PSN-L Email List Message

Subject: Re: Haunted system misbehaving again
From: chrisatupw@.......
Date: Thu, 21 Jun 2012 20:26:41 -0400 (EDT)







From: Larry Conklin lconklin@............
Sent: Wed, 20 Jun 2012 2:19
Subject: Re: Haunted system misbehaving again


Hi Larry,
****Not a silly question at all, but one I've considered in the past. Have=
=20
arefully checked and cleaned around the pendulum, etc to get rid of any=20
ossible webs or critters, with no noticeable effect.
****Regarding the enclosure, it isn't much bigger than it needs to be to=20
nclose the volume surrounding the sensor.  And I have a couple of power=20
esistors on an aluminum plate at the top of the box to try to create a=20
emperature inversion, but I've never been able to convince myself that=20
t makes any real difference. Maybe I need to make it a little hotter.=20

Hi Larry,

    The enclosure needs to be well insulated, with the heater inside=20

the top of the enclosure. It can work well with a few watts. Too much=20

power can stir up the air and give noise. Try switching the power off=20

completely when you get noise ? I reckon not more than about 20 watts=20

heating for my large Lehman enclosure. You only get reverse convection=20
noise when the air temperature falls below the ground temperature,=20
typically at night in the early hours of the morning. You can check for=20
this with one of the cheap indoor / outdoor electronic digital=20
thermometers.

 =20
***At one time I tried to take up as much of the empty space with baffles=
=20
ade of corrugated cardboard, but that didn't seem to help either and it=20
ade the cover of the enclosure surprisingly heavy.  I wound up=20
crapping that cover and making a new one out of celotex panels.  It=20
ould probably be a good idea to fill the empty space in this enclosure=20
ith something light and insulating, like styrofoam.

  Open cell polyurethane foam is a much better filler than sealed=20
cell polystyrene. Stick it on with lines of adhesive silicone rubber ?=20
****But none of those possibilities really fit the history of the problem v=
ery well.=20

cell polystyrene. Stick it on with lines of adhesive silicone rubber ?=20
****But none of those possibilities really fit the history of the problem v=
ery well.=20
=20
# I could write a small book on all the things I've considered and tried, s=
everal of=20
which have seemed to cure the problem, only to have it come back months lat=
er.=20
=20
    While it may be relatively unlikely, you could have several causes ?
=20
****I really don't have any real reason to suspect the electronics, other t=
han I=20
don't have any other ideas. The basement is humid, so at one time I took th=
e board=20
out and baked it under a hundred watt light bulb for a couple of days, clea=
ned it=20
and put it back. No effect.
=20
=20
    You can check for power line interferance with a MW / LW AM radio. Tune=
 it in between=20
stations and then listen for cracks and pops in the background noise. You c=
ould have a serious=20
problem / intermittent contact somewhere in your house wiring, or in an app=
liance - like a deep=20
freeze chest ? =20
=20
    It is fairly easy to test most of the electronics. Put a digital=20
multimeter onto the tilt test point and earth. Very carefully /=20
gently, insert two wood wedges between the sensor plates and the=20
oscillator plate so that the tilt output voltage is still ~0 V.=20
This effectively zeroes the input vibration signal, but all opamps=20
except U6 are still operating, so any noise on the recorder is in the=20
circuit itself. The only opamp not tested is U6, the NE5534 feedback=20
coil driver. They are not expensive.    =20
=20
****Since I can see corresponding glitches in the output of both channels,
if there is a problem with the board it either has to be upstream of the po=
int where=20
the two channels separate, or possibly in the power supply. But, with the l=
imited test=20
equipment I have, I don't see anything amis in the power. And at one time, =
I decided=20
it was worth the trouble, and replaced the first stage op amp. No dice.
So.... What other ideas have you got?
=20
    Check that the two suspension foils have not been visably damaged,=20
showing folds, wrinkles.... Check for continuity of the feedback wiring and=
=20
coil with a multimeter.=20
    Now, the most likely problem, what are the levelling screws resting on=
=20
and what are the screw ends like ? Pointed, rounded, etc ?
The seismometer itself changes in length and width by a tiny amount as it's=
=20
temperature changes - the ground / floor changes at a different rate and mu=
ch=20
more slowly. I glue 1/8" thick SS plates to concrete with pool cement, or=
=20
similar and cap the ends of the adjusting screws with dome headed stainless=
=20
steel nuts. You can also use glass or most hard, really smooth materials fo=
r=20
the mounting surface.=20
 =20
On 6/19/2012 6:52 PM, Larry Cochrane wrote:
> Silly question Larry. Are you sure the noise is not being caused by=20
> spiders, also known as spiderquakes, or some other insect getting=20
> inside the enclosure? I guess it could also be air currents inside the=20
> enclosure. Do you have a lot of free air space inside your sensor box?=20
=20
=20
****I think I remember your having posted an account of your "rug affair". =
=20
In fact I think that it prompted me to improve the leveling screws on my=20
system, which at the time were rather flimsy. I replaced them with=20
larger, stiffer screws and I think that it may have helped, or even=20
"solved" the problem for a while.  I'm now using fairly substantial=20
screws and the system is (and already had been) sitting on a couple of=20
1/8 inch thick strips of aluminum that are epoxied to the basement=20
floor.  The assumption being that the plates are a more rigid surface=20
than the bare concrete would be.

****The only pivot points I can identify are the pendulum hinge itself=20
(consisting of thin metal shim material, phosphor bronze or some such). =20
But there is certainly the possibility that the contacts between the=20
feet and the plates might shuck around a little with temperature changes?

****I also have had a series of cover experiments.  Original cover was a=20
wood box that sat on the frame of the sensor.  Very bad idea. Rebuilt it=20
to fit over the sensor and sit on the floor.  Better, but probably=20
vulnerable to transmitting noise to the instrument.  Added "stuffing" to=20
fill up the unused volume and suppress air movement. Not much noticeable=20
change.  Current cover is light weight, built of insulating board and=20
sitting on the floor.  At times I've duct taped it to the floor, but at=20
the moment it's just sitting there.  Maybe putting the tape back would=20
be a good idea.=20



=20
    NOT in my opinion. You DON'T want to either have the cover so=20
well sealed as it reacts to air pressure changes, or that there are=20
'leak channels' which blow in jets of air - although quite slowly.=20
I am strongly in favout of sticking a continuous 1/2" strip of the=20
very soft open cell foam door / window strip to the base and then=20
putting a weight on top of the enclosure to compress it onto the=20
ground. =20


****But the ovea dry rall behavior of the problem makes a lot of these=20
possibilities seem unlikely.  In one subset of episodes of the problem,=20
the noise starts abruptly and then continues indefinitely for days of=20
weeks at a time.  The data shown on my web page is typical of this=20
"class".  (And I noticed that this episode started on March 30, and is=20
still ongoing, although there were couple of periods of a week or so=20
duration since them where things were much better.)  Other episodes have=20
emerged gradually over the course of several hours, continued for days=20
or even weeks at a time and then subsided.  I'm at a loss to come up=20
with a theory that is consistent with both kinds of behavior.  And none=20
of the experiments I've tried have made any clear and lasting impact on=20
the problem, although I've tried a few things that I've thought fixed=20
it, only to have it crop up six months later.

    This is starting to sound more like a failing opamp / component or dry =
solder joint.=20

On 6/20/2012 9:41 AM, RSparks wrote:
> Hi Larry,
>
> My experience parallels yours with loose mounted components.  I once=20
> went through a rug to concrete below and failed to make good contact=20
> with one screw on the concrete.  I was effectively floating on the rug=20
> with one of the three seismometer legs!   Very noisy!
>
> Based on this experience, I am careful to make all mechanical contacts=20
> solid connections in some way.  I even drilled holes into the concrete=20
> and mounted my seismometer solidly.  I thought there was a small=20
> improvement in the noise level after taking this somewhat extreme=20
> measure.
>
> I also avoid pivot points in favor of bending joints to avoid noise. =20
> Friction is present in our seismometers if there is a moving joint and=20
> mechanical friction is always noise.
>
> Maybe look more carefully at any mechanical connection that might be=20
> loose, maybe  not only in the leveling screws.  Look also at the=20
> mounting of the surrounding air/insulating box, making sure it is not=20
> rocking from the ever present 0.2 Hz peak noise or more likely, the=20
> thermal air currents in your room.




****But the ovea dry rall behavior of the problem makes a lot of these=20
possibilities seem unlikely.  In one subset of episodes of the problem,=20
the noise starts abruptly and then continues indefinitely for days of=20
weeks at a time.  The data shown on my web page is typical of this=20
"class".  (And I noticed that this episode started on March 30, and is=20
still ongoing, although there were couple of periods of a week or so=20
duration since them where things were much better.)  Other episodes have=20
emerged gradually over the course of several hours, continued for days=20
or even weeks at a time and then subsided.  I'm at a loss to come up=20
with a theory that is consistent with both kinds of behavior.  And none=20
of the experiments I've tried have made any clear and lasting impact on=20
the problem, although I've tried a few things that I've thought fixed=20
it, only to have it crop up six months later.

    This is starting to sound more like a failing opamp / component or dry =
solder joint.=20

On 6/20/2012 9:41 AM, RSparks wrote:
> Hi Larry,
>
> My experience parallels yours with loose mounted components.  I once=20
> went through a rug to concrete below and failed to make good contact=20
> with one screw on the concrete.  I was effectively floating on the rug=20
> with one of the three seismometer legs!   Very noisy!
>
> Based on this experience, I am careful to make all mechanical contacts=20
> solid connections in some way.  I even drilled holes into the concrete=20
> and mounted my seismometer solidly.  I thought there was a small=20
> improvement in the noise level after taking this somewhat extreme=20
> measure.
>
> I also avoid pivot points in favor of bending joints to avoid noise. =20
> Friction is present in our seismometers if there is a moving joint and=20
> mechanical friction is always noise.
>
> Maybe look more carefully at any mechanical connection that might be=20
> loose, maybe  not only in the leveling screws.  Look also at the=20
> mounting of the surrounding air/insulating box, making sure it is not=20
> rocking from the ever present 0.2 Hz peak noise or more likely, the=20
> thermal air currents in your room.



    I don't follow Larry Cochrane's point about the two 100muF capacitors. =
There aren't=20
any in his circuit - assuming that you are using his board ?=20
    I hope that this helps,=20
    Regards,=20
    Chris Chapman







= From: Larry Conklin lconklin@twcny= ..rr.com
Sent: Wed, 20 Jun 2012 2:19
Subject: Re: Haunted system misbehaving again

Hi Larry,

****Not a silly question at all, but one I've considered in the past. Have=
=20
carefully checked and cleaned around the pendulum, etc to get rid of any=20
possible webs or critters, with no noticeable effect.

****Regarding the enclosure, it isn't much bigger than it needs to be to=20
enclose the volume surrounding the sensor.  And I have a couple of power=20
resistors on an aluminum plate at the top of the box to try to create a=20
temperature inversion, but I've never been able to convince myself that=20
it makes any real difference. Maybe I need to make it a little hot=
ter. 
Hi Larry,
    The enclosure needs to be w=
ell insulated, with the heater inside 
the top of the enclosure. It ca=
n work well with a few watts. Too much 
power can stir up the air and g=
ive noise. Try switching the power off 
completely when you get noise ?=
 I reckon not more than about 20 watts 
heating for my large Lehman enc=
losure. You only get reverse convection 
noise <=
/TT>when the air temperature falls below =
the ground temperature, 
typically at night in the early hours of the morning. You c=
an check for 
this with =
one of the cheap indoor / outdoor electronic digital 
thermometers.
  
****At one time I tried to take up as much of the empty space with b=
affles=20
made of corrugated cardboard, but that didn't seem to help either and it=20
made the cover of the enclosure surprisingly heavy.  I wound up=20
scrapping that cover and making a new one out of celotex panels.  It=20
would probably be a good idea to fill the empty space in this enclosure=20
with something light and insulating, like styrofoam.
  Open cell polyurethane foam i=
s a much better filler than sealed 
cell polystyrene. Stick it on with lines of adhesive silicone rubber ? 
****But none of those possibilities really fit the history of the problem v= ery well. 
=  
= # I could write a small book on all the things I've= considered and tried, several of
= which have seemed to cure the problem, only to have it c= ome back months later.=20
<= /FONT> 
&= nbsp;   While it may be relatively unlikely= , you could have several causes ?
<= /FONT> 
*= ***I really don't have any real reason to suspect the electronics, other th= an I
d= on't have any other ideas. The basement is humid, so at one time I took the= board
o= ut and baked it under a hundred watt light bulb for a couple of days, clean= ed it
a= nd put it back. No effect.
<= /FONT> 
 
    You can check for power line interferance w= ith a MW / LW AM radio. Tune it in between
stations and= then listen for cracks and pops in the background noise. You could have a = serious 
problem / intermittent contact somewhere in your house wiring,= or in an appliance - like a deep
freeze chest ?  
<= /FONT> 
&= nbsp;   It is fairly easy to test most of t= he electronics. Put a digital
m= ultimeter onto the tilt test point and earth= .. Very carefully /
g= ently, insert two wood wedges between the sensor plates and the
o= scillator plate so that the tilt output = ;voltage is still ~0 V.
T= his effectively zeroes the input vibrat= ion signal, but all opamps
e= xcept U6 are still operating, so any noise on the recorder is in the
c= ircuit itself. The only opamp not tested is U6, the NE5534 feedback
c= oil driver. They are not expensive.&nbs= p;    
<= /FONT> 
*= ***Since I can see corresponding glitches in the output of both channels,
i= f there is a problem with the board it either has to be upstream of the poi= nt where
t= he two channels separate, or possibly in the power supply. But, with the li= mited test
e= quipment I have, I don't see anything amis in the power. And at one time, I= decided
i= t was worth the trouble, and replaced the first stage op amp. No dice.
S= o.... What other ideas have you got?
<= /FONT> 
&= nbsp;   Check that the two suspension foils= have not been visably damaged,
s= howing folds, wrinkles.... Check for continuity of the feedback wiring and&= nbsp;
c= oil with a multimeter. 
&= nbsp;   Now, the most likely problem, what are the levelling scre= ws resting on 
a= nd what are the screw ends like ? Pointed, rounded, etc ?<= /div>
T= he seismometer itself changes in length= and width by a tiny amount as it's
t= emperature changes - the ground / floor changes at a different rate an= d much
m= ore slowly. I glue 1/8" thick SS plates= to concrete with pool cement, or
s= imilar and cap the ends of the adjusting scr= ews with dome headed stainless
s= teel nuts. You can also use glass or most ha= rd, really smooth materials for
t= he mounting surface. 
&= nbsp; 
O= n 6/19/2012 6:52 PM, Larry Cochrane wrote:
&= gt; Silly question Larry. Are you sure the noise is not being caused by
&= gt; spiders, also known as spiderquakes, or some other insect getting
&= gt; inside the enclosure? I guess it could also be air currents inside the =
&= gt; enclosure. Do you have a lot of free air space inside your sensor box? =
<= /FONT> 
<= /FONT> 
*= ***I think I remember your having posted an account of your "rug affair".&n= bsp;
In fact I think that it prompted me to improve the leveling screws on my system, which at the time were rather flimsy. I replaced them with
larger, stiffer screws and I think that it may have helped, or even
"solved" the problem for a while.  I'm now using fairly substantial screws and the system is (and already had been) sitting on a couple of
1/8 inch thick strips of aluminum that are epoxied to the basement
floor.  The assumption being that the plates are a more rigid surface =
than the bare concrete would be.

****The only pivot points I can identify are the pendulum hinge itself
(consisting of thin metal shim material, phosphor bronze or some such).&nbs= p;
But there is certainly the possibility that the contacts between the <= br> feet and the plates might shuck around a little with temperature changes?
****I also have had a series of cover experiments.  Original cover was= a
wood box that sat on the frame of the sensor.  Very bad idea. Rebuilt = it
to fit over the sensor and sit on the floor.  Better, but probably vulnerable to transmitting noise to the instrument.  Added "stuffing" = to
fill up the unused volume and suppress air movement. Not much noticeable change.  Current cover is light weight, built of insulating board and =
sitting on the floor.  At times I've duct taped it to the floor, but a= t
the moment it's just sitting there.  Maybe putting the tape back would=
be a good idea.
 
    NOT in m= y opinion. You DON'T want to either have the cover so 
well sealed as it reacts to= air pressure changes, or that there are <= /div>
'leak channels' which= blow in jets of air - although quite slowly.
I am strongly in favo= ut of sticking a continuous 1/2" strip of the
very soft open cell foam d= oor / window strip to the base and then
putting a weight on <= /FONT><= FONT color=3Dblack size=3D2 face=3Darial>= top of the enclosure to compress it onto the
ground.  <= /div>


****But the ovea dry rall behavior of the problem makes a lo= t of these
possibilities seem unlikely.  In one subset of episodes of the problem= ,
the noise starts abruptly and then continues indefinitely for days of
weeks at a time.  The data shown on my web page is typical of this "class".  (And I noticed that this episode started on March 30, and is=
still ongoing, although there were couple of periods of a week or so
duration since them where things were much better.)  Other episodes ha= ve
emerged gradually over the course of several hours, continued for days
or even weeks at a time and then subsided.  I'm at a loss to come up <= br> with a theory that is consistent with both kinds of behavior.  And non= e
of the experiments I've tried have made any clear and lasting impact on the problem, although I've tried a few things that I've thought fixed
it, only to have it crop up six months later.

    This is starting to sound more like a fai= ling opamp / component or dry solder joint. 

On 6/20/2012 9:41 AM, RSparks wrote:
> Hi Larry,
>
> My experience parallels yours with loose mounted components.  I o= nce
> went through a rug to concrete below and failed to make good contact <= br> > with one screw on the concrete.  I was effectively floating on th= e rug
> with one of the three seismometer legs!   Very noisy!
>
> Based on this experience, I am careful to make all mechanical contacts=
> solid connections in some way.  I even drilled holes into the con= crete
> and mounted my seismometer solidly.  I thought there was a small =
> improvement in the noise level after taking this somewhat extreme
> measure.
>
> I also avoid pivot points in favor of bending joints to avoid noise.&n= bsp;
> Friction is present in our seismometers if there is a moving joint and=
> mechanical friction is always noise.
>
> Maybe look more carefully at any mechanical connection that might be <= br> > loose, maybe  not only in the leveling screws.  Look also at= the
> mounting of the surrounding air/insulating box, making sure it is not =
> rocking from the ever present 0.2 Hz peak noise or more likely, the > thermal air currents in your room.
= &nb= sp;
=     I don't fol= low Larry Cochrane's point about the two 100muF capacitors. There aren't
= any in his circuit - assuming = that you are using his board ?
=     I hope that= this helps,
=     Regards,
=     Chris Chapm= an


[ Top ] [ Back ] [ Home Page ]