[sdiy] 100 MHz EMI, what can it be?

[Magnus Danielson]

From: "Czech Martin" <Martin.Czech at Micronas.com>
Subject: RE: [sdiy] 100 MHz EMI, what can it be?
Date: Mon, 14 Apr 2003 13:33:43 +0200

> It's actually not so much the circuit, (it did not fool me too long)
> the scope allone will catch the RF by it's long ground clip cable.

When I do measurement near or above 1 GHz I pay particular attention to the
ground-lead. When doing TDR's for instance, few probes actually cut it and the
best result is acheived by actually soldering the coax straight to the board,
second best is to solder in a SMA contact (always have loads of SMA-contacts
and coaxes with SMA on only one end lying around). If your into radio-stuff you
want to pay even more attention to things like impedance matching, due to the
harder requirements on dynamics.

> The RF is much higher than the scope input amplifier noise.
> I think there was a Pease Porridge about "how can you have so clean
> screenshots when mine are so noisy".

A good read. There are also good books to study from which gives you a really
good idea of what you need to do for real life measurements.

> For mV measurments the ground clips of scopes are pretty ueseless.
> But I havn't seen anything better, it is a mechanical problem also.
> Of course, this depends on bandwidth, a lot.
> A 20MHz scope simply won't see much of the 100 MHz RF stuff.

The input bandwidth dampens stuff out. Low input bandwidth can actually be a
good feature for some stuff.

> Anyway, I'm a good candidate for a "banana, XLR, telephone plug, shielded or
> not shielded, balanced or not balanced measurement comparison".
> The air is full of RF, the railway will add 16 2/3 Hz power line noise and
> magnetic fields to that. Could you ask for more for such a test? ;->

You must recall that the balancing effect is a low-frequency property. It
usually planes out at about 100 MHz. There you *really* want the transmission
mode go over to a coax-behavior. It is only when you spent sufficient time over
a wide range of frequencies that you start to learn how fortunate it is that
audio is 20 Hz to 20 kHz. It's almost like DC and many of the techniques can be
made to work very well.

It is really fun to hunt leaking frequencies with E and H field probes (your
groundclip on probe is a simple H-field probe) and especially the nearfield
H-field probe (diffrentially coupled double-loop) is really usefull.

> I think I will get to that on easter, at least I hope so.

Good luck!

Cheers,
Magnus - probably emitting a 5,000 000 000 MHz buzz unintentionally