Daqarta for DOS Contents
DDisk recording, this option is locked out in Board mode, so you will need to make sure that is off first (Virtual Source active).
You will be prompted for the file name, with a default of SAMPLE.000. You may accept the default with ENTER, enter another directly, or use the File Access system to select an existing file or review names. If you enter or select a file that Daqarta determines to be an InDat file instead of a DDisk file, the DskRd option will exit automatically and InDat invoked instead.
When the file name has been accepted, the start of the file will appear on the screen. The DDisk Position Readout box will appear to the right of the Cursor Readout box below the trace area. This shows the file position at the start of the current trace, so it will show 0 initially.
Some options will be disabled while reading the file. In particular, the Board (and Virtual Source) will be locked out, including those control menus, since the data source is now the file. So don't be startled if you have the Virtual Source control menu on-screen and it is suddenly replaced by the Trace Info menu when you load the DDisk file. Likewise, you won't be able to load other data files with InDat. But some options like Number of points N are only locked while the DDisk file is Paused, as it is when first loaded.
When the file is loaded, its sample rate will be used to set the Daqarta X-axis. If it is a true Daqarta file, most other parameters will assume the values in effect during the original recording, and the original comment line will be shown. These settings are only in effect while that file is being read, after which the original settings and comment will be restored.
Options like sample rate that are properties of the recording will be disabled during DDisk Read. The cursor will not be allowed to move to disabled items, although those items will still be displayed in their respective control menus.
Pause mode. If you unPause, the file will "run" by itself and the Status will show PLAY. You can then Pause when you see something that merits a closer look. Whenever the display is Paused, the above manual movement controls are available for fine control, etc.
triggering can be used just as for a live trace, so you can see a stable display if your data is repetitive. You can set a suitable trigger level and perform waveform or spectrum averaging as the file replays, for example.
sample rate as when it was recorded. If you recorded at a high speed, this will often be too fast, so you can set a negative Speed Factor to slow down by that amount: A factor of -10 means that the file will be replayed at one tenth the recorded speed. On the other hand, if you have recorded some very slow phenomenon at a low sample rate, perhaps over hours or days, you can speed up the replay with a positive factor.
trigger level to help find areas of interest if you know that the waveform will be above some background level: Daqarta will ignore all data below the trigger level, so this becomes a "skip to next peak" function. If you want to Pause after the next peak is found, rather than seeing all peaks in the file in rapid succession, use the Single Sweep function.
triggering active you are not seeing ALL your data, since some will be skipped to get to the next trigger level. Use the Trig key to disable triggering if you want to see everything. However, the resulting Free-run display may go by too fast to be of much use. If you use the Speed Factor to slow things way down, the display will flash a full screen of data at a time, then flash the next full screen. Sometimes this may be just what you want, but there is another option: Scrolling via Overlap.
You control the apparent scrolling rate with Overlap Points in the DDISK READ control menu. If you set this value to one less than the number of points N per screen, then the next screen will advance by only a single point. As you reduce the overlap from this value, the display appears to scroll faster and faster. Using this control and the Speed Factor, you can tailor the apparent rate to be as fast as you can reliably scan your data.
spectrogram data presentation. Each vertical time-line in the spectrogram represents one spectrum of N data points. There are 512 spectra in a full-width spectrogram, so with no overlap this would show N * 512 samples of data. But with overlap each time-line does not advance as far through the data. At maximum overlap (one less than N) each line is only one sample later than the start of the prior line, so the final line would be only 511 samples later than the first. Overlap thus has the effect of expanding the spectrogram display.
If you want the spectrogram to cover a certain amount of time, adjusting the Overlap control is not very intuitive. Instead, the Sgram Width control below it allows you to control the spectrogram X-axis time directly. You can enter or scroll the Sgram Width to a desired time value, and all the relevant math will be done and the corresponding Overlap value updated. Likewise, whenever you change the Overlap value, the Sgram Width is updated as well.
Note that the spectrogram is not updated automatically just by changing the Width or Overlap control; the next screen shown will reflect the change. You can hit HOME after each change to cause a redraw from the start of the file if you are trying to get a short file to just fill one screen.
This is actually quite simple: If the first view of the spectrogram stops before the end of the screen, read the time at which it stopped from the X-axis and enter that as the Sgram Width, then hit HOME to redraw with the new overlap.
Use of DDisk recordings with Sgram Width or Overlap controls on DskRd playback often allows much higher overlap values than you would get during normal untriggered operation in RTime mode, especially with slower systems. It also allows constant values, whereas the RTime overlap may change as the processing time changes with changing data.
Record a signal such as a triangle or sine wave that has no step discontinuities (unlike a square or ramp). Set the signal frequency to get a few cycles displayed on each trace at the sample rate you require. For example, if you are sampling at 20 kHz and viewing 512 points, the trace is 512 / 20000 = 0.0256 or 25.6 msec wide. If you want to have 3 cycles in this interval, each needs to be 25.6 / 3 = 8.5 msec or 117 Hz... but this is really not critical and can be done by eye. Anything in the 100 to 200 Hz range should be fine. This will also make it easy to spot glitches using the scrolling method mentioned above. A triangle wave is best for that sort of glitch visualization, but a sine wave is better for the automatic scan. Use the triangle if you expect to do both.
For the purposes of this test, the Sine or Triangle waveforms of the DEMO driver are good choices. The output is generated directly from the counts of a system timer, so if anything happens like missed sample interrupts during a disk transfer, the delay will be reflected as a glitch in the recorded wave.
Note that this is NOT the same as using DEMO with sine waves generated by the STIM3A Stimulus Generator. Why not? Since in that case sample output and input are synchronized, anything that interferes with the sampling process will cause both processes to be affected in exactly the same way.
For example, if 50 samples are missed, the actual output waveform would not change during that interval, giving a serious glitch. However, this would be completely overlooked by the ADC input, since no samples would be recorded either. When sampling resumes after the transfer, everything picks up right where it left off and the trace would look perfectly fine... but it would be a lie!
The glitch-scan method is to detect any discontinuity by its effect on the spectrum, where it will cause a broad high-frequency "splatter". The recorded signal usually will not have an exact integer number of cycles per trace, so you need a window function (such as Hanning) to reduce the apparent trace edge discontinuities seen by the FFT.
Since a glitch could appear anywhere in the recording, including at either edge of a trace where it would be reduced or eliminated by the window, you need to play the recording back with enough overlap to assure that any glitch will fall near the central region of the window on some sweeps. With N = 512, setting Overlap Points to 384 means that the window area moves through the recording by 1/4 of its width at a time, insuring that nothing is missed. This is MUCH faster than you could scroll with visual scanning.
Here is the step-by-step procedure:
At this point you should see a strong fundamental frequency, plus (for a triangle or not-so-pure sine wave) a "toothy" family of harmonics that decay toward a ragged noise floor at higher frequencies. Notice the approximate shape and level of the noise floor... you are going to be looking for a jump above this baseline.
Of course, since smaller glitches will result in smaller jumps in the noise floor, this method does not guarantee that your recording is perfect... just that there are no major problems. You can use this to screen out faulty setups, then follow up with a slow visual scan for confirmation once things look good.
Questions? Comments? Contact us!We respond to ALL inquiries, typically within 24 hrs.
25 Years of Innovative Instrumentation
© Copyright 1999 - 2006 by Interstellar Research