Do I need to reduce source impedance when using a SCXI 1100?

[DW]

I have a peak detecting circuit (for a sine wave -- 25 channels)consisting of a diode, capacitor to ground and a 1 Mohm resistor connected to a SCXI analog input pin. If I sample too quickly with channels that are alternating high and low, a charge travels from the SCXI to my peak detector charging the capacitor. What value of capacitor should be used in order not to reduce the time resolution of the system.

Is the RC constant equal to the 1Gohm * my capacitor?

Do I need the 1 Mohm resistor or is that added source resistance that is not needed?

[RIO-Todd]

Hey DW,

I don't think you're going to need the 1 Mohm resistor within your circuit. The capacitor by itself should be able to perform the peak detection.

I would make sure that other inputs to the SCXI module have a low source impedance. Any time you sample at a fast rate and have sources with high impedances you will run into charge injection. A 1 Mohm resistor in series with an analog input will definitely increase the source impedance and cause charge injection.

Have a good weekend.

Regards,

Todd D.
NI Applications Engineer

[DW]

Thanks Todd.

Do you know what kind of Capacitance would be useful? Currently my little 10 pf is being influenced by charge injection (cross channel effects). Does the 220 pf input cap have an effect?

The input is a 1 Mhz Sine function.

[RIO-Todd]

Hey DW,

Thinking about this additionally, I'm not sure that only having the capacitor in the setup will solve the problem either. Charge injection is caused by not having a path to ground for the multiplexer to discharge. Placing a capacitor in parallel with an input doesn't provide an effective path to ground when scanning at a fast rate (I don't think the cap is allowing the multiplexer to discharge).

If you take the 10 pf capacitor away and look at the signal without the cap, do you see charge injection?
I'm just curious to see whether the cap itself is causing the problem.

Regards,

Todd

[DW]

Are you suggesting putting a resistor in parallel with the capacitor (to allow the mux to discharge)?

This is the current setup:

diode 1M res.
---|>|----------<><><>----- +
_|_
___10pf cap to gnd SCXI1100
|
|_________________ -


1. You suggest that the 1M res. is not needed. (I believe it is there to prevent shorts that fry the diode or the opamp before it).
2. If I increase the 10 pf to 110 pf, it helps significantly, while reducing the resolution. Is it appropriate to look at the 1Gohm impedance (SCXI) as the R in an RC circuit?
3. If I put a resistor in parallel with the capacitor, it will allow the SCXI to discharge, and lower the RC of the circuit. What value do you sugg
est? I'm using a 1Mhz signal and am trying to get a constant peak-level output. (I'm going to try inserting a large resistor across the capacitor.

Thanks.