Fullscreen HTML5Med Res (??MB)HTML4

These two new digital storage oscilloscopes from Tektronix are right at the leading edge of technology. The TDS 784A at left, has a bandwidth of 1GHz & 4-gigasamples/second maximum sampling rate. The TDS 744A at right operates at real-time speeds up to 500MHz & with a maximum sample rate of 2Gs/s. Both have liquid crystal shutters to provide colour displays & have an unsurpassed ability to catch & display rare glitches in signal waveforms. Tektronix TDS 784A TruCapture oscilloscope Tektronix has really taken the bit in its teeth over the last few years in developing the art of digital storage oscilloscopes. Now it has taken another big step forward with its TDS 784A & TDS 744A scopes which can display up to 400,000 acquisitions per second. This is a huge improvement over previous digital scopes. By LEO SIMPSON While digital scopes have come a long way over the last few years, they still have drawbacks in the way they display signal waveforms. Partly this is due to the sampling system which shows the waveform as a series of dots. On a signal which has superim­posed noise, the resultant waveform can be quite jagged and quite different from what would be displayed on a conven- tional analog oscilloscope. The truth is that the both oscilloscopes show the waveforms differently and both conceal information. Actually, a major shortcoming of digital storage oscilloscopes (DSOs) has been the small fraction of time they spend capturing waveforms. This is quite different from the impression that you get when the display is updated at 60 times per second. For ex­ample, if the DSO is set at an appro priate sweep speed to display a 10MHz clock signal, each refreshed display will show about five clock signals or half a microsecond (500ns). This means that in 60 displayed waveforms, only 30 microseconds of signal will be acquired by the scope in one sec­ond. This is 30 parts per million or just .003% of real time. So while things appear to happening rapidly on the screen, in reality the scope is sitting there doing nothing most of the time and many “events” could occur which are just not captured. Analog scopes do a lot better in terms of their “display refresh” rate; ie, the number of times the screen display is updated. The best analog scopes can refresh the display at sev­eral hundred times a second (at a sweep speed of higher than 1µs/div) but then they also have trouble dis­playing rare events; March 1995  85 Displaying a 3MHz signal of a Tektronix 2465 analog scope shows a waveform which is clean and apparently free of any glitches. The same 3MHz signal displayed on a Tektronix 2467B, one of the world’s fastest analog scopes which has an enhanced CRT. Here a glitch is apparent in the form of a “runt” pulse (about half the full height), although the reproduction of this photo may not show this. the writing speed of the phosphor used in cathode ray tubes (CRTs) is too slow for single glitches to be observed by the user, even if a viewing hood is employed. The only way to see very fast glitch86  Silicon Chip es buried in a repetitive signal with an analog scope is use one that includes an electron multiplying plate between the deflection plates and the phos­phor of the CRT. Examples of such scopes are the Tektronix 2467B and 7104 but these are expensive scopes indeed. Of course, some high-end digital scopes can be programmed to find glitches in repetitive signals but you have to know what you are looking for in order to do the programming. And since the digital scope spends so little time actually acquiring the signal, you might have to wait a long time before the glitch actually is found, if at all. And while analog scopes can be bet­ter at finding glitches, you have to spend unconscionably long times glued to the screen in order to actually see them. Sometime in the future, digital scopes must equal the glitch finding ability of the best analog scopes but according to theory, this would re­quire a display system capable of sev­eral thousand full screen acquisitions per second. The instrument would then have to rasterise these acquisi­tions at nearly 200 million pixels per second (compare that to today’s VGA screens at about 55 million pixels per second. 1024 x 768 x 70). In addition, the data move:gient between the ac­ quisition system and the display would need to be around 200 mega­bytes per second. Now while these parameters are technically feasible, there is no digital scope available to­day which comes within cooee of them. All of which leads up to how Tektronix has gone about achieving the desired result by taking another approach – changing the architecture of the digital scope. Briefly, these changes are as follows. first, the rasterisation capability of the display system is duplicated in the acquisi­tion system, next, the rasteriser is al­lowed to use a portion of the high speed acquisition memory to build display images; and third, the acqui­ sition hardware is allowed to start acquisitions without the intervention of the instrument’s firmware and to calculate its own trigger positions. This new architecture is used in the Tektronix TDS700A TruCapture dig­ital scopes in a mode called “InstaVu” acquisition. When this mode is ena­bled, the data moved from the acqui­sition system is a complete rasterised image of many triggered acquisitions of the input signal. By the way, perhaps we should briefly explain the term rasterisation as it pertains to digital scopes. It refers to the display system. In a conven­tional analog scope, the input signal is applied directly to the deflection plates of the CRT and so the signal on the screen is an “analogue” of the input; it is also a vector display with the electron beam tracing out the signal on the screen in response to the deflection voltages on the plates. A raster signal, by contrast, is the same as a computer video display; the electron beam scans the whole screen at rates similar to a computer VGA display and the beam is modulated on and off by the video signal to produce the individual pixels (picture ele­ments). In essence, the DSO converts the input signal to digital data and stores it in high speed video memory. Getting back to the plot, we talked about moving a complete rasterised image from the acquisition system to the display. Transferring this 500 x 256 pixel map requires a lot more data to be transferred between the two systems but the raster is only moved at the refresh rate of the scope’s display and contains information from tens of thousands of acquisitions. Doing it this way makes the data transfer rate manageable and in fact, it equates to 417Kb/sec. Tektronix has had to develop a considerable range of new semiconduc­tor hardware to achieve its new architecture and among these is a new kind of demultiplexer which integrates 360,000 transistors into a CMOS IC with 304 pins. It dissipates about 2.5 watts when running at full speed. Normally, the only function of this IC would be to demultiplex (ie, switch) data from the analog to digital converter and store it in a high speed static RAM. One third of this new demultiplexer is devoted to that job. The remainder is split between a high speed rasteriser and a digital signal processor (DSP). The DSP is included for, among other things, mathematical algorithms and trigger position cal­culations. We could discuss this new technology at greater length but none of it really means much until you see the results. To this end, four screen pho­tos are included with this article, showing how different scopes behave when displaying a 3MHz waveform with buried glitches. Briefly, all but the best analog scopes never reveal the glitches and nor does the Tektronix 544A colour digital scope (reviewed in Silicon Chip, November 1993) but the TDS784A and TDS744A, with On a Tektronix TDS 544 digital colour scope, the 3MHz signal results in a waveform which is similar to that shown on the Tektronix 2465 analog scope. Note that it has been sampled at a rate of 500 megasample/second. Finally, this is the 3MHz signal depicted on a Tektronix TDS 784A digital colour scope in InstaVu mode. Here the runt signal is clearly visible, made doubly by the colour display (although not reproduced in this B&W photo). Note that the acquisition rate is also 500Ms/s, the same as for the TDS 544A, but the number of acquisitions is a great deal more (75,896 versus 1156). their extremely high sampling rates and high acquisition rates, do reveal the glitches and do so even more dra­matically with the aid of a colour screen. Most impressive. For more information and prices on these new digital oscilloscopes, con­tact Tektronix Australia Pty Ltd, 80 Waterloo Road, North Ryde NSW 2113. Phone (02) 888 7066. SC March 1995  87