NFC field generator

[judge]

I knocked up a LTSPice model from the AO4294 datasheet and this web site, and the results of the simulation were pretty spectacular, so I need to make up a bread-board friendly adapter for it and try it out.

[judge]

This is what I have so far. Naturally it doesn't work. If I connect the source resistor (R3) the signal at pin 8 of the 74LS00 disappears. If I increase the value of R3, the amplitude of the signal at pin 8 will increase too.

Before I abandon this and explore other oscillator circuits, I would really like to understand why connecting the source of the MOSFET has this effect.

[PA0PBZ]

I would really like to understand why connecting the source of the MOSFET has this effect.

The input capacitance of your mosfet is about 2.4nF or 5 Ohm at 13.5 MHz

[judge]

On to a Pierce oscillator, or at least that what this is supposed to be. The oscillation dies down after about 10s, but until then I am getting good range from the coil (L1), with the LED tags lighting up at about 1cm from the coil.



The values of RFC, C2 and C3 are pure guesswork. 'RFC' is not really an RFC, it is just an inductor I happened to have lying around. C2 I set to 2x the load capacitance, minus a bit for stray capacitance. I also tried 1x the load capacitance and I also tried nothing - it doesn't make any difference. C3 was just a capacitor I happened to have lying around.

So I have a couple of questions:

• How do I calculate what values these things should have?
• Why do the oscillations die out?

[judge]

I think the answer to (2) is that the mosfet overheats...

[T3sl4co1l]

Would guess it's marginal and temp rise causes gm loss.  Crystal has much too high an impedance (~kohm) for this (the transistor is ~ohms between its terminals), and even without matching network(s) the drive level is too much for most crystals (i.e. it's likely to shatter).

Replace the crystal with an L+C and tune as closely as you can?  A buffer circuit would be neat but would require more components and matching LCs to realize.

Tim

[judge]

Would guess it's marginal and temp rise causes gm loss.  Crystal has much too high an impedance (~kohm) for this (the transistor is ~ohms between its terminals), and even without matching network(s) the drive level is too much for most crystals (i.e. it's likely to shatter).

Replace the crystal with an L+C and tune as closely as you can?  A buffer circuit would be neat but would require more components and matching LCs to realize.

Tim

Thanks for nudging me to use L+C, I was already concerned that I was probably abusing the crystal. Part of me wants to get something going with the crystal - so I can understand what I am doing. At the moment I am just blindly following recipes, so I stand no chance of figuring something out if it isn't working right. On the other hand, I'm pretty sure that using L+C is what I would end up doing anyway. I doubt that the exact frequency matters much in this application. I guess I'll find out!

[judge]

As I now understand it, the temperature rise is because the MOSFET is never turned fully on. If I replace the crystal with a series L+C resonant pair, I get the same behavior. If I replace it with a parallel resonant pair the MOSFET runs cool, but it oscillates at around 4MHz instead of the 13.56MHz the pair are tuned for. This is with a supply voltage of 5V.

I'm going to also try some JFET-based oscillators (using the J310), but maybe I should also play around with RF amplifiers like the BGA2866?

[Berni]

You likely want the oscillator and power output stage to be two separate circuits anyway.

Using logic gates around crystal is a pretty good way for making a stable oscillator. But a MOSFETs gate needs a fair bit of current to turn on and off that quickly. So you need a extra logic buffer between your oscillator and the transistors gate. A common trick is to parallel multiple gates of a logic buffer or hex inverter chip to get more output current capability.

The MOSFET choice seams pretty reasonable as it does have fairly low gate capacitance for how powerful it is, but that is still 2.4nF and so still needs a fair bit of current to switch at >10MHz

Also take care with doing >10MHz RF work on a breadboard. The crappy parasitics of it can certainly make this fast analog circuits act weird.