@radiolistener:
If you can get me a good, working one, with reasonably easy-to-find spares, dual-channel, one of them invertible, 20 MHz analog scope that can be delivered to my door for €150, no matter if you got it for $80 or for free, I'm ready to pay that to you, just send me a private message and a Paypal link.
@aldo 22
Some background, available to you, in my previous post about aliasing. Follow the link.
Signals have harmonics. Particularly square waves *need* to have harmonics or hey couldn't be square. These would be, I think, higher order odd harmonics. So a square wave at 10 MHz does have higher frecuency components that would be 3x, 5x, 7x the main frequency, and so on. These harmonics have increasily lower amplitudes. Please see this animation:
https://youtu.be/cUD1gMAl6W4Now, to sample a signal, Nyquist theorem says you need to take samples at at least 2x the frequency, or you wouldn't be able to reconstruct that signal with any fidelity. Worst case, if you don't know signal phase, you will need to sample that signal at least at 4x the frequency because of, well, aliasing. I think these sine waves you showed at 60 MHz show aliasing, because I doubt your signal source puts out a different amplitude at each cycle.
Now, Tektronix et al will say to you that 2,5x the frequency will be enough if you use sin(x)/x interpolation to reconstruct the signal from your samples. Anyway, it's expected that your device will be able to filter higher frequencies that could make aliasing appear, because they can't be sampled at least 2x, 2,5x, 4x, whatever, their frequency. So, to fight that aliasing, you need an anti-aliasing filter, that is, a low-pass filter that cuts frequencies higher that some cut point. That filter could be as simple as an RC network.
Now, probably you already understood that radiolistener is being more papist than the pope. First of all, that statement about needing to have at least 1Gsample per second or else you have a toy. I can't agree with that. Even without antialiasing filter, a device will be good enough up to some frequency. It's just that the device would be only good up to a quite lower frequency. Please correct me if you can prove I'm wrong. I would love to know.
I usually consider the frequency where a square wave becomes a sine one, to be a good rule of thumb about a device's bandwidth. That happens because the scope can't cope with the higher odd harmonics anymore.
That said, no scope is able to show a square wave up to its advertised frequency, at least in our hobbyst price range, I think. But radiolistener has a point, a serious device should have an antialiasing filter. To put in the market a device able to do 100 Ms/s without putting an atialiasing filter to cut off frequencies above 25-40 MHz would be an incredibly stupid thing to do IMO. We all know for sure DSO2512 can't be trusted up to 120 MHz. But, if it doesn't have any antialiasing filter, it can't be trusted even at 25 MHz even if it does sin(x)/x interpolation. So, a device that could be good to do some low-end but serious work, like looking at PSUs, becomes a toy. I wouldn't buy that device for a $100. I already have my €40 DSO154 as a toy to learn just the more basic things. I would look instead for something that can reliably work at 20 MHz, even if the price is 2x that of DSO2512.
I'm afraid the purpose in going to market without antialiasing filter could be only to scam newbies. As radiolistener said, that way you could claim you have higher bandwidth, but it would be no more than a scam, and a incredibly stupid one because it makes a potentially useful device just a toy. Thinking nobody would be able to notice it is even dumber. So there could be some base for comptent, yeah.
But before throwing any comptent, DSO2512 should be carefully tested to seet what it's good up to. That's never academic. I hope that, once TinySA comes to you, we all will know the truth, should you be willing to do these tests.