Unbelievable video, what a practical guide to help people get into amplifier design. As an aspiring builder of radar systems, I am grateful!

@MegawattKS

2 ай бұрын

Glad it was helpful! I finally added a "Part 2" for this, but it's in the Appendices :-) ; kzread.info/dash/bejne/n22Mks-fl8KenLQ.html

This series is the best high-level view on this topic. I am an electronic engineer and I think you destilled out all the key features without loosing yourself in the details as all of us engineers are prone to. Thanks 👍

@MegawattKS

2 жыл бұрын

This amplifier one was definitely the hardest - trying to distill 2 or 3 semesters of electronics into one episode. Good to hear it came across well. I've been concerned that this episode was pretty dense. Thanks for the reassurances !

Thanks again for a really excellent explanation - I'm looking forward to the next episode!

@MegawattKS

2 жыл бұрын

You're very welcome. Thanks for the support.

Looking forward to part2. I hope/wish you will add a design procedure process for RF amp from requirements involved, like how the drive power requirements of Mixer and the input signal strength from antenna dictates the choice of emitter current(Ie) and collector resistor along with mixer input impedence.

Awesome turor 🎉

Thank you for another great tutorial.

Thank you for these videos.

Great presentation, explained so clearly.

@MegawattKS

6 ай бұрын

Glad it was helpful !

@nick1f

6 ай бұрын

​@@MegawattKS Yes, it is very helpful. I am going through the entire series and I hope to make my own design of FM receiver. I like the idea of modular circuits, where we can swap and experiment with different blocks.

@MegawattKS

6 ай бұрын

Excellent. Designing and building your own should really help. Do you have a NanoVNA or TinySA and/or other test gear? Here's some info on the RF protoboards if you are interested: ecefiles.org/rf-circuit-prototyping/ @@nick1f

This is a great presentation that encapsulates everything I’ve been looking for. Do you by chance have a link where the slideshow file can be downloaded? It would be very useful for making annotations and notes during the study and design process. Thank you!

@MegawattKS

Жыл бұрын

Thanks. Yes, the slides are here: ecefiles.org/radio-design-101-slides/ There are also slides for some of the companion series as well as class notes here: ecefiles.org/

@aerofart

Жыл бұрын

@@MegawattKS Thank you so much!

great video thanks

Great stuff. Did you ever mention the specific gain and tuning for this amplifier or did I miss that?

@MegawattKS

5 ай бұрын

Thanks. The common-base amp usually gives about 15 dB of power gain (S21) at a collector current of 1 mA (5 mW consumption). I designed it with a fairly narrow (about 10 MHz) bandwidth so it had some selectivity, and adjusted the center frequency during testing. See this video in the companion video series on the NanoVNA and TinySA test gear for some details. There's a schematic with some hand-written analysis stuff at about timestamp 2 minutes 30 seconds into this vid (the gain shown calculated is before accounting for losses in the LC tuned circuit, etc): kzread.info/dash/bejne/eXWtxdqgktHJeLg.html

A great video thanks

@MegawattKS

2 жыл бұрын

You're welcome. Glad it was helpful.

useful information , best wishes brother

@MegawattKS

Ай бұрын

Thanks! Glad it was helpful. I finally got around to making a "Part 2" to this episode which goes into some of the background/theory in more depth. It can be found here: kzread.info/dash/bejne/n22Mks-fl8KenLQ.html (Thanks to your comment, I also added the link to it in this video's description so it's findable by others too :-) )

@theoryandapplication7197

Ай бұрын

@@MegawattKS i appreciate your effort, thank you again bro. indeed , i follow and watch all your esteem channel's videos, you deserve it. sincerely

@theoryandapplication7197

Ай бұрын

@@MegawattKS also i shared it ,

How do you calculate the input impedance to the common base amplifier (the 10ohms you mention at around 27:30)? I don't think the previous videos covered this.

@MegawattKS

Ай бұрын

Hi. The short answer is that it is roughly equal to 1/gm, where gm = Ic/(0.04) for the transistor used. So when biased at about 4 mA for Ic, Rin = 10 Ohms. A much longer answer is given in Part 2 of the video here: kzread.info/dash/bejne/n22Mks-fl8KenLQ.html . I queued that up to the part that shows the formulas. As mentioned there, the formulas given in the slides for the different configurations are found in textbooks - but the Part 2 video presents the model for the transistor that is used in those derivations :-)

@alleblanc39

Ай бұрын

@@MegawattKS that's great, thank you!

Is it possible to ask you more questions? I have quite a few questions about this and would love to pick your brain. This video series is awesome.

@MegawattKS

2 жыл бұрын

Sure. I try to reply to all comments. However, my response time can sometimes be a little slow. Our HVAC system went out recently so I got behind. But will try to check comments more often now that I got it fixed. :-)

@MegawattKS

2 жыл бұрын

I saw your question about the biasing, but now I can't see it anymore. To try to answer what I remember from it, the device used in the class (and the Radio Design 101 amp) is a MMBV5179 - a surface mount version of the classic 2N5179 RF device. It's an NPN bipolar transistor, so biasing assumes that Vbe drop is about 0.7V. BJTs are easier to work with than FETs are in that sense. See timestamps 22:09 and 26:22 for the BJT biasing procedure and example values. The 0.7V drop is of course not precise, but its close enough for most designs as noted in the class handout at 22:09.

I like the details and theories on LNAs, and believe they are really critical for getting deeper understanding of RF amps. But I notice that there are quite a few LNA gain blocks, which I think may simplify the design, save time and ease the learning curve, I use RF gain block ICs as LNA. (Such as TQP3M9009 (50M-4GHz Gain is 27dB, 0.8dB NF) or BGA2869 (DC-2.2GHz, Gain 31dB, 3dB NF) ). But as I google it or browse through youtube videos, I rarely see people using such ICs as LNA to place JFET and BJT based LNAs in their FM radio reciever designs. Am I on the right track? Or I am doing something wrong? I use those ICs for 2 reasons. 1) They simplify my design and save debugging time 2) Their performances are stunning. Just reading the datasheet of those RF gain block ICs, I think it is very difficult for me to build RF LNAs with discrete components to match the performance and noise figures of those commercially mature LNA ICs. I just need to mind about impedance matching between RF gain blocks (usually 50 ohms) and down stream mixers. I even plan to use those LNA ICs for IF amps as long as the IF freq is within proper working frequency of those ICs. Do you recommend using RF gain blocks/ICs to replace JFET/BJT LNA for FM radio receiver?

@MegawattKS

2 жыл бұрын

Good questions. We avoided those gain blocks in the class that this material is based on because we wanted to teach matching and filtering. But also, the gain blocks have some limitations when making the FM receiver. 1) They are broadband, so you don't provide the preselection filtering needed (see the latest video uploaded yesterday called Epilogue 1). Filtering comes for free when discrete RF amps are built, since the resonators are needed to address capacitances in the devices/etc. Also broadband amps tend to be power-hungry, and with wide bandwidth could potentially be more prone to oscillations. That said, they are fine to use. Things will still work. They just may not be ideal for the reasons above. Hope that helps. 73's

Looks like Tuned circuit is producing a oscillation that gets to the output. I was confused seeing a strong signal, tried turning off everything in the room, but it still had it on SA.

@MegawattKS

2 жыл бұрын

On the plus side, you are actually seeing the oscillation. In the class, we used 3 GHz spectrum analyzers because the transistors we used had a habit of oscillating well above 100 MHz - typically somewhere close to 1.5 GHz (which unfortunately the TinySA can't detect, so a transistor with a lower fT like 1 GHz is a good defense. Question on what you're seeing though: What frequency and power level? If its inside the FM band, its possible it is actually a strong station getting into the inductors. If not, try touching the coils to spoil their Q, and/or decreasing the supply voltage to make it stop. Oscillations are hard to track down - part of the 'art', although there are known reasons sometimes, and gathering clues by touching the coils and/or lowering the voltage can sometimes help localize where the problem might be coming from...

@MegawattKS

2 жыл бұрын

Common emitter and cascode circuits are probably more prone to oscillate due to high feedback and gain respectively, although common-base can do it too. If the base bypass capacitor in a CB or Cascode and its associated interconnect has too much inductance, some signal from the output can feed back to the base through Ccb (and not be AC grounded by the bypass due to excessive inductance). Just throwing out some ideas based on things we've experienced in the class. A solution in that case was to put a 10 Ohm resistor in series with the base bypass capacitor to "spoil the Q" of the LC circuit that forms between Ccb and the parasitic inductance. Again - these things can be hard to find and fix. Hope you have good fortune in your search.

thank you.

@MegawattKS

2 жыл бұрын

You're welcome :-)

is there a replacement/equivalent for the MMBT5179

@MegawattKS

Жыл бұрын

Yes. In the video, I actually used a surface-mount version of an S9018 transistor since that's what was available in my "Interstellar Electronics SMD" kit and it's Ft value was around a GHz (the 5179 is 1.4 GHz). One could probably even use a classic MMBT3904 - although the collector-to-base capacitance is a bit higher and the "Ft" value is lower (300 MHz min at 10 mA), so the matching networks will change a little and the amplifier gain will be a bit lower for a 100 MHz amp. Hope that helps :-) FWIW, two places that stock these (and the MMBT5179) are DigiKey and Mouser. Those are the distributors we have used a lot. I got the "Interstellar" kit from Amazon, but it contains lots of other parts and costs a fair amount.

38:00 Is it ok to have L1 parallel with L3 like that? Doesn't that affect the resonance frequency?

@MegawattKS

Жыл бұрын

It won't affect the center frequency Fo since the matching network presents a pure resistance so there is no effect on Fo. The L is resonating with the C in the matching network so it's not presenting inductance to the tank, at least at the resonant frequency. Off frequency the matching network will have an effect which modifies the Q to some extent. And hence the bandwidth...

@coffeecuppepsi

Жыл бұрын

@@MegawattKS thank you

"when triodes are glowing electrons are flowing" QRO,,

Is IMD relevant and if so, how to measure?

@MegawattKS

2 жыл бұрын

It's definitely an important performance parameter in some situations. In particular, receiver front ends, and transmitter power amps if they're using OFDM or QAM or other linear modulations modulations. Measuring is a bit involved - normally requiring two signal generators, a power combiner, and a good spectrum analyzer. My IC735 HF transceiver uses an LNA/Mixer set that consumes about a watt to make it's performance better in this regard !

8:25: Your triode gate batteries are backwards. 32:25: L2 and L3 are in parallel. Why doesn't this skew the resonance frequency of the tank?

@MegawattKS

2 жыл бұрын

Good catch. I thought that might cause some confusion, but decided to let it go since it's technically correct in the sense that the grid battery has to be plus side up to work correctly with the graph. The graph shows the value is negative in normal operation. So the implication is that the top of the battery is actually a negative voltage, even if the schematic looks like it's positive. Very confusing since it looks unusual. Sorry about that. The answer to the question about L2 and L3 is that L3 in the matching network has to resonate with C3 in the matching network. Hence, it has to disturb the total L when they are combined. Otherwise there would be no "residual L3" to resonate with C3 to do the match. To put it another way, if it did disturb Fo, then one would be tempted to take it out, and then the matching network would not be complete. So it has to decrease the total L, but doesn't shift the frequency. Hope that helps.

@hectorbarrera9496

9 ай бұрын

@@MegawattKS so if I understood, L2 + L3 = Total, but L3 needs resonate with C3?

@MegawattKS

9 ай бұрын

@@hectorbarrera9496 Yes. But to be precise, we should write it as Ltot = L2 || L3, where the || operator means "use the parallel resistor formula", not addition. Ltot will actually be lower than L3 when the two are combined in parallel.

@hectorbarrera9496

9 ай бұрын

@@MegawattKS Thank you

Link of part 2 of this episode 3 RF Amplifier

@MegawattKS

Жыл бұрын

Unfortunately I never did a follow-up on this one. But I probably should. I've been thinking about it and now that the Antenna Briefs series is done, hopefully I can revisit this. Maybe I can also cover a bit more on matching networks at the same time, which I've been thinking would be useful. Will give it some thought... Thanks for the inquiry !

@kamleshchavan7451

Жыл бұрын

@@MegawattKSThanks for the reply Please do it sir , secondly if you can please make video's on basic electronics e.g. multivibrators using transistors , different amplifiers of different class , for the beginner's in electronics

Sir still waiting for the part 2 of this video

@MegawattKS

7 ай бұрын

Sorry it took so long. I finally did one (though it doesn't cover everything). It is called "Appendix A - Transistors and Amplifiers at RF" and is at the end of the Radio Design 101 playlist. Hope it helps.

Would someone explain the possible output if the final RF stage input power doesn't exceed 1 watt?

@MegawattKS

7 ай бұрын

The devices and circuits talked about in the video are focused on receiver design. So they typically never see more than a few hundred microwatts (or a few milliwatts) of input power. They are also too small to handle watt level signals without damage. While the video series doesn't really address power amps, there is one slide in this video that shows how some of those are done (depending on the modulation type used in the system). See time 23:00 in the video. The left side shows a class-C power amp topology (but a higher power transistor is needed than the 2n5179) kzread.info/dash/bejne/h4mg07lmprbXabA.html

@DumbledoreMcCracken

7 ай бұрын

@@MegawattKS thank you very much for a straightforward set of clear instructional videos!

"if triodes are glowing electrons are flowing"

gm is transconductance and is used mostly for Valves and fets yes!!! A bjt???well technically yes but talking about Gm with bjts is unusual because you would be measuring Ic/delta vbe. That assumes that a bjt is a voltage device like a tube or fet...transistors are current devices. You wont find gm on a transistor data sheet.

@MegawattKS

3 ай бұрын

I agree that they were originally conceptualized as current-controlled devices, but there is no need to continue that tradition (except in biasing where it matters more). Indeed, a common more modern view is to treat them just like FETs - as voltage controlled current sources when in the active region. The value of doing that is that FETs, BJTs, and Vaccuum tubes can then all be analyzed with the same equations and circuit architectures (though the biasing will be different and the formulas for gm are different for each). I've recently added an Appendix A to this series where this is covered in some detail. Here's a link, with it queued up to the part that discusses this. It's covered in about a minute or two from this point: kzread.info/dash/bejne/n22Mks-fl8KenLQ.html

@paulphoad

3 ай бұрын

I had a look. I appreciate the work you've done, however, using transconductance with a device which has HFE or beta, both current amplifying specs is silly. No offence. You can get really close approximations for gain using dynamic emitter resistance and load impedance. Work out quiescent emitter current. Divide 26mV by this current to give re. Gain in simply collector load /re plus any unbypassed emitter resistance. Easy peasey. So a common emitter amp with a Q current of 1ma, with bypassed emitter resistor and a 1k collector load has a voltage gain of 1000/26 =38.5 Why make things hard??

@MegawattKS

3 ай бұрын

​@@paulphoad Hi. Apologies in advance for a long answer... One formula for a CE amp (with bypassed bias resistor RE) is Av = -Rc/re as you said. Agreed. But that is identical to the gm based formula Av = - gm Rc. This is true because re = Vt/Ic as you said, and gm = Ic/Vt in the gm based analysis. So re is simply 1/gm. Neither formula is based on a BJT having a current-gain spec. And neither method or formula is harder than the other. They are identical - except that one may look more familiar depending on how things were taught. However, it is true that some other formulas exist which factor in Beta and can look quite different than in the gm model based analysis. I use and taught the gm based model exclusively because it translates directly to FETs, so there is no need to change things when we moved to analog or RF integrated circuit design using CMOS processes. BJTs have that annoying base current - but that's captured easily by adding r_pi = (hfe*re)=(hfe/gm) to the model, which is the only topological difference in the BJT and FET models then. However, it goes beyond that. Tracing back to the original physics analysis of BJTs, the modulation of collector current occurs via the PN junction voltage at the base-emitter (with a correction factor alpha approx=0.99'ish capturing the base recombination, resulting in the Beta parameter). IMPORTANT NOTE: The use of the simplified formulas Vt/Ic or Ic/Vt is limited to textbooks. Real-world devices have a junction grading coefficient n as well. So re = nVt/Ic (or gm = Ic/(nVt)), At higher frequencies, even that breaks down and the industry uses measured Y or S parameters. Y parameters are directly compatible with the gm based model. The (2n)5179 device used in RF design and in this video series is a good example. Please see Figure 9 of the Motorola datasheet for the 2n5179 device which graphs Y parameters versus frequency. www.silicon-ark.co.uk/datasheets/2N5179-DATASHEET-MOTOROLA.pdf "gfe" in that figure at low frequency is gm (or 1/re). From the 0.04mA/V at 1.5mA bias, Fig 9 indicates that n is actually 1.5 in this real-world device, not the 1.0 value assumed in most textbooks for examples and problems. So the real world gain would be 1/1.5 lower. Moreover, neither model would work above 250 MHz, as the forward transconductance falls off due to extra physics (base transit time effects if I remember right) not captured in the simplified equations/models. I hope this helps explain why I don't use Beta in AC models and analysis (except to find r_pi in the model). Not trying to be difficult. Just trying to add to the discussion. 🙂

@paulphoad

3 ай бұрын

@@MegawattKS nice, well thought out. I don't teach gm with bjts but I see your point. Gm being reciprocal of re. Your aproach is interesting. If it works for you then perfectly ok. Couple of things. In Australia we generally use 0.6v vbe rather than 0.7volts. It's much more accurate. If I were to use the term gm then students would automatically assume an open circuit input z. Being use to fets etc. The input z of bjt is parallel base bias resistors in parallel with hfe(re +RE). So not anything like a gm device. You seem like a smart man and a good teacher but I don't like applying gm to a current amplifying device which has transresistance rather than transconductance. Happy Eastef to you and family Paul.

@MegawattKS

3 ай бұрын

Cool. I used 0.6 in some classes too, rather than 0.7. It's good to go against the proverbial "0.7" in any case. Students need to know that it's not a magic number that works everywhere (or at every current :-) ) On the gm thing, it's a choice for sure. A colleague who does our first electronics course told me he covers both. I think I first got into the gm model from using the "hybrid pi model" in the textbook we had in the 80s. (cf. en.wikipedia.org/wiki/Hybrid-pi_model ) Hope your Easter was a good one as well. Cheers! - Bill

You are repeatingly making the same mistake with current direction indicator made on the first tube triode drawing (5:39)!

@MegawattKS

Жыл бұрын

Sorry - I'm not sure I understand the issue. The current arrows shown follow the "conventional current" convention used in circuit analysis and design, where current flows from positive V++ , through the tube, to ground. While it is true that electrons leave the cathode and flow to the plate, that is "electron current" - not conventional current. Does that help? (See, for example: web.engr.oregonstate.edu/~traylor/ece112/beamer_lectures/elect_flow_vs_conv_I.pdf )

i dont understand.. all the videos done like this.. noone... noooooonnneeeee has a highlited pointer.. the old hand or arrow and it fades into the background... someday somewhere someone will use a seeable pointer.. anyways good videos so far besides the pointer hahah

@MegawattKS

2 жыл бұрын

Thanks for the feedback. Believe it or not, I totally agree. This is definitely a problem. In my case, it's due to PowerPoint and the MultiCam software I use to capture the voiceover+zoom/pan on the slides. I have tried several times over the last year to find a way to get a better pointer - but no luck. Even when I go to win+R and change the pointer for Windows as a whole, PPT overrides it during zoom, replacing it with the smaller hand thing because it thinks I need to be reminded that I can pan while zoomed in. And the most annoying thing is when it blanks the cursor after you pause too long. PPT software won't give you a stable way to create a large pointer (or I haven't found it). I'll keep looking though ! (I'm currently employing "Reading View" during production to maximize the pointer size - rather than "Presentation View" which has it's own issues. Microsoft: If you're listening, please help us all out by allowing the user to chose and size a pointer independently of what PPT thinks is best !)

@XPFTP

2 жыл бұрын

@@MegawattKS ive never done anything on linux for video capture. but i wonder if would work better on linux... ill have to look into it some day ... but hey over all videos are good. ive watching the rf amp video.. learning more from it then anything ive read or watched ... wanted to build a 6m amp. so hope after watching it a few times iam able to .. lol