Need both wideband (100MHz minimum) and low noise sensitivity (20uV) and lowish cost.Not asking much I know!!!!!!
We are used to that.
When talking about signals below about 1 mVpp, one of the very few options would be the Picoscope 4262, as has been already mentioned. It is only 5 MHz, but it is 16 bits and has a noise level of <5 µVrms (~81 µVpp). You seem to expect noise levels below 20 µVpp and for this I wish you luck. Of course, the noise can be lowered even further, by resolution enhancement up to 20 bits and digital channel filters, but all these measures reduce the bandwidth even further.
So, you’re unlikely find any 100, let alone 200 MHz instrument to show 20 µV resolution in the time domain, even if you were willing to spend serious money.
On the other hand, even 8 bit scopes, also with much higher bandwidth than just 200 MHz, with a
decent FFT, can measure signals that low in the frequency domain. Yet that doesn’t mean they can distinguish 5.00002 mVpp from 5.0 mVpp, if that’s what you want.
In other words: the 16000 count DMM with 1 count accuracy and 200 MHz bandwidth of is not going to happen.