Hi Dave --

Do you have a schematic of your analog section that you could put up as a
gif?  I assume your gain stage is before the filter chip.

It may be that with the gain at unity, your predominant noise source is the
Maxim filter chip and that's why you see the single-pole roll off of noise.
 These chips have a certain amount of noise that they generate themselves
and if your signal is below this, it will be lost in the noise of the
filter chip.  But this doesn't mean that the filter chip isn't doing any
good -- when a larger signal comes along, it should do its job.

It looks to me that with the gain at unity, the amount of noise you are
seeing is getting down pretty low in terms of A/D counts.  Some of it may
be A/D converter noise.  What does the noise look like if you remove the
Maxim chip and short the input of your RC filter to ground?

The maximum output capacitance spec for the Maxim filter is probably there
to avoid oscillation of their output stage.  Many analog IC's with
low-power (spelled wimpy) output stages aren't stable driving capacitive
loads.  But the series resistor at the input of your RC filter should keep
the Maxim filter chip from seeing the full capacitive load and get around
the problem.  If the value of that resistor is 10K or above, you shouldn't
have any problem.  And it may work with values down to 1K.

If you put a scope on the output of the filter chip, look to see if there
is any significant signal (oscillation) in the range of 1MHz.  A little bit
of signal is usually there due to pickup from AM radio stations and other
transmitters in the area.  If it's not more than 50mV or so, then you're
probably ok for oscillations.

Hope this helps.

Karl Cunningham


At 12:02 PM 11/21/2000 -0500, you wrote:
>I tried to do what you suggested and all I see is what looks to me to
>be 3db/oct noise slope in both cases:
>
>http://www.infiltec.com/seismo/nois10ka.gif   noise with 10k in place of
>coil - amp
>http://www.infiltec.com/seismo/nois10kf.gif    noise with 10k in place of
>coil - fft
>http://www.infiltec.com/seismo/nois1xa.gif        noise at unity gain, no
>coil - amp
>http://www.infiltec.com/seismo/nois1xf.gif         noise at unity gain, no
>coil  - fft
>
>So I guess my filter chip is not
>doing anything and all I have is the RC filter that I was
>using to get rid of the switching noise.  Hmmm.  Does
>this mean I have to give back my diploma to the Radio
>Shack school of electronics?
>
>I find it interesting that a Lehman works pretty well
>with only a single pole RC lowpass filter!
>http://www.infiltec.com/seismo/001117x.gif
>I guess my fallback for keeping my system simple is to
>replace my switched cap filter chip with an opamp chip
>configured as a 2 or 4 pole bessel filter.
>
>I am not sure yet why my MAX7401 8 pole bessel
>filter chip is not doing anything.  It could be a stupid
>wiring problem on my board.  But on re-reading the
>spec sheet it says "Maximum Capacitive load at OUT is
>min 50 pfd, typical 500 pfd".  That could be my problem since my
>RC filter cap the output is in the ufds.   I thought I set the Maxim
>corner frequency to 4 Hz, so the sampling frequency would
>be 100x=400 Hz.   I guess that when the spec sheet recommends RC
>filtering for the sampling noise, it assumes that you have
>much higher corner frequencies so you can limit your RC
>capacitor size to pfds?  If this Maxim output capacitance is the problem
>then only solution I can see is to put a unity gain amp
>between the filter chip and the RC.  Ideas?
>
>> > During the day the record shows many spikes
>> > similar to what happens when someone walks near the pendulum and
>> > deforms the slab. I get some pretty quiet data in the middle of the
>night
>> > when the Big Macs go home
>>
>>     Could be magnetic or mechanical or both. Did your fft represent the
>> quietest period or was it average over the day?
>
>My system noise record was from  9 UT or 4 AM local time.  Pretty quiet.
>During the day it would be similar but riding on spikes every few
>minutes.  These spikes seem to be related to slab tilting, since
>they look like what happens when I walk over to the pendulum.
>Here is an hour of system noise at noon local time.  The big spikes
>may be me walking past the pendulum to get a Big Mac.
>
>http://www.infiltec.com/seismo/nois-amp.gif     noise at 4am local - amp
>http://www.infiltec.com/seismo/nois-fft.gif       noise at 4am local- fft
>http://www.infiltec.com/seismo/nois12na.gif    noise at noon local - amp
>http://www.infiltec.com/seismo/nois12nf.gif     noise at noon local - fft

__________________________________________________________

Public Seismic Network Mailing List (PSN-L)