Digitizing vs analogue scope on low level signals

Analogue scopes do have lower noise on low level signals?

I have an HP54520A digitizing scope and when I set the scale to 10 mV/div (scopes lowest limit), there is noise present on the screen (on the first division on the 10 scale) and I can actually not being able to measure anything within the first division.

Is that a common thing with digitizing scopes and analogue ones do not suffer from this?

A second question: The frequency counter in this scope is wandering around in the last 2 bits all the time (even on average reading) so I cannot exactly measure the frequency as I do with an external frequency counter (HP 5245L).

Is that normal? Is the frequency calculated from analyzing the waveform instead of classic level triggering a counter, this is why?

Analog scopes can have lower noise and resolution. For example
I have a 7000 series Tek mainframe and a high G differential plugin
for it that can resolve uV, 10 uV/div, and a broadband noise of 16 uV,
1Mhz BW, at low source R. And filters to reduce noise thru BW reduction :


There are digitizing scopes with higher res A/D converters in them, and signal
averaging that helps. But not matching specialty analog plugins. Also speicalty
amps, some using CDS, to eliminate 1/f noise and offsets :


Your frequency measurement questions best answered by attached. Basically all
counters have LSD uncertainty.


Regards, Dana.

danadakk said:

Analog scopes can have lower noise and resolution.

Click to expand...

Not generally. May be if you refer to first generation DSO like said HP54520A. Front end of modern DSO can well compare with classical high end analog scopes, e.g TEK 485. In addition, by using averaging on periodical signals or digital filters, modern DSO can display low level signals in noise floor even better.

When you compare oscilloscopes, don't forget to configure similar analog bandwidth, noise level scales with sqrt(BW).

Beware of measurement errors with the quality of the probe cable and the length of the ground strap.

It is a well-known phenomenon that the ground strap acts a loop antenna with any magnetic pulses nearby then resonates with the coax capacitance to cause ringing.

To isolate your fault. report the noise Vpp;
1. with no probe attached.
2. with BNC input shorted if 1) is too high.
3. with probe tip to gnd and clip to circuit ground.
4. with probe clip and ground clip removed and ring and tip shorted to circuit ground.

If 4 is significantly better, then use only this way with the proper spring probe adapter to tip&ring.

Adding an external shield to coax can also reduce large noise pulses. This can be a gauze wire connected at one end.

For measuring ripple on DC, only use AC cap to coax only and terminated by 50 Ohms on DSO. This will reduce your measurement errors using rather than using 10:1 probe unless using a spring tip.

FvM said:

Not generally. May be if you refer to first generation DSO like said HP54520A. Front end of modern DSO can well compare with classical high end analog scopes, e.g TEK 485. In addition, by using averaging on periodical signals or digital filters, modern DSO can display low level signals in noise floor even better.

When you compare oscilloscopes, don't forget to configure similar analog bandwidth, noise level scales with sqrt(BW).

Click to expand...

My curioisty raised I looked at 485 specs, no noise specs at all. 465, 475 series same, no specs.

Then looking at 2020 Tek stuff, the 12 bit scopes ENOB ~ 6 - 8 bits. 1 mV sesnitivity kinds of scopes.
Same for Rhode-Schwarz....

Looking for a table of SNR or ENOB versus size of averaging sample set, cant find
but as I recall not a giant leap if one does 256 samples in a set.

And of course averaging only real effective if the signal has no correlation.
If it does the effectiveness drops.

Will contiunue to look for the table....I know I have it somewhere.


Regards, Dana.

If you have a good quality DSO, there is no reason why you cannot get textbook-looking waveforms with proper calibration and measurement tricks, err I mean skills.

My old TDS3054 has, I believe, an 8-bit vertical resolution. That's
pretty crude by modern signal standards but this is not a production
test instrument ("indication only") and the math functions can get
you some finer resolution by oversampling (though it's on you, to
decide whether you're throwing away important information or
"just noise").

An old CRT analog 'scope does not have a "binning" limited
resolution although it does carry its own inaccuracy-terms. It is
however likelier that you could see tiny features which would
get "swallowed" by digitization, even if accuracy is not real well
known.

For "chasing ghosts" a variable persistence analog 'scope can be
a real good tool. But such 'scopes tend to be old and low BW, and
modern 'scopes (too rich for my blood) do have some sweet
triggering (not just level & slope). I do keep one old blue Tek
boat anchor around, just 'cause it won't die, with all the plugins.

@dick_freebird

I have several 7000 mainframes (I am a hoarder) and almost all the plugins.
Sans most of the spectrum analyzers, just sold 2 7L5's, dont need them anymore.
And their scopes back to early 50's. I have the dual beam, 2 sections, 200 tubes,
major lab heater. Have not fired that up in a long time. 545 and 585 quite a scope
and all its plugins.

Crazy thing is in all my collecting almost all tube scope fired up w/o issues
after decades of storage. Cant say that about the transistor going forward versions.

Simply beautiful series of instruments.

The 7834 storage model just fried its power supply, so on my list to fix.

Still use the 7A22 diff plugin and the 7CT1 tracer. But I find most of my time
now on low end DSO 100 Mhz. Functions fine but quality nothing like the
old Teks.

I do have a 7104 full 1 Ghz analog, love that as well.

Nerdism is a serious disease, and I have a lot of it.


Regards, Dana.