Thanks a lot Shariar for explaining the purpose of 90 degree (quadrature) couplers and amplifier input mismatch. Also on the Doherty amplifier, the LDMOS inputs are fed by a Wilkinson divider, but, i do not know why, it is assymetrical (one trace is wider than the other one). It would be so interesting to have one of this board's engineers send us technical details (and i strongly suspect many of them are subscribed to TSP) !!!
2 роки тому+1
Preamplifier is for Receiving, a LNA is a Preamplifier and it is used for Receiving, The 1st Stage of amplification is Called the Driver or the Exciter.
Oh, and other than transformers, too. Speaking of which, I have a 2nd question! Is a passive DC "transformers" explicitly prohibited by physics, or have we merely failed to invent them? (Obviously they'd have to work on some other principle than AC transformers, since the fields wouldn't constantly be changing.) Thx :-)
OK, here is the question: I am a big fan of all your video, just a simple question: can you tell us more about your career background, and how you learn the stuff about every aspect of electronics ? I mean that every engineer in the industry usually working on a daily job, and for me, I am an engineer for automotive electronics, and I don't have a bunch of time to learn something else. So how could you manage with the time management ?
+OtherTheDave A Transformer works by converting the energy in two stages. First, electric energy gets converted into a changing magnetic field and then vice versa. This could also be done by converting the electric energy into other forms of energy like kinetic energy (for example with motors) or, a very ineffiecient method, light shining on a solar panel. Im not sure if you could consider the last one a passive DC transformer, but something like that could be heavily optimized in the future.
Right, that's how AC transformers work, and varying the ratio of windings on the primary and secondary coils is what lets you change, say, 2v 5a power to 5v 2a power (or any other ratio, as long as the product is 10w). I was wondering if a passive circuit could do the same thing for DC power.
Hi .. submitting my question and thanking you for coming back to us. Your videos are amazing and I learn so much from them. I have never designed any RF circuits before and the maximum frequency I've gone in my designs is 300MHz (mostly for processors, memories and communication ICs). The thing is, since I first saw your 2 hour video on the internals of an spectrum analyzer I became marveled with the RF world and I've been studying a lot about the components of such systems (SDR, QAM, OFDM, spectrum analyzers, etc). I can truly say I've learned a lot, which unfortunately is not the same as knowing a lot. On hand not knowing a lot sucks, but on the other hand, it means that I can improve and discover new things. I know that most of these blocks that compose the systems can be implemented digitally, as well as analogically. I am planning on designing a few systems: a spectrum analyzer (there are some open source projects online that can get me started) and a customized QAM communication system (I have experience designing protocols from the third OSI layer, with this project I'd be learning to design the first two layers as well). I know that there is a lot of value in designing every single block by hand (even the frequencies are not very high), but my question is .... is it realistic to design most of the blocks inside an FPGA (as much as possible ... NCO, downconvertes, upconverters, filters, etc) instead of using external components (I reccon that I must use external components, I know some of them are absolutely necessary)? Or do you recommend that I start using external components? I mean .. designing the PLLs, filters and all by hand or in an FPGA would make me learn more, but what is the chance of them working properly compared to using known components that work? I ask that because of the disappointment factor. Even though all of this stuff makes me happy to learn, spending years on the same project without advancing can be demoralizing. Sorry for the long post, but I find it extremely difficult to express myself with few words. PS: I started watching your videos last year when I HAD to finish my master thesis, but couldn't stop watching your videos. I'm happy to say that as of now I've got my master's degree. Thank you so much for the contribution to the community.
+TheSignalPathBlog good to see you again, one remark for the -3db you should say that "-3db = signal power / 2" as we split the signal in 2, the power is divided by 2 on each output. As you explain after the real loss is the insertion loss (0.022 dB)
I totally appreciate your comments about Bell Labs - many of the sites became Lucent, which was eventually sold off...pity. I grew up within sight of the Holmdel facility... I remember walking through that BLabs building in Murray Hill - back when I was a young engineer, they had a mock up of the Telstar satellite in the lobby...my dad worked there. I wound up at RCA in Somerville and I got to sit next to - literally in the cubicle next to - the guys who invented CMOS. I was nervous on the interview, and even more nervous when I got the job and had to perform my engineering skills next to these guys. They were gray-haired and very patient with me. At the time, I was at the top of my class and I felt like my knowledge was in peak condition. But now with the hindsight of, ahem, several decades, I realize how little I actually knew and the skill and patience and absolute joy of the old guys who mentored me - they were passionate about their craft. They fought daily like cats and dogs, but they loved the work, and they totally respected each other and were the best of friends. It was a "family" of sorts, which I think is rare to find in today's industry, with high turnover and rapid reward/failure cycles... Thanks for this tutorial. Very enlightening.
Good to see you back. Congrats on the new place; I can't wait for the tour. Also, am I the only person with a non-technical background that can maybe comprehend only about 5% of the things you talk about but still love the hell out of these videos? Seriously, I know nothing about this stuff, but for some reason I love these videos.
Questions: 1. Why/how do you find such valuable equipment during dumpster dives? Do you have a network of scouts? Is it mostly that a prestigious place like Bell Labs has a bunch of technical businesses nearby? What tips would you give for younger divers? 2. Estimate your success rate on repairing equipment. Do you sometimes attempt repairs and fail, and not show us? Or perhaps you succeeded but the equipment wasn't interesting? 3. Have you ever hidden any easter eggs on ASIC or other small chips you've made? If so, pics!
Hi! First of all, thank you for all your videous! They are always magnificent! And I'm looking forward to see more! For Q&A: How can one get into the microwave design engineering (R&D)? I mean, the problem is you have to have hands on experience, for instance, in designing amplifiers. But you need an expensive equipment, software, parts and everything. And you can't just use only computer simulators all the time. So, what to do? Suggestion for a new video: Can you tell us about making fixtures for mm-wave devices (and beyond!)? Maybe you can show us one and explain how it works, how it was made, what problems can be with such things and how to avoid them. Your channel is the best! Thanks again!
Great video, I always had troubles understanding circulators but you solved that in 30 seconds, I think it will be very educational to inject and follow a signal through the amplifier using a scope and SA. Even more so if you inject "faults" so you can see the advantage of the construction. How about a video tour through bell labs with you as the guide ? A possible interesting video subject: Is the "wave machine" still there ? There is a movie on the bell labs archive about that thing. That helped me very much to understand reflections. It would be great if you could do a remake of that. Fred
This is one of best tech video ever watched. What a clarity and technical content. Only the good tech guy can explain like this. I request you to post more technical videos to educate and have fun. Thanks. Welcome to tech group. The real tech group....
Great Video. I've been playing with some microwave amplifiers to use with Amateur Television DVB-T and DVB-T transmission. I love the way you can see what's going on in microwave equipment as the signal path is usually easy to follow and the boards have a kinda geeky beauty that isn't seen in lower frequency gear.
Geeky beauty? All microwave devices are elegant and beautiful and not a bit geeky. They reflect a physical ( visible) manifestation of the laws of physics and the more you understand u-wave physics the more you can see the beauty.
Great video, despite the improvised setup in your new home.RF design is a totally different world for me, anyhow your explanations, I'm able to follow easily.Well done.
Great video! I learned something new today: how to do AB+C on microwave frequencies. You also could have done a more in detail explanation about importance of matching between preamp and amp input (for stability and flatness) as well as of output matching/isolation (for protection). Can't wait to see the new lab in all its greatness.
Also as beautiful as the most microwave amp designs are, some of us sadly know what for, those isolated rectangular traces, all around the LDMOS, are. If I'm too pessimistic please tell me.
Thanks for a great intuitive video over high power balanced and doherty amplifier. If you can please elaborate about the temperature compensation bias network design and the use of DC-DC converter then it will be very helpful..
Great to see a new video, Shahriar. Congrats on your new home! I look forward to seeing what you've done with home automation, as I'm currently designing a new home myself. For the Q&A, I'd be interested to hear about a project that you've been involved in, about which you're most proud or which has been the most rewarding. Cheers!
Please tell us more about the Doherty output impedance matching network. I have read whitepapers about it, but still not understand the math well enough. This is the best video on youtube!!!
Hi Shahriar, good video, as usual. RF PCB layouts can be read as a book, if one understands the "language". My questions: What is the next "big thing" in the elecrtonic industry? Photon computers, IoT, robotics, bio medicine ? What are the main development areas Bell Labs is focusing on? BR U.
The last time I remember seeing dual-polarization antenna like that was on an LTE amplifier module from Alcatel Lucent which was half the size of that whole antenna and diplexer combined.
Great to have you back Shahriar! I look forward to your upcoming videos - I'm quite curious about the operating principles of circulators . I have a question for you: Are there any promising high-power devices or techniques on the horizon for more efficient RF/microwave/millimeterwave amplification? It seems like this is one area of high frequency design where engineers haven't been able to make much progress in over the years, despite many clever ideas and advances in manufacturing.
All of this is so great and interesting! Thank you! I've seen some directional couplers in my course and I thought that they are easily made on PCB by placing two parallel lines having exact sizes like width, length and distance between them. But instead the designers of the first amplifier bought a ready one. Is it because of the tolerances they would have to meet if they wanted to make one themselves or something else?
That is a good question. I guess that the tolerances will not be a problem, and neither the power handling. I can't really think of a reason but I'm sure Shahriar can tell us :).
A 3dB coupler is hard to achieve within in a simple two-layer PCB, due to the high required coupling and it is also sensitive to PCB variations and etching-tolerance: Inside the Anaren component are two tightly spaced striplines on top of each other, separated by a thin dielectric layer (broadside coupler). Also directivity is easier to achive in a homegeneous dielectric, than in PCB-air stack-up. Futrthermore the component gets smaller than on PCB by using higher dielectric constant material, and it isalso well shielded in itself.
Questions. First, nice to see you back, thanks for the video, as always top notch. Now for the questions, in the first amp the signal is split 0-90 and sent to one transistor and then it is spilt from 90-180 and sent to the second transistor, what is happening from 180 to 360 ? How broad banded are these splitters, my guess is that they 1/4 lambda lumped element and not broad banded at all. Next question, with the second amp (the Doherty) you said one transistor is operating in class AB, did you mean class A? and the second transistor is operating in class C. Don't you need 2 transistors to operate in class AB? Lastly these amp do not seem to be very linear, what modulation modes do they operate, my guess is that they would be used purely with some form of constant envelop angle modulation like FM?
You've misunderstood. The quadrature hybrid takes the input signal, and outputs a copy of the entire signal, 3dB down and with zero phase shift on one port. It also outputs a copy of the entire signal, 3dB down and delayed by a phase shift of 90 degrees. I.e., sine in, sine AND cosine out (each 3dB weaker than the input sine). The question "what is happening from 180 to 360" therefore makes no sense at all, the outputs are both complete, full waveforms.
The hybrids are based on broadside coupled quarterwave lines. A single stage coupler is in fact quite broad band: A balanced amplifier covering 1.8-2.7 GHz using such is quite possible. Regarding class-AB: this is a "pre-biased class-B compromize" to give softer turn-on characteristic and better linearity than class-B, and as such, it only conducts half periods. In the low frequency world I guess that implies you need to operate in push-pull, in order to faithfully reproduce the signal, but for RF, the output waveforms contain a fundamental component that is filtered out from the output network characteristics: Linearity is of course not perfect, but that's why you need predistortion ;-) (The coupler on the output is most likely feedback path to an adaptive predistorter.) Finally, the whole point for Doherty is to increase efficiency for high dynamic amplitude modulated signals 3G, 4G, 5G.
Thanks for the amazing and clear explanation. My question is that for the Doherty amplifier what would be the maximum power that can the class C add to the output power (I assume it's 3dB)? My second question is that the Doherty amplifier usually has bandwidth limitation so what kind of methodology can be used to increase the bandwidth in this case? My third question is that if you use a common base rather than a common emitter for the Doherty amplifier what would be the main advantage or disadvantage in this regard (in terms of power and phase noise)?
Nice to see you are still alive!=) -How were your grades in school/university? (Just curios if there is any connection between real knowledge and grades). -Do you think the simplified Arduino platform is generally a good or bad thing for students/young engineers? I am working on my master thesis right now and the setup I am using is based on Arduino and was build by the guy who finished his Ph.D. few months ago. Everything this guy done with electronics is just fundamentally wrong. He didn't even try to read datasheets for the component he was using. For example he uses LT1210 as voltage follower. His circuit doesn't have any bypass caps (You can't imagine how noisy this LT1210 was!) There is no compensation for output offset voltage (due to current through feedback resistor) . The relays he is using are too slow, not reliable for our application. The wiring is messy and on some places just wrong...and so far and so on. The main problem is,that this guy stated:"Everything is tested and working fine" And this is not the first time I see the case, where the Arduino platfrom with all premade libraries and shields gives someone a false feeling of having experience/knowledge in real engineering and electronics. -Do you have any ambitious hobby projects you are working on? -Do you rewatch your own videos? When I try to watch/listen my tutorial videos, it is always so cringy I immediately switch the sound off:D -Which other engineering related youtube channels do you watch?
Regarding your first question, I suggest you should read up on the effect known as "survivor bias". Veritasium has a reasonable summary. You are setting yourself up to reach some very invalid conclusions by canvassing successful ("survivor") engineer's grades; that's completely backwards stats. Also, it takes all sorts; there are people who have no passion and no studying ability who are bad engineers and get bad grades; passionate, intuitive engineers who breeze through university exams and are good real-world engineers; study-hards who learn exactly how to pass poorly-written exams but suck at real-world engineering; and finally savants who have the potential to be amazing real-world engineers, but find the exams so boring that they don't actually get particularly great grades. So yeah, every permutation is covered, not to mention all the shades of grey inbetween. Having said all that, it's a very interesting question; just saying it's not a "useful" question per se. Also, I know you weren't asking me; but I think Arduino is a mixed bag -- good to encourage a wider audience to be interested in electronics; but the way that their documentation hides that the Arduino board is just an ATmega328P breakout board is really anti-educational.
I would really love someone to do a video on general ideas of substitution. I only have limited time in my schedule for learning this. and all of it is split between learning, experimenting and my garden. I would love to know a general way to understand what I can substitute certain components with when I do not have that exact resistance/capacitance etc.. (470 pf is one I could not find very much information on)
typical output power in dBm, for wcdma, also referred to as effective radiated power, ERP, is in the order of 10 - 15 dBm for dense metro propagation model.
There is a thing called the link budget which accounts for the gains and losses on an rf system, wcdma relies heavily on power control, so i meant to say even though these RRHs transmit with 20, or 40 watts, the user equipment only receives, because of fading and other factors, around -35 dBm on the port of the antenna, which is plenty. These propagation models are based on the isotropic dipole, so the real figure is in reality, higher. And user equipments can not go any further than 600 mW, in the older days of AMPS yes, but not anymore.
Yes, I know about link budgets but you mentioned ERP :) And class 1 wcdma mobile peak power is +33 dBm = 2W. That's just in the specs so I don't know if they are actually made.
no, for sure. Remember Ec/no is very sensitive to BTS set output power, and BTS always tells the phone to "tone it down". 600 mW worst case scenario, and even that much is all the way down to 1x EVDO, for example, there is also link budget for BTS and link budget for UE. power balance is a big thing in wcdma! dont think in absolutes when translating dBm to Watts!
Great to have you back!!! I'm assuming there must be a significant efficiency differential between a pair of LDMOS amplifiers in the first configuration in comparison to the Doherty architecture, enough to warrant the hardware cost of adding the passive isolator? Can you give us a ballpark comparison?
Both PA's surely have a passive isolator, to protect it from "revererse IM": intermodulation arising from power leaking in to the PA output from nearby transmitters, and also to assure good matching in case of poor VSWR from ice on antennas, as well as hardware protection, in case it is powered up with no output termination connected. Isolators of this size typically have about 0.2 dB loss.
Great to see you back! I do also have a question for you: I just graduated as an electronics engineer and have a decent background knowledge in both electronics and mathematics. I must admit however that only have a very limited understanding about RF, impedance matching is almost as far as it goes. I would love to learn more about how to do RF designs and how to get started. I'm thinking to buy a descent RF spectrum analyzer with tracking generator in order to do some experiments and get some hands on experience. Do you have any recommendations for me? Topics to google, tests to do,... any suggestions are welcome! Your video are a wealth of knowledge by the way and got me really interested into the field of RF!
You can do a lot with a spec with integrated TG. However, there is no phase information so I would say that the first thing to get is a VNA, vector network analyzer. They're unfortunately not cheap and you also need an expensive calibration kit. For real understanding of the whole process I would recommend you get a simulation software, such as Keysight ADS och Ansoft HFSS, and design a distributed filter. Then manufacture and measure this board with a VNA. That's an excellent learning experience!
Happy to see new vid from you! Hope you get moving sorted out soon. It also does help to optimize inventory ;) As for question, what materials/articles would you recommend for a person getting VNA (HP 8753A)? I got one for upcoming repair project, but have no idea atm how to test it or even check if it works. What minimum gear would I need to check it's operation?
Great video as usual, I'm glad that the house with the new lab is almost finished, I'm looking forward to the upcoming videos. As a question: Would it be possible to have some high-resolution photo's available from the boards you analyse?
Great to see another video!Question: Instead of using the 20dB coupler to monitor the output, would it be possible to take the output from the second quad hybrid coupler output (the one after the PAs?) to monitor the output?
If you did that you would be measuring the reflected power, and while there is a relation to the output power it is not a constant relation but it will depend on the antenna, cable length and such.
Do you have any advice on how to advance in the workplace? I've contributed a lot to my company. I've re-designed all of the PCBs and their reliability, cost, and functionality have increased tremendously. Also, there have been several designs that I've developed independently from scratch that have exceeded their requirements. I love the hair, by the way. It's great to see you again.
Hey man, nice to see you back and congratulations on the new place. Question: How long have you been using Matlab is it any good. last one. How old is your Cat ?
Interesting ! as always. i have an cellphone-tower tranceiver module with the exact same data-power connectors as the first board. never seen before. Made by Ericsson as well ?
Very interesting design. On a semi-related note, do you have any advice for picking up equipment (specifically RF like spectrum analyzer or VNA)? Is it mainly persistent hunting on ebay?
So I arleasy ask you on twitter, but how you would transfer a lab like yours? What is the step to package the instrument in a good way? Second some day you can do a video about PLL and Frac/N divider (maybye a design of those would be nice)? Last would be good to some day, if possibile, do a videoblog from the bell labs!
q for Q&A: What are some typical precautions you take while testing / fixing / debugging RF devices so that you limit the rf exposure of your body to safe levels? At which point do you get concerned when dealing with rf devices? Can you recommend any good literature on this please? JE, Ottawa
Hooray! Always look forward to these videos. Does the isolator have any relation to a magnetron? If I understand correctly, both use magnetic fields to guide signals.
Good video. Any of these questions are available for you to consider. Thanks. These boards do not use all standard components, opting for microstrip line and specific dielectric boards. How are these circuits designed and laid out and why use these complicated filters, capacitors, and inductors, versus lump components? Is it only cost or also performance because sometimes they mix and match? Do they use Eagle, Altium, Kicad, etc to place these stripline components into the board? Where are they designed; like a simulator of some sort? I know they use specific boards and mills to produce these boards. What about layers, are they mostly 2 layer or do they get created with the same capability as common printed circuit boards? I am interested in designing analog circuits and RF circuits as I am too comfortable with digital domain. I am reading RF Circuit design by Chris Bowick as a resource to learn more about RF design. Not much that I have not witnessed before in coursework or the internet, but provides lots of basic worked out examples.
I found a video on a tool called genesys. Explains quite a bit of my questions. community.keysight.com/community/keysight-blogs/eesof-eda/blog/2016/09/05/how-to-design-custom-rf-microwave-and-analog-filters?sf35244374=1
Friend, how are you? I work in the electronic part of telecommunications... I have a question. I have a failure in an ericsson 01b5 rru that does not raise the PA in the cywin manager. I get dependency failed, it does not transmit or resive.. Could you please help me?
nice ! but why they need more power than typical few watts of phone ? i mean what's the point in being able to transmit without hearing response? i guess it's because powers of individual transmissions to individual phones somehow adds up all together cause they all use CDMA bitstreams with different sequences and one phone can only transmit one sequence at one time. can you go into more details about it? also it would be nice to hear more about whole 3G/4G modulations stuff !
Those are TRANSMITTING amplifiers in the base station, which provides the output power for ALL users in a cell-sector simultaneously On the receiver end a base station have low power LNA's as expected.
Hi, Shahriar. thanks for the great video! I have a question about both those amplifiers. There is only one transistor in each path, means there are two options of offsetting them. Option one is to have a B or AB class which means they only work on one half-period of the signal. If this is the case, then how is the rest of the signal on the output is reconstructed? Option two would be to offset the transistors into class A which means burning 50% of power on DC. That would work for the first amplifier you explained, but no way for this to work with the Doherty amplifier: handling both half-periods on the Class C path would require two transistors, and the saturating with one transistor path would be just Class A, not AB.
The parts of the Doherty amplifier work on different parts of the single half-period: main part and peak. If each part of the Doherty amplifier is built on a single transistor, then they will be only able to serve one half-period of the signal. At 24:00 Shahriar shows two graphs for Class AB and class C. My point is that it is not possible to make a single-transistor Class C amplifier that would serve positive and negative swings of the signal as shown on his graph. Also regarding his Class AB signal graph: with single-transistor it can only be achieved with Class A biasing, i.e. burning ~50% DC power at idle.
My suspicion is that the Doherty amp in this case works on a half-period, creating a distorted sinusoid on output, which is then filtered to get rid of unneeded harmonics.
Actually, the way of getting high efficiency in an amplifier is to minimize the voltage-current overlaps inside the transistors (and to compromize linearity): For class B and AB the current is half-clipped sinusoid or even better: square like (class-F etc.) The fundamental output power is then reconstructed through bandwidth-limiting output.
good morning friend, I saw a video on your channel that you use the cywing moshell software for rf radios like erisson... I work in a telecommunications company and I repair those boards electronically... I have a fault in an ericsson rru 01b5 that does not raise the pa on The manager gives me (dependency failed) and it does not give tx or rx....please could you help me please since my work depends on that
Q. Can the Doherty architecture be made to work with a quadratic hybrid on the output rather than a delay line (phase shifter). I am working on a Nokia PA which appears to do this??
My guess is that it is somewhere between the shown modules and the antenna, if you use the same antenna for transmit and receive you will need a duplex filter anyway and that will take care of the harmonics.
Really appreciate the description of the balanced amplifier and how the quadrature hybrid makes the input return loss look nearly perfect. Nice!
I am a big fan of your channel
Thanks a lot Shariar for explaining the purpose of 90 degree (quadrature) couplers and amplifier input mismatch.
Also on the Doherty amplifier, the LDMOS inputs are fed by a Wilkinson divider, but, i do not know why, it is assymetrical (one trace is wider than the other one).
It would be so interesting to have one of this board's engineers send us technical details (and i strongly suspect many of them are subscribed to TSP) !!!
Preamplifier is for Receiving, a LNA is a Preamplifier and it is used for Receiving, The 1st Stage of amplification is Called the Driver or the Exciter.
Such a great video and concise equation as proof. Next video: how to miniaturize the circulator, thanks & look forward!
Not a question, but it is great to you have you back online, creating interesting content.
The best English diction I have ever seen.
Don't forget to submit your questions for the Q/A session!
For the Q/A, other than inductors, does the magnetic half of electromagnetism ever play a role in solid-state devices?
Oh, and other than transformers, too. Speaking of which, I have a 2nd question! Is a passive DC "transformers" explicitly prohibited by physics, or have we merely failed to invent them? (Obviously they'd have to work on some other principle than AC transformers, since the fields wouldn't constantly be changing.)
Thx :-)
OK, here is the question:
I am a big fan of all your video, just a simple question: can you tell us more about your career background, and how you learn the stuff about every aspect of electronics ? I mean that every engineer in the industry usually working on a daily job, and for me, I am an engineer for automotive electronics, and I don't have a bunch of time to learn something else. So how could you manage with the time management ?
+OtherTheDave A Transformer works by converting the energy in two stages. First, electric energy gets converted into a changing magnetic field and then vice versa.
This could also be done by converting the electric energy into other forms of energy like kinetic energy (for example with motors) or, a very ineffiecient method, light shining on a solar panel.
Im not sure if you could consider the last one a passive DC transformer, but something like that could be heavily optimized in the future.
Right, that's how AC transformers work, and varying the ratio of windings on the primary and secondary coils is what lets you change, say, 2v 5a power to 5v 2a power (or any other ratio, as long as the product is 10w). I was wondering if a passive circuit could do the same thing for DC power.
Hi .. submitting my question and thanking you for coming back to us. Your videos are amazing and I learn so much from them.
I have never designed any RF circuits before and the maximum frequency I've gone in my designs is 300MHz (mostly for processors, memories and communication ICs). The thing is, since I first saw your 2 hour video on the internals of an spectrum analyzer I became marveled with the RF world and I've been studying a lot about the components of such systems (SDR, QAM, OFDM, spectrum analyzers, etc). I can truly say I've learned a lot, which unfortunately is not the same as knowing a lot. On hand not knowing a lot sucks, but on the other hand, it means that I can improve and discover new things.
I know that most of these blocks that compose the systems can be implemented digitally, as well as analogically. I am planning on designing a few systems: a spectrum analyzer (there are some open source projects online that can get me started) and a customized QAM communication system (I have experience designing protocols from the third OSI layer, with this project I'd be learning to design the first two layers as well).
I know that there is a lot of value in designing every single block by hand (even the frequencies are not very high), but my question is .... is it realistic to design most of the blocks inside an FPGA (as much as possible ... NCO, downconvertes, upconverters, filters, etc) instead of using external components (I reccon that I must use external components, I know some of them are absolutely necessary)? Or do you recommend that I start using external components? I mean .. designing the PLLs, filters and all by hand or in an FPGA would make me learn more, but what is the chance of them working properly compared to using known components that work? I ask that because of the disappointment factor. Even though all of this stuff makes me happy to learn, spending years on the same project without advancing can be demoralizing.
Sorry for the long post, but I find it extremely difficult to express myself with few words.
PS: I started watching your videos last year when I HAD to finish my master thesis, but couldn't stop watching your videos. I'm happy to say that as of now I've got my master's degree. Thank you so much for the contribution to the community.
+TheSignalPathBlog good to see you again, one remark for the -3db you should say that "-3db = signal power / 2" as we split the signal in 2, the power is divided by 2 on each output. As you explain after the real loss is the insertion loss (0.022 dB)
Great, I don't know of any electronics channel on this level. Please keep it up. Your time and knowledge is really appreciated.
I totally appreciate your comments about Bell Labs - many of the sites became Lucent, which was eventually sold off...pity. I grew up within sight of the Holmdel facility... I remember walking through that BLabs building in Murray Hill - back when I was a young engineer, they had a mock up of the Telstar satellite in the lobby...my dad worked there. I wound up at RCA in Somerville and I got to sit next to - literally in the cubicle next to - the guys who invented CMOS. I was nervous on the interview, and even more nervous when I got the job and had to perform my engineering skills next to these guys. They were gray-haired and very patient with me. At the time, I was at the top of my class and I felt like my knowledge was in peak condition. But now with the hindsight of, ahem, several decades, I realize how little I actually knew and the skill and patience and absolute joy of the old guys who mentored me - they were passionate about their craft. They fought daily like cats and dogs, but they loved the work, and they totally respected each other and were the best of friends. It was a "family" of sorts, which I think is rare to find in today's industry, with high turnover and rapid reward/failure cycles... Thanks for this tutorial. Very enlightening.
Good to see you back. Congrats on the new place; I can't wait for the tour. Also, am I the only person with a non-technical background that can maybe comprehend only about 5% of the things you talk about but still love the hell out of these videos?
Seriously, I know nothing about this stuff, but for some reason I love these videos.
I spent most of my career designing Doherty amplifiers, and I have to say, your explanation is great Shahriar. Love the videos, keep up the good work!
Questions:
1. Why/how do you find such valuable equipment during dumpster dives? Do you have a network of scouts? Is it mostly that a prestigious place like Bell Labs has a bunch of technical businesses nearby? What tips would you give for younger divers?
2. Estimate your success rate on repairing equipment. Do you sometimes attempt repairs and fail, and not show us? Or perhaps you succeeded but the equipment wasn't interesting?
3. Have you ever hidden any easter eggs on ASIC or other small chips you've made? If so, pics!
Hi! First of all, thank you for all your videous! They are always magnificent! And I'm looking forward to see more!
For Q&A:
How can one get into the microwave design engineering (R&D)? I mean, the problem is you have to have hands on experience, for instance, in designing amplifiers. But you need an expensive equipment, software, parts and everything. And you can't just use only computer simulators all the time. So, what to do?
Suggestion for a new video:
Can you tell us about making fixtures for mm-wave devices (and beyond!)? Maybe you can show us one and explain how it works, how it was made, what problems can be with such things and how to avoid them.
Your channel is the best! Thanks again!
also nice to see you back!
Great video, I always had troubles understanding circulators but you solved that in 30 seconds, I think it will be very educational to inject and follow a signal through the amplifier using a scope and SA. Even more so if you inject "faults" so you can see the advantage of the construction.
How about a video tour through bell labs with you as the guide ?
A possible interesting video subject: Is the "wave machine" still there ? There is a movie on the bell labs archive about that thing. That helped me very much to understand reflections. It would be great if you could do a remake of that.
Fred
You're back! Hooray!
Real nice to have you back, now i can sleep Again...
This is one of best tech video ever watched. What a clarity and technical content. Only the good tech guy can explain like this. I request you to post more technical videos to educate and have fun. Thanks. Welcome to tech group. The real tech group....
Great Video. I've been playing with some microwave amplifiers to use with Amateur Television DVB-T and DVB-T transmission. I love the way you can see what's going on in microwave equipment as the signal path is usually easy to follow and the boards have a kinda geeky beauty that isn't seen in lower frequency gear.
DVB-S & DVB-T
Geeky beauty? All microwave devices are elegant and beautiful and not a bit geeky.
They reflect a physical ( visible) manifestation of the laws of physics and the more you understand u-wave physics the more you can see the beauty.
Welcome back, Shariar!
you are the MAN!
Great video, despite the improvised setup in your new home.RF design is a totally different world for me, anyhow your explanations, I'm able to follow easily.Well done.
Good to se you back with a new video.
Great vid.
And great to see you back. You have been very sorely missed, my friend. :)
Congrats on the new home.
Glad you are back. You are inspiring many people through your excellent videos. Thanks for sharing your knowledge.
Love your channel. Searched out your channel after listening to your interview with the Amp Hour podcast. Fantastic material. Thank you.
So great to have you back! Very informative piece! Looking forward to duplexer/diplexer exploration ...
So am I: have it been performed and uploaded yet?
Nice to see you back. Great video
Great video! I learned something new today: how to do AB+C on microwave frequencies.
You also could have done a more in detail explanation about importance of matching between preamp and amp input (for stability and flatness) as well as of output matching/isolation (for protection).
Can't wait to see the new lab in all its greatness.
Also as beautiful as the most microwave amp designs are, some of us sadly know what for, those isolated rectangular traces, all around the LDMOS, are. If I'm too pessimistic please tell me.
Thanks for a great intuitive video over high power balanced and doherty amplifier. If you can please elaborate about the temperature compensation bias network design and the use of DC-DC converter then it will be very helpful..
Great to see you back
Great to see a new video, Shahriar. Congrats on your new home! I look forward to seeing what you've done with home automation, as I'm currently designing a new home myself. For the Q&A, I'd be interested to hear about a project that you've been involved in, about which you're most proud or which has been the most rewarding. Cheers!
Brilliant video, thanks so much
brilliant useful video
Motion sensing light switches are cool automation. A layer of security as well.
Please tell us more about the Doherty output impedance matching network. I have read whitepapers about it, but still not understand the math well enough. This is the best video on youtube!!!
Hi Shahriar,
good video, as usual. RF PCB layouts can be read as a book, if one understands the "language".
My questions: What is the next "big thing" in the elecrtonic industry? Photon computers, IoT, robotics, bio medicine ?
What are the main development areas Bell Labs is focusing on?
BR U.
In context it starts to make sense. Amazing that it works at such frequencies. Doesn't the metal back have all kinds of nasty field effects?
Great video as always!
Yay! New signal path video!
The last time I remember seeing dual-polarization antenna like that was on an LTE amplifier module from Alcatel Lucent which was half the size of that whole antenna and diplexer combined.
What a great video. I love watching stuff like this. RF and Microwave voodoo is the best!
Great to have you back Shahriar! I look forward to your upcoming videos - I'm quite curious about the operating principles of circulators .
I have a question for you: Are there any promising high-power devices or techniques on the horizon for more efficient RF/microwave/millimeterwave amplification? It seems like this is one area of high frequency design where engineers haven't been able to make much progress in over the years, despite many clever ideas and advances in manufacturing.
Yes! He's Back!!
Great, can't wait to see the next video.
GREAT VIDEO....THANKS!!!
All of this is so great and interesting! Thank you!
I've seen some directional couplers in my course and I thought that they are easily made on PCB by placing two parallel lines having exact sizes like width, length and distance between them. But instead the designers of the first amplifier bought a ready one. Is it because of the tolerances they would have to meet if they wanted to make one themselves or something else?
That is a good question. I guess that the tolerances will not be a problem, and neither the power handling. I can't really think of a reason but I'm sure Shahriar can tell us :).
A 3dB coupler is hard to achieve within in a simple two-layer PCB, due to the high required coupling and it is also sensitive to PCB variations and etching-tolerance:
Inside the Anaren component are two tightly spaced striplines on top of each other, separated by a thin dielectric layer (broadside coupler).
Also directivity is easier to achive in a homegeneous dielectric, than in PCB-air stack-up.
Futrthermore the component gets smaller than on PCB by using higher dielectric constant material, and it isalso well shielded in itself.
Questions. First, nice to see you back, thanks for the video, as always top notch. Now for the questions, in the first amp the signal is split 0-90 and sent to one transistor and then it is spilt from 90-180 and sent to the second transistor, what is happening from 180 to 360 ? How broad banded are these splitters, my guess is that they 1/4 lambda lumped element and not broad banded at all. Next question, with the second amp (the Doherty) you said one transistor is operating in class AB, did you mean class A? and the second transistor is operating in class C. Don't you need 2 transistors to operate in class AB? Lastly these amp do not seem to be very linear, what modulation modes do they operate, my guess is that they would be used purely with some form of constant envelop angle modulation like FM?
You've misunderstood. The quadrature hybrid takes the input signal, and outputs a copy of the entire signal, 3dB down and with zero phase shift on one port. It also outputs a copy of the entire signal, 3dB down and delayed by a phase shift of 90 degrees. I.e., sine in, sine AND cosine out (each 3dB weaker than the input sine). The question "what is happening from 180 to 360" therefore makes no sense at all, the outputs are both complete, full waveforms.
Doah, yes of course, thanks makes sense... thanks
The hybrids are based on broadside coupled quarterwave lines.
A single stage coupler is in fact quite broad band: A balanced amplifier covering 1.8-2.7 GHz using such is quite possible.
Regarding class-AB: this is a "pre-biased class-B compromize" to give softer turn-on characteristic and better linearity than class-B, and as such, it only conducts half periods.
In the low frequency world I guess that implies you need to operate in push-pull, in order to faithfully reproduce the signal, but for RF, the output waveforms contain a fundamental component that is filtered out from the output network characteristics: Linearity is of course not perfect, but that's why you need predistortion ;-)
(The coupler on the output is most likely feedback path to an adaptive predistorter.)
Finally, the whole point for Doherty is to increase efficiency for high dynamic amplitude modulated signals 3G, 4G, 5G.
Thanks for the amazing and clear explanation. My question is that for the Doherty amplifier what would be the maximum power that can the class C add to the output power (I assume it's 3dB)? My second question is that the Doherty amplifier usually has bandwidth limitation so what kind of methodology can be used to increase the bandwidth in this case? My third question is that if you use a common base rather than a common emitter for the Doherty amplifier what would be the main advantage or disadvantage in this regard (in terms of power and phase noise)?
Nice to see you back :-)
Nice to see you are still alive!=)
-How were your grades in school/university? (Just curios if there is any connection between real knowledge and grades).
-Do you think the simplified Arduino platform is generally a good or bad thing for students/young engineers?
I am working on my master thesis right now and the setup I am using is based on Arduino and was build by the guy who finished his Ph.D. few months ago. Everything this guy done with electronics is just fundamentally wrong. He didn't even try to read datasheets for the component he was using. For example he uses LT1210 as voltage follower. His circuit doesn't have any bypass caps (You can't imagine how noisy this LT1210 was!) There is no compensation for output offset voltage (due to current through feedback resistor) . The relays he is using are too slow, not reliable for our application. The wiring is messy and on some places just wrong...and so far and so on. The main problem is,that this guy stated:"Everything is tested and working fine"
And this is not the first time I see the case, where the Arduino platfrom with all premade libraries and shields gives someone a false feeling of having experience/knowledge in real engineering and electronics.
-Do you have any ambitious hobby projects you are working on?
-Do you rewatch your own videos? When I try to watch/listen my tutorial videos, it is always so cringy I immediately switch the sound off:D
-Which other engineering related youtube channels do you watch?
Regarding your first question, I suggest you should read up on the effect known as "survivor bias". Veritasium has a reasonable summary. You are setting yourself up to reach some very invalid conclusions by canvassing successful ("survivor") engineer's grades; that's completely backwards stats. Also, it takes all sorts; there are people who have no passion and no studying ability who are bad engineers and get bad grades; passionate, intuitive engineers who breeze through university exams and are good real-world engineers; study-hards who learn exactly how to pass poorly-written exams but suck at real-world engineering; and finally savants who have the potential to be amazing real-world engineers, but find the exams so boring that they don't actually get particularly great grades. So yeah, every permutation is covered, not to mention all the shades of grey inbetween. Having said all that, it's a very interesting question; just saying it's not a "useful" question per se.
Also, I know you weren't asking me; but I think Arduino is a mixed bag -- good to encourage a wider audience to be interested in electronics; but the way that their documentation hides that the Arduino board is just an ATmega328P breakout board is really anti-educational.
great explanation
Nice to see you back Shahriar! :)
Return of the cool cat :D
Vid is great as always, looking forward to more.
Looks like an Ericsson PCB. Not sure about the antenna element though. Andrew? At Powerwave we used dual stacked patches for multiband elements.
soooo glad to see a new video! :)
Thanks for this tutorial.
I would really love someone to do a video on general ideas of substitution. I only have limited time in my schedule for learning this. and all of it is split between learning, experimenting and my garden. I would love to know a general way to understand what I can substitute certain components with when I do not have that exact resistance/capacitance etc.. (470 pf is one I could not find very much information on)
typical output power in dBm, for wcdma, also referred to as effective radiated power, ERP, is in the order of 10 - 15 dBm for dense metro propagation model.
10-15 dBm is 10-30 mW, so are you talking about a some distance/angle from the antenna? Even the mobiles can do 1-2W.
There is a thing called the link budget which accounts for the gains and losses on an rf system, wcdma relies heavily on power control, so i meant to say even though these RRHs transmit with 20, or 40 watts, the user equipment only receives, because of fading and other factors, around -35 dBm on the port of the antenna, which is plenty. These propagation models are based on the isotropic dipole, so the real figure is in reality, higher. And user equipments can not go any further than 600 mW, in the older days of AMPS yes, but not anymore.
Yes, I know about link budgets but you mentioned ERP :) And class 1 wcdma mobile peak power is +33 dBm = 2W. That's just in the specs so I don't know if they are actually made.
no, for sure. Remember Ec/no is very sensitive to BTS set output power, and BTS always tells the phone to "tone it down". 600 mW worst case scenario, and even that much is all the way down to 1x EVDO, for example, there is also link budget for BTS and link budget for UE. power balance is a big thing in wcdma! dont think in absolutes when translating dBm to Watts!
Getting here almost 4 years late, but maybe that diplexer video has been made by now :)
Great to have you back!!! I'm assuming there must be a significant efficiency differential between a pair of LDMOS amplifiers in the first configuration in comparison to the Doherty architecture, enough to warrant the hardware cost of adding the passive isolator? Can you give us a ballpark comparison?
Both PA's surely have a passive isolator, to protect it from "revererse IM": intermodulation arising from power leaking in to the PA output from nearby transmitters, and also to assure good matching in case of poor VSWR from ice on antennas, as well as hardware protection, in case it is powered up with no output termination connected.
Isolators of this size typically have about 0.2 dB loss.
Great to see you back! I do also have a question for you: I just graduated as an electronics engineer and have a decent background knowledge in both electronics and mathematics. I must admit however that only have a very limited understanding about RF, impedance matching is almost as far as it goes. I would love to learn more about how to do RF designs and how to get started. I'm thinking to buy a descent RF spectrum analyzer with tracking generator in order to do some experiments and get some hands on experience. Do you have any recommendations for me? Topics to google, tests to do,... any suggestions are welcome! Your video are a wealth of knowledge by the way and got me really interested into the field of RF!
You can do a lot with a spec with integrated TG. However, there is no phase information so I would say that the first thing to get is a VNA, vector network analyzer. They're unfortunately not cheap and you also need an expensive calibration kit. For real understanding of the whole process I would recommend you get a simulation software, such as Keysight ADS och Ansoft HFSS, and design a distributed filter. Then manufacture and measure this board with a VNA. That's an excellent learning experience!
Happy to see new vid from you! Hope you get moving sorted out soon. It also does help to optimize inventory ;)
As for question, what materials/articles would you recommend for a person getting VNA (HP 8753A)? I got one for upcoming repair project, but have no idea atm how to test it or even check if it works. What minimum gear would I need to check it's operation?
Great video as usual, I'm glad that the house with the new lab is almost finished, I'm looking forward to the upcoming videos.
As a question: Would it be possible to have some high-resolution photo's available from the boards you analyse?
Great to see another video!Question: Instead of using the 20dB coupler to monitor the output, would it be possible to take the output from the second quad hybrid coupler output (the one after the PAs?) to monitor the output?
If you did that you would be measuring the reflected power, and while there is a relation to the output power it is not a constant relation but it will depend on the antenna, cable length and such.
Do you have any advice on how to advance in the workplace? I've contributed a lot to my company. I've re-designed all of the PCBs and their reliability, cost, and functionality have increased tremendously. Also, there have been several designs that I've developed independently from scratch that have exceeded their requirements.
I love the hair, by the way. It's great to see you again.
thank you so much !
About the Q and A, do a lab tour!! Show 360º of your lab, not a direct Q/A, but, will be interesting!
Hey man, nice to see you back and congratulations on the new place. Question: How long have you been using Matlab is it any good. last one. How old is your Cat ?
Interesting ! as always.
i have an cellphone-tower tranceiver module with the exact same data-power connectors as the first board.
never seen before.
Made by Ericsson as well ?
excellent !!
Love this video. It's just brilliant!
Thanks for sharing always very informativ even though I do not do HF electronic
Very interesting design. On a semi-related note, do you have any advice for picking up equipment (specifically RF like spectrum analyzer or VNA)? Is it mainly persistent hunting on ebay?
For the Q&A: Which Bell Labs location are you employed?
So I arleasy ask you on twitter, but how you would transfer a lab like yours? What is the step to package the instrument in a good way?
Second some day you can do a video about PLL and Frac/N divider (maybye a design of those would be nice)?
Last would be good to some day, if possibile, do a videoblog from the bell labs!
brilliant explanation !
q for Q&A: What are some typical precautions you take while testing / fixing / debugging RF devices so that you limit the rf exposure of your body to safe levels? At which point do you get concerned when dealing with rf devices? Can you recommend any good literature on this please?
JE, Ottawa
Hooray! Always look forward to these videos. Does the isolator have any relation to a magnetron? If I understand correctly, both use magnetic fields to guide signals.
Or a travelling wave tube.
Thanks 🙏👍💯😊
Very Cool....
Good video. Any of these questions are available for you to consider. Thanks.
These boards do not use all standard components, opting for microstrip line and specific dielectric boards. How are these circuits designed and laid out and why use these complicated filters, capacitors, and inductors, versus lump components? Is it only cost or also performance because sometimes they mix and match? Do they use Eagle, Altium, Kicad, etc to place these stripline components into the board? Where are they designed; like a simulator of some sort? I know they use specific boards and mills to produce these boards. What about layers, are they mostly 2 layer or do they get created with the same capability as common printed circuit boards?
I am interested in designing analog circuits and RF circuits as I am too comfortable with digital domain. I am reading RF Circuit design by Chris Bowick as a resource to learn more about RF design. Not much that I have not witnessed before in coursework or the internet, but provides lots of basic worked out examples.
I found a video on a tool called genesys. Explains quite a bit of my questions.
community.keysight.com/community/keysight-blogs/eesof-eda/blog/2016/09/05/how-to-design-custom-rf-microwave-and-analog-filters?sf35244374=1
Friend, how are you? I work in the electronic part of telecommunications... I have a question. I have a failure in an ericsson 01b5 rru that does not raise the PA in the cywin manager. I get dependency failed, it does not transmit or resive.. Could you please help me?
How are the the Anaran parts made? Stripline?
nice !
but why they need more power than typical few watts of phone ?
i mean what's the point in being able to transmit without hearing response?
i guess it's because powers of individual transmissions to individual phones somehow adds up all together cause they all use CDMA bitstreams with different sequences and one phone can only transmit one sequence at one time.
can you go into more details about it?
also it would be nice to hear more about whole 3G/4G modulations stuff !
*****
it means they need even less power !
with better antenna ;)
Those are TRANSMITTING amplifiers in the base station, which provides the output power for ALL users in a cell-sector simultaneously
On the receiver end a base station have low power LNA's as expected.
Hi, Shahriar. thanks for the great video! I have a question about both those amplifiers. There is only one transistor in each path, means there are two options of offsetting them. Option one is to have a B or AB class which means they only work on one half-period of the signal. If this is the case, then how is the rest of the signal on the output is reconstructed? Option two would be to offset the transistors into class A which means burning 50% of power on DC. That would work for the first amplifier you explained, but no way for this to work with the Doherty amplifier: handling both half-periods on the Class C path would require two transistors, and the saturating with one transistor path would be just Class A, not AB.
Maybe that is why there are 2 amplifiers on the Doherty amplifier board?
The parts of the Doherty amplifier work on different parts of the single half-period: main part and peak. If each part of the Doherty amplifier is built on a single transistor, then they will be only able to serve one half-period of the signal. At 24:00 Shahriar shows two graphs for Class AB and class C. My point is that it is not possible to make a single-transistor Class C amplifier that would serve positive and negative swings of the signal as shown on his graph. Also regarding his Class AB signal graph: with single-transistor it can only be achieved with Class A biasing, i.e. burning ~50% DC power at idle.
My suspicion is that the Doherty amp in this case works on a half-period, creating a distorted sinusoid on output, which is then filtered to get rid of unneeded harmonics.
Actually, the way of getting high efficiency in an amplifier is to minimize the voltage-current overlaps inside the transistors (and to compromize linearity):
For class B and AB the current is half-clipped sinusoid or even better: square like (class-F etc.)
The fundamental output power is then reconstructed through bandwidth-limiting output.
good morning friend, I saw a video on your channel that you use the cywing moshell software for rf radios like erisson... I work in a telecommunications company and I repair those boards electronically... I have a fault in an ericsson rru 01b5 that does not raise the pa on The manager gives me (dependency failed) and it does not give tx or rx....please could you help me please since my work depends on that
How do you calculate the power to be dissipated in those 50 ohm resistors that terminate the input and output quadrature combiners?
can those ldmos be used on other project or cellular application only
what is load modulation concept?
Q. Can the Doherty architecture be made to work with a quadratic hybrid on the output rather than a delay line (phase shifter). I am working on a Nokia PA which appears to do this??
Yes, it can.
how do you desolder those ldmos
What crystal do you use?
Hi, I wonder if you had worked with metalbacked PCBs. Do you think that it's a good optio to reach better grounding an heat sinking? Thanks!
Some paths seem longer than necessary. Wouldn't it work if it was just one tight line of components?
The longer paths are phasing and impedance matching traces, you can think of 1/4 wavelength transmission lines and such.
If it was a balanced amplifier that is true, se my reply above ;-)
Thanks for the video :)
Hi, just wondering how they get away without a low pass filter at the output? Or they just don't care about harmonics?
My guess is that it is somewhere between the shown modules and the antenna, if you use the same antenna for transmit and receive you will need a duplex filter anyway and that will take care of the harmonics.
The output of the isolator is connected to a large cavity filter (like the duplexer shown in the beginning).
hey i learned new things, thanks!!!
Highest signal to noise ratio videos on YT.
great vid!