The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Measurements >> RF Measurements >> IP2/IP3 Paper
https://designers-guide.org/forum/YaBB.pl?num=1031470045

Message started by Lieutenant Columbo on Sep 8^th, 2002, 12:27am

Title: IP2/IP3 Paper
Post by Lieutenant Columbo on Sep 8^th, 2002, 12:27am

Ken,
Thanks for the great paper on distortion. I was just getting ready to figure out how to simulate this on Spectre - and here is a step by step description.

You might be intersted in knowing that I believe I spotted a couple of typo's. The first are on page 4 - Table 1. It seems like the terms in parentheses (the delta P's) should be subtraction. Ditto for table 2. The other is in equation 30. The 'a' should be cubed.

Title: Re: IP2/IP3 Paper
Post by Ken Kundert on Sep 8^th, 2002, 12:29am

Thanks so much for you message. It is very nice to heard that you find what I wrote of value. I also really appreciate you taking the time to tell me about the typos. I have corrected the errors you found (all except for the last one) and updated the website. On the last one, I believe you are mistaken. If I changed a to a³ in (30) then it would be identical to (29), which would imply that iIP₃ and oIP₃ are the same, which in general they are not. When deriving oIP₃ we set Po₁=(a*alpha)² and when deriving iIP₃ we set Po₁=(alpha)². This results in a difference of a², which explains why (29) has the a³ and (30) has a.

Thanks,
Ken

Title: Re: IP2/IP3 Paper
Post by StoppTidigare on Mar 19^th, 2004, 5:09am

Hi all, I also have some questions about the IP2/IP3 paper.

Ken, you wrote that IPN= P+DeltaP/(N-1).

This is how I've understood it:
Say that one applies an input signal with power P to a nonlinear device. If I connect a power meter at the input I'll measure the power y1, which of course I can write as the function y1=x, where x is the input power in dBm. A linear function that starts at the origin.

If I measure the n-order component at the output it will certainly contain less power, and if I increase the input power it will increase as x^N.

If I linearise this function at x=1. I will be able to draw a tangent described by the line y2=N*x + C, where C is a constant to be determined.

The value of IPn is the point where the functions y1=x and y2=N*x + C cross.

This is done by finding the value of x where y1=y2, and then instert that value of x in either of the functions y1 or y2.

Starting first with the determination of C...

From your definitions of P and deltaP, we have
that the difference y1-y2 =delta_p at inputsignal x= P
This gives:
y1-y2 =Delta_p= x- (N*x +C), this is at x=P,so therfore
Delta_p=P-(N*P +C) <=>
Delta_p=P(1-N) -C <=>
C=-Delta_P + P(1-N)

So know we can write the function y2 as:
y2=N*x + P(1-N) - Delta_p

For which x are they equal ?
y1=y2 => x= N*x + P(1-N) - Delta_p <=>
x(1-N) = P(1-N) - Delta_p <=>
x= P - Delta_p/(1-N) <=>
x= P + Delta_P/(N-1)

What value y1 or y2 (IPN)is there at the crosspoint?
Plug x into the function y=x ,so

Am I correct ? I'm sure about the algebra but I've might misinterpeted the concept anyway.

You don´t show in your paper that IPN formula written above here is equal to formulas at page 4, for instance
iIP2=pi1 +(po2 - po12)

How should I proceed to derive that they are equal ?
(equal to IP2= P + Delta_P/(2-1)

Title: Re: IP2/IP3 Paper
Post by ibaiba81 on Aug 29^th, 2005, 8:06pm

hi i am new here.. can anyone tell how to simulate IIP3 in Hspice Cadence? Thanx :)

Title: Re: IP2/IP3 Paper
Post by ssk_288 on Jan 5^th, 2006, 9:05pm

Hi Ken,

I read your paper on IPn calculation. It saved me a lot of simulation time.

I am trying to calculate the 1-dB compression point of a LNA + Mixer. Is there any method other than the sweep power PSS analysis which would reduce the time of simulation for this anlysis also.

Thanks

Bye

Title: Re: IP2/IP3 Paper
Post by Ken Kundert on Jan 6^th, 2006, 9:53pm

iCP_1dB = iIP₃ – 9.6 dB

See section 4.3 in http://www.designers-guide.org/Analysis/intercept-point.pdf

-Ken

Title: Re: IP2/IP3 Paper
Post by ssk_288 on Jan 8^th, 2006, 6:22am

Hi Ken,

Thanks for the reply. But I have observed that this relation is not exactly satisfied when we separately do simulations for 1dB compression point and IIP3 calculation using the sweeping power in PSS analysis in spectreRF. I have observed the constant C in the equation
1dB = IIP3 - C
varies from 7 - 10 dB . What is the reason for this variation?

Thanks

SUDHIR

Title: Re: IP2/IP3 Paper
Post by Ken Kundert on Jan 9^th, 2006, 1:29pm

This equation is approximate as it assume that the third order nonlinearity dominates.

-Ken

Title: Re: IP2/IP3 Paper
Post by Patrick.B.Wu on May 9^th, 2006, 8:08am

Hi Ken,

I am new in RF stuff. Forgive me if I look stupid.

I am reading "Accurate and Rapid Measurement of IP2 and IP3" and comparing it with the Cadence on-line document "Vituoso SpectreRF Simulation Option Theory, Product Version 5.0". I am a bit confused on the formula to calculate IP3 with pss-pac analysis approach as I found they are different from each other in those two sources I mentioned above.

In source #1, we have IPn=P+delta(P)/(n-1), which is obviously correct by definition.
In source #2, we have IP3=VL1-(VS3+VS1)/2, which I have no idea how it comes.

Are we talking about the same approach? the same thing? and Are they consistent? Please give advice. Many thanks in advance.

-- Patrick, 9th May, 2006Α

Title: Re: IP2/IP3 Paper
Post by cybq on Jun 8^th, 2006, 5:20am

another important and useful equation:
Vimn=(Pin^n)/VIIPn^(n-1),

all the iterms are voltage, which are useful measurement within RFIC.

hope to be benifit to all.