Jim,

Good points. Nice summary.

I'd like to add something which I've found to help. For seismometer
work, the A/D converter inputs aren't always differential. And when they
are they sometimes sacrifice an A/D channel to do it.

By putting the differential stage on the output of the seismometer
instead of at the A/D one can accomplish the same thing with a
single-ended A/D input. This uses the same cabling as you mention -- a
twisted pair carrying very little current and a separate ground to carry
the currents between ground systems of the A/D and the seismometer. The
difference in this approach is that one of the twisted wires is driven
from the output of the seismometer while the other is connected to
ground at the A/D converter. In a way, the signals in the two conductors
go in opposite directions. But they very much benefit from being twisted
together. Since both are driven from low-impedance sources, shielding
the cable isn't normally necessary.

The differential stage is an op-amp at the seismometer end, which
electronically adds the ground voltage of the A/D to the seismometer
output signal. For instance, if the seismometer's output is zero but the
A/D ground is 1mV more positive than the seismometer ground, the
differential stage increases the seismometer's output by 1mV and the A/D
sees zero voltage.

One can see this concept in the stage comprised of U504 and resistors
R512-R515 in the schematic at:
http://www.keckec.com/seismo/images/short.gif
The SHORT PERIOD OUTPUT and the GND SENSE input go to the twisted pair,
while the GND RETURN is the separate conductor that carries the current
between the two ground systems.

There is an oversight in this schematic -- capacitor C513 should be
duplicated with another capacitor across R513 as well. Otherwise there
is a built-in common-mode-to-normal-mode signal conversion for
high-frequency AC signals appearing on the ground sense line.

Karl


On 11/04/2010 11:13 AM, Jim Santee wrote:
> Here are some basics to keep in mind:
> 1. Unbalanced lines such as RCA home entertainment cables and home owner
> style coaxial cables have one conductor. The return path is the shield
> which is supposed to be at ground potential. Ground being the common in
> the system, but ground having a huge amount of background noise from
> natural and man made sources. If the desired signal level is fairly high
> then modest background noise is not a problem. The desired signal over
> rides the ambient noise level. In cases where the ground is carrying a
> lot of noise or the there is a ground differential between the equipment
> in use, then you have noise injected into your system. The more gain you
> add to pick up the desired signal the more noise you bring into your
> system. This is one of several reasons why ignition noise and "10-4 good
> buddy" gets into audio systems. Ferrite beads and by pass capacitors can
> filter some of this out.
>
> 2. Balanced lines are a little different in that the you have two
> conductors independent of ground so picking up low level ground noise is
> reduced. The balanced line concept cancels out a lot of low level
> ambient noise. If the pair is twisted then even more low level noise is
> cancelled out. If you add shielding then even more noise is cancelled
> out. If you have a big check book and want double shielded cable then
> you can rest assured you have done your best. One point: Read the next
> paragraph and keep in mind having spent a ton of money on cable will not
> eliminate bad engineering practices on either end.
>
> 3. Recording and broadcast studios have used balanced line for
> generations. Between equipment audio sources are usually handled as
> balanced line with XLR type connectors with the signal levels of about
> .5 to 1 volt. This means the signal feeding the line is usually are a
> fairly high level so that the receiving equipment does not have to use a
> lot of gain to pick up the signal. By keeping the feed signal high and
> the receiving end gain low you minimize noise pick up. In these types of
> installations ambient and ground noise is very high so you need to beat
> the nose by using good engineering. Going one more level: High quality
> microphones use balanced lines. To keep the signal relatively high on
> very long audio lines the microphones have built in mini-amplifiers to
> feed a higher signal down the line to the mixer or audio board. The
> mixer or audio board introduces a small power source on the signal line
> called phantom power. The phantom power is what powers the audio amp in
> the microphone.
>
> 4. It is far better to feed a relatively high level low noise signal
> down a line to be processed by another piece of equipment with a
> relatively low gain setting. You are beating the ambient noise on the
> line by keeping the desired signal level well above the ambient noise.
> Having stated that, shoving an extremely high level signal down a line
> will cause interference to adjacent circuits and make life real bad.
> Balanced lines will not compensate for bad karma. Keep the line signals
> in the .5 volt range and you will not have a lot of bleed through.
>
> 5. In commercial communications systems and broadcast studios just about
> everything is shielded and balanced line inputs are used to prevent
> interference. In home entertainment systems cheap is what sells.
>
> 6. One closing point: Home PC's can have their own degrading issues
> which can reflect into other systems. Keep this in mind:. Unless you
> have a well shielded PC and monitor then keep it away from critical low
> level signal systems.
>
> Jim
>
>
> _____________________________________________________________________--
> Yes, I think I understand a tiny bit about this.
> But can you tell me about common mode verses differential.
> as referenced to ground.
>
> It seems to me AC can leak both ways into a system then through some
> mysterious force infiltrate a system in a differential form
> causing a dc or (low freq ac) drift.
>
> It seems to me at Seismic frequencies you never operate
> a proper balanced transmission line system because the wavelength is too
> long.
> You need like 1/2 wavelength or multiple thereof.
>
> Twisted lines only insure that common mode signals
> are distributed evenly on both differential wires.
> It does not stop common mode or even attenuate them.
> You must rely upon the CMMR ratio to attack the
> unwanted common mode stuff.
>
> Use twisted pair along with 100% shielding
> in my opinion is the only way.
>
> This is all laymen talk since I'm not an engineer.
>
>
> __________________________________________________________
>
> Public Seismic Network Mailing List (PSNLIST)
>
> To leave this list email PSNLIST-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 (PSNLIST)

To leave this list email PSNLIST-REQUEST@.............. with 
the body of the message (first line only): unsubscribe
See http://www.seismicnet.com/maillist.html for more information.