Pico Technology has a 26 year pedigree of innovative design, supply and support of USB controlled instruments; and in particular the low cost, high performance oscilloscope. From offices based in Tyler TX, Shanghai and its headquarters in the UK; Pico is an established global leader in PC controlled test and measurement solutions. In this interview, Pico’s RF Business Development Manager Mark Ashcroft introduces their SXRTO Sampler Extended Real Time Oscilloscope as a distinctive and newly cost effective third alternative to two pre-existing classes of high frequency oscilloscope; the real time digital storage oscilloscope and sampling oscilloscope.
Tech Briefs: Why does the world need a new format of oscilloscope?
Ashcroft:That comes down to un-fulfilled and rapidly growing demand in the high frequency oscilloscope application space. Pico supplies real time oscilloscopes for as little as $100, rising in price, performance and bandwidth to 1 GHz. Beyond this, for RF, microwave and gigabit data applications we supply our PicoScope 9300 sampling oscilloscopes, addressing signal frequencies up to 25 GHz. We have gained a ‘goto’ leader reputation in these markets and are intimately aware of high demand for low cost, easy to use, high frequency time domain instruments. Today, above a gigahertz, the number of competing oscilloscope suppliers falls dramatically (four world wide), pricing rises equally steeply and user compromise and frustration has very much set in.
TB: Why are high frequency oscilloscopes so expensive?
Ashcroft: Back in 1928 Messers Nyquist, Shannon, Kotelnikov and Whittaker all concluded that to sample any continuous analogue waveform into discrete sampled (digital) form we must have at least two samples within a cycle of the highest frequency that we wish to represent. In practice we have to go further than that if want to reveal waveshape detail or make meaningful repeatable measurements. In today’s real time oscilloscopes it is this that drives data bandwidth (Gb/s) to many times higher than analogue bandwidth (GHz). We also see a trend now towards higher resolution analogue to digital convertors; and both combine to drive rapidly escalating cost and/or compromise of maximum sampling rate! Thus the hf RTO embeds high cost A-D convertors, logic arrays and cache memory. Power needs and size also take a big hit.
TB: How then does your SXRTO pack 5 GHz bandwidth, 12 bit convertors and self-evident portability under a ticket of $14,995?
Ashcroft: The truth of that is a heavier reliance upon random Equivalent Time Sampling (ETS). All hf oscilloscopes use ETS to extend their costly or compromised sampling rate to well beyond maximum available data bandwidth. This is essential to the recovery of waveshape, serial data eye shape, precise timing, meaningful measurements, statistics, analysis and characterizations. But ETS can only do this if a waveshape or data stream is repetitive. Any uniquely occurring waveshape can only be captured using the super-Nyquist real time sampling mode and its limited data bandwidth.
The Pico SXRTO actually limits data bandwidth a bit further, to the benefit of cost, power and size. To close the gap, Pico has then developed the ETS technique to a new level of speed and time-linearity. The 9404 oscilloscope uses 12 bit A-D’s and a 500 MS/s data bandwidth behind each channel. In a 5 GHz oscilloscope, this evidently is sub-Nyquist sampling. Single event capture is compromised to usefulness only below 250 MHz. BUT those 12 bit A-Ds and ETS to 1 TS/s will match or exceed the performance of all available real time oscilloscopes at a fraction of their cost!
TB: So the SXRTO is the best solution in some applications, not so great in others – right?
Ashcroft: That’s correct? If your microwave waveshapes are repetitive, or a test can be designed to use repetitive signals then the SXRTO should be your choice. In fact we don’t care how slowly repetitive or even how irregularly repetitive our waveshape is. Internal full bandwidth trigger paths will catch each repetition and re-construct its precise wave or eye shape and timing; before, during and after the trigger event.
That said, if you need to capture unique events, or long data or modulation streams, the SXRTO limits you to 250 MHz frequency content or 500 Mb/s data rate and memory length tops out at 250kSa.
TB: So, how does the SXRTO compare with the Sampling Oscilloscope?
Ashcroft: The sampling oscilloscope remains a highly cost effective and typically very high bandwidth and very high resolutions instrument.
These oscilloscopes sample using the sequential ETS technique. They are completely reliant upon repetitive waveshape, but also impose some additional limitations. For example waveform samples cannot be taken until sometime after a trigger event and a trigger cannot in general be internally tapped off from incoming signal. A sampling oscilloscope therefore can be more complex to use and has no ability to capture the trigger or “pre-trigger” waveshape.
I can see your next question coming …“The signal is repetitive, so surely I can capture the next event?”
True, but if the timing of the next event is uncertain or jittery, then I have a problem. The truth of this is that the sampling oscilloscope is very capable in certain applications but in others we might have to arrange for a separate, leading and jitter free trigger signal to facilitate capture of our desired event. Fabulous instruments, but in practice with niche application and not the general purpose tool that you might wish it to be.
TB: It sounds as though the SXRTO is a step back from the leading edge of available technology?
Ashcroft: Yes indeed it is. In fact I, and Pico’s SXRTO design team, were designers of the very first digital storage real time oscilloscopes. We know that the sub-Nyquist oscilloscope is not new. What we have all perhaps forgotten is that until the costs of leading edge data bandwidth fall, the sub Nyquist SXRTO will have its place, and that over time, that place has and will continue to move up in bandwidth. I would argue that in 2019 that place for a lower cost oscilloscope option lies above 1 GHz. I would also observe that the costs, real estate and power demands of sub-Nyquist data bandwidth have all fallen very significantly indeed. Hence the elegant portability and price of our 9404-05 SXRTO.
TB: What do you see for the future of high frequency oscilloscopes?
Ashcroft: If I plot a graph of typical price versus bandwidth, and I plot two curves, one for time domain instruments (e.g. oscilloscopes and acquisition cards) and one for frequency domain instruments (e.g. spectrum and modulation analysers) then I see time domain instruments at higher and rapidly diverging cost. I also see frequency domain products driven by and keeping pace with ever higher demands for analysis bandwidth. The two product types are very different of course, but increasingly, where a frequency domain measurement can do the job, it can be the lower cost approach. Time domain instrumentation needs to respond to this. I believe that the SXRTO very much does that, and that in principal it is extensible to all existing RTO bandwidths and capture lengths. Further, the relatively low cost per channel of the SXRTO will suit embedded ATE and other multi-channel applications. I am certain that the SRTO will gain an important share of the hf T&M market.