Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

[Moved]: spectre stb analysis of an autozero amplifier

Status
Not open for further replies.

babun123

Newbie level 5
Joined
Apr 28, 2017
Messages
10
Helped
0
Reputation
0
Reaction score
0
Trophy points
1
Activity points
84
I have an amplifier with autozero implemented. The overall amplifier is implemented as a switched capacitor amplifier. In the amplification phase, if I break the loop (put 'diffstbprobe' or 'cmdmprobe') and run tran-stb analysis, due to the introduction of 'gmin' from each node to ground by Spectre, there comes a zero at dc and a pole depending on the value of gmin and autozero cap. So, how do I get a proper estimate for the actual loop gain of the amplifier in this situation. Because of these zero-pole pairs (due to 'gmin' problem), the phase response and phase margin also deviates from the actual. So, how to get around this problem and get the proper loop gain and phase plot in Spectre?
 

You could change the default GMIN value (1e-12 , AFAIR) to a (much) lower value.
 

You could run a series of transient analyses (GMIN goes
away after DC solution in many simulators) and derive
magnitude and phase from the waveforms. That would
chew up some time but (especially if you are chopping
in reality) might be the most realistic especially if clock
products (incl mixing w/ input) in the output interest you.
 

AZ amp usually involves two amplifiers. While one is driving, the other is correcting. A cap is used to stored the offset information. Take a closer look at your AZ amp. Pretty sure you can convert AZ amp into a linear amp.
 

Re: [Moved]: spectre stb analysis of an autozero amplifier

You could change the default GMIN value (1e-12 , AFAIR) to a (much) lower value.

Actually with gmin=1e-12 and autozero cap Caz=1pF, the pole should be at (1/2*pi) Hz, whereas in loop gain plot, the pole comes at about 30Hz. If I reduce gmin down to 10^-20, the pole moves down to about 10Hz, whereas it should have moved to a much lesser frequency. I cannot understand where from this deviation in pole frequency is coming.

AZ amp usually involves two amplifiers. While one is driving, the other is correcting. A cap is used to stored the offset information. Take a closer look at your AZ amp. Pretty sure you can convert AZ amp into a linear amp.

Actually my amp is not a traditional autozero amp. I just modelled the problem as that of an autozero amp for better understanding. So, I cannot separate out the autozero cap from the amp for stb analysis.
 

I fear, the question is pointless without a schematic.

eda1.png

Please find the attached image.
 

You could change the default GMIN value (1e-12 , AFAIR) to a (much) lower value.

The problem is that the low frequency pole due to 'gmin' should ideally come at gmin/Caz but simulation result shows little higher pole frequency. As I reduced gmin from 1e-12 to 1e-20, the pole frequency reduced by only a factor of 3.
 

The problem is that the low frequency pole due to 'gmin' should ideally come at gmin/Caz but simulation result shows little higher pole frequency. As I reduced gmin from 1e-12 to 1e-20, the pole frequency reduced by only a factor of 3.

Isn't that the proof that this low frequency pole is not due to 'gmin', is it?
 

Status
Not open for further replies.

Similar threads

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top