4
\$\begingroup\$

To be brief, I am as new as they get to creating SPICE models for components so I'm walking blind.

I was trying to create a SPICE model for the HSMS-2860 diode.

HSMS-2860 Schottky Diode Datasheet

Here is the .lib file I wrote for the diode, I got the parameters from the datasheet above.

.MODEL HSMS2860 D
+BV = 7
+CJO = 0.18E-12
+EG = 0.69
+IBV = 1E-5
+IS = 5E-8
+N = 1.08
+RS = 6.0
+VJ = 0.65
+XTI = 2
+M = 0.5

What I noticed is when I import this file into ANSYS Circuit Designer, it doesn't behave like a diode. When I apply a sinusoidal source to it (half-wave rectifier) I still encounter negative peaks on the output even though I have the peak input voltage amplitude below the reverse breakdown voltage of the model.

The goal of modeling this diode is so that I can determine a input impedance at a certain input power so that I can design a matching circuit to match that input impedance to my receiving antenna's impedance.

I'm sure the rabbit hole for SPICE modeling goes deep. I would appreciate some assistance with the matter.

\$\endgroup\$
3
  • \$\begingroup\$ What frequency excitation were you applying? \$\endgroup\$
    – The Photon
    Feb 17, 2021 at 1:10
  • \$\begingroup\$ @ThePhoton Sorry, forgot to mention that. I am exciting it with a 5.8Ghz Pure Sinusoid. \$\endgroup\$
    – EECE
    Feb 17, 2021 at 1:43
  • 1
    \$\begingroup\$ Two things that might be going on: The diode capacitance is high enough to give it low impedance at 5.8 GHz (this would happen if Ansys has different default values than typical Spice simulators); or you are doing an AC analysis that inherently linearizes the device behavior (this analysis simply doesn't capture rectifying behavior of diodes). I don't know enough about Ansys Circuit Designer to know which is more likely (or if there's some other pitfall you could have run into). \$\endgroup\$
    – The Photon
    Feb 17, 2021 at 6:43

1 Answer 1

4
\$\begingroup\$

While a practical design of an RF detector with the Schottky diode use distributed elements

enter image description here (see, for example, the AVAGO Technologies datasheet HSMS-286x Series Surface Mount Microwave Schottky Detector Diodes, Figure 21 5.8 GHz Matching Network for the HSMS-286x Series at 3 µA Bias), and these circuits should be developed using specialized design tools, the origin of "negative peaks" can be traced with a general-purpose SPICE simulator.

Sure there is always a reverse current through a diode, even for DC current, only it can be very small and you would not see it in your simulation plots. With AC current, a reverse current has two components: the reverse leakage current, which can be high for Schottky diodes when compared with p-n diodes, and the reverse recovery current, which is much smaller for Schottky diodes when compared with p-n diodes but still reveals itself when the signal waveform period \$T_{sig}\$ is on the order of Schottky diode's reverse recovery time \$t_{rr}\$ (switching time). The voltage sensitivity of HSMS2860 is halved at the frequency of 5.8GHz (ref value at 915MHz). At this frequency, the reverse recovery time effects become noticeable in the waveform of detected signals ("rectified" waveforms).

To mitigate the visible "negative peaks" in your "Half-Wave Rectifier" simulations, you can decrease a load resistance (impedance) in order to provide a sufficient reverse current.

For reference, the "Half-Wave Rectifier" circuit to be simulated:

OP's "half-rectifier"

, to easily see the effects of load, I first simulate your "Half-Wave Rectifier" with a 0.58GHz V1 source and 100 Ohm (R1) load:

enter image description here

You see "negative peaks" in graphs of both a voltage across the load and a diode current.

With 10 Ohm (R1) load:

enter image description here

"negative peaks" are noticeable, but become much smaller.

With 5.8GHz frequency (V1 source) and 10 Ohm (R1) load,

enter image description here

"negative peaks" are here again, but the current is 40 mA and I do not know if we can decrease the load without a risk of diode damage.

Summing up: the "negative peaks" are a legitimate phenomenon, it is a manifestation of finite reverse recovery time. You cannot get rid of it entirely. You have to only be aware of this phenomenon when designing your circuits.

\$\endgroup\$
2
  • \$\begingroup\$ Awesome! That was it, thanks a lot for your help. This is my first time using semiconductors at RF frequencies so I've got some reading to do. \$\endgroup\$
    – EECE
    Feb 17, 2021 at 20:58
  • 2
    \$\begingroup\$ @EECE To add a couple more details, the reverse recovery time in this instance is caused exclusively by the CJO parameter (junction capacitance). Set it to zero and the negative peaks go away. As mentioned in this answer, P-N diodes have a much higher reverse recovery time but it's primarily due to diffusion capacitance (non-applicable to Schottky diodes). The SPICE parameter for this effect is TT (transit time), and hence in Schottky diodes TT is set to zero which is also the SPICE default if you don't define it. \$\endgroup\$
    – Ste Kulov
    Feb 18, 2021 at 12:00

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.