Matching network between an unknown BJT and an unknown Tx antenna

[RoGeorge]

I have a TTL gate with open collector output (74N06 https://www.ti.com/lit/ds/symlink/sn7406.pdf ), and I want to use its transistor as a class C power amplifier for an AM modulator of only a few mW.  I have in mind the idea from Q3 onward, in this example schematic from the book "Build Your Own Low-Power Transmitters" by Rudolf F. Graf and William Sheets:



Except, in my schematic Q3 is not accessible.  Instead of Q3, I am using the open collector BJT from inside the 74N06 gate (in fact, the plan is to use 6 gates in parallel).  The carrier is fixed, 800kHz square wave, with 50% duty factor.

The transistor inside the logic gate is not characterized for RF, all I know is that it can withstand to max 15V and 20mA.


Another aspect, I would like to try a small loop antenna as seen in this video, but for a much lower frequency.  The antenna is a high impedance loop, and it will requires its own matching in two steps (at the antenna side).

I am linking the video just for the docs, doesn't necessarily need to be watched, it's about an antenna with a loop in a loop:  "Small Loop Antennas for FM / VHF / UHF Radio Receivers - Antenna Briefs #9" by MegawattKS



- Do I make 2 networks, one from Q3 to 50ohm, and another from 50ohms to antenna?
- I know only the general idea, to match the complex conjugate of the impedances for maximum power transfer, but how to do that for an unknown transistor from inside an IC?

Never did impedance matching before, where do I start? 

[rf-fil]

You can look up class C amp design formulas. There might be pre-made spreadsheets online as well. That should give you the starting point.

But, since this is only 800kHz, there's more: You can temporarily replace the output matching circuit components with variable versions and tune the circuit live, while observing the collector waveform with an oscilloscope, along with measuring the DC power consumption and the output power level. Then you can manually tune the circuit for best efficiency or best linearity, etc.

You can get large value tuning capacitors on ebay.

[profdc9]

As a rough estimate of the output impedance, you can calculate it using the power and a "fudge factor."  Recalling that for a DC load, V^2/R = P, we could estimate that

R = V^2 / P

Because the collector voltage isn't going to be 5 V the entire RF cycle, a fudge factor is usually used on the order of 6-10 because the actual output power is significantly lower, for example V^2 / 8 R = P and therefore R = V^2 / 8 P. 

so for example if the collector voltage is 5 V (for example for TTL), and the output power is 10 mW, the output resistance is R = (5^2)/(8 X 0.010) = 312.5 ohms.  Usually there is collector capacitance that adds some reactance as well that may have to be resonated out either with an inductor or using an inductive antenna.

You can design your output network to match this impedance to a 50 ohm antenna.  Alternatively, you can something like a 4:1 impedance transformer (2:1 windings) to get it somewhere within the neighborhood of 50 ohms.  With a higher turns ratio you will match a lower impedance better.  A variant of a circuit replaces L2 with the impedance matching transformer, rather than L2 being a RF choke (for example wound on a ferrite toroid like a transformer would be).  It can combine two components into one.

Since you are likely interfacing to an electrically small antenna (like a wire hanging off the circuit), the antenna will have a low radiation resistance and a high capacitive reactance, and you will probably want a very high turns ratio.  In order to resonate the antenna you need an inductor tuned to series resonate with the wire.  If you have a broken AM radio, the ferrite loopstick inside the radio is a good inductor you can salvage for this purpose.

[jwet]

RoGeorge- my only question is why?  Graf and Sheets used to publish a lot innovative stuff in Radio-Electronics magazine in the US.  I would look at their stuff and analyze their circuits and generally found them to be highly "optimized".  They couldn't be changed much and preserve their function.  What's wrong with the 2N3904 as the output stage.  Its operating class C.  What is your goal in this?

The problem with a 7405/6 OC driver isn't the output, its mostly a stout transistor, the problem is the input.  Its a TTL input- which is a emitter input, common base stage that will have a low driving impedance for RF and is not characterized for RF.  This will be the hangup.  Its interstage matching.

If you're trying to get more power, use a similar bipolar that can be heat sinked and goose its collector voltage a bit.  You can heat sink TO-92 transistors to get a bit more power.

On matching- If you don't have a Nano VNA, treat yourself to one.  They've saturated the market and you can get the H4 version for <$70 US- amazing.  Its a great learning tool and makes matching much more straightforward.

[RoGeorge]

RoGeorge- my only question is why?  Graf and Sheets used to publish a lot innovative stuff in Radio-Electronics magazine in the US.  I would look at their stuff and analyze their circuits and generally found them to be highly "optimized".  They couldn't be changed much and preserve their function.  What's wrong with the 2N3904 as the output stage.  Its operating class C.  What is your goal in this?

The goal is tinkering with radio.  For the fun of it, and because I have close to zero hands-on experience with RF.  Never matched an antenna, let alone designing a matching network.  In the last couple of days I have read a few chapters from the "RF Circuit Design" by Chris Bowick, and it is much clear now how to deal with matching networks.

(Side note, during this read I have realized how I can just raise the max voltage output of my AWG without a step up transformer, by simply tuning its 50\$\Omega\$ output to a much bigger, say 5k\$\Omega\$, load - that would give enough voltage swing to curve trace any Zener )


As for the why not a 2N3904, that's just me being silly.  I have found in the scrap boxes two TTL oscillator-cans on 4.9152MHz, from the 80's or so.  Hooked the PSU alligators to the GND/+5V pins, and the oscillators were working fine.   

Wanted to use them for something, anything.  4.9152MHz may seem a strange frequency, it's a multiple of 2ⁿ for RS232 baud rates.  Not interested to do any TTL RS232 project, but when divided by 6, the frequency matches one of the AM MW channels, 819kHz.  Here the MW broadcast band is mostly empty, has only 2 radio stations left, and not sure for how long (other EU states already discontinued their AM broadcasts, it is all FM or digital nowadays).

So, I thought I might make an AM modulator (for listening to audio books at the countryside) using "TTL ICs only", that's why trying to use a 74N06 instead of a BJT. 


Meanwhile, the more I read from "RF Circuit Design" by Chris Bowick, the less it makes sense to stick to that "TTL only" ad-hoc design goal.  I'll probably just use a discrete NPN instead of the open collector of a logic gate.

[mtwieg]

If the active device is an open collector output of a logic device, then it's not going to be acting as class B or C. For classes A, B, and C the input is sinusoidal, but with different conduction angles, and the collector current will also be sinusoidal (when the collector is conducting). But for a logic output it acts as a switch, not a current source. So you should instead aim to design a class D or E amplifier.

At such low frequencies you shouldn't need detailed RF parameters of the output device.

Make sure you mind the maximum output voltage of the logic device, you will need a collector bias voltage about 4x lower than that to avoid breakdown. Might be a good idea to use a clamping diode to prevent that while you tune things up.

[LM21]

The output impedance is quite close to the collector resistance. L2->R2. If you use it without resistor it breaks

But frequency and power are so low that you can forget conjugate matching and use a toroid transformer.

[RFDx]

I have a TTL gate with open collector output (7406), and I want to use its transistor as a class C power amplifier for an AM modulator of only a few mW.

Class-C is not possible but you can use the OC TTL gates as a switchmode amplifier in single ended configuration.  A 819 kHz diplexer network with Q ~ 4  is loading the amplifier at the fundamental frequency and harmonics with 50 Ohm. Collector voltage and current are therefore resembling a squarewave. With a 10V supply you get about 400mW into the 50 Ohm load RL. Resistor RD is both a safety resistor and the termination for all the harmonics. Amplifier efficiency is very good with very little dissipation in the gates, but overall efficiency is below 50% because RD has to dissipate half of the DC input power. Some additional lowpass filtering after the diplexer is recommended.

An even simpler configuration is shown in the second picture which will give you around 5mW with a 10 V supply. No need for any matching network, just a lowpass filter. Supply current is quite small (~22 mA @ 10V) and is limited by the two resistors. Efficiency is obviously terrible. AM modulation is easy to apply.

[antenna]

Check out this page and scroll down to figure 14.  This circuit works great for what you want and it wouldn't be too difficult taking a can out of a old radio and rewinding it for the AM broadcast band.  You could set it to any frequency that way and not be limited to crystals and dividing signals down.

https://www.nutsvolts.com/magazine/article/bipolar_transistor_cookbook_part_5