Troubleshooting Class E Amplifier

Thread Starter

Skaler87

Joined Sep 15, 2013
7
Hello,

First off, I would like to say that I am new to this forum and I apologize in advance if I do not adhere to any norms.

As far as my issue is concerned, I am working on a wireless power transfer system for my electrical engineering capstone/design thesis project. Over the past few months I have successfully designed and implemented many parts of the system, however recently I have been working on the practical implementation of a Class E power amplifier (designed for 13.56MHz to output about 20W of power) and I have come across some issues. Now, I have designed the amplifier using Agilent ADS and in simulation it works wonderfully. However, when I went into the lab to test the circuit, I found some odd behavior.

In particular, the issue I am having is that when I operate at the desired frequency of 13.56MHz, the output waveform at the drain of the transistor does not exhibit the switching waveform that is characteristic of the Class E amplifier. Also, changing the gate bias and therefore conduction angle of the amplifier does not really change anything. Instead, I end up with a DC offset and slightly amplified version of the signal I am giving to the gate of the transistor. Furthermore, when I try to increase either the gate bias or drain bias beyond 2V and 5V respectively, I end up with a strange amplitude modulated signal which seems to have components at about 5MHz and 80MHz. The strange behavior doesn't stop there. I find that when I run at lower frequencies (up to 5 or 6MHz) I do see the switching waveform and am able to control the conduction angle of the waveform using the gate bias and the amplitude of the waveform using the drain bias. That is, I get the expected characteristics at a lower frequency range.

I would like to know if anyone has any suggestions as to what could be causing such behavior as I have been unable to solve this problem for the past week or so. Unfortunately, my school does not have any professors who are experts in RF circuit design and so I am pretty much on my own in this regard. I have attached a schematic of the amplifier circuit as well as the waveforms that I expect in simulation. I would also like to add that the inductors I am using have self-resonant frequencies larger than 13.56MHz (the lowest SRF is 100MHz). Also, the transistor I am using is the MRF6V2150N LDMOS by NXP, which I have come across in a few IEEE papers which use the device in Class E PA design.

Any suggestions or insight would be greatly appreciated.

Thanks in advance.

classEcircuitwaveformPLUSeff.png fullClassE.png
 

Thread Starter

Skaler87

Joined Sep 15, 2013
7
My apologies. I was wrote the initial query on the train and could not provide a picture of the board at the time. The yellow arrow marks the input to the amplifier. At the moment I am simply using a function generator, but the goal is to use the oscillator and driver circuit seen to the left to drive the amplifier. Some things to note, due to the physical package of the transistor I had to use the wires to connect the terminals of the transistor to the board. In order from top to bottom the wires are the gate, drain, and source. The source is the bottom face of the transistor which is connected to the heat sink, which is then connected to ground through the wire that is attached to the screw. the output of the amplifier is the red wires seen running to the left which is the transmitting coil of the overall system. Everything is connected right now as I just got home from testing. But the transmitting coil is typically disconnected.

As far as a test schematic is concerned, it is simply the schematic attached above with a function generator and a load (I've been using about 50ohms since my coils have approx. 47ohm resistance at resonance). I was just doing some reading right now and read in some places that ceramic capacitors may not be the best for high frequency applications. Any thoughts on this?

20160413_235918.jpg
 

Lestraveled

Joined May 19, 2014
1,946
OK, your construction looks more like you are building a battery charger than a 13 MHz RF amplifier. There is more inductance in your leads than in your coils. Begin again. Get a sheet of blank printed circuit board stock. This will be your ground plane. Look up/google/research "manhattan prototyping". This is how your prototype should be built. I think there is a section on Manhattan prototyping in this website. This style of prototyping is for high frequency.

Your design might be fine but your construction is for DC circuits.
 

Thread Starter

Skaler87

Joined Sep 15, 2013
7
I really appreciate your insight Lestraveled. As I mentioned in my initial post, we have very little exposure to high frequency circuits and high frequency circuit design at my school, but this is the area I am most interested in. The highest frequency I have worked with prior to this is around 30kHz which is nothing really. I will most certainly look into Manhattan Prototyping. If you have any other suggestions, I would love to hear them. Thanks a bunch!
 

Lestraveled

Joined May 19, 2014
1,946
@Skaler87

When I am building with high frequency discretes, I start with a piece of PCB material. I cut up some double sided PCB stock into 1/8" squares and 1/8" wide strips. (Some bigger, some different sized). Where I need a pad, I tin the PCB at the spot and then take an 1/8" square and solder it to that spot. The squares are pads, the strips are buss bars. Prototyping RF circuits is fun and challenging. Go for it.

Check in. we will answer your questions.
 

Thread Starter

Skaler87

Joined Sep 15, 2013
7
Thanks for the tips.

The prototype isn't a requirement, but I really want to get it working after putting in so much effort. I'll have to work double time as final exams are right around the corner. I'll be sure to post any questions that come up. Thanks again!
 

Thread Starter

Skaler87

Joined Sep 15, 2013
7
@Lestraveled

So I ended up getting a hold of some scrap PCB material from my lab technician and created a layout of some pads (as opposed to cutting a PCB and gluing pads as I read there may be issues with capacitance) and had my lab technician etch them out on the board. When I tested the amplifier again, I found that it was better, but still not perfect. However, changing the construction method to this one has made it much simpler to test different components as I find it much simpler to solder and desolder using this method. I cannot express how grateful I am. Thank you so much!

After singling out some of the components I found that removing some of the capacitors resulted in a better output waveform. I was able to bias the drain and gate beyond the limitations I mentioned in my initial post, and was even able to get an amplified signal at the output (35Vpp output from 1Vpp input). The only thing that needs fixing now is that the amplifier is not switching but is always on. Based on the theory, the capacitor that I have connected at the drain to ground should be able to fix this, but it needs to be replaced since it is causing issues with the functionality of the amplifier. This confirms my doubt on the use of non-RF capacitors at such a high frequency. So now I plan on replacing all of the capacitors with capacitors better suited for higher frequencies.

I will be sure to post any updates as they take place.
 

SLK001

Joined Nov 29, 2011
1,549
13MHz isn't really high frequency. What does this capacitor look like right now? At the very least, you should be using chip devices where you can. These are closer to theoretical devices than leaded devices. At 20W of power, the ESR of a capacitor can cost you significant power. Look into obtaining an engineering capacitor kit from ATC (American Technical Ceramics) in your range. ATC makes some of the best caps for RF power applications in the business.
 

Thread Starter

Skaler87

Joined Sep 15, 2013
7
@SLK001

At the moment all of my capacitors are leaded. They are all polyester except for one ceramic. Basically I have a bunch of electrolytic and polyester capacitors that I have accumulated from lab kits that we have had to buy over the past 4 years. The circuits I have worked with prior to this are very low power(less than 3W) and low frequency (less than 30kHz) and I've read on some forums and online references that these capacitors will not work that well with higher frequencies and higher power. I will definitely look into the capacitor kits from ATC. Thanks a bunch!
 

Thread Starter

Skaler87

Joined Sep 15, 2013
7
So I've still been working on troubleshooting the amplifier over the past week or so but haven't been able to post an update because of final exams. I was able to replace the capacitors which made the overall circuit work better, but still I am finding that it is not working 100% as it should. I should point out that I was able to connect the amplifier to the rest of the system and transmit sufficient power over about 15cm to charge a smartphone, so it's doing something..it's just not functioning as efficiently as it was designed to.

The issue that I'm having is that the amplifier is not working like a Class E with the expected switching characteristic at the drain (as seen in my original post) but is more like a Class A/B/C in that it is always on. That is, the waveform at the drain is perfectly sinusoidal as opposed to being flat (or close to flat) for some duration of the period. Now, I came across something interesting today which I believe to be the source of the problem, but I am not entirely certain how I can go about solving it, or if it is indeed the problem.

The issue I saw today is that when I remove the RF signal input (coming from a function generator for the sake of testing) I still get a waveform at the drain. Not only that, the waveform at the drain of the amplifier is precisely the waveform I expect from a Class E amplifier, except it is at a different frequency than the one I am using (about 3MHz instead of 13.56MHz). Now the first thing that came to mind was that there must be some funny business going on with the inductors, and so I tried changing the values of the inductors I've used as RF chokes. As seen in my initial post, the RF choke inductors that I am using are 10uH. I managed to get a hold of some with a high enough current rating and SRF at about 4 or 5 times the carrier frequency. Oddly enough, changing the values of the inductors changes the frequency of this waveform that is being generated on its own. For example, decreasing the RFC values I found a switching waveform at about 6MHz.

Now of course this parasitic waveform is not controllable to the extent of the RF signal, but still I find that under certain biasing conditions it reaches a peak of 10V with no RF input. Moreover, when I apply the RF signal to the circuit under these different conditions and sweep the frequency, I am able to see a switching waveform at the frequency of my applied RF signal at the output so long as the RF signal frequency is below the parasitic waveform frequency. Now this switching waveform can be controlled up to a peak value of about 55V (as expected through simulation) but its frequency is limited by that of the parasitic waveform. If I continue to increase the frequency beyond that of the parasitic waveform, I find that either the waveform at the drain becomes a perfect sinusoid by the time I reach 13.56MHz, or the 13.56MHz signal is modulated with the parasitic waveform and it gives me a waveform where the envelope follows the switching pattern with the parasitic waveform frequency while the carrier is at 13.56MHz, or some other crazy behavior.

So as it stands, I believe there is an issue with the RF Choke. I came across a "RF Prototyping Guide" from UC Santa Barbara the other day which stated that as a rule of thumb RF chokes for the frequency that I am operating at should be typically in the range of 100's of microhenries. However, I am finding it difficult to find inductors large enough that also meet the current capabilities as well as the resonant frequency rating that I need.

Any insight would be greatly appreciated.
 
Last edited:
Top