Anti-Aliasing Filter

[NorthGuy]

Yes, quite a bit of a ripple, but, instead of smoothing it out in hardware, I just sample at much faster rater. The picture below (which is battery voltage during charging) is sampled around 5kHz (as I remember), then average of about a 100 cycles is calculated and plotted along with maximum and minimum values. I have a fast filter which cuts off anything above the Nyquist frequency for my sampling rate. But I preserve the rest. This gives me an ability to plot the ripple. At the same time, the software averages the values and displays them, so you get a very stable display too. The power gets calculated from raw values, which is more accurate.

If I used slower filter, I still could get the average, but I couldn't do the plot, and my power calculations would be less accurate.

 

[misterT]

Nice plot. OPs main problem may be that he does not know what kind of noise/disturbances he has on his signal. And what I understand, he doesn't even know what is the signal he is interested in (how dynamic the system is).
You need to know the system. What is the switching frequency. What kind of output filtering the PSU has. What kind of loads are connected. What is the update-rate of the control loop. He needs to do many design-iterations..
Everything I have said is just theory. Real world application is different. I would try to avoid complicated active filters.
That battery charger is a good example of a system you can easily measure, and then design a working filter.

 

[atferrari]

Take a resistive load ie a loadcell for example.. the analouge output will be very noisy...

Why so, Ian? Intrinsically or because of the expected environment?

 

[Tony Stewart]

A good way to measure battery health/end of life/ is to apply a pulse load and measure voltage drop, rather than simply cell voltage. This is a good maintenance prevention feature before it fails.

Depending on if and how you want to do this pulse , will determine your filter time constant which should be 5% of the pulse width.

 

[Ian Rogers]

Why so, Ian? Intrinsically or because of the expected environment?

Well actually... A bit of a learning curve for me.. I have always used 350 ohm bridges, but these things are terrible for wireless, too much power, so I have moved to 1k ohm... these things are very noisy!! It seems the higher the resistance the noisier they get.. I have now placed temp compensation but they are much harder to read..