Hi Chuck,

Here are some thoughts.

It sounds like your noise may be due to grounding or power supply 
issues. The comparator is going to take short current transients from 
the power supply each time it switches. Unless these transients are 
supplied from sources not shared with the analog circuitry there is 
likely to be noise getting through. By not shared, I don't necessarily 
mean different power supplies, but enough isolation in various part os 
the shared power supplies that noise in one section doesn't get into 
another.

Some op-amps have very poor power supply rejection at high frequencies. 
Some actually have gain from one of the power supplies to the output at 
some frequencies!

A traditional way to help this is with bypass capacitors from the power 
supplies to ground at the parts consuming the transient currents. The 
electrical path from each capacitor to the switching part it is 
connected to should be short (5-10mm is good), and the capacitors should 
preferably be multilayer ceramic of 0.1uF or more. There should also be 
some solid tantalum capacitors in the 20-50uF range from the power 
supplies to ground. Aluminum electrolytic capacitors, unless of a low 
ESR type, are probably not much use in reducing high-frequency noise.

One source is when these switching current spikes travel along 
conductors which are also used as analog grounds. The current spikes 
cause voltage drop across the conductors which shows up as a signal. 
This is more of a circuit layout problem than a schematic problem, but 
the overall goal is not to have any transient currents travel through 
conductors where a voltage drop along that path will affect the signal 
output. This isn't always easy or even possible, but is a good place to 
start. Sometimes isolating things with ferrite beads, as you mentioned, 
can provide enough impedance to reduce the high-frequency currents. 
Sometimes decoupling resistors (up to a few ohms) in series with power 
supplies in strategic places can provide an impedance for bypass 
capacitors to "break against" to improve filtering. These are often 
installed as a PI section, with a bypass capacitor from each side of the 
resistor to ground. Putting everything on a ground plane is a 
brute-force approach that usually improves things.

Karl


On 05/18/2010 03:11 PM, Chuck / Judy Burch wrote:
> 
> Thanks Matt, Karl and Chris for your responses.
> 
> I am currently using a modified version of the PSD described on page 4 
> of Linear Tech's Application Note #3.  I use an amplitude stabilized 
> Wien bridge oscillator (5000 Hz) for excitation.  The reference signal 
> goes to an LF1011 comparator that drives an LT1034.  The amplified 
> signal goes to the - input of  an LT1007;  the LT1043 switches the + 
> input between ground and the signal so that the LT1007 acts as a 
> synchronous detector.  (This circuit is also described in US patent 
> #3940693.)  This is followed by a 2 pole LPF.
> 
> This arrangement works fine.  The LPF eliminates the excitation 
> artifacts.  But the switching pulses still come through.
> 
> Shielding does not help the pulse problem, so I conclude that what I'm 
> seeing on the downstream part of my boards is magnetic or EM propagated 
> pickup.  Steel enclosures and/or ferrite beads might help, but I haven't 
> tried either.
> 
> On the argument that eliminating a noise source is better than trying to 
> filter or shield it, I wondered if other PSD designs might be 
> intrinsically quieter.
> 
> I will try slowing the rise-time of the clock signal going to the LT1043 
> and I have a Maxim DG419 switch on order to try as well.
> 
> I'll report in the event I have any success.
> 
> 
> Chuck
> __________________________________________________________
> 
> Public Seismic Network Mailing List (PSN-L)
> 
> To leave this list email PSN-L-REQUEST@.............. with the body of 
> the message (first line only): unsubscribe
> See http://www.seismicnet.com/maillist.html for more information.

__________________________________________________________

Public Seismic Network Mailing List (PSN-L)