Thanks for this short series on class E, very detailed and informative. I designed a class E stage myself, for a few Watts, and noticed that this class E configuration is very sensitive to load changes. Either the FET voltage or current increase quite a bit with Rload and operating in a real life situation you have to either keep SWR low or have good protection circuits for the FET. In the end I went for a slightly different configuration with the L from drain to supply having the same impedance as the load and use it in conjunction with a PI low pass filter to also get kind of class E operation. My measured efficiency is about 94% and the voltages and currents in the FET remain kind of the same, even with SWR of over 10:1, so no protection is needed. Try what happens in your LT spice simulation when stepping Rload ovee a wider range. It is very interesting.
Thank you very much for this video. I'm working on optimizing the final amp stage of my (tr)uSDX ham transceiver kit to get the efficiency up as high as possible while keeping power output around 4 or 5 watts in order to get maximum usable time on a battery charge. It's a fun process, but I wanted to better understand the variables and their relationships instead of just trying random changes until it worked better. The previous discussion about circuit Q and bandwidth is something I will have to delve into to make some necessary trade-offs. Understanding is great stuff.
Just a few remarks. The gate drive current is equal to Qg x Fsw. The second note is about the switching loss in the MOSFET. In LT spice, this is calculated from Id. But Id also contains the current required to charge the MOSFET output capacitance Coss. The loss you see is very often higher than the real loss specially if the turn off is faster than the rise time of the drain voltage. (you can see this on an oscilloscope by observing the gate plateau voltage or the absence of that plateau if Vds rises slowly because of limited current to charge Coss) You might also be interested in resonant gate drive circuits. Basically a class E amplifier driving a MOSFET..
Dear Sir. It is a very good video. But I would suggest you to consider the following: in order to reduce the skin effect, I would suggest to use the litz wire, in order to reduce the proximity and inter turns capacitance, I would suggest the "honey-comb" winding, and in order to reduce the relative number of turns, to use the Mn-Ni-Zn based ferrite core, and, finally, in order to avoid the core saturation, use open (not closed) cores.
@@Paxmax I got a Levell PM9A uV/pA meter and at uV (very high impedance compared to modern meters) and at pA range the needel moves from bottom to top of the range by just breathing. Idk what kind of setup I would need to make that range a bit more stable but hey it does actually work if I set up a camera from a distance and leave the room
Since the gate capacitance represents a storage of energy, is there a way to recover that energy and use much of it again later? What comes to mind is some form of inductance to store this energy during the turn off and use it again during turn on. One would of course still have loss in transferring this energy back and forth. In the second video you used a sine wave to drive the gate successfully which suggests an LC circuit for the driver where the inductor is in the collector circuit of the driver and the capacitor is the input capacitance of the final gate. But then isn't the class E, just a variation on a class C were external components are adjusted to assure 0 Voltage turn on and 0 current turn off?
How would you actually use this to amplify a signal? What does it look like if your signal has some information encoded in it beyond just the carrier? For instance you mentioned there is some rise time, does this impact the usefulness for amplifying CW signals? It seems reasonable to use this to amplify a frequency modulated signal as long as the frequency deviation doesn't exceed the bandwidth of the filter stage. Is there any way to use this with phase modulated or amplitude modulated signals?
Thanks for the video. Im still in my quest for a class E for 28MHz, but the BSS316 might work, and I also found the IRMLM0100. My goal is to build a 28MHz beacon either in CW morse or WSPR.
Then it should be ok. The transistor after all can support up to 0.5W of power dissipation, so even in a non-ideal situation, it should be ok for a 1w transmitter.
Any "pulsed DC", from a spectral point of view is still made up of sine-waves; if the pulse rate is low, then the AC-effects can be neglected - ex: with a 1Hz pulse you do not care about rise/fall time and switching losses; but with a 100MHz square wave the AC phenomenon are predominant.
Is there a minimum thermal resolution that you'd recommend for electronics? I see the B1L you use is 160x120 is over $300 USD. Referral link totally fine!
I can't say what a minimum usable resolution would be... in the end even a 1pixel pyrometer will do as long as the measured spot is small enough in reference to the component you need to investigate. The larger the resolution the easier it will be to observe a target in more detail, but of course, larger resolution is more expensive... so it will always be a budget related compromise. Another thing to look out for, if you are interested in small electronics (like SMD components), is the possibility to add an extra lens to be able to get close to the target - by default most cameras recommend a minimum distance of ~20-30cm, distance at which you cannot make out small components.
From what I could find, litz wire is recommended for uses-cases up to 0.5-2MHz; after that the benefit becomes negligible compared to solid wires... So the exact wire type that gives best results will depend on the usecase frequency.
Thanks for this short series on class E, very detailed and informative.
I designed a class E stage myself, for a few Watts, and noticed that this class E configuration is very sensitive to load changes. Either the FET voltage or current increase quite a bit with Rload and operating in a real life situation you have to either keep SWR low or have good protection circuits for the FET.
In the end I went for a slightly different configuration with the L from drain to supply having the same impedance as the load and use it in conjunction with a PI low pass filter to also get kind of class E operation. My measured efficiency is about 94% and the voltages and currents in the FET remain kind of the same, even with SWR of over 10:1, so no protection is needed.
Try what happens in your LT spice simulation when stepping Rload ovee a wider range. It is very interesting.
This is a very useful comment. Thanks!
Thank you very much for this video. I'm working on optimizing the final amp stage of my (tr)uSDX ham transceiver kit to get the efficiency up as high as possible while keeping power output around 4 or 5 watts in order to get maximum usable time on a battery charge. It's a fun process, but I wanted to better understand the variables and their relationships instead of just trying random changes until it worked better. The previous discussion about circuit Q and bandwidth is something I will have to delve into to make some necessary trade-offs. Understanding is great stuff.
The uSDX looks like a really interesting project! May I ask what sort of antenna you are using for it?
Just a few remarks. The gate drive current is equal to Qg x Fsw. The second note is about the switching loss in the MOSFET. In LT spice, this is calculated from Id. But Id also contains the current required to charge the MOSFET output capacitance Coss. The loss you see is very often higher than the real loss specially if the turn off is faster than the rise time of the drain voltage. (you can see this on an oscilloscope by observing the gate plateau voltage or the absence of that plateau if Vds rises slowly because of limited current to charge Coss)
You might also be interested in resonant gate drive circuits. Basically a class E amplifier driving a MOSFET..
fascinating series and it’s super interresting to see how you solve the limitations step by step!!!
Dear Sir. It is a very good video. But I would suggest you to consider the following: in order to reduce the skin effect, I would suggest to use the litz wire, in order to reduce the proximity and inter turns capacitance, I would suggest the "honey-comb" winding, and in order to reduce the relative number of turns, to use the Mn-Ni-Zn based ferrite core, and, finally, in order to avoid the core saturation, use open (not closed) cores.
Love how the Scope bounces around from milivolt and yoctovolt
That’s wild. Must be a nice scope
Just looking at it wrong will probably force it to change range.
@@Paxmax I got a Levell PM9A uV/pA meter and at uV (very high impedance compared to modern meters) and at pA range the needel moves from bottom to top of the range by just breathing. Idk what kind of setup I would need to make that range a bit more stable but hey it does actually work if I set up a camera from a distance and leave the room
@@AlpineTheHusky Avoiding disturbing entropy is a hell of'a hobby! 🤗
So happy to see this video!!
This videos triggers dopamine release in my monkey brain :)
Since the gate capacitance represents a storage of energy, is there a way to recover that energy and use much of it again later? What comes to mind is some form of inductance to store this energy during the turn off and use it again during turn on. One would of course still have loss in transferring this energy back and forth. In the second video you used a sine wave to drive the gate successfully which suggests an LC circuit for the driver where the inductor is in the collector circuit of the driver and the capacitor is the input capacitance of the final gate. But then isn't the class E, just a variation on a class C were external components are adjusted to assure 0 Voltage turn on and 0 current turn off?
How would you actually use this to amplify a signal? What does it look like if your signal has some information encoded in it beyond just the carrier? For instance you mentioned there is some rise time, does this impact the usefulness for amplifying CW signals? It seems reasonable to use this to amplify a frequency modulated signal as long as the frequency deviation doesn't exceed the bandwidth of the filter stage. Is there any way to use this with phase modulated or amplitude modulated signals?
Really interesting and useful, Fesz! Thanks!
How to desaign 144mhz 20-100W handy talky using class E RF amplifier sir?
Why there is no 100 GHz radio stations around? What are the limitations?
Thanks for the video. Im still in my quest for a class E for 28MHz, but the BSS316 might work, and I also found the IRMLM0100.
My goal is to build a 28MHz beacon either in CW morse or WSPR.
What sort of power level are you trying to achieve? the BSS316 is quite a low power transistor after all...
@@FesZElectronics 1W
Then it should be ok. The transistor after all can support up to 0.5W of power dissipation, so even in a non-ideal situation, it should be ok for a 1w transmitter.
I understood the skin effect and proximity effect in AC, but what is the situation in case of pulsed DC?
The same effects are present?
Any "pulsed DC", from a spectral point of view is still made up of sine-waves; if the pulse rate is low, then the AC-effects can be neglected - ex: with a 1Hz pulse you do not care about rise/fall time and switching losses; but with a 100MHz square wave the AC phenomenon are predominant.
@@FesZElectronics thank you for your response! I wish you a Happy New Year!
👍❤
Is there a minimum thermal resolution that you'd recommend for electronics? I see the B1L you use is 160x120 is over $300 USD. Referral link totally fine!
I can't say what a minimum usable resolution would be... in the end even a 1pixel pyrometer will do as long as the measured spot is small enough in reference to the component you need to investigate. The larger the resolution the easier it will be to observe a target in more detail, but of course, larger resolution is more expensive... so it will always be a budget related compromise. Another thing to look out for, if you are interested in small electronics (like SMD components), is the possibility to add an extra lens to be able to get close to the target - by default most cameras recommend a minimum distance of ~20-30cm, distance at which you cannot make out small components.
Many years ago, we used a limited resolution thermal camera that told us where to target with thermocouples.
Could air core inductors wound with litz wire be made smaller for the same efficiency?
From what I could find, litz wire is recommended for uses-cases up to 0.5-2MHz; after that the benefit becomes negligible compared to solid wires... So the exact wire type that gives best results will depend on the usecase frequency.