微波EDA网,见证研发工程师的成长!
首页 > 研发问答 > 微波和射频技术 > 天线设计和射频技术 > BJT frequency multiplier does not multiply

BJT frequency multiplier does not multiply

时间:04-04 整理:3721RD 点击:
Hi

The current task I have trouble with is building a bjt frequency doubler. Unfortunately I haven't been able to find a lot of practical hands-on design information but the circuit I built seems to be considered a working one in general. If someone knows a hands-on design guide somewhere please point me to it.

Right now the input is 13.5 MHz 1 V p-p and the output is supposed to be 27 MHz but I don't get anything usable out of it. The best I see is a distorted and much attenuated input frequency. First I had fixed resistors for the base bias whose values worked ok in a previous non-multiplier amp but when it made trouble I also tried 10k, 25k and 50k trimmers. The inductor is self-wound and should be in the right ballpark, however I also exchanged the resonant capacitor for slightly different values just to make sure. I did use a bread-and-butter 2222 transistor on purpose; although it might not be the best choice the working frequency is still way below the fT of 300 MHz.

Where's the pitfall? There must be one but I just don't see it.

Regards
Martin

Do you check the signal by an oscilloscope ? You cannot see the second harmonic but there is.
You should use spectrum analyzer.

FFT is what I can use right now and sure enough the second harmonic is there about 8 dB below the first. However I may not have correctly understood the concept of a frequency multiplier. I thought the tank circuit's purpose was to select the second harmonic (in this case) over the others. Is that a wrong assumption? Because if it did then I'd expect the second to stand out from the others.

Did you verify that the LC tank is resonating at 27 MHz? By design of the circuit, there will be surely a certain amount of fundamental. You should also consider that the tank is only selective with a relative large load impedance in the kOhm range.

Frequency Multiplier is basically to create Nth harmonics from a active device by proper biasing.So the circuit should create large harmonic contents by choosing a heavy Nonlinear Region of its Operating Point -for instance exponential region- of a Vbe transfer curve of a BJT or anything else such as.Even a diode can create harmonics but the second ( and other higher degree) harmonics cannot be higher than fundamental.The wanted Nth harmonic is filtered -for instance by a tank circuit as in your schematic- and the others are eliminated.I have checked your tank circuit and I couldn't find 27MHz resonance frequency ( I assume 12.75uH // 39pF).
I think you have to change the resonance circuit with high Q tank circuit elements ( Lowest possible L value, Highest possible C value) and take care the load impedance because it's connected directly in parallel to the tank circuit so it will degrade the Q factor so the selectivity.

Your schematic is more useful as a x3 frequency multiplier.

You can make the filter more selective and add a high input impedance amplifying stage to eliminate loading of the resonator.

I find I have this x2 simulation which is similar to your schematic.
The input should consist of narrow pulses. The wider its duty cycle, the more output resembles a square wave.



So, perhaps my post #6 (re x3) applies to a different topology.

Here are the relevant wave-forms of the circuit.

from this, a rectified sine wave will have a lot of 2 * Fin at the output:
http://www.utdallas.edu/~raja1/EE%20...lp/theory.html



So adjust the DC bias to form that output waveform. Also make sure you are DRIVING the transistor base with a big enough sine wave too!

You are trying to AVOID making the output look like a square wave, which would instead have a lot of 3 * Fin content.

Thanks for all your answers guys! I wasn't aware of some things and now focused on E-designs input. I was able to replicate the simulation in LTspice very well but in the real world no way.

I built the exact circuit with the exception that I didn't have any J310 but J112 and no 180 Ohm but 200. The tank circuit does resonate on 27 MHz now (yes at the beginning it was a bit off but it was 12.75 windings not nH), I checked with the signal generator and scope method, looking for the peak. This is also the only way I can do it I guess. Of course I had to temporarily solder the coil onto another pcb but as soon as it went onto the original pcb with the rest of the circuit (that I made in the meantime) the output was utter crap again with no trace of a selected second.

At the moment I have no idea what to even try. Any suggestions?

This is what I see. Hopefully the measurement makes sense and I didn't screw it up. Because the first harmonic is even a little bit stronger than the second I guess the selectivity of the tank circuit is somehow missing altogether. Furthermore it looks like the signal was very noisy too.

I didn't hear yet an answer about the output load, although the problem was addressed in many posts.

Another approach to make a frequency doubler with good fundamental suppression is to use any kind of full wave rectifying circuit.

Right. In this circuit I think the jfet buffer amp is to provide the high load, is it not? But I did try to load the output with 1, 10 and 100k anyway but with no obvious difference. Right now there's no further circuit behind yet to load it.

Just a minute ago I found one possible cause or mistake I'm not sure yet: It seems that it has to be driven with more power. If I turn the generator up to 6 dBm which is its maximum the output starts to look much better. First I thought that over 3 Vpp was massive however now I think that I probably have a measurement error or mistake somewhere because it seems to me that 6 dBm should in fact be less than 3.2 Vpp. Methinks this will be an impedance issue.

I first need to find more info on the full wave rectifier approach, might be worth considering though. I think however that those need transformer in and out which is a bit a downside.

Another topology of frequency doubler. Made from a phase splitter. Output frequency does not derive from ringing or oscillating LC loops. Output frequency is an exact integer multiple of input frequency.



Filtering is needed if you want sinewave output.

I tested the design and made e few changes to get better rejection of the fundamental.

* I added a 150 pF trimmer over the first coil to tune it better
* Changed the 2.2 uH to 1 uH and added the same 150 pF trimmer

I tested it with a 2N3904 because that is what was at hand. The rest of the circuit is the same.

The circuit performed well, and the fundamental was rejected by about 34 dB.

One thing that is not shown on the diagram is the RF bypass capacitor (100 nF) between supply and GND. Make sure you have one in place.

It works well with a square wave input as well.

Below is the gain plot for the physical circuit. The gain is somewhat higher with the modifications than predicted by the simulations of the original circuit. The output impedance is somewhere between 500-600 Ohm.

This is the simplest and one of the most efficient frequency doubler:
http://www.qsl.net/4f5aww/images/diodes-fig4.gif

To get some gain and better 2nd harmonic selectivity you can use this:
http://upload.wikimedia.org/wikipedi...ubler2.svg.png

Copyright © 2017-2020 微波EDA网 版权所有

网站地图

Top