Question about bandwidth of scope?

> Input impedance is the load the measurement places on the node being
> measured.  In the case you mention, a 1 megohm resistor, with a 30 pF
> capacitor in parallel.  If this is a 50 ohm impedance signal source,
> then the 1 megohm is not affecting anything (1 megohm in parallel with
> 50 ohms is still very very close to 50 ohms), but 30 pF and 50 ohms is
> about a 1 nanosecond time constant, so the capacitive load means the
> probe itself will significantly load down signals above 100 MHz
> (Z=1/2*pi*f*c = 53 ohms for 100 MHz !!!).
>
> 'Bandwidth' is where the measurement "rolls off" or is no longer
> accurately able to detect the amplitude at that frequency.
>
> '0 to 12 MHz +/- 3dB' means (to me) that the amplitude you measure can
> be off by 2:1 anywhere in the 12 MHz bandwidth, which is really lousy
(bad).

I think the 12MHz is the 3 dB point. At this point the input voltage
will only register on the scope screen as 50% of it's actual value.
3 dB must be with respect to a certain load, I suppose 50ohms or
perhaps 75ohms i'm not sure, which doesn't make sense, as at 12MHz
30pF=463ohms

(1) The frequency response of a scope input has nothing to do with the input
impedance.  The frequency response of a scope is measured by driving the
input with an ideal 50 Ohm signal generator.  The input impedance matters
little, and with most very high bandwidth scopes the input impedance is
often 50 Ohms.  The -3dB bandwidth is measured in this way.  That is the
response of the scope by itself.  No probe involved.  This is the ideal
conditions - scope alone.

(2) The input impedance of the front end of the scope is reactive, and has
both a resistive component and capacitance component.  Those are the figures
you quote.  The idea is to match/trim a probe to this input impedance.

(3) When you attach a probe (say 10:1), you adjust/trim the series
capacitance of the probe so that it maintains the 10:1 ratio across all
frequencies.  The Voltage divider is formed by the series impedance of the
probe and the front end impedance of the scope.  That is how the probe input
impedance ends up at 10M Ohms, while the scope is 1M.  However all probes
are designed to work/trim against a fixed range of input impedance.  So for
a scope with a 15pf input, a probe that can trim against a 30-60pf input
impedance would not trim right.  They tell you what the input impedance is
so you can select a probe that will match.

(4)  The 12MHz bandwidth you quote is very low.  For any digital work today
you need at least a 100MHz scope.  Most 10:1 probes today easily go to
100MHz or even 500MHz.  In your case the most limiting factor is the scope.
12MHz is 20-30 years out of date.  You cannot troubleshoot modern circuitry
with 5nS edge times using a scope with 12MHz bandwidth.  The 5nS edge rates
represent about 200MHz bandwidth.  Your signals could be ringing everywhere
causing massive errors in the logic and you would never even see it.

You can by a 100MHz scope on Ebay for about $100.

Regards,  Chris.