The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> SC filter: opamp clipping problem
https://designers-guide.org/forum/YaBB.pl?num=1122922313

Message started by Visjnoe on Aug 1st, 2005, 11:51am

Title: SC filter: opamp clipping problem
Post by Visjnoe on Aug 1st, 2005, 11:51am

Hi,

I'm currently trying to simulate a (model) of a SC (switched capacitor) integrator. In order to do this, I modeled the opamp by the most simple behavioral model (so excluding slewing etc. for the moment), namely:

Vout = Avo*(Vinp-Vinn) (single-ended opamp)
with Vout_max = VDD and Vout_min = 0

However, when I include this 'clipping' I can see in (transient) simulation that the opamp virtual node exhibits large voltage steps, but the node does not return to the initial value (let's call this VREF). The opamp gain is however large enough (+100dB)

Okay. Now, if i remove the opamp output clipping from the model, the transient simulations do show nice, virtual node waveforms (as presented in textbooks :)).

I can only assume that the feedback around the opamp does not work when you introduce a 'hard' non-linearity in the loop (the clipping).

So, now, that's all fine, my simulations agree with textbook pictures, yet, I must still be missing something, since, in reality, the opamp WILL exhibit clipping.

Does anyone see the flaw in my reasoning (or model)?

Kind regards

Peter

Title: Re: SC filter: opamp clipping problem
Post by Paul on Aug 1st, 2005, 2:05pm

Hello,

The first equation you use is only valid in the linear region of your Vout/Vin transfer function. When the opamp output is clipped, the amplifier input can no more be considered to be a virtual ground, because it is not operating in the linear region. You cannot calculate the input voltage difference as a function of the output voltage divided by the gain, only the output voltage you would obtain from the existing input voltage in the absence of clipping.

This also means that your input signal swing seems to be exceeding the specifications of your amplifier, i.e. the integrator will not operate correctly with the voltages you apply.

Paul

Title: Re: SC filter: opamp clipping problem
Post by Visjnoe on Aug 2nd, 2005, 1:03am


Thanks for your answer, Paul!


To me, it seems that (in reality, so assuming a transistor-level opamp implementation) a large input step to the integrator (Vstep > sqrt(2)*(Vgs-Vt) with Vgs-Vt the overdrive of the input pair) will cause the integrator's opamp to slew.

So, the integrator will be functional/disfunctional, depending on the allowed settling time.

Kind regards,

Peter

Title: Re: SC filter: opamp clipping problem
Post by Paul on Aug 2nd, 2005, 4:55am

Peter,

I don't see the relationship between clipping and slewing... Clipping is a DC spec, slewing a time-domain spec. Can you explain?

In the time domain, the slew-rate and settling time of the opamp must obviously comply with the sampling frequency specifications.

Paul

Title: Re: SC filter: opamp clipping problem
Post by Visjnoe on Aug 5th, 2005, 12:59pm


Hi Paul,

maybe my explanation was a bit confusing.
As you state, clipping is a DC spec and slewing is a
transient/time-domain spec.

You state however that if the integrator's input goes beyond its designed range, it will clip. I just wanted to add, that this can indeed happen, but it may also be that it will just experience a certain period of slewing before settling.
The slewing will occur due to the large input signal, which causes typically all the bias current of the OTA differential pair to be steered to one side.

Concluding: I agree there's no relation between clipping and slewing, but a 'large input signal' can lead to both.


Kind regards,

Peter

The Designer's Guide Community Forum » Powered by YaBB 2.2.2!
YaBB © 2000-2008. All Rights Reserved.