The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> LNA power consumption
https://designers-guide.org/forum/YaBB.pl?num=1154941429

Message started by aaron_do on Aug 7^th, 2006, 2:03am

Title: LNA power consumption
Post by aaron_do on Aug 7^th, 2006, 2:03am

Hi everybody,

I have a question about LNA power consumption. Suppose we have a system with say 16 channels, and the maximum input of the system is -20 dBm. At this signal level, the LNA's power consumption is much larger than at small signal. For example instead of 5 mA (-60 dBm) it could be 15 mA (-20 dBm = 31 mV pk into 50 ohm). This is because the bias point of the input transistor has changed.

Most papers I read quote the low end power consumption (5 mA). However, the standard might specify a blocker level much higher than the signal level...for example -30 dBm 802.11.

1) First question: if blocker level = -30 dBm, does that mean every other channel can have -30 dBm, so for 16 channels, total blocker ~ -30 + 12 dB = -18 dBm (approximately since actually there are only 15 blocking channels)

2) Secondly if this is true, won't this drain the power supply since the system must amplify all channels before channel selection.

thanks,
Aaron

Title: Re: LNA power consumption
Post by ACWWong on Aug 7^th, 2006, 7:00am

Hi Aaron,
My response to your points is:

1) Detailed reading of the specification will tell you, but your reasoning is sound. I'm not familar with 802.11, but I am familiar with IEEE type specifications... i'm sure somewhere buried in the spec/type approval test definition will be detailled blocker levels with offsets and no. of blockers.

2) Yes RF rectification of LNA inputs often occurs... admittedly it seems your example of 31mVpk causing 3 fold increase in current sounds in the realms of non-degenerated bipolar design... for MOS I find it hard to see -20dBm into 50 ohms can cause such a large change unless you have a very high Q input match....
Anyway if you know the blocker frequency offset, one can design the input bias to react to the blocker to limit the effect of rectification.... examples of this can be seen in action by the likes of Jeff Durec (was motorola, maybe freescale now ?). Google him to see his LNA biasing schemes which either "trap" the blocker energy or de-bias the LNA in influcence of large signals to limit current consumption increases.

When it comes to data sheets, it is note common for people to quote increased current in blocking conditions... but we all know signal dependant current consumpiton does exist... just on of those spec-man-ship things...

hope my ranting help...

cheers

aw

Title: Re: LNA power consumption
Post by aaron_do on Aug 7^th, 2006, 6:30pm

yeah...i just made up the actual numbers but i'm looking into weak inversion biasing so the increase in current can be significant. thanks for the help,

Aaron

Title: Re: LNA power consumption
Post by loose-electron on Aug 8^th, 2006, 10:16am

What you are saying about the LNA bias point current changing as a function of signal amplitude does not make much sense. Can you explain a bit more?

LNA's are generally Class-A amplifiers.

You generally have a handful of scenarios to meet:

Input referred noise
Linearity (under assorted IM scenarios, generally defined in the system specification)
Bandwidth (duh! :o )
Gain

The blocker in the passband is always going to be increasing the current consumed but that's a given. Usually most of the static current is there to meet linearity issues.

I've done LNA's for UWB, WCDMA, GSM, and the blocker-linearity issue tends to be the "fugly" one so to speak.

Am I missing something here?

thanks,
Jerry

Title: Re: LNA power consumption
Post by ACWWong on Aug 8^th, 2006, 11:12am

Agreed, usually the linearity is one which normally dictates the power consumption of an LNA.... because linearising the LNA means reduction of gain (increasing noise), and as you are have to meet gain and noise specs, you end up burning more juice for the linearity....
Blockers in the passband do change the current, gain (compression or in the case of weak invserion you gain more gain increase before final compression) and noise...
From the designs of LNA I have done, I have seen RF rectification that changes the effective bias point point of the LNA. But the extent to which that affects the current is dependant on the bias impedances/scheme and tend to suddenly appear beyond a certian signal level.

Anyway, with some good design, you should be able to keep any unwanted current increase to a minimum.

Title: Re: LNA power consumption
Post by loose-electron on Aug 9^th, 2006, 12:29pm

OK, good - now I understand.

I would think that linearity would get wiped out well before the point that you describe.
Perhaps not if it somethig that the linearity spec is pretty loose.