O'scope Noise Measurements

read AN70 from here

martin

with

Hi Rich, Please see my post in abse (alt.binaries.schematics.electronics), subject line -- Re: O'scope Noise Measurements I have a drawing of a method that minimizes noise.

"Rich"

For the simple reason they have too much input noise of their own.

First determine the * noise bandwidth * you need to measure.

Then estimate the minimum noise level you need to achieve.

Then use a low noise *pre-amplifier* that has that bandwidth PLUS enough gain to raise that minimum level to at least 10mV.

The scope then reads the output of the preamp.

Preferably use a wide band, true rms millivolt meter too.

Connection to the DUT is best done with a co-axial cable, either by soldering the ends or using a BNC plug & socket.

....... Phil

About the best you can easily do is get a piece of coaxial cable, BNC on one end into the scope, and solder the shield/center to the board on the other end, using a ground pad close to the signal pad. It helps to wind it through a ferrite core midway along the cable, to break up RF ground loops between the scope and the DUT. This technique of cource assumes a low-impedance signal that isn't changed much by the coax+scope capacitance.

Designing a few SMA connector footprints into your board layout, to snoop critical signals, is great, if you think about it beforehand.

Regular passive 10:1 probes are terrible for characterising wideband, low-level noise on a board. Fet probes are a lot better, but still cost you 10:1 on amplitude. There are exotic/expensive active differential probes, too.

John

What about using a 50R coax with BNC unterminated at the scope and at DUT tie shield to ground and tie signal thru 50R to center coax lead. Keep shield/center and 50R as short as possible. Will that not work for high gain/high Z scope measurements up to 20MHz? Regards, Harry

A 450-ohm chip resistor soldered on to the centre conductor of a flush-cut end of RG58 makes a pretty good X10 probe.

Cheers,

Phil Hobbs

Caddock makes some axial resistors that plug directly into an SMA female and make an instant 10:1 or 20:1 probe, 0.25 pF equivalent, good to 6 GHz or so and fine for hundreds of volts peak. HP used that idea in their 54006A probe kit.

There's something magical, I guess, about the resistance/distributed capacitance of those resistors. We've used them as pickoffs in our high-voltage pulsers.

Hey, take a look at the price!

John

Hi, John -

I could not find such a beast on Caddock's site. Perhaps I'm too blind. Do you have a link?

Thanks, John

They're the MG series; see pics in a.b.s.e.

John