# Measure impedance of rectifier using harmonic balance in ADS software

I am trying to use harmonic balance to find the impedance of the voltage double rectifier. After finding the impedance, I will use the impedance matching tool to find out the L matching.

As the pictures show, the frequency is 915MHz and power is 0dbm. I had got some result after simulation. Z0=50ohm, impedance of V1=Z0(4.108-j0.479), impedance of Vout=Z0(1-j7.163E-10).

But I am confused which point of impedance should be used to do impedance matching, V1 or Vout?

• The whole concept of impedance simply doesn't apply to nonlinear circuits. You can design a circuit that minimizes reflections, or minimizes harmonic output, but you won't be using the idea of impedance (of the rectifier) to do it. Commented Apr 6, 2019 at 16:00
• In any case, if you're still working on the energy harvesting project you posted about before, why do you need to minimize reflections from the rectifier? Your power levels will be tiny (or at least quite low), and an antenna isn't really subject to damage from reflections. Commented Apr 6, 2019 at 16:01
• Thanks. Yes, I am still working on that. It seems necessary to do impedance matching in order to maxmimize the output power. Every paper about rf energy harvesting I read have impedance matching Commented Apr 6, 2019 at 16:15
• I understand that the rectifier is nonlinear as it's impedacne vary as the different frequency and inputpower. Commented Apr 6, 2019 at 16:17
• Probably they ignore the rectifier and just match to the smoothing capacitor. Or just design empirically to maximize power transfer, with any explanation based on impedance just a post-facto justification for their design. Commented Apr 6, 2019 at 16:34

For the choice of energy harvesting, I see you selected a series capacitor that is smaller the expected capacitance of the diode, so it's impedance only affects the Q of the current at resonance and depending on the choice of diode ought to be around 1 Ohm with sufficient energy and higher with less.

The load R effects resonance as well as the ESR of the load Cap which demands the use of smaller shunt caps with SRF>> 1GHZ.

Thus your series resonant L is simply 0.3nH as expected from 100pF and 915MHz which is far too low.

Consider the impedances of C1 L1 and C2 with 1 Ohm diodes for maximum power transfer. What do you get?

I use the RLC nomograph for a ballpark estimate, then Falstad tools to add ERS, ESL, and sliders. Then compare with Frequency response then update model to include more parasitics. But a Smith Chart with scattering parameters ( s values) from Murata would be best for low ESL with 1:2 L:W ratio on SMD.

Here is my result of 30 minutes of design time ( pro bono) (free for what its worth) and another Sim link here

Your challenge is a design of an accurate stripline to mimic this circuit with a tradeoff for tolerance and selectivity given tolerances of dielectrics are 10% even with TDR testing on PCB maybe 5%.

Notice the Peak DC out matches the freespace Zo (377 Ohm source) peak voltage into R of 10 to 20k range.

• Thanks. So do you mean I need to replace the 100pf with larger value? Commented Apr 6, 2019 at 17:15
• No because 3nF is a trace of 3mm with 8:1 l:w ratio but I was saying 10uF is too inductive thus too low SRF and too high ESR. You have to reduce C1 << 10pF as I err'd in computing 3nH Commented Apr 6, 2019 at 17:19
• Thanks. If I ignore the rectifier, only consider the C1,C2,Load, what tools can I use to calculate the impedance or just use the equations in the handbook? Commented Apr 6, 2019 at 17:40
• resonance is the intersection of L1C1 lines on chart Q = Xc/Rs and Q= Rp/Xc for series parallel R's where |Xc|=|XL| at fo Commented Apr 6, 2019 at 18:11
• Thank you very much! Commented Apr 11, 2019 at 6:18