An ideal opamp goes tosaturation,no simulation ...!

Thread Starter

as21462

Joined Dec 29, 2006
22
Hi guys,..
I have a very simple coursework in Pspice, but because its the first time Im working with Pspice, it seems complicated to me.

I should investigate an inverting amplifier by an ideal opamp and some other staff.
One of the tasks is that by changing the resistor in feedback loop(using parametric sweep) after a certan value of R pspice will stop simulating,thats when it goes to saturation,..the lecturer asks us what can be done to solve the problem....actually no body knows what does he exactly want...we know that by reducing the voltage source of the circuit or even changing the bais voltage of inside of the opamp the problem can be solved, but in both of them we prevent the opamp to go to saturation,...is there any other ways to solve the problem and make the spice do the simulation and so show the clipped signal?!?....

or maybe can anyone introduce me a website which can tell me how does spice work in this problem,..maybe by finding out how does spice work,I can find out what to change in soice settings!

thanks,
as21462
 

Papabravo

Joined Feb 24, 2006
21,159
Hi guys,..
I have a very simple coursework in Pspice, but because its the first time Im working with Pspice, it seems complicated to me.

I should investigate an inverting amplifier by an ideal opamp and some other staff.
One of the tasks is that by changing the resistor in feedback loop(using parametric sweep) after a certan value of R pspice will stop simulating,thats when it goes to saturation,..the lecturer asks us what can be done to solve the problem....actually no body knows what does he exactly want...we know that by reducing the voltage source of the circuit or even changing the bais voltage of inside of the opamp the problem can be solved, but in both of them we prevent the opamp to go to saturation,...is there any other ways to solve the problem and make the spice do the simulation and so show the clipped signal?!?....

or maybe can anyone introduce me a website which can tell me how does spice work in this problem,..maybe by finding out how does spice work,I can find out what to change in soice settings!

thanks,
as21462
This seems relatively straightforward. The gain of the inverting circuit is minus the ratio of feedback to input resistance. If you define the input as a function of the feedback, then sweep the feedback resistor then you are doing the simulation at constant gain. Now by adjusting the gain you can approch saturation and see if it is the value of the resistor (large values equal vanishingly small currents) that is the source of the complaint or the gain itself. In the latter case I have no further suggestion, unless you can find a way to make the opamp "less than" ideal.

You could also try LTSpice. It stops not for the tide, nor any man.
 

Distort10n

Joined Dec 25, 2006
429
I have attached a simulation using TINA and I do not receive an error. All flavors of Spice are based on Berkley Spice from ages ago and each will have their quirks.
It could be as simple as PSpice not being able to find the DC operating point. Even with an AC analysis, Spice will calculate the DC operating point first. I am not sure about the 'logic' during a sweep.
You can try the Yahoo users group for LTSpice/SwitcherCADIII. It is a very active user group for Linear Technologies version of Spice:

http://tech.groups.yahoo.com/group/LTspice/

It is not about PSpice but the members may be able to offer some degree of help. Their LINKS section has a tremendous amount of information so you may want to read through that first. In fact look here:

http://dave.uta.edu/dillon/pspice/index.php

http://www4.ncsu.edu/~mbs/freeda_documentation/other_freeda_docs/SPICE/spice.pdf

You can literally get lost in their library of links about Spice.
 

Attachments

Thread Starter

as21462

Joined Dec 29, 2006
22
Thanks,..but we have to do it in spice,....I have attached a picture of the error it would give me,...when I change the first four boxes of the Pspice runtime settings, it will work, but very bizzar plots, that I think its just by ac accident...!!!
 

Papabravo

Joined Feb 24, 2006
21,159
Thanks,..but we have to do it in spice,....I have attached a picture of the error it would give me,...when I change the first four boxes of the Pspice runtime settings, it will work, but very bizzar plots, that I think its just by ac accident...!!!
LTSpice is just a spicy as any other spice. If it works properly then you can send the nice folks at pSpice a bug report. You would do well not to be contemptuous of the free advice you are offered. BTW there was no attachment. We call this attachmenesia.
 

JoeJester

Joined Apr 26, 2005
4,390
If your using the ideal op amp, I suspect you should never reach saturation. The output will always be -Vin * (Rf/Rin). If it doesn't, it's not ideal.

Now if you use a model other than ideal, you will go into saturation when the -Vin * (Rf / Rin) approaches the rail voltages, except when using the rail-to-rail op amp models, then -Vin * (Rf / Rin) must exceed the rail voltage.

Which version of pspice are you using?
 

Distort10n

Joined Dec 25, 2006
429
Hmmm...I didn't catch the ideal amplifier being used. This could be the problem too. TINA craps out at -10GV when using an ideal amplifier. Rf = 1T ohm and Rg = 500 ohm. The usual "Irregular Circuit" or "DC operating point not found."
 

JoeJester

Joined Apr 26, 2005
4,390
I just used Rf of 1T and Rg of 500 and the sine wave of 2V pk-pk at 1kHz. Tina works fine.

I'm using Tina version 7.0 Educational, XP media, 1GB ram, and a 1 GHz intel core solo processor.

Sometimes I start my transient analysis at zero volts, but it worked both ways .... starting with zero and calculating operating point.
 

Distort10n

Joined Dec 25, 2006
429
What do you get if you try Analysis -> DC Analysis -> Calculate Nodal Voltages?

In transient you can zero the intial values and it works fine.
 

Thread Starter

as21462

Joined Dec 29, 2006
22
I have found sth of solving the convergence problem,...and it works but by less accuracy,...but I dont know what exactly Im doing,...I changed Reltol,abstol,ventol....I multiply them by 10000,....I think its correct,...what do you think?!
 

JoeJester

Joined Apr 26, 2005
4,390
That is one way of solving a convergence problem as I have done in the past with the TINA simulation software. As far as being correct, you can compare that with your calculated values to see whatever error exists.

The DC analysis posted by knightofsolamnus is in the TINA program, which a free version pspice based program offered at TI.

Of course there are "paid" and "student" versions at http://www.tina.com
 

Distort10n

Joined Dec 25, 2006
429
I have found sth of solving the convergence problem,...and it works but by less accuracy,...but I dont know what exactly Im doing,...I changed Reltol,abstol,ventol....I multiply them by 10000,....I think its correct,...what do you think?!

I have been told that you can change GMIN to a larger value to help solve for convergence problems as well.
 

JoeJester

Joined Apr 26, 2005
4,390
knightofsolamnus,

As far as nodal voltages, when using a DC source of 1 volt, I get -2G at Vo and 0 at the Rf - Rin junction with the ideal opamp. Of course you'll get zero DC when using an AC source at those nodes.
 
Top