?
Designing circuits for frequencies in the upper UHF range and above requires in practice that you think about the various components in a circuits as sections of transmission lines. For instance "The ARRL UHF/Microwave Experimenter's Manual" will give some easy to read theoretical background to this. It was published in 1990, so there is not much point of copying the designs, since better components are available. The book might still be available from www.arrl.org For web searches, I would suggest including search expressions like "23 cm", "1.3 GHz" or "1296 MHz" should help you find some modern designs. Have you looked for application notes from various GPS chip set manufacturers. Even if you are not using the same chip-set, application notes about antennas and preamplifiers should still be usable. It should be noted that the line widths and lengths as well as the PCB material as PCB thickness should be the same if copied, unless you know how to scale them. Take a look at www.maxim-ic.com they might also have some application notes. Paul