Appendix A (Episode 3 - Part2) Transistors and Amplifiers at RF

This is Episode 3, Part 2, but is optional, so it is placed in the Radio Design 101 video series as Appendix A. The material here is not required for design and construction of our FM radio receiver. Rather, it is provided for those who want more background in the analysis of small signal RF amplifiers, and the operation of transistors used in their design. We review Episode 3 (Part 1), and then show how BJT transistors are created and how they are modeled. The goal is to provide a broader foundation for those interested in different amplifier types (such as common-emitter and cascode), and in the underlying mathematics from which gains, input and output impedances, and operating frequency limitations can be understood.

[@theoryandapplication7197]

the best video in electronic i have ever seen , thank you sir

[@MegawattKS]

Wow. You're welcome. Thanks for the comments!

[@KerryWiegand]

Great video! Thanks for sharing it!

[@theoryandapplication7197]

wonderfull , great explanations

[@RodrigoForrequi]

Again a great deep video, that make think what a great way to learn from (watch KZbin). I really like your way to explain, sometimes only scratch a subject, and makes me want to check how to derive it. I'm following your material to learn and design these "projects" of your course in my lab. I made a PCB of the amplifier and expect to receive it in a few weeks. Really excited to begin the tests. Thank you for the nice videos. It's better to have a slide full of gold than a animation with meanless metaphoras.

[@MegawattKS]

Thank you for your kind words. I'm glad to hear the videos are helpful and that the presentation style works ! Hope your amplifier performs well when the board arrives and tests begin !

[@VividSolutions]

Superb. Thank you for your time presenting this so beautifully. It brings back vivid memories from my college days when I initially fought to grasp this stuff. It is great to recap it 35+ years on, espically as it seems so much clearer now, probably because you explain it so well. Please don't stop!!

[@MegawattKS]

Thank you for the kind and encouraging comments. They are appreciated very much !

[@W1RMD]

That was awesome! W2AEW and The Offset Volt have a great videos on cascode amplifiers.

[@MegawattKS]

Thanks. Good references and channels. I might do one on cascode amps at RF. Oddly, KZbin doesn't have any that appear with that as a search term. But I'll watch those first to be sure not to duplicate what's out there...

[@W1RMD]

I like multiple versions on the same subject anyway as I feel I learn better that way. So go ahead a do a video anyway. Thank's, it's been rewarding to watch all of your videos.

[@waltsailing2009]

Great video… I think this was termed the hybrid pi ac small signal model… one thing that can help understand this is the formula Ic=Icq+gm*Vbe is better written as Ic=gm*Vbe+Icq, It’s the same, but rearranged. And why, because the math now is simpler, in that it is exactly the same form as Y=m*X+b, which is just the equation for the straight line with y intercept of Icq, and m, which is the slope which is a straight line, hence linear. This is for Y is the collector current X is the base emitter voltage, and the lines slope is m, m=gm, and the y intercept is b=Icq, and that is one of the first formulas of math that everyone studies in middle or early high school.

[@MegawattKS]

Thanks! Two excellent points. The hybrid pi model is the first one I learned, and still my favorite. Apparently there's also a "T model" now that has been introduced into some textbooks for some reason - but the hybrid-pi version is much more faithful to the physics and is more conceptual, IMO. Good call also on pointing out the line equation. I'm looking at the y=mx+b formula now. Is Icq the Y intercept here? If I extend the yellow line I drew, it looks like b would be negative. But that would be using x = v_BE (the total BE voltage). Using the small-signal v_be for X as you did, it will definitely work ! - though we'd have to move the yellow line to the left, away from being the tangent-line and relabel the x axis. Hmm... So many subscripts ! Looking up tangent-line, I found this. byjus.com/tangent-line-calculator/ I guess we could write it as I_C - I_CQ = gm(V_BE - V_BEON), giving I_C = gm(V_be) + I_CQ , which is what you wrote 🙂 I think this is why I'm an engineer first and a math person only of necessity ;-) BTW - I see you're a sailer. Brings back memories of my youth - sailing with my Dad in his Cal 2-27.

[@waltsailing2009]

What better way to mess with ham radio than maritime mobile….so yes, we do a bit of sailing and also a bit of ham radio. Decided to play with some receiver design stuff and your channel popped up. Nice videos on the subject.

[@robballantyne3]

Excellent! Thanks for all your efforts, I appreciate you trying to communicate stuff which can be difficult. I have one question (if it makes sense and you have time) and a suggestion (if you want an idea for a further video). You mentioned in passing that R_o (the output / emitter resistor) in the model could largely be ignored but said that wasn't really the case for FETs. I'm curious why. My suggestion is based on my curiosity about the various types of transistors (BJT, MOSFET, JFET) what are their strengths and weaknesses and in what kinds of designs would they be most applicable. (I'm aware there are other more exotic beasts but I suspect they have more clear uses: HEMTs)

[@MegawattKS]

Thanks for the comments and suggestion! On the FET r_o parameter, my main experience is with MOSFETs in IC design. Once the "process node" shrunk to sub-micron, an effect called channel-length modulation became stronger and r_o decreased. Here's a page with some good diagrams that discusses this. Notice that 1/r_o is the slope of Id vs Vds and the slope is significantly larger than in a typical BJT. So r_o is smaller. buzztech.in/channel-length-modulation-in-mosfet-vlsi-design/ The solution in analog/RF amps is to use the Cascode configuration. In addition to the frequency response improvement through lower Cin, it also raises the output resistance :-)

[@Mark_KE8YCV]

Dr. Kuhn, These are wonderful, but did you upload this one with low resolution? Most of the detail in the diagrams is indecipherable.

[@MegawattKS]

Hmm... I just checked and my computer was streaming it at 720p. There is also a full 1080(HD) option in the KZbin settings menu. Hopefully that will help. But - there's also the full set of slides in PDF form here :-) ecefiles.org/radio-design-101-slides/

[@Mark_KE8YCV]

@@MegawattKS Oh my gosh! What a goldmine! Thanks!

[@rjordans]

Great video again, thanks! Question though, is the closing bracket in your Miller capacitance calculation in the right place there? If expect it to be before the Cbe

[@MegawattKS]

Good catch. Yes - you are right - _except_ I think you mean before Ccb, not Cbe? Is that what you mean? Fortunately I did correct it during the voiceover, and the one on the website is corrected also. Actually, it needed an absolute value sign on Av as well: Cin = Cbe + (1 + |Av|)Ccb since technically Av is negative for a common-emitter amp...

[@rjordans]

@@MegawattKS ah yes, correct!

[@samadams6487]

I'm a little confused by your section where you have an npn transistor as a voltage control voltage source and I have always learned that it was a current control source. Could you elaborate?. I am aware that an FET transistor is in fact a voltage controlled voltage source in the same better that vacuum tubes were considered voltage-controlled sources.

[@MegawattKS]

Thanks for an interesting and kind-of deep question. At the physics/device level, all can be considered voltage controlled _current_ sources. (not voltage controlled voltage sources) As you said, FETs and tubes are both voltage-controlled. This is by virtue of the electric fields produced at the control terminal (relative to the cathode or source) modulating the current flowing through the other terminals. But as mentioned in conjunction with the BJT diagram, any voltage placed across the base-emitter terminals creates an electric field as well - in the base-emitter depletion region. This is fundamentally what modulates current in a bi-polar junction transistor. So fundamentally it's voltage controlled as well. (Actually MOSFETs can be modeled very similar to BJTs in this sense. They have NPN or PNP structure, but an insulating gate control terminal.) BUT, with that said, BJTs can also be viewed and modeled at a circuits level as being a current controlled current source. Either conceptualization can lead to a usable circuit model. Not the best way to view them IMO though, since it obscures their basic similarity in application circuits to FET and tube devices. I assume the reason people started thinking in the ICIS terms (instead of being a voltage-controlled current source) was to acknowledge the small base current - and perhaps because Shockley et.al. used "current amplification" as a way of saying it is able to amplify something ? But for modern design, treating all as VCIS devices allows a uniform treatment of all the devices in apps such as amplifiers, and is therefore 'best', IMO. 🙂

[@samadams6487]

@@MegawattKS in my electronics course in college all of the models and even all of the models from my father's textbooks which at that time covered both vacuum tubes and transistors all used current controlled current sources I've never seen until you showed me a model of a BJT transistor using a voltage controlled source. I admit I've been away from it for a while but since I'm retiring next year as I have been in the communications industry for some time, and as I have a ham radio hobby I thought I would start building both with transistors and with vacuum tubes and I have acquired a large number of vacuum tubes as well as a Conar 224 vacuum tube tester. If you have any references to sources that go into details about using voltage control sources for the standard BJT transistor I would be very interested and you are providing those resource Links or textbook references thank you very much

[@MegawattKS]

@@samadams6487 I also learned transistors (BJTs) from the current-controlled view. That is what the textbook used. But as a designer at RF, I switched to the voltage control model I think a good way to think about this is through "two-port parameters". The 'hybrid' parameters are historically used for BJTs. This set of 4 parameters uses current as independent variable on the input and voltage on the output - so we get a current controlled current source. However, RF BJTs typically use "Y parameters" with voltage as independent variable on both input and output (and currents as the dependent variables). Hence, voltage controlled current sources. See for example the first several pages of this set of notes from Niknejad: rfic.eecs.berkeley.edu/~niknejad/ee142_fa05lects/pdf/lect4.pdf And page 4 of the 2n5179 datasheet here: www.silicon-ark.co.uk/datasheets/2N5179-DATASHEET-MOTOROLA.pdf

[@MegawattKS]

Note that Niknejad's pdf talks about making the selection of y-parameters based on his interest in feedback circuits - but there is more to the story The 2n5179 and other RF devices moved to Y parameters because the h-parameter model wasn't good at high frequency. Y-parameters were _measured_ parameters, so they provided a better design base. For example, page 4 of the 2n5179 datasheet shows both g and b values for things like y21 (called gfe and bfe in Figure 9 in the lower left. 'g_fe' is the same as gm. (It's called g_fe because it's the forward path y-parameter measured in common-emitter configuration.) But note how it is only constant up to about 250 MHz. The susceptance values (like b_fe), give info from which the parasitic capacitances can be estimated. Some people even abandon the circuit model and just computer gains/etc based directly on the y-parameters as Niknejad does in his pdf. But that's not absolutely necessary. The circuit model still has utility in terms of providing insights - so when designing, I stick with it (or the derived parameters shown in the video for the amplifier configurations, together with estimates of the C values). Oh - we should also mention S-parameters. Y parameters became hard to measure when in the GHz range. Hence, the industry moved to S-parameters. The 2n5179 datasheet even shows both 🙂

[@samadams6487]

@@MegawattKS I hate to ask you this but could you put all that text from the answer above into a text file and attach it to the other attachments in this particular podcast because I don't seem to be able to make a copy of it off of my phone parrot perhaps I might be able to do so from a computer at home but I'm away from my home right now and I'm not entirely sure whether I can grab these notes of yours here and they look more interesting to look into I thank you for your excellent responses. I haven't seen whether or not I can see the PDF I'll go look for it in your notes section