| Products > Test Equipment |
| How much noise floor and other things matter in oscilloscope usability |
| << < (16/63) > >> |
| David Hess:
--- Quote from: Fungus on December 24, 2021, 08:03:33 pm --- --- Quote from: bdunham7 on December 24, 2021, 07:50:14 pm ---Meh, ringing on step responses isn't usually Gibbs unless something has been designed or set wrong. --- End quote --- It can be a mixture of both. --- Quote from: bdunham7 on December 24, 2021, 07:50:14 pm ---You can have ringing (and even pre-ringing) on an analog scope. --- End quote --- But you can't have Gibbs. ::) --- End quote --- I have wondered about that. Preshoot in analog oscilloscopes comes from the delay line implemented as either a lumped-element transmission line or counter braided differential coaxial line. Both have a sharp cutoff frequency and faster propagation of high frequencies than low frequencies, but that sharp cutoff reminds me of the truncated Fourier coefficients which lead to the Gibb's phenomena. --- Quote from: Fiorenzo on December 24, 2021, 09:12:21 pm ---With 1mv/div It gives about 160mV RMS and 1,2mV peak to peak. Whatching other scopes they seem to have about 4 time less noise. Am I wrong? --- End quote --- It is possible but as I pointed out, noise is not the only consideration. My preferred oscilloscope has about 120 microvolts RMS input noise because it comes with high input common mode range and differential inputs which are more useful, similar to the AM502 mentioned below. --- Quote from: Fiorenzo on December 24, 2021, 07:30:31 pm ---Trying to do a recap: at this point It seem to me that a "low noise" oscilloscope is important when working with FFT analysis, power supply ripple, audio signal, and high impedance circuits? --- End quote --- Like I wrote earlier, except for FFTs where noise is a direct limitation, it is not very important. Oscilloscopes are noisy because of the compromises they have to make. When making measurements, other sources of error usually overwhelm noise. Ground loops with single ended probes and the probe ground lead will pick up all kinds of noise in excess of an oscilloscope's front end noise. This suggests that money is better spent on better probes than an oscilloscope with the lowest possible noise which cannot be taken advantage of anyway. Power supply ripple is separate from power supply noise, and a DSO can be triggered and use averaging to remove the noise and keep the ripple. Measuring power supply noise on the other hand will often require a low noise preamplifier which is not difficult to build. In audio applications, an oscilloscope has so much distortion that only gross measurements will be accurate, so noise is not relevant. If you are interested in general purposes low noise measurements within a 1 MHz bandwidth on any oscilloscope, then you might find a Tektronix AM502 differential amplifier to be useful. They are easy to repair because unique parts can be sourced from the Tektronix 5A22 and 7A22, and they work well with 1x oscilloscope probes. Using one does mean acquiring a TM500 series power supply mainframe though. --- Quote ---In regard of the mso5000 with its high sample rate of 8GSa/s I am not sure if in the balance It is an advantage due to its apparently noisy front end. As an ignorant, at the beginning I thought: the Rigol is better because It has better specs, so I bought it. Could you suggest me a different model if you think It could be better? --- End quote --- Noise specifications are generally lacking so it is difficult to make a recommendation based on them. Something to consider about the 70 MHz Rigol MSO5072/MSO5074 is that it is bandwidth upgradable to 350 MHz via firmware meaning that it has the higher noise 350 MHz front end whether the bandwidth is limited to 70 MHz or not. So it must have a higher noise than the example I gave of 100 MHz oscilloscopes. What is going on here is that for a given device technology, higher bandwidth yields higher noise density. So for instance a JFET front end which supports a bandwidth of 100 MHz could have a noise density of 3.5 nV/Sqrt(Hz), while a JFET front end which supports 350 MHz would have a noise density several times higher, whether the bandwidth is limited or not. For the same bandwidth, a lower bandwidth oscilloscope can have lower noise than a higher bandwidth oscilloscope. I suspect that is a major part of what is going on with the relatively high noise of the MSO5072/MSO5074. Its noise should be compared to other 350 MHz instruments even though it is limited to 70 MHz. |
| bdunham7:
--- Quote from: David Hess on December 25, 2021, 12:40:56 am ---It is possible but as I pointed out, noise is not the only consideration. My preferred oscilloscope has about 120 microvolts RMS input noise because it comes with high input common mode range and differential inputs which are more useful, similar to the AM502 mentioned below. I suspect that is a major part of what is going on with the relatively high noise of the MSO5072/MSO5074. Its noise should be compared to other 350 MHz instruments even though it is limited to 70 MHz. --- End quote --- I also would put up with a bit of noise to get differential inputs. The isolated scopes that I have are also a bit noisier and I don't complain. As for the Rigols, I really think the issue is how they manage the lowest ranges by using a digital expansion of a higher range. Other scopes would have an additional 10X analog gain. The scope I used for comparison is 200MHz+, the more comparable SDS2000X+ models are 500MHz+ and have slightly better noise levels than mine. Comparing 300-500MHz class scopes with the old Tek 2465B that I have--which gets very close to what you said is the ideal noise level-- the better Siglent is maybe 50% noisier (hard to compare because the Tek doesn't have 1mV or 500uV/div settings), the cheap Siglent is 2X as noisy and the Rigol is about 8X. It's like going from a fine-tip ball point pen to a Sharpie and then to a highlighter or magic marker. Of course that is not using ERES or 10-bit or any other DSO tricks. |
| G0HZU:
My background is RF so I generally don't use a scope very often but when I do it is nice to look at small signals. I have a really old HP Infinium digital scope with 500MHz bandwidth and it has the noise performance I would expect. At full 500MHz bandwidth and with the input set to 50R termination and 1mV/div it shows about 100uV rms noise when the Vrms measurement is enabled. When fed with noise as the signal under test it can typically measure wideband or narrowband noise signals with acceptable results down to about 200uVrms. This scope is fairly limited in terms of features (and memory depth) compared to modern scopes but it does at least have the noise performance I'd expect from a scope like this. |
| knudch:
Random Noise is one thing.. If you @Fiorenzo would try on your Rigol a thing like this: (done 50ohm termination) Discrete "false signals is something else |
| G0HZU:
Here's my old HP Infinium scope set to 1mV/div with the 30MHz bandwidth limit enabled. The Rigol scope was showing about 1mVpkpk on a 20MHz bandwidth setting which seems really noisy in comparison to my HP scope from the 1990s. My advice to Fiorenzo is to consider something else if the poor noise performance of the Rigol is bothering you. That level of noise would put me off buying a Rigol scope although I doubt I would ever buy one anyway. |
| Navigation |
| Message Index |
| Next page |
| Previous page |