Newsgroups: sci.electronics,comp.robotics
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!hookup!uwm.edu!vixen.cso.uiuc.edu!howland.reston.ans.net!EU.net!CERN.ch!dxcern!jeroen
From: jeroen@dxcern.cern.ch (Jeroen Belleman)
Subject: Re: Help: noise on op-amp output from digital signals
Message-ID: <1994Nov29.082317.29454@dxcern.cern.ch>
Organization: CERN European Laboratory for Particle Physics
References: <mark.stephens-2211940850140001@mstephens.gsfc.nasa.gov> <1994Nov23.210813.21410@mksol.dseg.ti.com> <mark.stephens-2811940930280001@mstephens.gsfc.nasa.gov>
Date: Tue, 29 Nov 1994 08:23:17 GMT
Lines: 53
Xref: glinda.oz.cs.cmu.edu sci.electronics:109646 comp.robotics:15775

In article <mark.stephens-2811940930280001@mstephens.gsfc.nasa.gov>,
mark stephens <mark.stephens@gsfc.nasa.gov> wrote:
>While caps of various sizes helped, I'm still getting noise, even with out
>the digital part turned on (the 7660 might be the culpret here).  And
>witht the digital turned on, the analog signals still manage to "see" the
>LED switching.
>
>Sigh,  I need to step back and approach this consistantly and logically.  

<Lots of stuff deleted>

Here are a few hints on how to go about this. It is difficult to keep
clean analogue signals in the presence of dirty fast logic like TTL.

Three coupling mechanisms are the most important causes for interference.
(Not necessarily in the order given.)

- Common impedance coupling. To trace that, look where the (signal)
  currents flow. If any high level or fast signal has a piece of wire
  or PCB trace in common with a low level signal, you are in trouble.
  Resolving these ultimately leads to the familiar star-wired ground
  and power nets. An innocent looking piece of wire may sometimes
  have enough impedance do develop a sizable voltage across its
  ends!

- Inductive coupling. Again, look where the currents flow. Find loops
  with high dI/dt. Minimize the loop area. For example, your LED sits
  in a loop formed by the wiring, LED driver, power supply and bypass
  capacitors. You might think of the LED current as flowing from the
  power supply, through the driver, through the LED, through ground
  and bypass caps back to the power supply. Keep that loop's area small.
  Twist the wires to the LED. Connect both its wires to the circuit in
  proximity to the driver.

- Capacitive coupling. Now look for nodes with high dE/dt. Keep them
  close to a ground plane, to confine the fields. Put grounded screens
  around them. Also look for high impedance nodes, such as amplifier
  inputs and the like. Keep them far from the previous kind. Keep
  them close to ground. Shield them.

If you keep the above rules in mind when laying out a circuit, most
of the time, everything should come out fine. Sometimes, however,
you may find that some of these rules conflict with other circuit
requirements. There are many tricks to minimize one or another of
the sources of interference which apply in particular cases. I won't
go into that now.

One more hint: Many negative voltage stabilizers oscillate unless
they are properly bypassed close to the package.

Good luck,
Jeroen Belleman
jeroen@dxcern.cern.ch
