# Low calculated supply in Class E PA design

Status
Not open for further replies.

#### unreal2695

##### Newbie level 3
Hi:

I need to design a PA driving an IMD working at 13.56MHz. But due to the weak coupling between the TX and RX coil, the equivalent load impedance(including all ESRs) ends up around 6 Ohm. According to equation Vdd=(Rl*Pout/0.577)^0.5, the supply voltage is calculated to be only 500mV. Is my calculation correct or will the low supply be a problem in the PA (e.g. for the switching transistor)?

Thanks!

#### thylacine1975

##### Full Member level 6
Hey unreal2695,

Your question made me shudder... as a junior engineer I once spent *months* trying to design a *reliable* PA for driving a commercial 13.56 MHz RFID system. And... I failed

Let me pass on a lesson I learned the hard way and see if it (indirectly) answers your question... I'm not sure where the equation you describe comes from, but the low value of load impedance you mention is familiar. (As you know) these systems rely on power transfer between the TX and RX coil, and for reasons of small desired size, near-field operation and 'efficient' power transfer, loop antennas tend to dominate. At 13.56 MHz, the radiation resistance of practical loops is tiny and comparable (or swamped) by the ohmic resistances.

Being an over-enthusiastic radio amateur, I immediately tried to make my antenna look like 50 ohms via a variety of matching networks. Then, equations _like_ yours easily guide the choice of power transistor, rail voltages etc etc. and I could treat the system like a (familiar) HF radio system. Or so I thought. The problem is that the exact impedance presented by the transmit antenna varies WILDLY in response to the proximity of other objects, the receiver loading effects etc. This effect wreaked havoc with my carefully tuned matching networks, calling for complicated, hand tweaked (read: non-manufacturable) automatic tuning systems. Thankfully (?) we lost the contract before I had to deliver!

Adding insult to injury is that I (openly) mocked the suggestion of a senior engineer at the project outset that I drive the system with a class-B amplifier. You know, I think he was right.

I concluded relying on reactances and resonant effects with RFID systems was a BAD idea. If I had to do it all again, I'd treat the TX coil like a loudspeaker of a few ohms impedance and, uh, reach for a class B audio-amplifier inspired design
All that being said, if someone actually made a relying-on-resonances Class C/E/whatever design work, please post it here so I can learn where I went wrong!

unreal2695

### unreal2695

Points: 2

#### unreal2695

##### Newbie level 3
Hey thylacine1975:

Thank you very much for the answer. You are absolutely right in this application, the reactance seen by the PA can vary significantly and mostly due to the different coupling coefficient and loading. But I found some paper that makes a "robust" class E by means of transductor compensation (If interested, see this "Closed loop transductor compensated class E driver for inductive links"). Anyway I'm now thinking to make a class B amplifier like what you said.

Thanks again.

Status
Not open for further replies.