The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> RF Design >> Bandwidth in Linear PA
https://designers-guide.org/forum/YaBB.pl?num=1278414763

Message started by purplewolf on Jul 6^th, 2010, 4:12am

Title: Bandwidth in Linear PA
Post by purplewolf on Jul 6^th, 2010, 4:12am

whats the relation between bandwidth of a single ended linear PA vs diffrential PA .. There is no specific LC tank in the diffrential PA like class B as compared in single ended design..
is diffrental offers twice the BW as compared to single to single ended????????i dont get it?

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 19^th, 2010, 2:23am

in the single ended design i have classical formulas for LC tank design with which i can adjust my bandwidth..
but in diffrential i have transformer only doing impedance transformation from source to load... how do i come to know whats the BANDWIDTH of my load network

Title: Re: Bandwidth in Linear PA
Post by pancho_hideboo on Jul 19^th, 2010, 8:23am

purplewolf wrote on Jul 19^th, 2010, 2:23am:

in the single ended design i have classical formulas for LC tank design with which i can adjust my bandwidth..

What do you mean by "classical formulas for LC tank design" ?

purplewolf wrote on Jul 19^th, 2010, 2:23am:

but in diffrential i have transformer only doing impedance transformation from source to load...

Actual transfomer is no more than mutual coupled inductors which forms double resonator, that is, coupled LC-tank.
So it is same as "classical formulas for LC tank design", although I don't know what you mean by "classical formulas for LC tank design".

Also see http://www.designers-guide.org/Forum/YaBB.pl?num=1240229676

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 19^th, 2010, 11:59am

Q = R/wL
R = 50
Q = f/BW
wL = wc

From TH Lee book.

Title: Re: Bandwidth in Linear PA
Post by pancho_hideboo on Jul 20^th, 2010, 9:55am

purplewolf wrote on Jul 19^th, 2010, 11:59am:

Q = R/wL
..........
wL = wc
From TH Lee book.

What on earth do you mean by these ?

In your formula, Q is proportional to R. It is very funny.

Title: Re: Bandwidth in Linear PA
Post by Ken Kundert on Jul 20^th, 2010, 1:45pm

In an LC tank when the resistance is in parallel with tank, the Q is proportional to the resistance.

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 20^th, 2010, 5:15pm

yups... Ken is absolutely correct

Title: Re: Bandwidth in Linear PA
Post by spring on Jul 20^th, 2010, 6:13pm

purplewolf wrote on Jul 6^th, 2010, 4:12am:

Hi ,purplewolf
Can you post you schematic?

Title: Re: Bandwidth in Linear PA
Post by RFICDUDE on Jul 20^th, 2010, 7:30pm

Your original question asked about single ended versus differential impedance as it applies to amplifier matching to some desired load (I assume this is what you meant?).

The only difference between differential and single ended matching is that under the same conditions the differential parasitic capacitance of the amplifier may be reduced compared to the single ended amplifier. Why? because a differential amplifier looks like two single ended amps driven differentially. The size of the two devices is likely to be smaller than the single ended equivalent (although it does not have to be the case). The shunt capacitances of the two single ended amplifiers appear to be in series for the differential amplifier thus making the capacitance look smaller for the diff amp. There is no real advantage to this over the single ended amplifier.

The resistive losses may be a little less too because the resistance of each half of the diff amp looks like they are in series for the differential output. So, it may be that the resistance across the tuned tank is less for the differential case and perhaps this is why some may consider the Q to be higher for the diff amp?

The paradox seems to be that BW decreases as Q increases for a resonant circuit, so I too am perplexed by the assertion that BW is wider for a differential amp versus single ended.

However, there can be a LC tank for class B PA. Most textbook differential class B circuits are pseudo differential with a center tapped transformer. The center tapped transformer represents an equivalent half circuit for the class B amp, so the circuit really looks like two single ended circuits driven differentially. The transformer usually offers some RF common mode reject such that there still is some difference between the differential and common mode gain of the amp.

I'm not sure if this helps answer your brief question :-/

Title: Re: Bandwidth in Linear PA
Post by pancho_hideboo on Jul 20^th, 2010, 8:16pm

purplewolf wrote on Jul 20^th, 2010, 5:15pm:

yups... Ken is absolutely correct

No.
It seems your Q is for parallel RLC-tank.

Q=ω₀*C/G=R/(ω₀*L)

But if you consider parallel LC-tank where R means loss of L, Q is never proportional to R.

However your equation is still funny.

purplewolf wrote on Jul 19^th, 2010, 11:59am:

..........
wL = wc
From TH Lee book.

Should it be 1/(ω₀*L)=ω₀*C ?

purplewolf wrote on Jul 6^th, 2010, 4:12am:

There is no specific LC tank in the diffrential PA like class B as compared in single ended design..
is diffrental offers twice the BW as compared to single to single ended????????

Frequency characteristics of amplifier with LC tank is bandpass.
Why can differential PA give twice BW as compared to single to single ended ?

Ideally differential PA with no LC tank should show lowpass characteristics, unless there are DC-Block capacitors.

If you use two Single to Single ended Class-B Amplifies with LC tank as differential PA, BW is same as one Single ended Class-B Amplifies whose frequency characteristics is bandpass.

purplewolf wrote on Jul 19^th, 2010, 2:23am:

but in diffrential i have transformer only doing impedance transformation from source to load...

Actual transfomer is mutual coupled inductors which forms double resonator, that is, coupled LC-tank.

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 21^st, 2010, 2:15am

for individual inductor the Q = wL/R, where R account for the losses of inducto..
but the loaded quality factor of tank is Q = f/BW =R/wL=50/wL
wL =wC <=> 1/(ω0*L)=ω0*C <=> 1/(ω0*C)=ω0*L. .. makes no difference.

Dont mix up the two Q's...

I just assume that the bandwidth is either twice (or may be half) incase of diffrential as compared to single ended since it is combination of two single ended..I havent found any mathematical proof about its BW in literature .. Long story short ,I have absolutely no idea about it.

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 21^st, 2010, 2:30am

here are the two schematic

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 21^st, 2010, 2:31am

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 21^st, 2010, 3:49am

change the file extension to .tiff if theres problem in viewing

Title: Re: Bandwidth in Linear PA
Post by pancho_hideboo on Jul 21^st, 2010, 7:25am

purplewolf wrote on Jul 21^st, 2010, 2:15am:

Dont mix up the two Q's...

No, you don't understand Q of resonators.

Consider a definition of resonator's Q.
Resonator's Q is defined as ratio of reactive energy and loss energy at resonance for any resonator.

Q of parallel LC-tank resonator is Q=ω₀*L/R, here R means loss of L. This Q is not inductor's Q.
Q of parallel LCR-tank resonator is Q=ω₀*C/G, here G means parallel conductance of LC-tank.

purplewolf wrote on Jul 21^st, 2010, 2:15am:

wL =wC <=> 1/(ω₀*L)=ω₀*C <=> 1/(ω₀*C)=ω₀*L. .. makes no difference.

Again see your description surely.

Do you think "wL =wC" is surely correct ?
I think your "wL =wC" makes clear difference compared to 1/(ω₀*L)=ω₀*C <=> 1/(ω₀*C)=ω₀*L.

Your "diffrential.jpg" uses ideal transfomers which are not physical.
So your "diffrential.jpg" should show lowpass characteristics, since there are no DC-Block capacitors and no resonators.

But actual transfomer is mutual coupled inductors which forms double resonator, that is, coupled LC-tank.

purplewolf wrote on Jul 21^st, 2010, 2:15am:

I just assume that the bandwidth is either twice (or may be half) incase of diffrential as compared to single ended since it is combination of two single ended..

Again see http://www.designers-guide.org/Forum/YaBB.pl?num=1278414763/9#9

Study characteristics or behaviors of double resonators or coupled resonators
Generally BW will be narrower if you use double resonators.
But if you permit two peak characteristic over passband, you can increase BW using double resonators.

purplewolf wrote on Jul 21^st, 2010, 2:15am:

I havent found any mathematical proof about its BW in literature..

You seems to be new comer in RF-engineering world.
Such things are decribed in any RF books.
Sorry I don't have international books which are written in English.

Read true classical-RF circuit text books which are oriented for discrete RF circuits not RF-IC's.
Or search old application notes on discrete PA by Motorola(=Currently On-Semi).

Title: Re: Bandwidth in Linear PA
Post by Markaka Electronics on Jul 21^st, 2010, 9:14am

agree with ken

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 21^st, 2010, 11:11am

thnx pancho...

Title: Re: Bandwidth in Linear PA
Post by purplewolf on Jul 25^th, 2010, 9:08am

very good document...

Title: Re: Bandwidth in Linear PA
Post by pancho_hideboo on Jul 25^th, 2010, 9:40am

I don't have international books of this topics which are written in English.
So I searched relative useful documents in English.
See http://www.oersted.dtu.dk/upload/institutter/_oersted/uddannelse/emi_kurser/31415/Ch2Rfcirct05.pdf

purplewolf wrote on Jul 25^th, 2010, 9:08am:

very good document...

I have more good text books in my native language.