Sin(x)/x interpolation and Digital Filters in Oscilloscopes

[robrenz]

The explanation makes sense and I thought that might be what you were trying to say but was not sure. And it is a good point

[amspire]

It would be nice if the Chinese DSO manufacturers made sampling accessories ... the 10s of GHz sampling ADCs are too expensive for their target market, but a high frequency sampling head doesn't really need to be that expensive.

Of course they can't even be bothered designing active probes, so fat chance.

I agree. A 20GHz sampling head could probably be sold for $50 if it was made in reasonable quantity. I remember reading a Sampling scope designer who claimed that it was not hard to make a 10GHz sampling head with off-the-shelf components (if you know exactly what you are doing). There is skill in the design, but after that, not many components. 

Triggering is another matter. The performance of a sampling scope is all about an accurate trigger with jitter in the picosecond range. The modern sampling scopes often have to offer a range of triggering options to cover different situations.

Still, a cheap fast sampling head would be a great start.

There is a simple design that can be used for input frequencies up to about 1GHz from memory. Can't find the article, but here is the circuit. It can be used with any scope with a X-Y view option. Speed is not important.

Edit:  Google to the rescue again. By searching for the text in the image, I found the original article which is great - the first time I have seen it.

http://www.redrok.com/Circuits_1GHz-samplig-Oscilloscope-Front-End.pdf

300ps risetime, and you only need a 10MHz scope. The RC time constant of the sampling bridge is 100ps.

[Mechatrommer]

It would be nice if the Chinese DSO manufacturers made sampling accessories ... the 10s of GHz sampling ADCs are too expensive for their target market, but a high frequency sampling head doesn't really need to be that expensive.
Of course they can't even be bothered designing active probes, so fat chance.

I agree. A 20GHz sampling head could probably be sold for $50 if it was made in reasonable quantity.

is this implying "realtime" sampling? not "equivalent sampling". rigol 1052e can do "equivalent" 25GSps if its "realtime sampling" i highly agree, but then we need another device to read the memory and display that on monitor right? another price.

[amspire]

A sampling head goes in front of another oscilloscope or digitizer. A 20GHz head could have over 20Ghz bandwidth to the sampling gate, and a sampling aperture in the tens of picoseconds.

The DS1052E has a 50MHz analog front end and a sampling aperture around 2ns.

[Marco]

I remember reading a Sampling scope designer who claimed that it was not hard to make a 10GHz sampling head with off-the-shelf components (if you know exactly what you are doing).

Dunno, I think you have to fit the critical part in a fraction of the wavelength ... doable at 1 GHz, getting a bit hard at 10 (maybe there are some wizards who can solder bare die components by hand, but not me).

[Marco]

is this implying "realtime" sampling? not "equivalent sampling". rigol 1052e can do "equivalent" 25GSps

It's equivalent sampling, but the front end on a rigol doesn't have the bandwidth to make it worthwhile. A sampling head is a separate sample and hold circuit, generally with less attenuation options and input protection to make it easier to achieve high bandwidths.

[Smokey]

It's old thread time!

An interesting paper from LeCroy about sinec interpolation.  It's obviously showing off LeCroy stuff but he beginning is an interesting bit about the limits of sinc interpolation and when it's valid in theory.

http://cdn.teledynelecroy.com/files/whitepapers/wp_interpolation_102203.pdf