The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Applying same MC mismatch parameter to all instances of a circuit in a test
https://designers-guide.org/forum/YaBB.pl?num=1435715118

Message started by SimonC on Jun 30^th, 2015, 6:45pm

Title: Applying same MC mismatch parameter to all instances of a circuit in a test
Post by SimonC on Jun 30^th, 2015, 6:45pm

Hi, I am simulating an opamp. In my test setup, I have 6 instances of the same opamp to validate all the parameter at once. Is there a way to tell a Monte Carlo simulation (Cadence ADEXL) to apply the same mismatch parameter to all instances?

In my case it is important because one of the instance serves to calculate the input offset and that value is used to make sure the open loop AC simulation on another instance stays within the correct output range by cancelling the offset. The problem with the Monte Carlo analysis is that all the open loop measurements are completely wrong because the offset between the two instances are different. I found how to specify which device to include/exclude, but I didn't find a way to tell the simulator that all my OpAmp instances are in fact a single instance.

Title: Re: Applying same MC mismatch parameter to all instances of a circuit in a test
Post by sheldon on Jul 1^st, 2015, 1:30am

Simon,

First to make sure I understand your approach, you are using 6 test
benches to measure six parameters effectively. In one test bench you
measure the offset voltage and pass it to another test bench to cancel
the offset voltage so the output is not saturated when you measure
the open loop gain.

Some comments,
1. Usually for an op amp one test bench is sufficient. You can measure
many parameters from a single test bench for an op amp.
2. In general, the method you are using to measure the open loop
gain is not recommended. There is a good treatment of the topic
in Ken Kundert's book, "The Designer's Guide to SPICE & Spectre",
Chapter 3. Technically these really focus on loop gain.
3. The simplest method is to measure the open loop gain in your
closed loop test bench, vac(out)/(vac(ii_noninverting)-vac(inverting).

Sheldon

Title: Re: Applying same MC mismatch parameter to all instances of a circuit in a test
Post by SimonC on Jul 1^st, 2015, 7:12am

Sheldon,
I am using a single test bench that contains 6 instances of the same DUT. Each instances are arranged slightly differently to measure all the parameter I want for that DUT. The measured offset from one of the instances is passed to the other open loop configuration through a vcvs source to make sure they don't saturate.

I just bought "The Designer's Guide to SPICE & Spectre" as it should be a good book to improve my simulation in general

I don't understand your third point. vac(out)/(vac(ii_noninverting)-vac(inverting)) under feedback will only allow me to find the BW of the amplifier for the feedback gain. But it doesn't allow me to get the DC open loop gain. Can you elaborate on the limitation of that method?

Title: Re: Applying same MC mismatch parameter to all instances of a circuit in a test
Post by SimonC on Jul 1^st, 2015, 8:14am

Sheldon,
I just found the feedback needed to use your method (3). It is working now.

Title: Re: Applying same MC mismatch parameter to all instances of a circuit in a test
Post by sheldon on Jul 1^st, 2015, 5:00pm

Simon,

For reference.

Sheldon

Title: Re: Applying same MC mismatch parameter to all instances of a circuit in a test
Post by Frank Wiedmann on Jul 2^nd, 2015, 12:47am

For loop gain simulation in Spectre, I would recommend stb analysis. A lot of information on loop gain simulation is available from my webpage https://sites.google.com/site/frankwiedmann/loopgain.