Dear NFC community members,
I am working on designing a small RFID reader circuit that is based on the PN532 chip and I would like to get some feedback. My background isn't RF engineering, so I am not sure about many things.
I'll summarize the main questions first, and then describe in details what I've done:
I am using miniVNA Tiny for taking measurements.
Q1: which parameter (SWR, RP, RL) should I look at for figuring out the self resonance of a circuit?
Sometimes the sign flip of RP correlates well with the minimum of RL and SWR, but not always.
Q2: Why after tuning my antenna to 13.56 Mhz (based on SWR or RL measurements) do I see a huge shift in resonance frequency when I bring a tag close to the antenna? Typically, two peaks appear: one at ~12.8Mhz and one at ~14.3Mhz? Is this normal?
Before:
After a tag is nearby:
Full story below (and more questions below):
I basically followed the reference design and white papers describing how to tune the antenna.
The circuit has the EMC module (560nH, 220pF), series caps (C1) and parallel caps (C2).
At first, I tried to follow the general guidelines which were:
1. Measure antenna inductance (La)
I did this by measuring the reactance at 1Mhz. La = Xs / w0 = 1.6uH
This seems reasonable.
2. Measure antenna self resonance.
I looked either at the sign flip of RP or SWR.
I noticed multiple peaks, which one should I consider to be the resonance?
3. From the self resonance, estimate the antenna capacitance:
f0=1e6; % measured at 1 Mhz
w0 = 2*pi*f0;
La = Xs/w0;
Fself = 472270937;
Ca = 1/((2*pi*Fself)^2 * La);
4. Calculate C1 and C2.
I followed equations from white paper:
Ra =7.7; % Measured at 1 Mhz
Xs = 9.8; % Measured at 1 Mhz
nfc_freq = 13.56*1e6;
w = 2*pi*nfc_freq;
Qa = w*La/Ra
Rmatch = 30;
Rq = max(0, 0.5 * (w*La / Rmatch - Ra))
% in my case, no series resistors were needed. Q was ~17
% EMC filter:
L0 = 560 * 10^-9; % 560 nH
Femc_cutoff =14338865.060467; % 14.4 Mhz
C0 = (1/(Femc_cutoff*2*pi)).^2/Lemc; % 220 pF
% Calculate the series equivalent
Xtr = 2*w * ( L0 * (1-w^2*L0*C0) - Rmatch^2/4 * C0) / ((1-w^2*L0*C0)^2 + (w*Rmatch/2*C0)^2);
Rtr = Rmatch / ( (1-w^2*L0*C0)^2 + (w*Rmatch/2*C0)^2);
% Calculate the parallel equivalent circuit:
Lpa = La;
Cpa = Ca;
Rpa = (w*La)^2 / (Ra + 2*Rq);
C1 = 1/( w* (sqrt(Rtr*Rpa/4) + Xtr/2));
C2 = 1/(w^2 * Lpa/2) - 1/(w * sqrt(Rtr*Rpa/4)) - 2*Cpa;
------
The final result was:
For C1 (26.33), use 4.70 + 22.00 (=26.70), error = 0.370680 pF
For C2 (149.62), use 0.00 + 150.00 (=150.00), error = 0.380073 pF
I tried out these values, but the end result was really out of range.
Did I miss something out in my calculations?
I then played around by changing caps (without the EMC bandpass filter).
I noticed two major trends:
1. When I increase C1, I get a narrower profile (smaller arc on Smith Chart)
2. Increasing C2 generally doesn't change the radius of the arc much, but shifts the resonance frequency downward.
I ended up with this:
Does this look reasonable? If so, why do I get such a huge detuning when a tag is near by?
