ESC

Arthur Pini

's profile
image
Consultant

I am a technical support specialist and electrical engineer with over 50 years experience in the electronics test and measurement industry. I have supported oscilloscopes, real-time spectrum analyzers, frequency synthesizers, digitizers and arbitrary waveform generators for leading manufacturers.


Arthur Pini

's contributions
  • 05.05.2015
  • Signal processing boosts digitizer performance
  • Hello Sean, Thanks for your comment. As you point out there are many possible algorithms for computing a moving average. Each has its own advantage. The user has the option of using whichever suits his needs but they may need to write the code to implement it themselves. The point of the article is that this processing function is available to improve the measurement result..
  • 05.05.2015
  • Signal processing boosts digitizer performance
  • Hi Martin, Moving averages can be done in both ways. When used in financial analysis the second method you suggested (x-4, x-3, x-2, x-1, x.) is used. This produces a relative shift between the input and output signals. In scientific and engineering analysis the former, symmetrical form, ( x-2,x-1, x, x+1, x-2) is used. The symmetry keeps the input and average values 'in phase'. Spectrum's SBench 6 software, used in this example specifically defines the algorithm as being symmetrical: " Moving Average The moving average (smooth) function takes a an average around the current sample of the sample itself and a defined number of neighbor samples. As a result random noise is reduced: The „Window Width“ parameter define how many samples around the current sample are used for averaging. A value of 5 for example will result in the following average function: new_sample[x] = (sample[x-2] + samples[x-1] + sample[x] + samples[x+1] + sample[x+2]) / 5"
  • 04.09.2015
  • Limit the range of a waveform measurement
  • Hello Hithesh, Read operations, like those that appear in the left most burst in figure 2, are either 0 or 180 degrees relative to the strobe signal. So a read operation is occurring at the beginning (left side) of the trace. Write operations occur with a fixed delay between data an strobe. In figure 2 that delay is about 770 ps. The right side of the z2/z3 traces show a write operation. Art
  • 02.12.2015
  • Annotate oscilloscope displays for better documentation
  • Hi Steve, There is a trade off between font size and waveform visibility. On the oscilloscope the screen is 10 to 15 inches diagonally bur when you paste it into a word document in portrait mode it is only about five inches so you do lose resolution. Martin's suggestion helps but it does take up a lot of space having images in landscape mode. When writing, I generally transfer the important information from the image into the text. Art
  • 09.22.2014
  • 50 Shades of Timebase
  • An important thing to keep in mind is that equivalent time (ET) sampling only works with repetitive waveforms. The sample spacing in ET sampling depends on the precision of the scope's time to digital converter (TDC). The TDC is a fast counter that usually measures the time between the trigger event and the nearest sampling clock which is called the horizontal offset. In ET sampling it uses this information to classify the horizontal offsets of multiple acquisitions and uses those with desired ET sampling period to create a composite waveform as shown the article. In addition to a 'smoother' waveform rendition the other beneficiary is timing parameter measurements. An improvement in the accuracy of a timing parameter, such as period, indicated by a decreasing standard deviation occurs for oversampling of factors from 4 to 10 times the RT sampling rate. This is very similar to the reported improvement noted for the displayed waveform. Beyond ten times oversampling with ET there is little further improvement. ET was a more important feature at a time when ADC's had limited maximum sampling rates. With maximum RT sampling rates at 160 GS/s it offers less of an advantage
  • 08.16.2014
  • View noisy signals with a stable oscilloscope trigger
  • Hi Dave, Most oscilloscopes offer high and low frequency reject trigger modes. High frequency reject inserts a low pass filter in the trigger path; while low frequency reject inserts a high pass filter. Most set the cutoff frequencies around 50 kHz.
  • 06.24.2014
  • Digitizers: Finer resolution is better
  • Hello Olehwi, Thanks for your comment. The sampling rate for all three data records is 5 GS/s (200 ps sample period). The lower resolution waveform spends multiple sample periods at the same amplitude because it lacks the resolution to render the intermediate points. These are ideal waveforms so there is no vertical noise to 'dither' the amplitude values hence the quantified appearance. Regards Art