| Electronics > Projects, Designs, and Technical Stuff |
| Idea for reasmpler to view high frequency waveforms on a soundcard. |
| << < (3/4) > >> |
| SiliconWizard:
--- Quote from: thinkfat on July 06, 2019, 07:47:46 am ---Keep in mind that this scheme only works for repetitive signals and since you will be reconstructing your low frequency waveform from multiple, consecutive high frequency waveforms they strictly speaking have to be identical, too. I don't see a practical value in this. --- End quote --- This is basically the purpose of what are often called "sampling oscilloscopes". They can find many uses, for instance to make measurements for very high frequency oscillators, eye pattern measurements, and so on. It allows the use of high resolution/relatively low sampling rate converters. Doing the same kind of acquisition as repetitive single-shot doesn't even necessarily make sense, and would require equipement that would be hugely expensive (to deal for instance with GHz or even THz signals), or that doesn't even exist... Just an example: https://www.picotech.com/oscilloscope/9200/picoscope-9200-sampling-oscilloscopes Of course to deal with bandwidths of a couple MHz, that doesn't really make sense, and this is more of a curiosity than anything else, although again it can allow to get high resolution captures if done right (something that would already be pretty expensive for an amateur's budget when using real-time sampling). For a quick overview: https://www.electronicdesign.com/test-amp-measurement/what-s-difference-between-real-time-and-sampling-oscilloscopes |
| thinkfat:
--- Quote from: SiliconWizard on July 06, 2019, 02:40:26 pm --- This is basically the purpose of what are often called "sampling oscilloscopes". ... --- End quote --- Which means, they're not general purpose DSOs and resampling is not simply a neat trick to achieve higher bandwidth. They are specialized tools that come in handy when a particular waveform is not of interest but certain general, non-realtime characteristics of a signal. Gesendet von meinem Nokia 6.1 mit Tapatalk |
| Benta:
Don't forget: you're not going to defeat Shannon-Nyquist with this, the sampling theorem stands. Resampling to lower frequency is indeed possible, though of limited use. BUT: the bandwidth of the sampled signal has to be limited to the Nyquist frequency. On top of that, your sample-hold circuit needs to work correctly at those high frequencies. Not an easy task. |
| David Hess:
--- Quote from: thinkfat on July 06, 2019, 07:47:46 am ---Keep in mind that this scheme only works for repetitive signals and since you will be reconstructing your low frequency waveform from multiple, consecutive high frequency waveforms they strictly speaking have to be identical, too. I don't see a practical value in this. --- End quote --- The predictable frequency response of a well designed sampler make them useful for calibrating other instruments including reference level pulse generators. If you have 3 samplers, then can be used to calibrate themselves without any external calibration source. --- Quote from: magic on July 06, 2019, 09:32:18 am --- --- Quote from: David Hess on July 05, 2019, 10:36:32 pm ---Here is the same idea but with 1 GHz bandwidth: https://www.electronicdesign.com/boards/1-ghz-sampling-oscilloscope-front-end-easily-modified --- End quote --- I have seen this article a while ago while researching high speed sample and hold solutions. I think there are certain problems with it which the author glosses over a bit. --- End quote --- The author used a very simplified and low performance sampler implementation even given his construction requirements. --- Quote --- --- Quote ---The actual sampling event happens in the track-to-hold transition. It takes place during the few hundred picoseconds in which the master Schottkydiode bridge resistance switches from low to high. Such switching occurs within a small, central part of the full 8-V applied step. --- End quote --- And here's the first problem: if sampling time isn't long enough for the capacitor to fully settle, its state will be measurably dependent on waveform preceding the sample, because the gate is ON for a very long time before taking the sample. Notably, I would expect a step response similar to trace B in the article. For sampling time of 300ps (1GHz BW), which is merely 2·τ, the initial step would reach 90% of full scale and then exponentially decay towards 100%. I don't know if adding speedup capacitor is a legitimate solution. --- End quote --- In a sequential sampling design like this, that is irrelevant because sampling only occurs on adjacent points on the waveform and the sweep can be slowed as much as necessary to allow the capacitor to fully charge. --- Quote ---There is probably a reason why Tektronix made efforts to produce very short pulses rather than just fast-falling pulses in their S-1/S-2 sampling heads. --- End quote --- Later Tektronix samplers (starting with the S-4?) use traveling wave gates which rely on the transition time of the trailing edge of the sampling strobe to produce the sampling gate time. --- Quote ---I wonder if it would be feasible to use a tiny capacitor directly between the sampling gate and ground, and a noncapacitive resistor branching-off to the buffer amp, such that only the small cap needs to be charged during those short picoseconds. The buffer amp would compensate for the loss caused by distributing charge from the sampling cap to other capacitances in the system. --- End quote --- Toss out the capacitor entirely and use an open circuit 1/2 wavelength transmission line tuned to the sampling gate time. --- Quote ---Another advantage of that approach is that it doesn't put the guts of a low speed JFET opamp in a wideband signal path. Who knows what's the frequency response of parasitic capacitance of TL-072 input pin and the supply bypass capacitors in series with it :scared: Another problem with TL-072 is it dependence of capacitance on common mode voltage. Ideally, it should run on as high voltage supply as possible and/or be replaced with something better-behaved. OPA1642 supposedly is good. --- End quote --- Tektronix used a common source FET amplifier (no common mode effects) followed by shunt feedback so they effectively differentiated the output from the sampling bridge allowing accurate measurements even with low sampler efficiency due to the time constant being too long for the sampling gate time. Bandwidth now only depends on sampling gate time and this also allows random sampling because adjacent samples may be uncorrelated. As you point out, the RC time constant is really not fast enough (and not well controlled) however this does not matter in a sequential sampler when the sweep speed is restricted. The input capacitance of the TL072 dominates here. I think the bandwidth limit of this configuration is limited to about 500 MHz if random sampling is used. Check out the designs of the sampler in the Tektronix 7854 which uses a similar configuration to get to 400 MHz. The rule here is to get as low a driving impedance and sampling capacitance as possible for maximum bandwidth. --- Quote from: Benta on July 07, 2019, 08:35:48 pm ---Don't forget: you're not going to defeat Shannon-Nyquist with this, the sampling theorem stands. Resampling to lower frequency is indeed possible, though of limited use. BUT: the bandwidth of the sampled signal has to be limited to the Nyquist frequency. On top of that, your sample-hold circuit needs to work correctly at those high frequencies. --- End quote --- The bandwidth has to be limited based on the equivalent sampling rate which can be arbitrarily high and is ultimately limited by noise. And even if the bandwidth is not limited producing aliasing, the resulting histogram is still correct. My Tektronix analog sampling oscilloscopes happily produce good results from DC to 10+ GHz with a sampling rate of 50 kSamples/second and equivalent time sampling rate up to at least 200 GSamples/second depending on the sweep configuration. Waveform amplitudes are accurate to the full bandwidth of the sampler even when massive undersampling and aliasing is present. |
| magic:
--- Quote from: David Hess on July 08, 2019, 04:01:48 am --- --- Quote from: magic on July 06, 2019, 09:32:18 am ---And here's the first problem: if sampling time isn't long enough for the capacitor to fully settle, its state will be measurably dependent on waveform preceding the sample, because the gate is ON for a very long time before taking the sample. Notably, I would expect a step response similar to trace B in the article. For sampling time of 300ps (1GHz BW), which is merely 2·τ, the initial step would reach 90% of full scale and then exponentially decay towards 100%. I don't know if adding speedup capacitor is a legitimate solution. --- End quote --- In a sequential sampling design like this, that is irrelevant because sampling only occurs on adjacent points on the waveform and the sweep can be slowed as much as necessary to allow the capacitor to fully charge. --- End quote --- It matters. You are thinking about S-1, where the gate is only opened for the duration of one sample. This design opens the gate for a few nanoseconds, completely annihilating whatever charge remained from the previous sample. Precisely, it's opened when the trigger is armed and closed when the circuit has been triggered and the sample delay timer expired. Then it stays closed for half µs to transfer the sample to another capacitor and it opens again. Each sample is therefore taken completely individually and anew. To get good results, the state of the capacitor right before the gate is closed must reflect what the waveform is now, not a nanosecond ago. So a really low time constant is needed, much shorter than effective sample time. It's a tradeoff. The sampling head must track much faster than Tek's, but its drive circuitry can be slow and easy to build. --- Quote from: David Hess on July 08, 2019, 04:01:48 am ---Tektronix used a common source FET amplifier (no common mode effects) followed by shunt feedback so they effectively differentiated the output from the sampling bridge allowing accurate measurements even with low sampler efficiency due to the time constant being too long for the sampling gate time. Bandwidth now only depends on sampling gate time and this also allows random sampling because adjacent samples may be uncorrelated. --- End quote --- Yep, that's what they did and it cannot be done here. At the same time, this circuit could do random sampling, just connect some external random sweep management logic to the timebase input. What you say about advancing a sequential sweep slowly enough to render τ irrelevant is another option, but one which requires very short drive pulses to isolate the sample from preceding waveform. |
| Navigation |
| Message Index |
| Next page |
| Previous page |