Stability in LNAs

Thread Starter

dlsirisha

Joined Nov 22, 2019
1
There are 3 LNA topologies - Common Source, common gate and cascode structures. Cascode structure with a tank as the load impedance is considered as the most stable design. I am not able to decide among the common source (LNA with source inductor degeneration) and common gate LNAs which designs are more stable.
 

Janis59

Joined Aug 21, 2017
1,201
Just now I am researching the ways how to make a 60 dB capable LNA at 1...2 GHZ from very bottom of DC. Thus the topic is electrizing me.
So, my impression, only way to overcome well over the 40 dB limit for good 3-transistor CE-CB combination cascade is cascode cascade with pre-CE cascade (stays written 60 dB).

And as in my case I need a 150 mA signal output then some paralelled CC in output cascade. At least the Spice model says it may work well, because of last CC cascades return the K(U)=0,99 thus never creates the oscillations. I am planing for first 3 bjt of sort the BFU910X (F(t)=90 000 MHz and last 3 bjt of sort the BFP620 (F(t)=65 000 MHz).

More headache is about the capacitor-less input biasing I need for DC senses. I cannot to immaginate any other system like two identical cascades where one input is signal gnd and another is signal hotwire. And circuit gnd is something well insulated off the signal gnd. So, the whole circuit is sth like 7 transistors.

P.S. RE:""with a tank as the load impedance""
I cannot work with a tank as I need a flat response in whole possible frequencies including DC.
But, if any tank is set, the Q factor will be multiplied with factor of 10-100 if to sweep away from FR4 pcb material.

PPS Ah, ya. Forget about fet~s (pHEMT) as the only having GHz range are 200-600-2300 USD price for piece. In to return to those I named pays 20-50 cents per pcs.
 
Last edited:

Janis59

Joined Aug 21, 2017
1,201
FR4 is very bad material for LNA even for input area, as the cross-reflections from glass thread to thread causes the wide spectrum blue noise. Take the Rogers instead - one of Duroids, or ceramics (but it is more costly). As noted for tank area FR4 is detrimental.
 

Janis59

Joined Aug 21, 2017
1,201
At Your classification You forgot an another rather sophisticated "sliding bias with parallel channel" design for large signal cases like here https://www.mwrf.com/sites/mwrf.com/files/uploads/2016/10/20/30C_F1.gif

Then vertical circuit of pure cascode alias CE-CB https://www.google.com/url?sa=i&rct=j&q=&esrc=s&source=imgres&cd=&cad=rja&uact=8&ved=2ahUKEwjxseSviK7mAhUMC-wKHUysAZYQjRx6BAgBEAQ&url=https://www.qsl.net/va3iul/LNA Design/LNA_design.htm&psig=AOvVaw13hJO7KiFn6697DljKp0Qp&ust=1576169846293467. Then cascode combined with CE OR CB pre-amplif input like here https://www.google.com/url?sa=i&rct=j&q=&esrc=s&source=imgres&cd=&cad=rja&uact=8&ved=2ahUKEwi4-ZDHkq7mAhUGEBQKHYfpD8cQjRx6BAgBEAQ&url=http://iopscience.iop.org/article/10.1088/2053-1591/aaafea&psig=AOvVaw0OMkvngRHB-xHtHEa81-9V&ust=1576172579577336 . Or, I quess, with adittion with CC before everything for input Z multiplication to beta (as my main concern is high input impedance as the input wiring length will be shorter than lambda divide to 40 thus the matching may be ignored (seems). Then famous 30-40 dB circuit with 2 or three CE with negative loop via emitter resistor or combined with negative via 2.nd bjt emitter resistor and positive via 3rd bjt emitter (what allows set the input impedance so low as 5 Ohms however last is bit stucky at very high end of mm waves https://www.qsl.net/va3iul/ and particularly https://www.qsl.net/va3iul/High_Frequency_LNA_schematics/High_Frequency_LNA_schematics.htm . At least those 2-bjt version I had been used extensively for antenna amplif and that always works well (but 30 dB only) - as here https://www.qsl.net/va3iul/High_Frequency_LNA_schematics/LNA_schematic_05.gif

Here as well is not bad collection of different approaches to LNA thing https://www.uydudoktoru.com/dosyalar/elektronik/fmvericiarsivi/YeniArsiv/Homebrew_RF_Circuit_Design_Ideas.htm
 
Last edited:

Janis59

Joined Aug 21, 2017
1,201
Thanks, but it are not readable on normal computer. Its kind of some kinder kindle or other quirky reader playing page in post-stamp size on my screen without button how to resize, and no button for download. Sad, because those books are not bad, may be used for students well.
 
Last edited:

Janis59

Joined Aug 21, 2017
1,201
Re: Bertus ""When you do ask questions, you may look stupid.""
Reminiscence from the time I worked at environmental police for government and Health ministery was issuing a certain letter something to ask we do for them. But the wording and grammar was so knitted (and Latvian oveall is language with extra sophisticated grammar having 1000 pages in small font only for describing the laws) so I was completely not able to realize WHAT they want from us. So I went to my big boss and asked - probably I may just CALL them and ask in normal conversation what they mean. His answer was - You awful mad man, never even think about such, because they will THINK probably we have difficulcy with understanding a simple questions. :) :) Then he arranged to answer them citing practically all their text with minor changes and added yes on no, thus the case was closed. Probably such attitudes are the main reason wrecking many "failed states"???
 

Janis59

Joined Aug 21, 2017
1,201
And last: today I am meditating about where to get the input cable gnd point which may not be a circuit gnd point.
Otherhow capacitor into input is strongly prohibited, but without of it the biasing may flush into closet as soon the input will be plugged.
So ideally it would be the duplicat of input cascade (thus operating from base to the base), but its double complicated, thus double expensive for mass product. From just resistor divider?? But then thermical instability... What else may be used with absolute zero voltage between input base and this point? Diode plus resistor?? Will it pass well in wide T range?
 
Top