[SOLVED] Flicker noise simulation for Schottky diodes

[saad]

Hello everyone,
I am trying to simulate the noise of a Schottky diode power detector. Theoretically, the flicker noise should approach zero if there is no dc bias but the simulation results I am getting using ADS are quite the contrary. Any comments and experiences are welcome.

PS: I am attaching the plot of Id and output noise vs the DC bias voltage.

 

[pancho_hideboo]

I am trying to simulate the noise of a Schottky diode power detector.
Theoretically, the flicker noise should approach zero if there is no dc bias but the simulation results I am getting using ADS are quite the contrary.

What diode model do you use for Schottky diode ?

Diode_Model (PN-Junction Diode Model) ?
**broken link removed**

PS: I am attaching the plot of Id and output noise vs the DC bias voltage.

It seems you use HB Noise Analysis.
It gives Cyclostationary Noise that varies as a function of time, but in a periodic fashion.
And Time-varying operating points modulate noise.

It is different from Noise results based on DC Analysis.

Show me HB Noise Analysis settings.

Y axis is Id. Is this Id Id[0] ?
X axis is HB.Vb. Is this HB.Vb a forward bias voltage ?

Where do you observe vout.noise ?
Do you execute signle noise frequency analysis ?
If so, what is noisefreq value ?

Show me schematic.

Show me results of Noise Contribution.

 

[saad]

Thanks for the reply.

What diode model do you use for Schottky diode ?

I am using Diode_Model from ADS library. It is a detector diode by Avago Technologies (HSMS-2860), where the model parameters are available in its datasheet.

Except the flicker noise parameters (Highligted in red box of attached image), which I have inserted after reading through various resources.

Show me HB Noise Analysis settings.

HB and Noisecon settings are in attachment.
RFfreq=2.4 ,
LOfreq=RFfreq - IFfreq
IFfreq=IFfreqKHz/1e6

IFfreqKHz is in KHz range.

Y axis is Id. Is this Id Id[0] ?

No, it is the diode current from DC analysis

X axis is HB.Vb. Is this HB.Vb a forward bias voltage ?

Yes.

Do you execute signle noise frequency analysis ? If so, what is noisefreq value ?

As it is a frequency converion circuit, I use IF Frequency for noise simulation (single point analysis).
However, Input frequency under Misc tab of Noiscon is RF frequency.

Show me schematic.

It is a diode with RC filter at the output and with input matching circuitry.

Where do you observe vout.noise ?

Noise is observed at the output node i.e. across the RC low pass filter

Show me results of Noise Contribution

See attachment.

 

[pancho_hideboo]

Show me HB Noise Analysis settings.

HB and Noisecon settings are in attachment.
RFfreq=2.4 ,
LOfreq=RFfreq - IFfreq
IFfreq=IFfreqKHz/1e6
IFfreqKHz is in KHz range.

Do you execute signle noise frequency analysis ?
If so, what is noisefreq value ?

As it is a frequency converion circuit, I use IF Frequency for noise simulation (single point analysis).
However, Input frequency under Misc tab of Noiscon is RF frequency.

Lower Heterodyne Down Converter.
You do Small Siganl Noise Analysis under two Large Signal Tones Drive.
So you had better set FreqForNoise slightly different frequency from IFfreqKHz.
For example, FreqForNoise=(1+1e-2)*IFfreqKHz kHz or FreqForNoise=(1-1e-2)*IFfreqKHz kHz.

Y axis is Id. Is this Id Id[0] ?

No, it is the diode current from DC analysis

You have to plot HB.Id[0] v.s. HB.Vb.
DC.Id v.s. Hb.Vb is not appropriate.

Show me HB.Id[0] v.s. HB.Vb.

Show me results of Noise Contribution.

See attachment.

There are contributions from two diodes.
http://www.hp.woodshot.com/hprfhelp/4_downld/products/diodes/hsms2850.pdf

Confirm validty of definition HB.Vb and Id.

 

[saad]

Lower Heterodyne Down Converter.
You do Small Siganl Noise Analysis under two Large Signal Tones Drive.
So you had better set FreqForNoise slightly different frequency from IFfreqKHz.
For example, FreqForNoise=(1+1e-2)*IFfreqKHz kHz or FreqForNoise=(1-1e-2)*IFfreqKHz kHz.

Only one of the tones (i.e. LO) is large signal (PLO=-5 dBm, PRF=-40 dBm) and now I can understand why was the flicker noise non-zero for zero Vb.

I have changed the noise frequency accordingly but I still do not understand, why do I need to change it. Could you please elaborate?

You have to plot HB.Id[0] v.s. HB.Vb.
DC.Id v.s. Hb.Vb is not appropriate.

Show me HB.Id[0] v.s. HB.Vb.

I am attaching the plot. Could you comment on the behaviour of noise as it goes up and then comes down before saturating. Should it not increase monotonically with increase in current?

There are contributions from two diodes.
http://www.hp.woodshot.com/hprfhelp/4_downld/products/diodes/hsms2850.pdf

Confirm validty of definition HB.Vb and Id.

I have two diodes.

Confirm validty of definition HB.Vb and Id.

I dont understand this bit.

 

[pancho_hideboo]

I am attaching the plot.

This is not mag(HB.Id[0]).
Show me mag(HB.Id[0]).

Only one of the tones (i.e. LO) is large signal (PLO=-5 dBm, PRF=-40 dBm)

You don't have to use P1tone for PRF if you treat it as small signal.
If you treat PRF as small signal source, use Term instead of P1tone.

I have changed the noise frequency accordingly but I still do not understand,
why do I need to change it.
Could you please elaborate?

[Two Tone HB Analysis]
IFfreq=0.6kHz
RFfreq=2.4GHz
LOfreq=RFfreq-IFfreq
FreqForNoise=IFfreq

NoiseOutFreq = NoiseInFreq + m*RFfreq + n*LOfreq
NoiseInFreq=RFfreq
m=0, n=-1
NoiseOutFreq=IFfreq

Frequency conversion matrix for noises could be singular at FreqForNoise=IFfreq.
It is not well conditioned for FreqForNoise=IFfreq.

[One Tone HB Analysis]
If you use Term as PRF,
NoiseOutFreq = NoiseInFreq + n*LOfreq
NoiseInFreq=RFfreq
n=-1
NoiseOutFreq=IFfreq
Frequency conversion matrix for noises could not be singular at FreqForNoise=IFfreq.

 

[saad]

This is not mag(HB.Id[0]).
Show me mag(HB.Id[0]).

In the above plot, I redefined Id as Id=HB.SRC1.i[0] where SRC1 was the DC source.

Nevertheless, I am attaching the HB diode current by placing a series current probe in diode branch.

 

[pancho_hideboo]

I have two diodes.
............................................
I dont understand this bit.

Show me schematic and definition of Vb and Id.

I am attaching the plot.
Could you comment on the behaviour of noise as it goes up and then comes down before saturating.
Should it not increase monotonically with increase in current?

See noise current models in **broken link removed**

Noise current increases with HB.Id[0].
However impedance of diode, "rd" decreases with HB.Id[0].

So noise voltage shows peak at some HB.Id[0] value.

See https://www.edaboard.com/threads/353969/#post1514681#5
However x-axis has to be 0, LO, 2*LO, 3*LO, 4*LO,....

For One Tone HB Noise Analysis where LO is P1tone and RF is Term, NoiseOutFreq=0Hz could be a singular point.

On the other hand, for Two Tones HB Noise Analysis where both LO and RF are P1tone, LOfreq=RFfreg-IFfreq, NoiseOutFreq=IFfreq could be a singular point.

 

[saad]

Super Thanks a ton.