After nice bode plot examples by @Tautech
About aliasing and scope BW.
If we have 500MSa/s sampling and oscolloscope front end is nominally 200MHz or 100MHz and front ends do not have quite rare "brickwall" response we have mostly slowly decaying shape after nominal frequency limit (-3dB points). After then front end can still transfer more or less attenuated lot of higher frequency components to ADC input.
If they are included to ADC sampling they produce aliases if these components are over Nyquist frequency if we have ideal Sinc construction and infinite data. But we do not have. Depending about what level of accuracy we talk and depending Sinc function parameters etc we can say some rough thumb rule that mostly visually somehow acceptable limit is around 70 - 85% from Nyquist limit. Depending Sinc and depending also how critically we look this. But, if we need really example measure with higher accuracy risetimes we need throw away this kind of over simplified thumb rules.
No need now go dfeeper but who want, can easy find some calculation examples for find real needs related to needes risetime but also samplerate.
Back to main road.
This image is example what show that two scopes many ways same (same Sinc performance etc) and both 500MSa/s. Other front end is 100MHz and other 200MHz. It can clearly show that 100MHz BW can give some protection against aliasing. Naturally, because 100MHz model ADC (both have same ADC) see less these frequency components what go to area where aliasing start.
Both scopes get equal signal. All is equal exept front end before ADC.
There is some possible that 100MHz model have also some differencies in signal handling after ADC but, once produced aliases can not remove by software and only what poroduce aliasing is ADC so around all what matters is what signal ADC see. (this have totally nothing to do with screen pixel "aliasing" so do not mix this alising here at all - it is only cosmetic for eye). THis test signal most important think is rise and fall time. Not so much its shape other way. These edges are these what are important for harmonics what go far over both scopes Nyquist frequency and very specially when sampolerate is 500MSa/s (SDS1202X-E both channels in use)
Images marked A,B,C and D show this effect.
This effect of course dissappear if I remove these high frequency components from signal before scope input. Example using slower rise times.
Here need note that this used signal is far over these both scopes performance. If you need accurately measure this kind of signal these both oscilloscopes are not suitable. They can not hadle well this kind of signals higher frequency components specially if sample rate is dropped to 500MSa/s. With 1GSa/s there is not so much aliasing but still signal rtise times are far over these both scopes. But with 500MHz Nyquist and even 400MHz (0.8 Nyquist) ADC do not see lot of these higher components due to front end attenuation also with 200MHz model)
If our signal do not exeed scope performance this problem (aliasing) do not exista. And here come this - is it really wise to hack BW up if it leads high or very extreme aliasing. If want better oscilloscope by own modifications, one good place is analog front end before ADC and there making BW shape more like flat top and then very fast decaying so that 0.8 * Nyquist is highly attrenuated.
It is very different if we are talking oscilloscopes what do real repetitive aka equal time sampling for continuous signals. There we can tweak front end up without first hit aliasing problems.
There in image is last parts where oscilloscopes are in DOTS mode.
These oscilloscopes dots mode is not at all like repetitive sampling (equal time sampling) dots.
After scope and actual signal is in "aliasing position" this dots mode also is affected due to aliasing but it still looks like it draw only single line (sequential dots on the screen looks more or less like line depending dots density) but line is more or less bended (some times they may look very weird) but in some cases it do not give this warning about aliasing .. like lines display modes give due to lines "wobbling" areas. (some times you see only example corners very fat in lines draw mode (linear or sinc) but if signal itself do not have this wobbling it is always ign about aliasing what user need regognize as warning about possible aliasing. But this dot mode do not give directly this warning. (exept if user have experience and know that there is now something wrong)
Of course wise user immediately check aliasing turning changing from dots mode to vector (lines) mode and then linear or Sinc. Some cases linear is better for digital pulse/squarte type signals even if they are not so nice looking cosmetically. (not Sinc overshoot. Compare images marked A and C where A show lor of Sinc overhoot.)
Home
Help
Search
Login
Register
