[SOLVED] Oscilloscope for SMPS

Hey guys
Please suggest some good oscilloscope for SMPS work. I will be mostly working in the frequency range of 200-500kHz. I am going through some data-sheets but I would like to hear from experienced people.

I really like Saleae Logic: you use it with your computer and it is small and compact. I have the first gen Logic Analyzer only but the newer ones have dual logic and analog channels.

I am looking for very good professional oscilloscope. Saleae logic looks good though.

I'd spend more thought on the probes than the mainframe.
Current loops and tip loading can really make a mess of
measurements, give you more "ghosts to chase" and so
on. This extends to product and evaluation prototype
design, you really would like to use those cute little
'scope grounding jacks rather than clip-lead ground,
might like FET probes to look at whether all the SW
node ringing is the board, the powertrain or all an
artifact of the extra 10pF to ground you added just
by looking at it (kinda Heisenberg-y).

De-Q'ing the tip with a series resistor can help with
the "false ringing" at the cost of some bandwidth.
Finding what's real and what's not, can be a cut-
and-try-series exercise.

A 500MHz, 5GSPS 'scope with 200MHz passive probes
like my old, used TDS3054 is good enough for most
things. Although data export is via floppy and only
my one old lab computer still has one of those....

If your SMPS if off-line, you have the additional concerns
of HV safety (the equipment's - engineers are replaceable).
Look at the front panel and probe voltage ratings, some
newer "PC" 'scopes skimp on this aspect (and you may
have to wade into the poorly translated fine print to get a
proper whiff).

Thats great I will give importance to probes also.

Look at rise time of signal you wnat to see.
Highest Rise time = 0.35/Bandwidth..generally
******************************************
Oscilloscope selection

Digital scopes:
Remember you need a sampling rate of 2.5 x the frequency you want to see.
You need a analog bandwidth of 2 x the frequency you want to see.

But if your memory length is not big enough, you still wont be able to depict the signal.

If youhave it on 100us/div (1ms in total scope screen window), then a memory length of 100 words means that you capture data point every 1us..so that’s a 1Ms/sec sample rate…..even if you scope spec says its got a 100Ms/sec sample rate!


Rise time:
For rise time
Generally you need a scope bandwidth of 0.35/(rise time).

For rise time, think of a sine wave and the steepest rising gradient part, and the length of that, and what the bandwidth of the sne would be with that rise time “section”