Designer’s Guide Community

unevb

Junior Member

Offline

Posts: 18

Common Source LNA stability
Apr 26^th, 2010, 9:58pm

I am designing a single ended common source cascode LNA with inductor degeneration for a 5.5GHz receiver

Is there any intuitive explanation about how the k-factor is dependent on the circuit paramaters (gm, vds or L,C values). When I burn a current 'I' in the LNA, k>1 across 10MHz to 10GHz, it is unconditionally stable. But if I reduce the current to 'I/10' , all other circuit performance are meeting spec, but the k drops below 1 for most the the frequency of interest.

What is the relation between circuit parameters and k-factor or stability? If k is poor what can i improve in the circuit?

Back to top

IP Logged

pancho_hideboo Senior Fellow Offline Posts: 1424 Real Homeless	Re: Common Source LNA stability Reply #1 - Apr 29^th, 2010, 8:54am unevb wrote on Apr 26^th, 2010, 9:58pm: If k is poor what can i improve in the circuit? - Put shunt resistors for input or output or both which are ideally effective only for frequency range of poor K. - Resize transistor - Neutralization http://www.designers-guide.org/Forum/YaBB.pl?num=1264582305/5#5
Back to top	Kita━━━━━━(ﾟ∀ﾟ)━━━━━━ !!!!! http://www7.plala.or.jp/ungeromeppa/flash/kita.html http://www.youtube.com/watch?v=mjIxGh55bMM&feature=related IP Logged

unevb

Junior Member

Offline

Posts: 18

Re: Common Source LNA stability
Reply #2 - May 5^th, 2010, 10:46am

Why is k-factor an issue at RF but not an issue at low frequency (audio-range) circuits? For a single device k<1 at low frequency (where we compute MSG) and after the threshold frequency k>1 and we care about MAG, so I would think k-factor to be important at all frequencies.

If we are using a device which has k>1 at a mm-wave frequency eg. 60GHz �then it is said to be "unconditionally stable" when used at 60GHz. Does this account for possible oscillations at lower frequencies when k becomes less than 1?

Not many IEEE Papers on LNA circuits talk about the k-factor of the circuit. I have connected a 90nm device I am using , in common source configuration and plotted k-factor for different dimensions and it is never unconditionally stable over entire frequency range. Is there something that can be done at the architecture level to take care of this? I have not seen integrated circuits using input shunt resistors because of noise considerations.

Thanks!

Back to top

IP Logged

pancho_hideboo

Senior Fellow

Offline

Posts: 1424
Real Homeless

Re: Common Source LNA stability
Reply #3 - May 5^th, 2010, 11:00am

unevb wrote on May 5^th, 2010, 10:46am:

Not many IEEE Papers on LNA circuits talk about the k-factor of the circuit.

K-factor is not used except for LNA generally.

K-factor is not used for PA design where large signal is concerned.

I don't use K-factor even for LNA.

See the followings.
http://www.designers-guide.org/Forum/YaBB.pl?num=1218635777/5#5
http://www.designers-guide.org/Forum/YaBB.pl?num=1266403928/4#4

unevb wrote on May 5^th, 2010, 10:46am:

Why is k-factor an issue at RF but not an issue at low frequency (audio-range) circuits?

Consider loads for RF and audio.
Former is a resonator in many cases where impedance change drastically with frequency.
Latter is a resistor in many cases where impedance almost does not change with frequency.

pancho_hideboo wrote on Apr 29^th, 2010, 8:54am:

unevb wrote on Apr 26^th, 2010, 9:58pm:

If k is poor what can i improve in the circuit?

- Put shunt resistors for input or output or both which are ideally effective only for frequency range of poor K.
- Resize transistor
- Neutralization

http://www.designers-guide.org/Forum/YaBB.pl?num=1264582305/5#5

unevb wrote on May 5^th, 2010, 10:46am:

I have not seen integrated circuits using input shunt resistors because of noise considerations.

It seems you only know "EDA Tool Play" and technical paper.

We make bias circuitry and feed back network having frequency characteristics to prevent unpreferable oscillation in actual circuits.

Shunt resistor which is effective for some frequency range is also this purpose.

Cascode structure could be candidate for some case.

BTW, have you been able to understand the following ?
http://www.designers-guide.org/Forum/YaBB.pl?num=1272508752/3#3

MAG, MSG and U are not dependent on reference impedance even if S-parameter expressions are used.

Back to top

« Last Edit: May 6^th, 2010, 2:06am by pancho_hideboo »

Kita━━━━━━(ﾟ∀ﾟ)━━━━━━ !!!!!
http://www7.plala.or.jp/ungeromeppa/flash/kita.html
http://www.youtube.com/watch?v=mjIxGh55bMM&feature=related

IP Logged

unevb Junior Member Offline Posts: 18	Re: Common Source LNA stability Reply #4 - May 5^th, 2010, 6:33pm Hi, Thanks for the reply. It is true, I am still at the very early stage in RF design and right now beginning to pick up "EDA-tool" play and learn from Technical papers. Thanks for your patience in answering my questions. Regards,
Back to top	IP Logged

RFICDUDE

Community Fellow

Offline

Posts: 323

Re: Common Source LNA stability
Reply #5 - May 7^th, 2010, 7:32pm

By "unpreferable oscillation" we mean that the real part of the impedance is "negative" over some frequency range which can only happen if there is positive feedback.

You are wise to be concerned about low frequency stability when working with microwave and millimeter wave circuits. Often times the low frequency gain is so high (as compared to the desired high frequency gain) as to cause positive feedback at low frequencies. This can result in sinusoidal oscillations at HF, VHF or UHF frequencies, or (even worse) wideband chaotic instability.

As Pancho suggested, low frequency terminations to prevent positive feedback at low frequencies is effective (a few books cover this). It can be more problematic in RC circuits. Bandpass tuned circuits may be less of a problem because the low frequency gain is naturally reduced by the reduced load impedance at low frequencies.

Back to top

IP Logged