In a message dated 20/01/2008, PETERS_RD@.......... writes:

Chris,
Have you actually built an instrument  with the AD698 chip and determined 
that it is unsuitable for  seismometers?
I remember you telling me this some time ago; but it turns  out that Denny 
Goodwin put together a circuit for me, also some time  ago.  Only in the last 
month did I try his breadboarded unit.    He had been unsuccessful with it due 
to a solder bridge that I found only  after looking at the board carefully with 
magnification.
Hi Randall,
 
    If you read up the specifications, you will find  that the output of the 
AD698 is PULSED ! The pulse length is modified to  give the temperature 
compensation. I don't know what the ??designers?? at AD  thought that they were 
doing. DigiKey list the AD698 DIP version at $70.88. The  AD598 does not have this 
problem, but the DIP version still costs $52.53 for 1  off. The SOIC versions 
are about half this in small quantities, when stocked. 

You will remember that for much of  the two decades I have been using my 
fully differential
capacitive sensors  for internal friction research--that the work-horse for 
that work was the  NE5521 chip that is no longer manufactured. Well, I have 
done a direct  comparison of the AD698 with the NE5521 and do not find a great 
enough  performance difference to agree with your claim!  In fact, with the  
prototype new vertical that I recently built (the details of which I will soon  
share with list-serve readers), it picked up the Charlotte Is earthquake with  
this chip, even though the instrument was sitting on a lab bench here in the  
physics building. So I don't think you can make a defensible claim that  the 
chip is unsuitable for seismometers.  
    The correction is temperature dependant, so there  will likely be one 
temperature at which the correction is zero, presumably at  one end of the range.
    If you digitise the output directly, you will get  an uncompensated 
signal and occasional glitches due to the variable zero  output periods. To get the 
compensated output you need to provide a low  pass filter to integrate the 
signal. This may slow up the response if you want  to get low noise.
    Trying to use a chip with a stepped level  output in a low noise 
application seems to be 'simply buying  trouble'. I am quite happy to avoid using it, 
particulaly when I can make  up a good detector for a small fraction of the 
cost. _http://www.keckec.com/seismo/_ (http://www.keckec.com/seismo/) 

Insofar  as expense is concerned, the webpage I recently viewed at Analog 
Devices  indicates a price of about $25 (straight from the companry) for the 
version  that I would use (in large quantities).  I haven't enquired about single  
chip prices, although I know from Larry Cochrane that if bought from some of  
the 'distributors' it is indeed hideously expensive. They love to mark-up  
units, evidently in this case by a huge amount.


You might cross check those prices again? The  ''low'' AD website price I saw 
quoted was for the PLCC version in 1,000 off  quantities. The 15 off price 
for the DIP version is $60. 
    
    Regards,
 
    Chris



   





In a message dated 20/01/2008, PETERS_RD@.......... writes:
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000=20
  size=3D2>Chris,
      Have you actually built an instrum=
ent=20
  with the AD698 chip and determined that it is unsuitable for=20
  seismometers?
I remember you telling me this some time ago; but it turn=
s=20
  out that Denny Goodwin put together a circuit for me, also some time=20
  ago.  Only in the last month did I try his breadboarded unit. &n=
bsp;=20
  He had been unsuccessful with it due to a solder bridge that I found only=20
  after looking at the board carefully with magnification.
Hi Randall,
 
    If you read up the specifications, you will fin=
d=20
that the output of the AD698 is PULSED ! The pulse length is modified t=
o=20
give the temperature compensation. I don't know what the ??designers?? at AD=
=20
thought that they were doing. DigiKey list the AD698 DIP version at $70.88.=20=
The=20
AD598 does not have this problem, but the DIP version still costs $52.53 for=
 1=20
off. The SOIC versions are about half this in small quantities, when stocked=
..=20

<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000=20
  size=3D2>       You will remember that for m=
uch of=20
  the two decades I have been using my fully differential
capacitive sens=
ors=20
  for internal friction research--that the work-horse for that work was the=20
  NE5521 chip that is no longer manufactured. Well, I have done a direct=20
  comparison of the AD698 with the NE5521 and do not find a great enough=20
  performance difference to agree with your claim!  In fact, with the=20
  prototype new vertical that I recently built (the details of which I will=20=
soon=20
  share with list-serve readers), it picked up the Charlotte Is earthquake w=
ith=20
  this chip, even though the instrument was sitting on a lab bench here in t=
he=20
  physics building. So I don't think you can make a defensible claim th=
at=20
  the chip is unsuitable for seismometers.  
    The correction is temperature dependant, so the=
re=20
will likely be one temperature at which the correction is zero, presumably a=
t=20
one end of the range.
    If you digitise the output directly, you will g=
et=20
an uncompensated signal and occasional glitches due to the variable zero=20
output periods. To get the compensated output you need to provide a low=
=20
pass filter to integrate the signal. This may slow up the response if you wa=
nt=20
to get low noise.
    Trying to use a chip with a stepped level=20
output in a low noise application seems to be 'simply buying=20
trouble'. I am quite happy to avoid using it, particulaly when I can ma=
ke=20
up a good detector for a small fraction of the cost. http://www.keckec.com/seismo/
<=
FONT=20
  style=3D"BACKGROUND-COLOR: transparent" face=3DArial color=3D#000000 size=
=3D2>Insofar=20
  as expense is concerned, the webpage I recently viewed at Analog Devices=20
  indicates a price of about $25 (straight from the companry) for the versio=
n=20
  that I would use (in large quantities).  I haven't enquired about sin=
gle=20
  chip prices, although I know from Larry Cochrane that if bought from some=20=
of=20
  the 'distributors' it is indeed hideously expensive. They love to mark-up=20
  units, evidently in this case by a huge amount.


    You might cross check those prices again? The=20
''low'' AD website price I saw quoted was for the PLCC version in 1,000 off=20
quantities. The 15 off price for the DIP version is $60. 
    
    Regards,
 
    Chris