Designer’s Guide Community

new_guy

Junior Member

Offline

Posts: 10

LNA Test Key
Aug 04^th, 2007, 1:20am

Hi,

I'm doing a standalone LNA for testing. I know the testing equipment is 50 ohms based, currently through literature, I found 3 kinds of termination.

1) Spara Matching, match S22 to 50 ohms (my view: safest but would putting a matching network degrade the power gain by 50% or -3 dB, maximum power matching theory)
2) Putting a common drain buffer (my view: how to deemb the effect after measuring its Sparamaters???)
3) the LNA has a LC tank load, saw some literature, putting a dc blocking cap before the GSG bias tee (my view:LC tank resonates at my opearting freq leaving inductor's series resistor, even if from calculation i make the series resistance 50 ohms, how would putting a dc blocking cap match series resistor to 50 ohms of bias tee. The series cap would cause a anti-clockwise shift on the smith chart. Right???)

Hope it makes sense. Thanks for anyone's views.

Regards
Newbie

Back to top

IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: LNA Test Key
Reply #1 - Aug 4^th, 2007, 2:10am

Hey,

this is what i did and it seemed to work very well.

1) Add a 950 ohm in series with 50 ohm load in parallel to your LC tank load (or whatever load you have). I'm assuming the 1k resistance won't really affect your LNA.

2) Tap the output in parallel with the 50 ohm.

3) measured S22 will tell you the actual resistance of the 950 and 50 ohm resistors.

4) use these values to de-embed the voltage gain. Voltage gain is roughly 975/25*S21, but again you have to use S22 to find the actual values of the 950 and 50 ohm resistor.

5) NF can be measured by directly tapping a separate GSG from the normal LNA output. As long as the 50 ohm output doesn't change the DC operating point of the LNA you're ok.

BTW from what you said all three are a little difficult to de-embed. With knowledge of S22 you should be able to de-embed the 50 ohm load as per you 3rd method....use S22 to find Zout. assume S21 is gm(Zout//50) and de-embed the 50.

cheers,
Aaron

Back to top

« Last Edit: Aug 4^th, 2007, 6:49pm by aaron_do »

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

new_guy Junior Member Offline Posts: 10	Re: LNA Test Key Reply #2 - Aug 5^th, 2007, 9:00pm Hi Arron, I'm not too sure wat u meant, have drawn out the diagram, was wondering if its correct? Thanks Regards Newbie
Back to top	Drawing1.jpg IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: LNA Test Key
Reply #3 - Aug 5^th, 2007, 10:11pm

Yeah that's roughly what i did, but you need a blocking cap. In my case, i used a resistor load of 925 ohm, so i just split it up into 900+50//50. The advantage of this is the shape of the output in the frequency domain doesn't change. It would change with a buffer or LC match.

BTW if you want to measure noise, you need to add a separate output pad at the node between the transistor, the 950 and the LC tank. With a DC blocking cap too...

cheers,
Aaron

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

didac

Senior Member

Offline

There's a million
ways to see the
things in life

Posts: 247
manresa,spain

Re: LNA Test Key
Reply #4 - Aug 6^th, 2007, 3:09am

Hi to all,
In testing I've used the methods 1 and 2. They work well if you've got the option of manufacturing the matching network or the buffer as separate circuits to characterize them, also if you use a buffer it's probably better to measure Voltage gain with a PSA instead of s-parameters with a VNA. The advantatge of using a pi-match or similar is that you can characterize it with the VNA, obtain a touchstone file and directly put it the simulator to see how it affects the real implementation, also using some math you can de-embed it more easily than a buffer.
Hope it helps,

Back to top

IP Logged

new_guy Junior Member Offline Posts: 10	Re: LNA Test Key Reply #5 - Aug 13^th, 2007, 2:14am Hi everyone, Thanks for all your helpful insights. I'm not really sure on adding resistors in the RF path, keep having the feeling RF signals would be attenuated, right? Thanks Regards New_guy
Back to top	IP Logged

didac Senior Member Offline There's a million ways to see the things in life Posts: 247 manresa,spain	Re: LNA Test Key Reply #6 - Aug 14^th, 2007, 6:29am Hi, I think that you are right, but this attenuation can be compensated easily by math when doing the measurements, what I'm not sure about is the effect of the resistors in the Q of the inductor and in the NF measurement. Hope it helps,
Back to top	IP Logged

aaron_do Senior Fellow Offline Posts: 1398	Re: LNA Test Key Reply #7 - Aug 14^th, 2007, 5:42pm Yes, better to make sure before you use it. For me i had no load inductor, so i conveniently split up the load resistor. If you add an extra resistor maybe you'll have more difficulty de-embedding.
Back to top	there is no energy in matter other than that received from the environment - Nikola Tesla IP Logged