Suggested by stj on the EEVblog forum. Thank you!
| Every sample | |
| Fast picture | |
| Z input | |
| Time plot |
The XY plot pairs the samples the scope took at its own sample rate, the whole window of each acquisition (up to a million pairs at the slow time bases), and counts every pair into the XY density map, which shows graded under the newest one with the persistence, the way a CRT's phosphor integrates the beam. A long window's pairs are now shared among all four counting threads and counted in integers: on a test PC a dense million-pair frame counts in 88 ms of one thread where it took 249 ms, and four threads share it; the 13, 25 and 152 waveforms a second our MHO934 showed in XY at 1 ms, 100 us and 10 us/div are measured again on the scope with this build.
The newest frame is drawn by the same counting into a map of the XY square and shown pixel by pixel, not stroked as a line: 0.26 ms instead of 1.6 ms for a 10,000-pair figure, 1.1 ms instead of 13.5 ms for a dense one, 11 ms instead of 38 s for a dense million-pair zoom (a test PC). Where the screen trace is a min/max envelope, whose pairs are not one instant on both channels, the picture is drawn from the samples themselves, so a fast signal no longer turns into the corners of its figure.
The XY sheet (the Acquire sheet's Z button) has a Z row: Off, CH1 to CH4 or the math as the source; Blank turns the beam off where Z is at or below a level, as a CRT's Z blanking, and Linear runs from dark at Low to full brightness at High; Negative reverses either. Over SCPI: :NOVAos:XY:Z:SOURce, :MODE, :POLarity, :LOW, :HIGH and :YT; the stock :TIMebase:XY:Z still picks a channel.
With Time plot on, every channel's trace but the Z channel's own takes the same brightness sample by sample, triggered or in roll: a gate or marker channel lights the part of a signal it marks, blanked or in four steps of brightness. So to the question whether XY only: both.
The engine counts every XY pair with its Z weight on every counting thread; the screen draws the newest frame the same way, the XY sheet sets Z, SCPI reads and sets it, and the time plot takes Z in the main mode and in roll.
A known Lissajous counted pair by pair (every sample once, shares of a frame equal to the whole), blanking, linear brightness and both polarities in the XY map and on the time plot, the roll trace on the simulated engine, the SCPI forms on both families, the sheet in every language, and the cost of a picture against a budget.
Measured on our MHO934 with its two generator sines, before and after: waveforms a second in XY at 1 ms, 100 us and 10 us/div against the time plot's at the same settings, and a Z-modulated Lissajous filmed.
Try it on your XY work (a Lissajous, a curve tracer, a vector display) and tell us what the brightness modulation still lacks.
A Z input as on an analog scope: a channel whose voltage sets how bright each point of the trace is drawn (stj, msg6378097); Whether Z belongs to XY only or to the time plot too, and whether the XY mode is fast enough to be worth it (ebastler, msg6378127); Z brightness on the rolling time plot too (stj, msg6378181).
can you add a Z-input to modulate the brightness of the trace like on good analog scopes?
Would you use that in XY mode only, or are there uses in Y(t) mode? If the former, is XY mode in general fast enough to be useful on the Rigol?
i was thinking about XY mode, but it could have potential use maybe in rolling mode.
Try it and tell us what to change: reply on NovaOS's thread on the EEVblog forum.