Anti-Aliasing Filter

[guerrez]

Hi,
I would digitalize a sinewave of 3kHz using a 16bit ADC with sample rate of 450 kSPS, that is 450 kHz, Nyquist frequency of 225kHz. What characteristic have to have the anti-aliasing filter? how many stages (poles) I have to implemet. I mean, how steep has to be?


Thank you!

 

[Harald Kapp]

The characteristic should be almost flat in the pass band. That calls for a Butterworth filter chracteristic.
The corner frequency of the filter needs to be higher than 3 kHz in order not to attenuate the signal too much. I suggest at least 2*fsignal -> 6 kHz.
From 6 kHz to the Nyquist frequency of 225 kHz you have approx. 5 Octaves. A first order filter has -6 dB/Octave attenuation, which gives -30 dB at the Nyquist frequency in this case. It will depend on the harmonics content of your signal whether this is enough.
a 16 Bit ADC has a resolution of 96 dB. For a filter to reduce harmonics to or below that level at the Nyquist frequency, you will need at least -18 dB/Octave attenuation which can be realized by a 3rd order filter. Using a 4th order Butterworth low pass filter with a 6 kHz corner frequency you'd be on the safe side.

Harald

 

[ElectronWorks]

probably best if you simulate it. Have a look at this website for LTSpice tutorials and a guide to active filter design:http://www.simonbramble.co.uk/lt_spice/ltspice_lt_spice.htm

 

[Raven Luni]

If you fancy some heavy reading, this book is an absolute gem when it comes to learning the theory. Its available as a free PDF download (It's so good I'm about to buy a hard copy):

http://www.dspguide.com

 

[Harald Kapp]

Much easier than using SPICE (any version) is a dedicted filter program like e.g. here: http://www.ti.com/tool/filterpro

I'm a fan of LTSpice myself, but in this case a spezialized tool is simply better.

Harald

 

[guerrez]

Thank you for your answers, are all very useful! Now I am using the filterPro utility!

 

[guerrez]

I am reproduceing the schematic of the attached file. I have modified properly the antialiasing filter.
Now I want to decrease the impedance that see the secondary windings. As it written, it is 100kOhm, but my LVDT is designed for a load of 10kOhm. Can I add a small resistor between the two resistors of 47.5 kOhm (so would be in parallel), without changing the parameters of the filter before the the Instrumentation Amplifier?
Thanks!

 

[Harald Kapp]

Somehow my a answer was lost.
Repeat:
Yes, you can. Add 11.1 kOhm in parallel to the input to get 10 kOhm impedance.

Harald

 

[guerrez]

Yes, perfect, thanks!

 

[guerrez]

At the moment I have used the dual supply for the instrumentation amplifier and the power amplifier. The simulator's results are all good, but now I would like to use this schematic using a only single positive supply (I can adjust it between 0 to 24). How could I modify the schematic? Thanks!

 

[guerrez]

I don't think that there are devices that can convert +24 V into -15 with at least 2A for power supply ( the LVDT that I use required a lot of current), so I have thought to change that power opamp with for example the OPA541 or LM1875, that admit single supply. Eliminate the capacitor before the power amplifier so the imput of the power amplifier will be positive-only. Ando only after the power amplifier eliminate offset with the capacitor.
Unfortunately I have found only the model of OPA541, but appear not correct with a single supply. Do you think mine is a good solution?

 

[Harald Kapp]

You can't.
The OPA544 is not a rail-to-rail amplifier. Output voltage is between V(-)+4V ...V(+)-4V, depending on load conditions. So to achiever an output voltage of 0V...XXV you need at least -4V supply (-5.5V at high load).

http://www.ti.com/lit/an/slod006b/slod006b.pdf tells you everything you want to know about OpAmps. Chapter 4 deals with single supply OpAmp circuits.

Harald

 

[guerrez]

Yes, I have checked the datasheet, you have right!
I already have found a better power amplifier, the OPA548, that admit a greater voltage input: (V–) – 0.2 ; (V+) – 2.3

 

[guerrez]

I have simulated the circuit with the OPA548, but if I use a load of 285 Ohm resistor I have seen that it doesn't arrive to 20V, also if I use 24 V of power supply.

I have seen something like TDA2030, that use also a single supply.
But it works like a classical OPAMP, with the same configuration (inverting and non inverting)? Becouse I read only amplifier, not operational amplifier.

 

[Harald Kapp]

You could add a booster stage to the OpAmp as for exa,ple here: http://www.ti.com/lit/an/snoa600a/snoa600a.pdf

Or you can use the TDA according to its application.

But: Since you are driving a square wave only, why use an OpAmp at all? You could just as well use a simple push-pull output stage http://www.ecircuitcenter.com/circuits/pushpull/pushpull.htm


Harald

 

[guerrez]

This is my first project, so I have tried to start with something already done!
However I also study these other possible solutions!

I have attached two images, in which I used the method of the virtual ground.
The output of the opamp is exactly what I want in 2 cases: a sinewwave from 0V to 3V and the other one a sinewave from 0V to 20V, all of them starting with a sinewave from -10V to 10V.

The simulation output is exactly what I want, but I don't understand how does a DC voltage of 5V can move up the wave of 10V and the same for the other picture!

Do you think that is correct the simulator that I use (Tina TI) ?

Thanks!

 

[Harald Kapp]

Tina ia o.k.
It is 5V DC + the superimposed AC. Thus, the output is 5V+gain*Vin.

It might be a good idea to start with some very basic OpAmp circuits to get an understanding of the OpAmp's operation. For example here http://www.electronics-tutorials.ws/opamp/opamp_1.html or google "operational amplifier tutorial".Having understood ist opearion, an OpAmp is a very versatile device. However, the more complex the circuit you start with is, the more it will be difficult to understand what happens (or why the circuit doesnt do as expected).

Harald