High-Speed Digitizers

cancel
Showing results for 
Search instead for 
Did you mean: 

Can you explain RIS and how it is done on the digitizers?


0 Kudos
Message 1 of 4
(5,665 Views)
Random interleaved sampling (RIS) is a form of equivalent-time sampling (ETS) for sampling repetitive signals so that the effective sampling rate is higher than the real sampling rate. RIS is accomplished by acquiring a series of points along a repetitive waveform for each occurrence of a
trigger. The waveform is then reconstructed from the data acquired over many cycles.

In RIS, the arrival of the waveform trigger point occurs at some time
randomly distributed between two sampling instants. The time from the
trigger to the next sampling instant is measured with a highly accurate time-to-digital converter (TDC), and this measurement
allows the waveform to be reconstructed.
Message 2 of 4
(5,664 Views)
Why does it have to be random?

Lets say if I want to oversample at 3*200Mhz.
Why not fill in first pass first bin, second pass - second bin, and thid pass - third bin?

Because your current system does RIS the expected number of passes for 3x oversampling
is 5.5 and for 20x oversampling it is about 72.

Is it possible to write my own program that will fill bins in order?

Thank you for your time,
Nikolai.
0 Kudos
Message 3 of 4
(5,405 Views)
RIS requires that your input signal be periodic but that there be some jitter in the periodicity.  Imagine that you set the digitizer to trigger when the input signal rising edge crosses 0 V.  The analog edge trigger circuitry will detect when this occurs very precisely.  However, the signal will not be sampled until the next sample clock edge.  The device can determine the time difference between the trigger point and the sample point.  The next time the device acquires a record of the this signal, it is likely that the 0 V crossing point occurs at a slightly different time relative to the sample clock edge.

The important point is that the digitizer's sample clock does not change.  Further, we work hard to keep the jitter of this clock very low.  So to acquire enough bins to oversample the periodic input signal, we rely on the jitter of the input signal.  Because this "phase noise" of the input signal is random, the order that the bins get filled, and the time it takes to fill all bins, is random.  You could have several acquisitions that fall into the same bin before you fill all the bins. 

Unless you can precisely control the phase and jitter of the input signal, you won't be able to write a program that fills the bins in order.

One way you might speed up the acquisition is to change the NISCOPE_ATTR_RIS_NUM_AVERAGES attribute.  On most devices, the default is 4, which means it attempts to fill each bin four times, then averages the points in each bin.  You can try setting that to a smaller number.
Message 4 of 4
(5,381 Views)