RF Amplifier impedance matching technique question

[alftel]

We have design which includes LNA as an input point, and there is a possibility of impedance mismatch on both sides (RF input and load). One of the accepted techniques for impedance matching is to implement 50 Ohm attenuation.

1) Is it ok to implement 50 Ohm 0dB attenuator before input DC isolation capacitor and after bias tee DC isolation capacitor?

2) Should these attenuators be guarded by another DC isolation capacitors on opposite sides (i.e. guarded by DC isolation capacitors on both sides)?

I was thinking to use this part from Susumu - PAT120TR-ND part from Digikey - Susumu 0dB Attenuator

Any suggestions/recommendations/previous experience with such approach?

 

[volker@muehlhaus]

Is 0dB attenuator a typo? It will improve the matching by 0dB.
n dB attenuator will give 2*n dB improvement in return loss.
LNA noise figure will increase by n dB.

Attenuators have a DC path to ground, so if there is an external DC you need to add a series capacitor.

 

[FvM]

Not a typo, there's really a 0 dB part available, placed if you want to bypass antenuation function in your PCB. You should have at least 6 dB, better 10 dB attenuation to create some isolation.

 

[alftel]

Thanks Valker and FvM! The reason why I was thinking to implement 0dB attenuator is to present a perfect 50 Ohm matching component on both sides of LNA (Susumu part is 0dB and 50Ohm impedance on both sides with frequency range 0-10GHz). Ideas was as following:

Am I wrong with this concept? I just do not want any impedance mismatch on both sides outside of the circuit to "affect" LNA - not because it will fix potential problem(s) but rather avoid LNA misbehavior.

 

[FvM]

At best the "0 dB attenuator" acts as a few mm of 50 ohm transmission line.

 

[alftel]

So what you are saying is that this concept does not make any sense? It looks like is based on what I read here : Techniques for Improving Impedance Mismatch . I would be shy to use attenuation before LNA in my design. I guess better design and better cables will be the best approach. Thnaks to everybody for help and support!

 

[volker@muehlhaus]

As said, the 0dB attenuator is a thru connection. To improve matching by an attenuator, you need attenuation. And that has a lot of negative side effects, e.g. noise figure.

Impedance matching can better be done by lossless components (L,C).

Another question is if your LNA needs to be matched at the input. Often, LNA input is optimzed for noise figure which can result in poor input matching. That's not unusual.

 

[starchern]

0dB part is just for replacing part with attenuation. I guess it's just a zero ohm(a short). If you want to use attenuator to do the interstage matching, 3dB is OK for the design (at least 6dB return loss)

 

[alftel]

LNA is Internally matched to 50 Ohms - Minicircuits GALI-39+

Forget about 0dB - will something like 0.1-0.2 dB attenuator with 50 Ohm impedance make any sense in this case? Will LNA "see" 50 Ohm regardless to mismatch beyond attenuator?

 

[volker@muehlhaus]

Forget about 0dB - will something like 0.1-0.2 dB attenuator with 50 Ohm impedance make any sense in this case? Will LNA "see" 50 Ohm regardless to mismatch beyond attenuator?

Did you read my post #2? The improvement in return loss from a 0.2dB attenuator will be 2 * 0.2dB better return loss. Just 0.4dB better. Makes no sense.

There is no impedance matching magic in these attenuators. They just reduce signal (incoming and reflected) by the attenuation value. That's all.

 

[Georgy.Moshkin]

One possible solution is to make tunable matching network. It may be realized using stubs with varactors. Or mechanical tuning, some dielectric slab or other tunable element. Or 2x SPDT switch with internal SWR meter which protects LNA to 50 ohm resistor to ground until external device is "measured" to be within safe. Maybe you need to do a little research and decide if tunable matching is really necessary. Minicircuits provide S-parameters on theirs website, so you can check both mismatch and possible stability problems.