The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Modeling >> Behavioral Models >> integrator abnormal behavior
https://designers-guide.org/forum/YaBB.pl?num=1319581083

Message started by NTR on Oct 25^th, 2011, 3:18pm

Title: integrator abnormal behavior
Post by NTR on Oct 25^th, 2011, 3:18pm

Hi,
I am implementing integrator with UGBW:2Ghz, Dc gain:46dB, Fc:10MHz. by the help of LPF H=1.257e010/s + 6.283e007.
then Im running " lsim(H,input,time)" to see integrator response for arbitrary input with time.
I am getting response as shown in fig.
The blue one is integrated signal and down blcak signal is my input.

Here my doubt is what causes such behavior. the signal saturates after particular period of integration instead of continuing integration of the signal.

Thank you
/NTR

Title: Re: integrator abnormal behavior
Post by raja.cedt on Oct 26^th, 2011, 7:35am

hello,
could you please explain where did you model, like matlab or spice..? One reason could be there may be some max value on the o/p, which you are observing like saturation, generally integrators can't be run in open-loop..

Thanks,
raj.

Title: Re: integrator abnormal behavior
Post by NTR on Oct 27^th, 2011, 3:23am

Hi Raj,

I Model this using MATLAB.
I think it is independent of input amplitude & it depends on settling time of filter( 1/F_cuttoff). After this time(1/F_cuttoff) LPF is saturating.

Here I am using integrator in closed loop only.

How can I integrate all my input pulse train even after settling time (saturation ) of filter?

Thank you
/NTR

Title: Re: integrator abnormal behavior
Post by boe on Oct 27^th, 2011, 9:40am

NTR,
I'm guessing you don't have a pole at DC; I didn't try it, but I suspect you use H=1.257e010/(s + 6.283e007). This is a low pass, not an integrator.
- B O E

Title: Re: integrator abnormal behavior
Post by NTR on Oct 27^th, 2011, 9:50am

Hi BOE,
ya this is low pass filter. In general lpf acts as integrator at higher frequencies. But I wonder how people are using then LPF as integrator?
from my understanding I think LPF acts as integrator untill it gets saturated(up to settling time of low pass filter).

Thank you
/NTR

Title: Re: integrator abnormal behavior
Post by boe on Oct 27^th, 2011, 10:23am

NTR,
sorry that my first answer was apparently not elaborate enough:
Of course, any practical (real) implementation of an integrator will be lossy, i.e. it will not have infinite DC gain. However, a gain of 200 or so is far from a good implementation for an integrator (i.e. it has significant non-idealities).
If the DC gain is large enough, saturation will be the limiting factor; however, your model (just the transfer function) does not include saturation, so you get the finite DC gain as limiting factor.
- B O E
[PS:] I called it a low-pass filter to highlight the non-idealities of the integrator.

Title: Re: integrator abnormal behavior
Post by NTR on Oct 27^th, 2011, 10:49am

BOE,

Is there any way to choose best lpf(i.e with low cutoof freq) to maintain UGBW 2GHz with reasonable DC gain. I know that UGBW=DC gain* Fc. Here I have more freedom in choosing Fc. but how can I choose best one?.
(I am using this integrator in non coherent energy detectors to estimate energy of received bit where input for this integrator is signal spans in range of 0-2GHz ).

Title: Re: integrator abnormal behavior
Post by boe on Oct 28^th, 2011, 2:01am

NTR,
essentially this is a question of trade-off between effort and cost (area, power, ...) for building a better integrator vs. achieved/required performance of your system.
A (first-order) low-pass filter acts as an integrator over time intervals Δt ≪ τ, so you need to figure out over what time you have to integrate to achieve the performance you need.
- B O E