clipping recovery time

[MGray]

I am working with a signal that consists of a high amplitude burst followed about 100us later by a burst that is about 1/10th the amplitude.  The bursts last for about 50us each, and all data requires sampling at at least 10MHz.  

Using a 5105 PCI card, I am interested only in measuring the low amplitude burst portion of the signal.  Given the broad dynamic range of the total signal, it would help to allow some degree of clipping of the high amplitude burst so that the second burst would occupy more of the digitizer's dynamic range.  While I don't expect that clipping is ever a good thing, what is known about the recovery time from a transient clipping event?  Does it depend strongly on the difference between original signal and input range setting?  Is there a "safe" range of clipped levels and recovery times?

thanks,  

[johnsold]

I do not know about the 5105, but I would not rely on clipping any amplifier (not specifically designed for it) followed by normal linear behavior.

You did not specify the actual voltages of your signals, but I would consider an external clipping circuit, possibly using Schottky diodes.

Lynn

[MGray]

Lynn, the high amplitude part of the overall signal is in the 1-2V pk range. thanks, Mike

[johnsold]

Mike,

What is the amplitude of the small signal? What input range on the DAQ device do you want to use for the measurement of the samll signal?

Lynn

[MGray]

Lynn, the signal segment of interest is in the 100mV rms range, but I can scale with preamp gain.  I was using the 1V setting on 5105, as it is rated to have the highest SINAD of the 1MOhm settings. -Mike

[johnsold]

Mike,

This is the kind of circuit I had in mind. The diodes would be Schottky diodes such as 1N5711.  The resistor would be in the 100 to 1000 ohm range.  With a 2 V peak input and 1000 ohms, the peak voltage out would be less than 0.5 V peak. At 170 mV peak (the signal of interest) the output would be within 1 mV of the input.  The time constant would be on the order of 4-5 ns or less than 10 degrees of phase shift for a 5 MHz signal component.

Lynn

[MGray]

Thanks Lynn, I'll give this a look.