Design of GaN Power Amplifiers (Part 1) by Dr. Edna Hickey

hello and welcome to design of gand power amplifiers part one with dr. Edna Hickey I&#39;m Mike Hamilton your host for this I Triple E microwave theory and technique society webcast which is sponsored by the MTTs Education Committee before we start I&#39;ll mention a few housekeeping items first this presentation will be archived recording should be posted approximately 24 hours after we finished the presentation we&#39;ll send all registrants and email when the archived webinar goes up so you can revisit it or share it with your colleagues second we encourage questions we&#39;ll answer them after the talk but you can submit them at any time during the discussion into your question in the Q&amp;A box in the webcast window and don&#39;t forget to click Submit third some words about the interface you can enlarge slides by clicking on the rectangle at the top right of the live slide window you can also enter full screen mode if you desire refresh or reload the current web page if you encounter any problems with regards to audio if you&#39;re listening over your computer speakers you can adjust the media player volume you may also need to adjust your system&#39;s master volume the portion of the presentation today will be open in the media player so please make sure that that window is also visible also the icons at the bottom of the webinar window include a resource list clicking that link will start the process to download copies of the slides to be presented today now let&#39;s introduce our speaker dr. Edna Hankey has pioneered the development of state of the art RF microwave and millimeter wave components at Westinghouse and Northrop Grumman for 34 years circuits includes low-noise amplifiers low noise low noise oscillators mixers power amplifiers space shifters attenuators limiters frequency multipliers low phase noise millimeter wave fiber optical links and miniature integrated assemblies and subsystems he previously worked in cryogenic electronics research at Martin Marietta he now consultant lectures on nonlinear and linear and wireless transmitter receiver circuits and systems since 1983 he has lectured over 3,000 professionals throughout the world for Besser Assoc in the continuing education of Europe yields nine patents one George Westinghouse Innovation Award and has authored and presented numerous papers on RF microwave and millimeter wave sir so now is my pleasure to turn the virtual podium over to dr. Edna Hankey for part one of a two-part webinar series on gallium nitride power amplifier design it thank you for the wonderful introduction I really appreciate that and it&#39;s really a pleasure to present the design of gallium nitride power amplifiers part one what we&#39;ll do is we&#39;ll discuss the the gallium nitride power amplifier design and first we&#39;ll have a little introduction to power amplifiers then what&#39;s why is the gallium nitride transistor so important for the for the power amplifier will talk about the material properties now since since we&#39;re talking about very high power there&#39;s going to be some power dissipated in the transistor so we want to make sure that we get the heat out of the transistor properly so that the junction temperature is low now for reliability is very there&#39;s very important characteristics that we have to really be be accustomed to for reliability so we&#39;ll be discussing many many factors for high reliability operation now the next question is well where can we where can we can we get these transistors where can we where can we work with these transistors we can either work with it with an actual actual transistor or we can we can do a mimic design or we can actually use the use of transistor with it with the design included in the mimics in the mimic circuits and we can work with the foundries where we can actually do the various designs using their design rules and then we&#39;ll talk about the guidelines for reliable operation now biasing of the transistor is critical when we have to do it with a certain procedure for hybrid liability then what we&#39;ll do is we&#39;ll actually do a step by step design example for a Class A operation and we&#39;ll talk about all the very important steps though that are required to have a good have a good design with high efficiency high reliability and low cost now let&#39;s see these are some of the very important parameters that we would like to have for our power amplifier we like to have high power efficiency and reliability and we like to have good frequency range and with what&#39;s up with high power we like to have circuitry that are not very complex easy matching circuits would have very wide bandwidth and that&#39;s one of the one of the very important features that we have for the gallium nitride transistor and also there are a lot of new systems that are using amplitude modulation different types of amplitude modulation like 64 QAM QPSK where the amplitude is bearing so we need to operate the transistor in a special way for linear and we like to have the transistor for act linearly linearly what that means we like to gain not to change with input power we want that we don&#39;t want the phase changing and that&#39;s what&#39;s important for for for linear operation of course we like to have we like to have low cost now let&#39;s take a look at some of the some of the gallium nitride transistors well we can actually look right here at a chip this is actually a chip where we have our input our input gate our output drain and the source which goes to ground now right here these would be connected with bond wires and for high frequency operation generally these are actually connected to the back of the chip for low inductance because if you have too much inductance you&#39;re going to reduce the gain again on the circuit now you can see these little gate fingers in here these are the gate fingers in here and they have to be a certain length compared to the to the to the wavelength of the frequency of operation because if they&#39;re too long the change is just not going to work some currents are going to be going one way and some currents are going to gone the other way so we have to we have to tailor the electrical lengths of these little fingers here considering the frequency of operation now here we see of the transistor here&#39;s another transistor down here which is very high power you can see there&#39;s a lot of a lot of gates a lot of gates here and up so this is for higher power application now here we see here we see these all these transistors this is a gallium nitride monolithic integrated circuit here and you can see oh here&#39;s your transistors here all the matching circuits that require as you can see a lot of times what&#39;s done is just a little length of transmission line lengths are used to match the circuit and here&#39;s our here&#39;s our input matching circuit our input matching circuit here so this is an example of a gallium nitride monolithic integrated circuit here and we can see now here the transistors are actually put into a package you can see the lids taken off here here&#39;s the input and here&#39;s the output and then these are placed with a screw the screw the circuit down here and you can actually here&#39;s here&#39;s packages with the lid with the lid connected and these are ceramic packages and then these are low costs low cost packages right down here so this is some of the some of the transistors that we can see here now now let us examine what the world it&#39;s very important for for our transistor operation well what we need to do is we need to look at what&#39;s what&#39;s required for all the RF output power well we can see up here the RF output power is a function of the d-max which would be typically twice the power typically twice the power supply voltage minus V Men that&#39;s niebolt each we like that to be small compared to v-max okay then we have to multiply that by the maximum current external current and we have to divide that by eight so that&#39;s that&#39;s the RF power so you can calculate how much power you can get out with transistors just by looking at the datasheet by looking at these various various characteristics now here we have an array of different of different semiconductors here first we had our silicon we had a wonderful silicon bipolar transistor and our LT MOS transistors here and then in the 80s we have our gallium arsenide our Colin gallium arsenide becomes very prevalent here and now we have our gallium nitride transistors here so let&#39;s take a look now this the gallium nitride can either be placed on silicon or silicon carbide that&#39;s a typical substrates that the gallium nitride is manufactured on so let&#39;s take first of all let&#39;s look at the bandgap why is it being gap important the game the bandgap determines how well the substrate works at high temperatures we don&#39;t we want the substrate to be semi insulating we don&#39;t we don&#39;t want a lot of current going from the top to the bottom of the substrate and if we have a high band gap that means it&#39;s going to work very well at high temperatures early early in the 70s they tried to make an integrated circuits out of silicon and for military applications and they found that when they run things up of 70 and 89 degrees centigrade temperature the the transistor didn&#39;t work very well but the gallium arsenide works a lot better for gallium arsenide integrated circuits here so you can see this has a very high positive thing for the high temperature operation now let&#39;s talk about the breakdown the critical breakdown field well if we have a high breakdown field we&#39;re going to have high voltage operation that&#39;s our v-max that 30 B max okay and then the thermal conductivity well gallium nitride is very similar to silicon okay but now when gallium arsenide is put the gallium nitride is put on the on the silicon carbide then it has a very high thermal conductivity so that so the temperatures want to stay nice and cool now we have our saturated velocity that&#39;s our max behold that&#39;s RI Max and so we have all the important characteristics for high power high V max hi max and since we&#39;re working at a very high voltage here typically twice the power Seibel the V Max is very small compared to I max so we&#39;re going to get we&#39;re going to get up we&#39;re going to get good voltage swing across the transistor here now the disadvantage of the gallium nitride transistor is the mobility the mobility is not nearly as high as the gallium arsenide but it&#39;s a little trick we can use we can use a high electron mobility transistor structure which boosts which boost the mobility of the transistor to allow it to work with higher gain and higher frequency of operation now there&#39;s two types of silicon carbide that we can use we can use the 4 H or the 6 H now the 6 H so it&#39;s used for light emitting diodes it&#39;s just one contact and it goes to ground well that would not be very appropriate for our for our mimic circuit here because we want the substrate to be insulates I mean insulated so we want to use the poor eh it has a better crystalline structure match to gallium nitrite and it&#39;s a lot easier to make so that&#39;s that&#39;s how the transistors are made they&#39;re made with the port the the silicon carbide is always a foraged 4-h characteristic that we have now well what about using the gallium nitride transistor for a low noise operation well here we can see with this particular device this is a Corvo device point one eight micron micrometer gate line the short of the gate line the lower the noise the higher the gain and we can see that at expan frequency has a noise figure way under DP with a with loss of gain so this works very nicely this works very nicely now let&#39;s talk about how we can use a low noise amplifier generally there&#39;s going to be a transmitter associated with it with it with it with the system and the leakage power from the trans when the transmitter is on the low noise amplifier generally is not working and we don&#39;t want the transmitter leakage power to to burn out the low noise amplifier so the usually is a limiter place between the transmitter leakage power and and the low noise amplifier also there&#39;s another consideration that the low noise amplifier would have to work with extraneous signals that are not desired it could be that could exist and it could exist so the low noise amplifier canopic has to be operating in the presence of these other things now recovery time is critical in many applications for the low noise amplifier when the transmitter turns off we want to ln8 the work when this when it&#39;s being hit by high power from extraneous sources when the high power is off we want the low noise amplifier to work so there is a problem if we have if we have a if we have a gallium nitride transistor substrate mature and material that has a lot of dislocations or traps what happens is the electrons flows into the trap and it stays there for a long time and then it comes out so the recovery time can be quite long for a low noise amplifier getting back to its normal state if there&#39;s a lot of dislocations or traps in the material and it could be in the order of milliseconds but for for good quality gallium nitride substrates the recovery time is in the order of 5 nanoseconds and the other advantage of this device here will have very high input ip3 so we can see that our gallium nitride trans our gallium nitride transistor meets many of the up many of the objectives we have high power good efficiency and high reliability now let&#39;s talk about what was important for reliability I remember when I was working at Westinghouse and we had a we had a spike linkage come in that was one nanosecond wide and it was 1 watt of power and it blew out our blew out our Ellen A&#39;s so you can have you can have very short pulses with with with high power that can blow out the transistor so you have to make sure that that even for short pulses you can exceed the breakdown voltage now another thing is a lot of times somebody may forget to hook up the antenna to the to the system so the the the Luna the the power amplifier is not going to see a match load and it could see a high voltage standing wave ratio so you get twice the voltage that appear across the transistor so you have to take care of that and a lot of manufacturers will state that the the power amplifier has to work with the short circuiting phases it has to be able to survive it doesn&#39;t have to work properly it just has to work and can&#39;t can&#39;t it has to be able to work with that so that&#39;s very important now the other thing is the junction temperature we have to keep that below the recommended value and another thing that&#39;s critical is the gate current for the transistor if we have high power coming in to the if the power is increased the gate current can increase and what we have is what&#39;s called metal migration whenever you have too much current running through us through a through a conductor you can get metal migration what happens is to carry the conductor carriers move around so that there&#39;s a space where it looks like an open and then there&#39;s spaces where the look the current density is very high through that particular part of the part of the channel so eventually it burns it burns up and gate opens up this doesn&#39;t not happen right away I hadn&#39;t we had examples where we&#39;re where our power amplifier was being hit by too much input input power and and we took the transistor support and sure enough the gates were open so we had to reduce the input power and I&#39;ll show in an example here we can actually put a gate resistance in the DC bias apply in the gate and that sort of tremendously reduces the gate current for high reliability and also we have to be able to turn on internal the power supply with the right sequence so we&#39;ll be talking about that later so those are the key factors for reliability now here here we have our silicon carbide our transistor mountain on silicon carbide and you&#39;ll see that there solder there solder here now there&#39;s two ways to connect chips we can end use epoxy or we can use solder and it&#39;s generally recommended for higher power applications to use solder as opposed to as opposed to the epoxy you can use epoxy for like low-noise amplifiers or or for amplifiers don&#39;t have a lot of power and now we have to put this on it we have to solder this to some sort of a base base plate so the two characteristics that are important is we want have high thermal conductivity for the base plate and also we want to make sure that the temperature variation of the elongation is similar to that of the of the of the of the silicon carbide because if they&#39;re a lot different they can actually pop off I remember when we first were working with our gallium arsenide chips and we soldered them to two aluminum and we temperature cycle then the chips the chips came off because of the different that the large difference in them between the two materials so now if we take a look at the various if we take a look at the various materials that we have generally copper malik all copy molar copper copper is generally used for the foot for the for the circuit that the transistor is soldered to and you can see that it has a pretty good thermal conductivity now the thermal expansion is not exactly perfectly matched to that but it works it seems to work fairly well now there&#39;s a new material aluminum diamond which has a tremendous high thermal conductivity and it&#39;s also better matched to the transistor so people are starting to use this and you can ask people actually make these up make the make these carriers we can actually map mount your mount your transistor on onto these with screws so so here you see your aluminum diamond now when you design the mimic circuit you want to make sure that you don&#39;t put all the transistors too close together like for example in this particular in this particular example here because what can happen is the temperature gets much too high so the layout has to be done differently to make sure that your transistors aren&#39;t too close together so that they don&#39;t interact with on another increase of temperature and this provides a 20% reduction in the temperature compared to this circuit right here which is pretty good now the next question is well okay I want to design a gallium nitride circuit so where do I go so these are these are some examples of US companies that that provide the provide the transistors there are there are some over in China and there so there&#39;s a Germany has some good has some good boundaries also so here&#39;s the the wolf full-speed which is the which is cray and they have a quarter micron and 0.4 micron and where they where they got this miss this lead was with light emitting diodes late emitting diodes and light bulbs so they have a lot of a lot of processing capability in that in that particular area now try Quinn has been involved an RF MD non Corvo bought them and we have we have the higher shorter gate lengths 0.15 and 0.25 micron okay now Sumitomo is very very popular for very high power at lower frequencies and they can really put out a lot of a lot of power now make on they bought nice next which is a transistor that&#39;s put on silicon as opposed to the silicon carbide and we have Northrop Grumman and hrl for the for the higher frequencies now here are some of the foundries that we&#39;re talking about here here&#39;s the hrl they&#39;re up north of Brahman and Raytheon Raytheon has a wonderful process 490 gigahertz operation they can put out a lot of power with really good efficiency so they have a very high they have a very good very good process Raytheon and here&#39;s our triQuint which is very very high quality now Dae also has has a process that works very nicely at 30 gigahertz 46% power added efficiency so these are some of the places that you can go to if you want to do if you want to do foundry work and also there are places in Germany and also places over in China now let&#39;s see here well let&#39;s talk about the gallium nitrite structure so here we have our t gate the t gate is used to have a short gate length here and then have more current capacity up in this area here now the first thing is we have a Schottky barrier which is a metal to semiconductor so we have metal to semiconductor that forms our Schottky barrier circuit here and this is aluminum gallium nitrite that&#39;s that&#39;s going to be our Schottky barrier circuit now that&#39;s put on top of gallium nitride gallium nitride now when you go through the physics of the circuitry what you find out that the electrons like to flow in this gallium into the circuit right into the gallium nitrite so in this blue area right here that&#39;s where the electrons flow so why is that important well gee electrons can flow a lot faster through a semiconductor that doesn&#39;t have dopants in it compared to a semiconductor with dopants so that&#39;s the idea you get very high very high electron flow which gives you a very higher higher mobility out of the circuit here and then here&#39;s your buffer and here&#39;s your circuit here&#39;s your circuitry right here so that&#39;s the example of a high electron mobility transistor now here we see a typical a typical structure right here and this shows the depletion region that&#39;s the area where there&#39;s no electrons available for conduction and if you look at the electric field the electric field is very high in this area very high in this area which means the voltage of the breakdown voltage is good it&#39;s not going to be that that&#39;s super but now by putting this field plate one here putting this field plate on top look how it spreads out the depletion region and it reduces see it reduces the electric field which is great this is going to this is going to provide higher breakdown voltage what we want for the transistor here and also it&#39;s less dispersion in other words there&#39;s less electrons that are put into these traps here the imperfections of the semiconductor here so there&#39;s less dispersion with that so we won&#39;t have the the disadvantages of the traps associated with the circuit we can also see that with the IV curves for the transistor we can see this is without the field plate and this is what the field play so if you have to work something with this thing if you&#39;re lute you&#39;re not going to be able you&#39;re not going to be able to get the high max instead of going up to here you&#39;re only going to be gone off to here so you&#39;re not going to get as much output power ll of the circuit here now this characteristic here is for for a material that&#39;s not that&#39;s super all all materials are not like this this just just shows you the this just shows the extreme when you have a lot of a lot of traps associated with the would be with the semiconductor now here&#39;s the reliability curves we have so for example if we have a tens 10 to the six mean time to failure requirement then we can operate this about two hundred and sixty sixty degrees centigrade okay so that&#39;s pretty pretty good reliability now here&#39;s these here&#39;s the gallium nitride on the silicon and we can see they have the different layers they have the different layers here you can see we have a transition layer because the silicon and the gallium nitrite don&#39;t have the same crystalline structure so you have to have a transition layer to to reduce the stress to reduce the stress between the two layers and here we can see here&#39;s our here&#39;s our aluminum gallium nitride and here&#39;s our electron gas right right in here in the gallium nitride circuitry right here now may calm has improved the reliability of these circuits the reliability than it used to be as good as the as the other services as this silicon carbide but they&#39;ve improved the process now and they can work with a we can work off the 225 degrees now with with pretty good high reliability now here&#39;s here&#39;s an example of a circuit that works up at 90 gigahertz and these you have to do you have to the circuitry has to be done a little bit differently when you have these very short gate lines to make sure that you have proper operation without with the higher frequencies with this very short very short gate lamp so what they do is they have they have this layer right here this layer right here helps to confine and the electrons to run in this channel right here okay and then these layers here are used to help make the circuit work with the very short gate length to suppress a whole electron diffusion and improves the electron confinement in the gallium nitride channel so here&#39;s an example of the latest of hrl 40 nanometer device here let&#39;s see how it works here we can see here&#39;s our actually here&#39;s our actual transistor right here and let&#39;s see how it works here&#39;s our here&#39;s our here&#39;s our power added efficiency of the circuit here it&#39;s 40% which is pretty good and we have about 9 DB of gain and our output power is around 27 27 or so DB DB M so this really has pretty good pretty good performance typically power amplifiers working at these higher frequencies do not have very good efficiency okay this is a first circuit I&#39;ve seen and has pretty good efficiency that working at this at these higher frequencies this is a new development for them now here&#39;s the bae circuit here this works without a field plate they have processed this to work very nicely without a field plate and you can see this is a load pull circuit that we have for the circuit here and though and behold we get 53 percent drain efficiency with 45 percent power added efficiency with 4.9 watts per millimeter at around 29 gigahertz so this works this works very nicely so the government supported the development of this of this process for BAE and here&#39;s the Fujitsu this said this has very high very high power 3.6 watts per millimeter all the way up it at 90 gigahertz or so and you can see they have double layer silicon nitride passivation to reduce the traps and here&#39;s the here&#39;s the here&#39;s the gallium nitride channel and then this is the other other layers that we have here now the disadvantage of this circuitry the efficiency is not that not not that fantastic it&#39;s only twelve point three percent in eighty eighty six gigahertz Raytheon has a circuit that has much higher much higher efficiency at these frequencies now the next thing we want to examine is the turn on and turn off of the of which of the transistor first we want to a lot of times with the power supply when you first turn the transistor on and off it will it will have spikes in it so it&#39;s a good idea not to hook up the transistor to this to the power supply first turn the power supply on turn the voltages down and then hook them up the first thing you want to do is you want to now increase the gate voltage to the pinch off voltage then you want to turn the drain voltage up to the required value and then you want to adjust the gate voltage to your to the desired current now to turn it off you want to just do the opposite then you want to reduce so you want to reduce the gate voltage to want to reduce the gate voltage to the pinch off voltage then you want to turn the drain voltage off then you want to turn the gate off now a lot of a lot of regulators will actually have these built in to turn the transistor on and off properly now let&#39;s take a look at operating the transistor classe ah what we&#39;re going to do is we&#39;re going to have a we&#39;re going to have a current the current generator here that&#39;s going to generate the current through the transistor here and we want to adjust the voltage we want to adjust the load resistance so with this given current it fills up the voltage voltage across the transistor so the load voltage is just the current swing over the voltage swing and then our our RF power is going to be the RMS voltage on the RMS current so here we see it&#39;s a voltage swing times the current swing over eight and the DC power this is where you actually have to add the V minimum and V the maximum the minimum and AI minimum to the circus so the efficiency is 50% now the advantage of the gallium nitride transistor is this v-max can be very high compared to the best values typically you&#39;re going to get your 50% efficiency out of the circuitry here so what we&#39;re going to do now is we&#39;re going to now look at look at the design of a design of the circuit of the circuit here this is 30 watt 30 watt transistor and we want to now go through the complete design of the circuit so what I&#39;m going to do is I&#39;m going to actually look at the look at the circuitry here and we&#39;ll go through the various steps involved in the design of the circuit now the first thing we want to do is we want to look at the stability of the circuit and this is derived from the small signal s parameters and the stability doctor wants to be greater than one so we can see at our operating frequency we&#39;re in trouble because less than one so it has the possibility of oscillating and also around 13 and 1/2 gigahertz and that&#39;s a possibility of oscillating so it can actually look the stability circles now since we&#39;re interested in since we&#39;re interested in all we have to add be sisters to the circus so we want to add them on the front of the circuit so we&#39;re going to now this is the output stability plane so what that means is if we put any impedance here the input impedance will have a reflection coefficient larger than one and it&#39;s going to be possible to oscillate now on the input stability planes if we put a resistance in here then the output reflection coefficient will be greater than one so since we&#39;re going to add loss to the circuit we want to put it on the front not on the output so what we&#39;ll do is we&#39;ll turn this off and then we&#39;ll look here so we want to add a resistor here if we take point O four times 50 it&#39;s about two ohms so now if we put two ohms in the circuit here you can actually move that right out and then if we look at our stability here we can see that we have improved that so I&#39;m going to add about four I&#39;m going to increase to five ohms for better stability now we have to worry about this circuit right here so now if we look at our stability circles we can see now we have to put something in parallel here so now let&#39;s now go to the circuit here and we&#39;re going to input put and put this resistor in the circuit and this is a quarter wavelength line at 12 and a half gigahertz open short and it puts the resistance in this circuit now let&#39;s now look at the stability here and we&#39;ll add our parallel resistor here and we&#39;ll put 50 ohms in here so now you can see the circuit is very stable and we&#39;re in pretty good shape now the next step the next step of the design is to is to look at our load line now we&#39;re going to look at our load line here and here is our load line here so this is the operating point here so what we have to do is we have to buy a transistor so it&#39;s Class A so we&#39;re now we&#39;re going to we&#39;re going to increase the volt you can change the gate voltage so we&#39;re about here for the maximum current then we have to increase our power now here the low brisas is seven ohms so it&#39;s not why it&#39;s not big enough we have to increase it until we get the full voltage swing and now when we put our input power in here or in pretty good shape now you can see we have a little bit of hysteresis that says the current and both these aren&#39;t exactly in phase okay now let&#39;s see how we&#39;re doing with our with our current and voltage so we can see we have a 5 volt 5 volt knee voltage up to 60 50 60 volts and our current small and we have our maximum current so we should be in pretty good shape for a good e pretty good performance out of the circle let&#39;s see how we&#39;re doing whoa we&#39;re getting 45 DBM of power which is 30 1.6 watts we&#39;re getting 51 percent efficiency and this is our 1 DB gain compression point and we can see the phase is varying about eight tenths of a degree so that&#39;s pretty good so the next step is to actually match the 50 we have this we have to look at our 15 ohm so now let&#39;s open up the project open up the project and that&#39;s going to be our output match so here&#39;s our circuitry here we have our 15 ohm which is what the transistor wants to see and we have a 50 ohm generator so we have to add we have to add circuitry to the circuit so now let&#39;s look at our matching circuit here so let&#39;s add some electrical length until the conductance is equal to 1 and then we&#39;ll add our shunt capacitance so now we&#39;ve got our input match so now we&#39;ve done a nice job with our input match now the next step is to match I&#39;m sorry that&#39;s the output match now we have to match the input circuit here so first we have to look and see what the impedance actually is so this is input match step 1 so this is the actual impedance here they&#39;re normalized resistance is point 1 to 3 now whenever we want to match this with the 2 step matching circuit here so what we&#39;ll do is we&#39;ll well first the intermedia impedance is going to be the square root of this times 1 which would be about 17 point 8 ohms so the first step now is to match is to match the next step now is to match this to 17.8 ohms so that&#39;s going to be our next match to circuitry here so now we&#39;re going to match that and also I match that so we now we have this match to 17.8 ohms then step three is going to be to match that 250 ohms so now we have this now we&#39;re going to match this 250 ohms so that&#39;s going to be the next step here it&#39;s going to be the next step to match that 250 ohms so our electrical length will add that and now we have the complete circuit matched so now let&#39;s look at the final results for the complete circuit so here&#39;s the here&#39;s the complete circuit now we have our 45 DBM of power our 51% power added efficiency so you can see the performance is practically identical to what we had when we had when we had individual circuits associated with that and now the next thing I want to look at is the temperature rise for the transistor now here&#39;s the actual temperature rise it gets pretty hot when you&#39;re not using it and then when we actually have the operation it&#39;s running cool so that&#39;s the one of the disadvantages for Class A operation when you&#39;re not using it you should turn the transistor off and now we can actually see here&#39;s the actual gate and if we ask this is for for higher output for higher output characteristics so if we add a resistor in here I guess this isn&#39;t working properly oh if we add the resistor in here it&#39;s going to here&#39;s here&#39;s without the resistor and then this is with the resistor so anyway we saw that the power amplifier worked pretty good very close to the characteristics that we had so we can actually put in a we can actually put a time constant in here to account for if we have short pulses with short duty cycles to see what happens so basically we did a pretty good design with our with our with our parent for fire design so in phase two and the next phase will actually be looking at Class B operations other classes of operation D F minus 1 and J modeling of the transistor and the Dority amplifier so thank you so much for your kind attention [Music] ok great thank you good there was a lot of lot of info there and ecologist everyone for for the issue with not being able to see the pointers on the on the slides I know that can be a challenging thing so we&#39;ve got time for a few questions here and before we start that though remember that you can still submit your questions through that QA panel so let&#39;s go ahead and get on to this so the first question here is on slide 21 reliability is extrapolated with only one one data point so how is that how is that justified or do you have any insight into that how do we get to this okay slide 21 yeah see here let me help oops here give me just one second here let me get right to it oh oh hold on one second there you go nope there you go yep any thoughts on it oh there&#39;s those points were just put on there to show that when you work at 225 degrees channel temperature that you have the MT to failure of that particular value that&#39;s not that that&#39;s not how it was generated that&#39;s just that&#39;s just goes to show you what the operation what the reliability would be at 225 so for example here at 300 if we drew that up here then that would be the reliability over there that&#39;s just to show that&#39;s just to show a particular point there&#39;s there some underline how the reliability was determined yeah okay great okay so here&#39;s a question about efficiency what happens to efficiency as you start to combine transistors to get more power how do you get the best efficiency yes did you then I didn&#39;t hear that go ahead okay so right so what happens to efficiency as you start to combine transistors to get more power how do you get the best efficiency okay well generally generally how you do how you get how you combine transistors is you take a group of them maybe four or five transistors and like say say you had twenty of them see you have to combine 20 units so you break them up into five units of 5:3 have four units of five now what you&#39;ll do is you&#39;ll combine the five units with another five units together and then you combine the other five units to the other five units together and that&#39;s them with the quarter wave length transformer then you combine the five and the five with the other five and the five together and then you have another matching circuit to get to your to your 50 ohms so that&#39;s typically how it&#39;s done it&#39;s called cluster matching cluster matching and it&#39;s very very low it&#39;s very low loss because you just have you just have the transmission lines in there though impedance transmission lines and it&#39;s free it&#39;s very it&#39;s a very nice way to do it broadband with low loss that&#39;s how that&#39;s how everybody doesn&#39;t combine the outputs sure okay here&#39;s another question could you comment on the major differences in terms of design methodology of gallium nitride hint versus gallium arsenide hbt or CMOS yes now the difference between the the difference between the gallium arsenide and the gallium nitride is that you have to use many more transistors with the gallium arsenide because generally with the gallium arsenide you can only bias them at 9 volts at the the high electron mobility transistors you can only operate them at 9 volts because a breakdown voltages or maybe only 15 volts or so so what that means is you have to use many more transistors to get the same output power now when you do that your input capacitance is going to be a lot higher when you have to use so many transistors so with the gallium arsenide circuitry you&#39;re going to have to use many more transistors and the input capacitance is higher which means you&#39;re not going to have as good bandwidth with the gallium nitride transistors you are able to work with a very high voltage very high voltage operation which means you won&#39;t have to have a lot of current and a lot of current means you don&#39;t have to use as many devices to get the output current so that&#39;s the major difference between between them okay let&#39;s see there&#39;s several questions here regarding traps and memory effects could you make some is I specifically for gain could you make some comments about that if by the way the memory facts share memory effects and effects of traps oh okay [Music] well I guess memory effects is when when you&#39;ve run the circuit one direction it works differently than if you go the other direction because it remembers what happens in the past and that creates a problem with digital modulation because they have to have especially memory effect cancellation effect with it with the linearization with the Alain techniques and also with it with the traps with the traps that that also creates the dispersion where where the where the it takes time it takes time for things to happen it takes time for things to happen when you change when you change things okay I have I&#39;ll see there&#39;s a part of the part of one of the questions and asks about low frequency limitations and in in regards to traps so is there a is there a low frequency limit for these does it cause low frequency noise oh the traps are the frequency noise well I whenever we talk about noise I always think about one over F noise one over F noise and I believe that with with silicon with silicon silicon has Braille oh one over F noise because the material is they have many years of processing the material so they have very little one over F noise so for example if you have an oscillator or if you have a low noise amplifier and you&#39;re looking close very close to the carrier that one over F noise is a predominant effect but now with the gallium arsenide with the gallium arsenide the one over F noise is not as good because of the traps and I&#39;m sure with the silicon with the gallium nitride we would see that also you would see the the trap effect and that that definitely affects the one over F noise characteristics so when you get close to the carrier the noise comes up and I remember when I designed my first oscillator at x-band I could either use gallium arsenide or or silicon and I picked silicon because it has lower one over F noise because the crystalline material is much much better for that characteristic and it worked very nicely very nicely and ty also remember traps when I had an oscillator and I was using I was using material and it worked fine but then but then when I had a material that had traps associated with it when I changed the voltage this the circuit in tune to the new frequency because it took a while because of the traps you know so I had to go to the manufacturer and get a better quality material and then we were okay so traps do affect the low frequency noise sure that&#39;s great okay all right well we are out of time there&#39;s still a number of questions here and the presenter will follow up with those offline and also oh I would love to answer your questions and there&#39;s an opportunity to do that with the part 2 of this as well which is surely in April of 2019 so everybody please be on the lookout for that so also as we said earlier this session will be archived on the Society website at MTT org and all the registrants will get an email reminder with the website address when that&#39;s available for attendees who would like to receive the pdh credits then please follow the link in the webcast view and use the code that&#39;s provided on the last slide of this presentation once again I&#39;d like to thank dr. Nikki for this excellent and informative presentation we look forward to part two of this webinar series in April of 2019 and special thanks to our audience today for for joining us we hope you found today&#39;s event valuable and that you&#39;ll return for future I Triple E microwave theory and technique society webcasts thank you and have a great day

hello and welcome to design of gand power amplifiers part one with dr. Edna Hickey I&amp;#39;m Mike Hamilton your host for this I Triple E microwave theory and technique society webcast which is sponsored by the MTTs Education Committee before we start I&amp;#39;ll mention a few housekeeping items first this presentation will be archived recording should be posted approximately 24 hours after we finished the presentation we&amp;#39;ll send all registrants and email when the archived webinar goes up so you can revisit it or share it with your colleagues second we encourage questions we&amp;#39;ll answer them after the talk but you can submit them at any time during the discussion into your question in the Q&amp;amp;A box in the webcast window and don&amp;#39;t forget to click Submit third some words about the interface you can enlarge slides by clicking on the rectangle at the top right of the live slide window you can also enter full screen mode if you desire refresh or reload the current web page if you encounter any problems with regards to audio if you&amp;#39;re listening over your computer speakers you can adjust the media player volume you may also need to adjust your system&amp;#39;s master volume the portion of the presentation today will be open in the media player so please make sure that that window is also visible also the icons at the bottom of the webinar window include a resource list clicking that link will start the process to download copies of the slides to be presented today now let&amp;#39;s introduce our speaker dr. Edna Hankey has pioneered the development of state of the art RF microwave and millimeter wave components at Westinghouse and Northrop Grumman for 34 years circuits includes low-noise amplifiers low noise low noise oscillators mixers power amplifiers space shifters attenuators limiters frequency multipliers low phase noise millimeter wave fiber optical links and miniature integrated assemblies and subsystems he previously worked in cryogenic electronics research at Martin Marietta he now consultant lectures on nonlinear and linear and wireless transmitter receiver circuits and systems since 1983 he has lectured over 3,000 professionals throughout the world for Besser Assoc in the continuing education of Europe yields nine patents one George Westinghouse Innovation Award and has authored and presented numerous papers on RF microwave and millimeter wave sir so now is my pleasure to turn the virtual podium over to dr. Edna Hankey for part one of a two-part webinar series on gallium nitride power amplifier design it thank you for the wonderful introduction I really appreciate that and it&amp;#39;s really a pleasure to present the design of gallium nitride power amplifiers part one what we&amp;#39;ll do is we&amp;#39;ll discuss the the gallium nitride power amplifier design and first we&amp;#39;ll have a little introduction to power amplifiers then what&amp;#39;s why is the gallium nitride transistor so important for the for the power amplifier will talk about the material properties now since since we&amp;#39;re talking about very high power there&amp;#39;s going to be some power dissipated in the transistor so we want to make sure that we get the heat out of the transistor properly so that the junction temperature is low now for reliability is very there&amp;#39;s very important characteristics that we have to really be be accustomed to for reliability so we&amp;#39;ll be discussing many many factors for high reliability operation now the next question is well where can we where can we can we get these transistors where can we where can we work with these transistors we can either work with it with an actual actual transistor or we can we can do a mimic design or we can actually use the use of transistor with it with the design included in the mimics in the mimic circuits and we can work with the foundries where we can actually do the various designs using their design rules and then we&amp;#39;ll talk about the guidelines for reliable operation now biasing of the transistor is critical when we have to do it with a certain procedure for hybrid liability then what we&amp;#39;ll do is we&amp;#39;ll actually do a step by step design example for a Class A operation and we&amp;#39;ll talk about all the very important steps though that are required to have a good have a good design with high efficiency high reliability and low cost now let&amp;#39;s see these are some of the very important parameters that we would like to have for our power amplifier we like to have high power efficiency and reliability and we like to have good frequency range and with what&amp;#39;s up with high power we like to have circuitry that are not very complex easy matching circuits would have very wide bandwidth and that&amp;#39;s one of the one of the very important features that we have for the gallium nitride transistor and also there are a lot of new systems that are using amplitude modulation different types of amplitude modulation like 64 QAM QPSK where the amplitude is bearing so we need to operate the transistor in a special way for linear and we like to have the transistor for act linearly linearly what that means we like to gain not to change with input power we want that we don&amp;#39;t want the phase changing and that&amp;#39;s what&amp;#39;s important for for for linear operation of course we like to have we like to have low cost now let&amp;#39;s take a look at some of the some of the gallium nitride transistors well we can actually look right here at a chip this is actually a chip where we have our input our input gate our output drain and the source which goes to ground now right here these would be connected with bond wires and for high frequency operation generally these are actually connected to the back of the chip for low inductance because if you have too much inductance you&amp;#39;re going to reduce the gain again on the circuit now you can see these little gate fingers in here these are the gate fingers in here and they have to be a certain length compared to the to the to the wavelength of the frequency of operation because if they&amp;#39;re too long the change is just not going to work some currents are going to be going one way and some currents are going to gone the other way so we have to we have to tailor the electrical lengths of these little fingers here considering the frequency of operation now here we see of the transistor here&amp;#39;s another transistor down here which is very high power you can see there&amp;#39;s a lot of a lot of gates a lot of gates here and up so this is for higher power application now here we see here we see these all these transistors this is a gallium nitride monolithic integrated circuit here and you can see oh here&amp;#39;s your transistors here all the matching circuits that require as you can see a lot of times what&amp;#39;s done is just a little length of transmission line lengths are used to match the circuit and here&amp;#39;s our here&amp;#39;s our input matching circuit our input matching circuit here so this is an example of a gallium nitride monolithic integrated circuit here and we can see now here the transistors are actually put into a package you can see the lids taken off here here&amp;#39;s the input and here&amp;#39;s the output and then these are placed with a screw the screw the circuit down here and you can actually here&amp;#39;s here&amp;#39;s packages with the lid with the lid connected and these are ceramic packages and then these are low costs low cost packages right down here so this is some of the some of the transistors that we can see here now now let us examine what the world it&amp;#39;s very important for for our transistor operation well what we need to do is we need to look at what&amp;#39;s what&amp;#39;s required for all the RF output power well we can see up here the RF output power is a function of the d-max which would be typically twice the power typically twice the power supply voltage minus V Men that&amp;#39;s niebolt each we like that to be small compared to v-max okay then we have to multiply that by the maximum current external current and we have to divide that by eight so that&amp;#39;s that&amp;#39;s the RF power so you can calculate how much power you can get out with transistors just by looking at the datasheet by looking at these various various characteristics now here we have an array of different of different semiconductors here first we had our silicon we had a wonderful silicon bipolar transistor and our LT MOS transistors here and then in the 80s we have our gallium arsenide our Colin gallium arsenide becomes very prevalent here and now we have our gallium nitride transistors here so let&amp;#39;s take a look now this the gallium nitride can either be placed on silicon or silicon carbide that&amp;#39;s a typical substrates that the gallium nitride is manufactured on so let&amp;#39;s take first of all let&amp;#39;s look at the bandgap why is it being gap important the game the bandgap determines how well the substrate works at high temperatures we don&amp;#39;t we want the substrate to be semi insulating we don&amp;#39;t we don&amp;#39;t want a lot of current going from the top to the bottom of the substrate and if we have a high band gap that means it&amp;#39;s going to work very well at high temperatures early early in the 70s they tried to make an integrated circuits out of silicon and for military applications and they found that when they run things up of 70 and 89 degrees centigrade temperature the the transistor didn&amp;#39;t work very well but the gallium arsenide works a lot better for gallium arsenide integrated circuits here so you can see this has a very high positive thing for the high temperature operation now let&amp;#39;s talk about the breakdown the critical breakdown field well if we have a high breakdown field we&amp;#39;re going to have high voltage operation that&amp;#39;s our v-max that 30 B max okay and then the thermal conductivity well gallium nitride is very similar to silicon okay but now when gallium arsenide is put the gallium nitride is put on the on the silicon carbide then it has a very high thermal conductivity so that so the temperatures want to stay nice and cool now we have our saturated velocity that&amp;#39;s our max behold that&amp;#39;s RI Max and so we have all the important characteristics for high power high V max hi max and since we&amp;#39;re working at a very high voltage here typically twice the power Seibel the V Max is very small compared to I max so we&amp;#39;re going to get we&amp;#39;re going to get up we&amp;#39;re going to get good voltage swing across the transistor here now the disadvantage of the gallium nitride transistor is the mobility the mobility is not nearly as high as the gallium arsenide but it&amp;#39;s a little trick we can use we can use a high electron mobility transistor structure which boosts which boost the mobility of the transistor to allow it to work with higher gain and higher frequency of operation now there&amp;#39;s two types of silicon carbide that we can use we can use the 4 H or the 6 H now the 6 H so it&amp;#39;s used for light emitting diodes it&amp;#39;s just one contact and it goes to ground well that would not be very appropriate for our for our mimic circuit here because we want the substrate to be insulates I mean insulated so we want to use the poor eh it has a better crystalline structure match to gallium nitrite and it&amp;#39;s a lot easier to make so that&amp;#39;s that&amp;#39;s how the transistors are made they&amp;#39;re made with the port the the silicon carbide is always a foraged 4-h characteristic that we have now well what about using the gallium nitride transistor for a low noise operation well here we can see with this particular device this is a Corvo device point one eight micron micrometer gate line the short of the gate line the lower the noise the higher the gain and we can see that at expan frequency has a noise figure way under DP with a with loss of gain so this works very nicely this works very nicely now let&amp;#39;s talk about how we can use a low noise amplifier generally there&amp;#39;s going to be a transmitter associated with it with it with it with the system and the leakage power from the trans when the transmitter is on the low noise amplifier generally is not working and we don&amp;#39;t want the transmitter leakage power to to burn out the low noise amplifier so the usually is a limiter place between the transmitter leakage power and and the low noise amplifier also there&amp;#39;s another consideration that the low noise amplifier would have to work with extraneous signals that are not desired it could be that could exist and it could exist so the low noise amplifier canopic has to be operating in the presence of these other things now recovery time is critical in many applications for the low noise amplifier when the transmitter turns off we want to ln8 the work when this when it&amp;#39;s being hit by high power from extraneous sources when the high power is off we want the low noise amplifier to work so there is a problem if we have if we have a if we have a gallium nitride transistor substrate mature and material that has a lot of dislocations or traps what happens is the electrons flows into the trap and it stays there for a long time and then it comes out so the recovery time can be quite long for a low noise amplifier getting back to its normal state if there&amp;#39;s a lot of dislocations or traps in the material and it could be in the order of milliseconds but for for good quality gallium nitride substrates the recovery time is in the order of 5 nanoseconds and the other advantage of this device here will have very high input ip3 so we can see that our gallium nitride trans our gallium nitride transistor meets many of the up many of the objectives we have high power good efficiency and high reliability now let&amp;#39;s talk about what was important for reliability I remember when I was working at Westinghouse and we had a we had a spike linkage come in that was one nanosecond wide and it was 1 watt of power and it blew out our blew out our Ellen A&amp;#39;s so you can have you can have very short pulses with with with high power that can blow out the transistor so you have to make sure that that even for short pulses you can exceed the breakdown voltage now another thing is a lot of times somebody may forget to hook up the antenna to the to the system so the the the Luna the the power amplifier is not going to see a match load and it could see a high voltage standing wave ratio so you get twice the voltage that appear across the transistor so you have to take care of that and a lot of manufacturers will state that the the power amplifier has to work with the short circuiting phases it has to be able to survive it doesn&amp;#39;t have to work properly it just has to work and can&amp;#39;t can&amp;#39;t it has to be able to work with that so that&amp;#39;s very important now the other thing is the junction temperature we have to keep that below the recommended value and another thing that&amp;#39;s critical is the gate current for the transistor if we have high power coming in to the if the power is increased the gate current can increase and what we have is what&amp;#39;s called metal migration whenever you have too much current running through us through a through a conductor you can get metal migration what happens is to carry the conductor carriers move around so that there&amp;#39;s a space where it looks like an open and then there&amp;#39;s spaces where the look the current density is very high through that particular part of the part of the channel so eventually it burns it burns up and gate opens up this doesn&amp;#39;t not happen right away I hadn&amp;#39;t we had examples where we&amp;#39;re where our power amplifier was being hit by too much input input power and and we took the transistor support and sure enough the gates were open so we had to reduce the input power and I&amp;#39;ll show in an example here we can actually put a gate resistance in the DC bias apply in the gate and that sort of tremendously reduces the gate current for high reliability and also we have to be able to turn on internal the power supply with the right sequence so we&amp;#39;ll be talking about that later so those are the key factors for reliability now here here we have our silicon carbide our transistor mountain on silicon carbide and you&amp;#39;ll see that there solder there solder here now there&amp;#39;s two ways to connect chips we can end use epoxy or we can use solder and it&amp;#39;s generally recommended for higher power applications to use solder as opposed to as opposed to the epoxy you can use epoxy for like low-noise amplifiers or or for amplifiers don&amp;#39;t have a lot of power and now we have to put this on it we have to solder this to some sort of a base base plate so the two characteristics that are important is we want have high thermal conductivity for the base plate and also we want to make sure that the temperature variation of the elongation is similar to that of the of the of the of the silicon carbide because if they&amp;#39;re a lot different they can actually pop off I remember when we first were working with our gallium arsenide chips and we soldered them to two aluminum and we temperature cycle then the chips the chips came off because of the different that the large difference in them between the two materials so now if we take a look at the various if we take a look at the various materials that we have generally copper malik all copy molar copper copper is generally used for the foot for the for the circuit that the transistor is soldered to and you can see that it has a pretty good thermal conductivity now the thermal expansion is not exactly perfectly matched to that but it works it seems to work fairly well now there&amp;#39;s a new material aluminum diamond which has a tremendous high thermal conductivity and it&amp;#39;s also better matched to the transistor so people are starting to use this and you can ask people actually make these up make the make these carriers we can actually map mount your mount your transistor on onto these with screws so so here you see your aluminum diamond now when you design the mimic circuit you want to make sure that you don&amp;#39;t put all the transistors too close together like for example in this particular in this particular example here because what can happen is the temperature gets much too high so the layout has to be done differently to make sure that your transistors aren&amp;#39;t too close together so that they don&amp;#39;t interact with on another increase of temperature and this provides a 20% reduction in the temperature compared to this circuit right here which is pretty good now the next question is well okay I want to design a gallium nitride circuit so where do I go so these are these are some examples of US companies that that provide the provide the transistors there are there are some over in China and there so there&amp;#39;s a Germany has some good has some good boundaries also so here&amp;#39;s the the wolf full-speed which is the which is cray and they have a quarter micron and 0.4 micron and where they where they got this miss this lead was with light emitting diodes late emitting diodes and light bulbs so they have a lot of a lot of processing capability in that in that particular area now try Quinn has been involved an RF MD non Corvo bought them and we have we have the higher shorter gate lengths 0.15 and 0.25 micron okay now Sumitomo is very very popular for very high power at lower frequencies and they can really put out a lot of a lot of power now make on they bought nice next which is a transistor that&amp;#39;s put on silicon as opposed to the silicon carbide and we have Northrop Grumman and hrl for the for the higher frequencies now here are some of the foundries that we&amp;#39;re talking about here here&amp;#39;s the hrl they&amp;#39;re up north of Brahman and Raytheon Raytheon has a wonderful process 490 gigahertz operation they can put out a lot of power with really good efficiency so they have a very high they have a very good very good process Raytheon and here&amp;#39;s our triQuint which is very very high quality now Dae also has has a process that works very nicely at 30 gigahertz 46% power added efficiency so these are some of the places that you can go to if you want to do if you want to do foundry work and also there are places in Germany and also places over in China now let&amp;#39;s see here well let&amp;#39;s talk about the gallium nitrite structure so here we have our t gate the t gate is used to have a short gate length here and then have more current capacity up in this area here now the first thing is we have a Schottky barrier which is a metal to semiconductor so we have metal to semiconductor that forms our Schottky barrier circuit here and this is aluminum gallium nitrite that&amp;#39;s that&amp;#39;s going to be our Schottky barrier circuit now that&amp;#39;s put on top of gallium nitride gallium nitride now when you go through the physics of the circuitry what you find out that the electrons like to flow in this gallium into the circuit right into the gallium nitrite so in this blue area right here that&amp;#39;s where the electrons flow so why is that important well gee electrons can flow a lot faster through a semiconductor that doesn&amp;#39;t have dopants in it compared to a semiconductor with dopants so that&amp;#39;s the idea you get very high very high electron flow which gives you a very higher higher mobility out of the circuit here and then here&amp;#39;s your buffer and here&amp;#39;s your circuit here&amp;#39;s your circuitry right here so that&amp;#39;s the example of a high electron mobility transistor now here we see a typical a typical structure right here and this shows the depletion region that&amp;#39;s the area where there&amp;#39;s no electrons available for conduction and if you look at the electric field the electric field is very high in this area very high in this area which means the voltage of the breakdown voltage is good it&amp;#39;s not going to be that that&amp;#39;s super but now by putting this field plate one here putting this field plate on top look how it spreads out the depletion region and it reduces see it reduces the electric field which is great this is going to this is going to provide higher breakdown voltage what we want for the transistor here and also it&amp;#39;s less dispersion in other words there&amp;#39;s less electrons that are put into these traps here the imperfections of the semiconductor here so there&amp;#39;s less dispersion with that so we won&amp;#39;t have the the disadvantages of the traps associated with the circuit we can also see that with the IV curves for the transistor we can see this is without the field plate and this is what the field play so if you have to work something with this thing if you&amp;#39;re lute you&amp;#39;re not going to be able you&amp;#39;re not going to be able to get the high max instead of going up to here you&amp;#39;re only going to be gone off to here so you&amp;#39;re not going to get as much output power ll of the circuit here now this characteristic here is for for a material that&amp;#39;s not that&amp;#39;s super all all materials are not like this this just just shows you the this just shows the extreme when you have a lot of a lot of traps associated with the would be with the semiconductor now here&amp;#39;s the reliability curves we have so for example if we have a tens 10 to the six mean time to failure requirement then we can operate this about two hundred and sixty sixty degrees centigrade okay so that&amp;#39;s pretty pretty good reliability now here&amp;#39;s these here&amp;#39;s the gallium nitride on the silicon and we can see they have the different layers they have the different layers here you can see we have a transition layer because the silicon and the gallium nitrite don&amp;#39;t have the same crystalline structure so you have to have a transition layer to to reduce the stress to reduce the stress between the two layers and here we can see here&amp;#39;s our here&amp;#39;s our aluminum gallium nitride and here&amp;#39;s our electron gas right right in here in the gallium nitride circuitry right here now may calm has improved the reliability of these circuits the reliability than it used to be as good as the as the other services as this silicon carbide but they&amp;#39;ve improved the process now and they can work with a we can work off the 225 degrees now with with pretty good high reliability now here&amp;#39;s here&amp;#39;s an example of a circuit that works up at 90 gigahertz and these you have to do you have to the circuitry has to be done a little bit differently when you have these very short gate lines to make sure that you have proper operation without with the higher frequencies with this very short very short gate lamp so what they do is they have they have this layer right here this layer right here helps to confine and the electrons to run in this channel right here okay and then these layers here are used to help make the circuit work with the very short gate length to suppress a whole electron diffusion and improves the electron confinement in the gallium nitride channel so here&amp;#39;s an example of the latest of hrl 40 nanometer device here let&amp;#39;s see how it works here we can see here&amp;#39;s our actually here&amp;#39;s our actual transistor right here and let&amp;#39;s see how it works here&amp;#39;s our here&amp;#39;s our here&amp;#39;s our power added efficiency of the circuit here it&amp;#39;s 40% which is pretty good and we have about 9 DB of gain and our output power is around 27 27 or so DB DB M so this really has pretty good pretty good performance typically power amplifiers working at these higher frequencies do not have very good efficiency okay this is a first circuit I&amp;#39;ve seen and has pretty good efficiency that working at this at these higher frequencies this is a new development for them now here&amp;#39;s the bae circuit here this works without a field plate they have processed this to work very nicely without a field plate and you can see this is a load pull circuit that we have for the circuit here and though and behold we get 53 percent drain efficiency with 45 percent power added efficiency with 4.9 watts per millimeter at around 29 gigahertz so this works this works very nicely so the government supported the development of this of this process for BAE and here&amp;#39;s the Fujitsu this said this has very high very high power 3.6 watts per millimeter all the way up it at 90 gigahertz or so and you can see they have double layer silicon nitride passivation to reduce the traps and here&amp;#39;s the here&amp;#39;s the here&amp;#39;s the gallium nitride channel and then this is the other other layers that we have here now the disadvantage of this circuitry the efficiency is not that not not that fantastic it&amp;#39;s only twelve point three percent in eighty eighty six gigahertz Raytheon has a circuit that has much higher much higher efficiency at these frequencies now the next thing we want to examine is the turn on and turn off of the of which of the transistor first we want to a lot of times with the power supply when you first turn the transistor on and off it will it will have spikes in it so it&amp;#39;s a good idea not to hook up the transistor to this to the power supply first turn the power supply on turn the voltages down and then hook them up the first thing you want to do is you want to now increase the gate voltage to the pinch off voltage then you want to turn the drain voltage up to the required value and then you want to adjust the gate voltage to your to the desired current now to turn it off you want to just do the opposite then you want to reduce so you want to reduce the gate voltage to want to reduce the gate voltage to the pinch off voltage then you want to turn the drain voltage off then you want to turn the gate off now a lot of a lot of regulators will actually have these built in to turn the transistor on and off properly now let&amp;#39;s take a look at operating the transistor classe ah what we&amp;#39;re going to do is we&amp;#39;re going to have a we&amp;#39;re going to have a current the current generator here that&amp;#39;s going to generate the current through the transistor here and we want to adjust the voltage we want to adjust the load resistance so with this given current it fills up the voltage voltage across the transistor so the load voltage is just the current swing over the voltage swing and then our our RF power is going to be the RMS voltage on the RMS current so here we see it&amp;#39;s a voltage swing times the current swing over eight and the DC power this is where you actually have to add the V minimum and V the maximum the minimum and AI minimum to the circus so the efficiency is 50% now the advantage of the gallium nitride transistor is this v-max can be very high compared to the best values typically you&amp;#39;re going to get your 50% efficiency out of the circuitry here so what we&amp;#39;re going to do now is we&amp;#39;re going to now look at look at the design of a design of the circuit of the circuit here this is 30 watt 30 watt transistor and we want to now go through the complete design of the circuit so what I&amp;#39;m going to do is I&amp;#39;m going to actually look at the look at the circuitry here and we&amp;#39;ll go through the various steps involved in the design of the circuit now the first thing we want to do is we want to look at the stability of the circuit and this is derived from the small signal s parameters and the stability doctor wants to be greater than one so we can see at our operating frequency we&amp;#39;re in trouble because less than one so it has the possibility of oscillating and also around 13 and 1/2 gigahertz and that&amp;#39;s a possibility of oscillating so it can actually look the stability circles now since we&amp;#39;re interested in since we&amp;#39;re interested in all we have to add be sisters to the circus so we want to add them on the front of the circuit so we&amp;#39;re going to now this is the output stability plane so what that means is if we put any impedance here the input impedance will have a reflection coefficient larger than one and it&amp;#39;s going to be possible to oscillate now on the input stability planes if we put a resistance in here then the output reflection coefficient will be greater than one so since we&amp;#39;re going to add loss to the circuit we want to put it on the front not on the output so what we&amp;#39;ll do is we&amp;#39;ll turn this off and then we&amp;#39;ll look here so we want to add a resistor here if we take point O four times 50 it&amp;#39;s about two ohms so now if we put two ohms in the circuit here you can actually move that right out and then if we look at our stability here we can see that we have improved that so I&amp;#39;m going to add about four I&amp;#39;m going to increase to five ohms for better stability now we have to worry about this circuit right here so now if we look at our stability circles we can see now we have to put something in parallel here so now let&amp;#39;s now go to the circuit here and we&amp;#39;re going to input put and put this resistor in the circuit and this is a quarter wavelength line at 12 and a half gigahertz open short and it puts the resistance in this circuit now let&amp;#39;s now look at the stability here and we&amp;#39;ll add our parallel resistor here and we&amp;#39;ll put 50 ohms in here so now you can see the circuit is very stable and we&amp;#39;re in pretty good shape now the next step the next step of the design is to is to look at our load line now we&amp;#39;re going to look at our load line here and here is our load line here so this is the operating point here so what we have to do is we have to buy a transistor so it&amp;#39;s Class A so we&amp;#39;re now we&amp;#39;re going to we&amp;#39;re going to increase the volt you can change the gate voltage so we&amp;#39;re about here for the maximum current then we have to increase our power now here the low brisas is seven ohms so it&amp;#39;s not why it&amp;#39;s not big enough we have to increase it until we get the full voltage swing and now when we put our input power in here or in pretty good shape now you can see we have a little bit of hysteresis that says the current and both these aren&amp;#39;t exactly in phase okay now let&amp;#39;s see how we&amp;#39;re doing with our with our current and voltage so we can see we have a 5 volt 5 volt knee voltage up to 60 50 60 volts and our current small and we have our maximum current so we should be in pretty good shape for a good e pretty good performance out of the circle let&amp;#39;s see how we&amp;#39;re doing whoa we&amp;#39;re getting 45 DBM of power which is 30 1.6 watts we&amp;#39;re getting 51 percent efficiency and this is our 1 DB gain compression point and we can see the phase is varying about eight tenths of a degree so that&amp;#39;s pretty good so the next step is to actually match the 50 we have this we have to look at our 15 ohm so now let&amp;#39;s open up the project open up the project and that&amp;#39;s going to be our output match so here&amp;#39;s our circuitry here we have our 15 ohm which is what the transistor wants to see and we have a 50 ohm generator so we have to add we have to add circuitry to the circuit so now let&amp;#39;s look at our matching circuit here so let&amp;#39;s add some electrical length until the conductance is equal to 1 and then we&amp;#39;ll add our shunt capacitance so now we&amp;#39;ve got our input match so now we&amp;#39;ve done a nice job with our input match now the next step is to match I&amp;#39;m sorry that&amp;#39;s the output match now we have to match the input circuit here so first we have to look and see what the impedance actually is so this is input match step 1 so this is the actual impedance here they&amp;#39;re normalized resistance is point 1 to 3 now whenever we want to match this with the 2 step matching circuit here so what we&amp;#39;ll do is we&amp;#39;ll well first the intermedia impedance is going to be the square root of this times 1 which would be about 17 point 8 ohms so the first step now is to match is to match the next step now is to match this to 17.8 ohms so that&amp;#39;s going to be our next match to circuitry here so now we&amp;#39;re going to match that and also I match that so we now we have this match to 17.8 ohms then step three is going to be to match that 250 ohms so now we have this now we&amp;#39;re going to match this 250 ohms so that&amp;#39;s going to be the next step here it&amp;#39;s going to be the next step to match that 250 ohms so our electrical length will add that and now we have the complete circuit matched so now let&amp;#39;s look at the final results for the complete circuit so here&amp;#39;s the here&amp;#39;s the complete circuit now we have our 45 DBM of power our 51% power added efficiency so you can see the performance is practically identical to what we had when we had when we had individual circuits associated with that and now the next thing I want to look at is the temperature rise for the transistor now here&amp;#39;s the actual temperature rise it gets pretty hot when you&amp;#39;re not using it and then when we actually have the operation it&amp;#39;s running cool so that&amp;#39;s the one of the disadvantages for Class A operation when you&amp;#39;re not using it you should turn the transistor off and now we can actually see here&amp;#39;s the actual gate and if we ask this is for for higher output for higher output characteristics so if we add a resistor in here I guess this isn&amp;#39;t working properly oh if we add the resistor in here it&amp;#39;s going to here&amp;#39;s here&amp;#39;s without the resistor and then this is with the resistor so anyway we saw that the power amplifier worked pretty good very close to the characteristics that we had so we can actually put in a we can actually put a time constant in here to account for if we have short pulses with short duty cycles to see what happens so basically we did a pretty good design with our with our with our parent for fire design so in phase two and the next phase will actually be looking at Class B operations other classes of operation D F minus 1 and J modeling of the transistor and the Dority amplifier so thank you so much for your kind attention [Music] ok great thank you good there was a lot of lot of info there and ecologist everyone for for the issue with not being able to see the pointers on the on the slides I know that can be a challenging thing so we&amp;#39;ve got time for a few questions here and before we start that though remember that you can still submit your questions through that QA panel so let&amp;#39;s go ahead and get on to this so the first question here is on slide 21 reliability is extrapolated with only one one data point so how is that how is that justified or do you have any insight into that how do we get to this okay slide 21 yeah see here let me help oops here give me just one second here let me get right to it oh oh hold on one second there you go nope there you go yep any thoughts on it oh there&amp;#39;s those points were just put on there to show that when you work at 225 degrees channel temperature that you have the MT to failure of that particular value that&amp;#39;s not that that&amp;#39;s not how it was generated that&amp;#39;s just that&amp;#39;s just goes to show you what the operation what the reliability would be at 225 so for example here at 300 if we drew that up here then that would be the reliability over there that&amp;#39;s just to show that&amp;#39;s just to show a particular point there&amp;#39;s there some underline how the reliability was determined yeah okay great okay so here&amp;#39;s a question about efficiency what happens to efficiency as you start to combine transistors to get more power how do you get the best efficiency yes did you then I didn&amp;#39;t hear that go ahead okay so right so what happens to efficiency as you start to combine transistors to get more power how do you get the best efficiency okay well generally generally how you do how you get how you combine transistors is you take a group of them maybe four or five transistors and like say say you had twenty of them see you have to combine 20 units so you break them up into five units of 5:3 have four units of five now what you&amp;#39;ll do is you&amp;#39;ll combine the five units with another five units together and then you combine the other five units to the other five units together and that&amp;#39;s them with the quarter wave length transformer then you combine the five and the five with the other five and the five together and then you have another matching circuit to get to your to your 50 ohms so that&amp;#39;s typically how it&amp;#39;s done it&amp;#39;s called cluster matching cluster matching and it&amp;#39;s very very low it&amp;#39;s very low loss because you just have you just have the transmission lines in there though impedance transmission lines and it&amp;#39;s free it&amp;#39;s very it&amp;#39;s a very nice way to do it broadband with low loss that&amp;#39;s how that&amp;#39;s how everybody doesn&amp;#39;t combine the outputs sure okay here&amp;#39;s another question could you comment on the major differences in terms of design methodology of gallium nitride hint versus gallium arsenide hbt or CMOS yes now the difference between the the difference between the gallium arsenide and the gallium nitride is that you have to use many more transistors with the gallium arsenide because generally with the gallium arsenide you can only bias them at 9 volts at the the high electron mobility transistors you can only operate them at 9 volts because a breakdown voltages or maybe only 15 volts or so so what that means is you have to use many more transistors to get the same output power now when you do that your input capacitance is going to be a lot higher when you have to use so many transistors so with the gallium arsenide circuitry you&amp;#39;re going to have to use many more transistors and the input capacitance is higher which means you&amp;#39;re not going to have as good bandwidth with the gallium nitride transistors you are able to work with a very high voltage very high voltage operation which means you won&amp;#39;t have to have a lot of current and a lot of current means you don&amp;#39;t have to use as many devices to get the output current so that&amp;#39;s the major difference between between them okay let&amp;#39;s see there&amp;#39;s several questions here regarding traps and memory effects could you make some is I specifically for gain could you make some comments about that if by the way the memory facts share memory effects and effects of traps oh okay [Music] well I guess memory effects is when when you&amp;#39;ve run the circuit one direction it works differently than if you go the other direction because it remembers what happens in the past and that creates a problem with digital modulation because they have to have especially memory effect cancellation effect with it with the linearization with the Alain techniques and also with it with the traps with the traps that that also creates the dispersion where where the where the it takes time it takes time for things to happen it takes time for things to happen when you change when you change things okay I have I&amp;#39;ll see there&amp;#39;s a part of the part of one of the questions and asks about low frequency limitations and in in regards to traps so is there a is there a low frequency limit for these does it cause low frequency noise oh the traps are the frequency noise well I whenever we talk about noise I always think about one over F noise one over F noise and I believe that with with silicon with silicon silicon has Braille oh one over F noise because the material is they have many years of processing the material so they have very little one over F noise so for example if you have an oscillator or if you have a low noise amplifier and you&amp;#39;re looking close very close to the carrier that one over F noise is a predominant effect but now with the gallium arsenide with the gallium arsenide the one over F noise is not as good because of the traps and I&amp;#39;m sure with the silicon with the gallium nitride we would see that also you would see the the trap effect and that that definitely affects the one over F noise characteristics so when you get close to the carrier the noise comes up and I remember when I designed my first oscillator at x-band I could either use gallium arsenide or or silicon and I picked silicon because it has lower one over F noise because the crystalline material is much much better for that characteristic and it worked very nicely very nicely and ty also remember traps when I had an oscillator and I was using I was using material and it worked fine but then but then when I had a material that had traps associated with it when I changed the voltage this the circuit in tune to the new frequency because it took a while because of the traps you know so I had to go to the manufacturer and get a better quality material and then we were okay so traps do affect the low frequency noise sure that&amp;#39;s great okay all right well we are out of time there&amp;#39;s still a number of questions here and the presenter will follow up with those offline and also oh I would love to answer your questions and there&amp;#39;s an opportunity to do that with the part 2 of this as well which is surely in April of 2019 so everybody please be on the lookout for that so also as we said earlier this session will be archived on the Society website at MTT org and all the registrants will get an email reminder with the website address when that&amp;#39;s available for attendees who would like to receive the pdh credits then please follow the link in the webcast view and use the code that&amp;#39;s provided on the last slide of this presentation once again I&amp;#39;d like to thank dr. Nikki for this excellent and informative presentation we look forward to part two of this webinar series in April of 2019 and special thanks to our audience today for for joining us we hope you found today&amp;#39;s event valuable and that you&amp;#39;ll return for future I Triple E microwave theory and technique society webcasts thank you and have a great day

Transcript for:Design of GaN Power Amplifiers (Part 1) by Dr. Edna Hickey

Transcript for:
Design of GaN Power Amplifiers (Part 1) by Dr. Edna Hickey