Effect of DC Supply voltage ripple on LNA Noise figure and Gain

[ah.vinod]

Hi all,

I need to know effect of DC Supply voltage ripple on LNA Noise figure and Gain. I am designing a TV Receiver where the front end is a LNA. The LNA is connected to 3.3V DC Supply which has ripple of 15mVpp. I am failing the sensitivity requirements for TV receiver by 6dB. SO could it be because of ripple on LNA Supply. The noise figure of LNA is 1.5 while gain is 15dB.

Regards
Vinod

 

[WimRFP]

Hello Vinod,

I assume that the frequency content of the ripple is outside the frequency band of interest. If so, the sensitivity should not decrease, however as gain of an LNA circuit has some voltage dependency, you may get phase noise. This may reduce the dynamic range of your system.

If you have a tiny fast switching DC/DC converter close to your LNA, this may induce noise that will exceed the amplifier's noise up to GHz frequencies (from bad experience).

If possible, try to shut down possible interfering circuits and notice the result on sensitivity. You may also check for instability of the LNA, also for out of band frequencies.

 

[jiripolivka]

DC voltage ripple may affect the MEASURED noise figure but only if such voltage can modulate LNA gain at a (low) frequency which can b e close to that one used in noise-figure meter.
DC voltage ripple will make no harm if your receiver uses FM to transmit information . I would assume your LNA has a RF response > 10 MHz, then DC voltage ripple of 10-50 mV can cause no harm.
You can also add a 10-100 uF capacitor across the DC terminals, to further reduce the ripple.

 

[tony_lth]

You can use battery if you doubt the ripple, as foresaid I don't think that's the ripple caused the problem.

Maybe you can upload you sch picture and pcb picture. There are so many reasons can cause NF degrade 6dB.

 

[ah.vinod]

Hi all,

Thank u for clarifying my doubt on ripple effect on LNA. I have one more query. The LNA is followed by a Silicon Hybrid Tuner IC and Demodulator IC. The Tuner and demodulator is also powered from 3.3V DC Supply with ripple of 15mVpp. The demodulator output is digital signal which is captured and BER (Bit error Rate) is calculated. Based on BER, we find sensitivity. So Could the bad sensitivity be because of ripple on DC supply of Tuner?

 

[vfone]

15mVpp ripple for 3.3Vdc supply, I don't think this is the issue for low NF in your case.

Using BER to find Thermal Noise is ok, but BER could be affected in an exponential manner by different datarates or by the system bandwidth, so you have to be careful choosing these parameters (e.g. BER doubles, for double of datarate).

Usually the Hybrid TV tuners use differential inputs (like Si2170 from Skyworks)
https://www.silabs.com/products/audiovideo/HybridTVTuners/Pages/default.aspx
so is not impossible to get a mismatch between LNA and tuner input, which degrades the BER.

 

[WimRFP]

Hello,

Are you able to measure the gain from RF to IF (with AGC off or at fixed setting)? If so, you can see whether the gain is as expected and you can determine the input noise by measuring the power spectral density at the IF output.

I agree with vfone that several unknowns may affect accuracy of measuring noise via BER.

To rull out any modulation due to the voltage ripple on gain or LO in the tuner, you may add a carrier at the RF input and observe the spectrum at IF level. The spectral peak generated by the RF generator alone (directly coupled to the analyzeer) should look similar as the spectral peak measured at the IF output (at frequency of IF).