Hi,
I want a maximum sampling rate of 5 ksps for voltage and 5 ksps for current.
OK...
With 5ksps.. I´d definitely use an appropriate low pass filter.
Read about nyquist theorem.
When I hear "transients" then I think about narrow peaks with a timing of less than 10us, maybe less than 10ns...
As said: only you know the requirements...
Hint:
Just to give you a "picture": think about the transient as a triangle pulse with constant risetime and constant fall time.
I recomend to use paper and pencil. Draw a chart with X = time, and Y = volts (amperes)
Draw a little vertical line every 200us (5kHz) just to mark the sampling events.
If you have use a tranparent foil and put it over the timing chart.
Now draw your "transient" signal onto the foil.
Focus on the points where the sampling event meets the transient signal curve. This is what your software will see. Only the points ,not the line inbetween.
Now slowly shift the transient signal (foil) to the right with respect to the sampling events. See how the poins change.
If it was my application .. and I just wanted to see the peak of the transients, but with a 5kHz sampling rate, then I´d use some "peak detector". In simplest case a capacitor charged via a diode.. and slowly discharged by a resistor.
All this can simply be simulated with excel..
On the PT100 there is no sense of that high speed cause the PT100 itself is has a much slower response.
Especially here a low frequency low pass filter gives more stable and more reliable readings.
Also here applies nyquist.
Especially with wired sensors there is a good chance that mains frequency is introduced. 50Hz or 60Hz.. it will cause your temperature reading to fluctuate (randomly)...but you know that the temperature will not change that fast. Thus suppress the 50Hz, focus on the expected teperature change rate and get more stable and reliable readings.
Klaus