|
The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl Design >> Analog Design >> Opamp spec for switced capacitor filter https://designers-guide.org/forum/YaBB.pl?num=1213163106 Message started by manodipan on Jun 10th, 2008, 10:45pm |
|
Title: Opamp spec for switced capacitor filter Post by manodipan on Jun 10th, 2008, 10:45pm Hi Guys, I have to design a opamp based switched capacitor filter having some Fclk(clk frequency) and maximum signal frequency spec.How to derive the opamp spec(DC Gain,GBW product ,Slew rate) for a certain load capacitance?can u guys suggest some good reference on that?? |
|
Title: Re: Opamp spec for switced capacitor filter Post by buddypoor on Jun 11th, 2008, 3:12am Hi manodipan, perhaps the attached document can help. Regards and good luck. |
|
Title: Re: Opamp spec for switced capacitor filter Post by manodipan on Jun 11th, 2008, 3:39am Hi Buddypoor, Thanks a lot for the thesis.I have another doubt:In SC filter design we generally design one continuous time Opamp based circuit first from the filter specs(passband gain,stopband rejection etc.),then convert each resistors by switched capacitor and the assumption is that clock frequency>>signal frequency.So that in technical terms is kind of oversampling the signal,which i don't want to do always.In the case of clock frequency close to signal bandwidth or in case Bandpass filter of high center frequency where the requirement dictates very high clock frequency how to design the SC filter?? |
|
Title: Re: Opamp spec for switced capacitor filter Post by buddypoor on Jun 11th, 2008, 4:14am manodipan wrote on Jun 11th, 2008, 3:39am:
1,) It is the great advantage of S/C filter design that structures often are even simpler than RC topologies (for example: positive or difference integrator to be realized only by proper switch phasing,without equivalence to the time continuous case). Therefore, simple resistor replacement is not always economic. 2.) The clock frequency has to be always larger to the pole frequency of the filter - at least by a factor of 50 (normally: 100...200). Otherwise, the formulas for resistor-capacitor equivalence cannot be applied. And more than that, there is a large distortion of the filter transfer curve for low ratios clock-to-frequnecy. For example: Resistor replacement by the EULER approximation leads to infinity at f=f,clock. This can be corrected more or less by the S/H action of the last stage - but this creates the typical sinx/x distortion. In short: Low clock rate leads to more distortions and deviations from the desired response. 3.) If the clock-to-frequnecy ratio is low, it is very difficult to remove the clock signals from the output. For a ratio which is sufficient high an addtional time continuous filter often is not needed. 4.) SUMMARY: Donīt use a clock rate which is too low. Your filter cannot work properly. |
|
Title: Re: Opamp spec for switced capacitor filter Post by manodipan on Jun 11th, 2008, 10:21pm Hi Buddypoor, So my question is for a 3MHz center frequency Bandpass filter with BW of 2MHz(Q=1.5) ,what sort of clock frequency and architecture will be ok?? |
|
Title: Re: Opamp spec for switced capacitor filter Post by buddypoor on Jun 14th, 2008, 9:05am Sorry for the late answer, I was absent for some days. However, my answer will not satisfy you. I must confess that I am not at the "front end" (is this the correct expression ?) of development of S/C circuits, but a center frequnency of 3 MHz would require a clock rate of at least 100...150 MHz !!! I doubt if this would be realistic. And donīt forget that the time constant for loading the switched cap must be smaller than the switching period (in your case app. 10 ns maximum). Sorry, thats all I can say at the moment. |
|
Title: Re: Opamp spec for switced capacitor filter Post by manodipan on Jun 15th, 2008, 2:24am Hi Buddypoor, Thanks for ur response,u don't need to be sorry because you have other reponsibilities also.So i have seen that for high frequency bandpass filters,the clk frequency is 4 times the center frequency and Q is very high .But in my case Q is low and center frequency is high,clearly clk frequency of ~100MHz will consume lot of power.So probably it would be 4 times center frequency as specified in JSSC,April 1989,10.7 MHz bandpass filter paper by B.S.Song. I had another question on DFT.Generally in transient reponse for fractional ratio between signal and sampling frequency the no of transient points for integer no. of cycles is not a power of 2.But it is better to do DFT of 2^n points(where n is an integer) for fast calculation(it does FFT).Now is that statement ok?or there is something wrong and what is the purpose of windowing?? Another doubt is Given some input signal frequency the output of system contains several harmonics along with signal due to nonlinearity of the system.Now For a filter i can get ac response and get the rejections at different frequency,but can we get the same information by doing FFT of transient response because it will consist of signal and other frequency components??it would be better if u clarify my doubts..again thanks for ur help.. |
|
Title: Re: Opamp spec for switced capacitor filter Post by buddypoor on Jun 21st, 2008, 5:20am manodipan wrote on Jun 15th, 2008, 2:24am:
If you want to know the rejections resp.dampings at different frequencies, the only way is to perform an ac analysis. The FFT is something different - it gives you an indication of the spectrum content for a specific time function. There is a big difference between ac response of a system and the spectrum of a certain wave. |
|
The Designer's Guide Community Forum » Powered by YaBB 2.2.2! YaBB © 2000-2008. All Rights Reserved. |