Budget DSO with 1mv/div or better sensitivity

[Mechatrommer]

dso averaging function will also help to remove random noise, but without proper or external triggering, its useless. with noise riding on a signal, if the signal is high enough to rise noise to certain level, built in dso triggering and averaging can still be usefull.

[robrenz]

@ David Hess

Thank you for taking the time to document that for me.     It is now on my to do list

[David Hess]

Just keep in mind that the original design of the vertical output circuit is not bad for what it was intended for.  The only reason the ripple became visible in my case is that I was using DSO averaging and external triggering to remove the broadband noise.  Under normal conditions, the broadband noise almost completely covers the ripple which comes from both the power supply and the sweep.

I stopped fooling with it once I made the simple change with the resistor and +50 volt supply because it worked so well at that point.  At some point in the future, I will add the transistor current source and maybe cascode and make it adjustable so I can calibrate the offset.

[pfm]

does dso averaging/resolution enhancement techniques remove noise from the original signal as well ? If the signal itself has noise which needs to be measured (and not hiden/removed) and displayed as well, then after dso averaging would it get removed in other words give an incorrect result ?

[David Hess]

Averaging removes asynchronous components of the signal including noise because they change over multiple acquisitions.  Resolution enhancement just low pass filters the signal.

What you see in the oscillographs that I posted is ripple from the linear power supply that is synchronous to the line frequency which was used as the trigger.  The noise which would have completely covered it up was removed by averaging multiple acquisitions.

If the sweep had been running then there would have been ripple from that as well and if the sweep was synchronous to the line frequency which is easy enough to arrange, it would have shown up clearly.  Or I could have triggered off of the sweep and used averaging to remove both the ripple from the power supply and any broadband noise.

Since all of the ripple in the vertical output could be controlled by increasing the common mode rejection of the differential to single ended converter circuit, I just picked the easiest one to measure and used that as my benchmark.

[Mark_O]

Let me see if I understood this right. If one were to use a very low noise low distortion external amplifier with enough bandwidth then you could use the scope at 1X setting and divide the min sensitivity of the scope by the gain of that amplifier. So a 10X amplifier would essentially allow the 2mv/div to be used as a 200uV/div. Ofcourse the conversion would have to be done manually. Is that correct ?

No need for manual conversion on the Rigols.  Even the cheapest 1000Z series unit provides Probe Attenuation coefficients of 0.01X-1000X, in 1-2-5 steps.  Many Agilents also provide 0.1X-1000X, in 1-2-5 steps, which supports up to 10X pre-amplification. 

[Siglent unfortunately is only 1X, 5X, 10X, 50X..., though it would be pretty trivial for them to support other factors.  It's just scaling.  But I have some home-made 20X 50-ohm probes that would be a PITA to use on them.]