From: Ben Bradley 
To: psnlist 
Sent: Sun, 30 Sep 2012 3:04
Subject: Re: ageing, baking, and creep


On 9/28/2012 5:41 PM, Brett Nordgren wrote:
....
> Magnetically supported verticals are going to suffer more from=20
> variation in support force with temperature than those using springs.
> These numbers x 1g give the variation in apparent vertical=20
> acceleration due to temperature change.
>
> NdFeB magnet  -1200 ppm/degC  (most commonly used)
> SmCo magnet  -300 ppm/degC (more $$ and not as strong)
> 17-7PH stainless spring -240ppm/degC   (we use)
> Special thermally compensated spring alloys (weaker and $$$)=20
> <10ppm/degC   maybe

Hi Ben,=20
    There is so much 'wrong' with this device, both in principle and in the=
 construction, that I hardly=20
know where to start. =20

For a "reference design" in the following discussion, consider this=20
"gravitometer" device from Scientific American - it has been discussed=20
before on this list:
http://www.scientificamerican.com/article.cfm?id=3Ddetecting-extraterrestri=
a

Basically it's a force-feedback vertical with permanent-magnet=20
suspension and temperature regulation. I did some research on temperature=
=20
coefficients of different permanent magnet types (it turns out the=20
ceramic magnets used in the device have the WORST coefficient), and=20
found Alnico (I forget if it's "Alnico 5" or "Alnico 8") to have the=20
lowest of standard available types. But even more, I found that it's=20
possible to make permanent magnets with (near) zero temperature=20
coefficient, and indeed very large ones are used in some particle=20
accelerators. But as one might expect, it's a specialty item and surely=20
very expensive, and I didn't find a source selling any such magnets.=20
Perhaps for professional-grade instruments this may be worth looking into.

Here's an abstract on such a magnet:
http://ieeexplore.ieee.org/xpl/login.jsp?tp=3D&arnumber=3D5087293&url=3Dhtt=
p%3A%2F%2Fieeexplore.ieee.org%2Fiel5%2F4915369%2F5087117%2F05087293.pdf%3Fa=
rnumber%3D5087293
Setting the temperature at a point where the temperature coefficient=20
flattens out (where the integral is zero) would provide the least change=20
with temperature and the best stability of the field.=20

    An interesting idea for magnet design, but even Alnico 5 has a tempco o=
f -200 ppm.=20
Which is still a VERY LONG WAY from the ~0.01 ppm stability required and yo=
u do=20
need a cylindrical design. Stabilising the temperature of something the siz=
e of a magnet=20
to milli degree levels could be 'quite challenging' - particularly with a '=
hot coil' inside !=20
Maybe you could use a permanent magnet boosted by thermally controlled curr=
ent=20
through a coil ?=20

But what about electromagnets? The magnetic field of an electromagnet is=20
determined by the number of turns (and their geometry), and the current=20
through the coil. A coil for both the "stator" (fixed element used to=20
provide force opposing gravity) and the "rotor" (the moving arm) would=20
provide a force proportional to the product of the currents in the=20
stator and rotor coils. Both currents would be regulated, and the other=20
modulated by the position sense signal to maintain a fixed position.

Probably the biggest disadvantage would be the heat produced by the=20
coils. This could be stabilized by putting the mechanism at the top of a=20
vessel to reduce thermal currents. Very low frequency waves would cause=20
the coil dissipation to change slightly with the force-feedback current=20
changing to keep the arm in place.   =20
    Maybe use a LVDT or a LCDT precision position sensor. Filter out the ve=
ry long=20
period to DC signal to provide the position feedback and differentiate the =
higher=20
frequency signal to provide the dynamic damping ?=20
    Sorry, but I think that this would 'the hard way to try to do it' !!=20

    Regards,=20

    Chris Chapman
       =20


=20




From: Ben Bradley <benbradley@...........>


To: psnlist <psnlist@..............>


Sent: Sun, 30 Sep 2012 3:04


Subject: Re: ageing, baking, and creep















On 9/28/2012 5:41 PM, Brett Nordgren wro=
te:
....
> Magnetically supported verticals are going to suffer more from=20
> variation in support force with temperature than those using springs.
> These numbers x 1g give the variation in apparent vertical=20
> acceleration due to temperature change.
>
> NdFeB magnet  -1200 ppm/degC  (most commonly used)
> SmCo magnet  -300 ppm/degC (more $$ and not as strong)
> 17-7PH stainless spring -240ppm/degC   (we use)
> Special thermally compensated spring alloys (weaker and $$$)=20
> <10ppm/degC   maybe

Hi Ben, 

    There is so much 'wrong' with this device, both in principle and in the=
 construction, that I hardly 

know where to start.  



For a "reference design" in the following discussion, consider this=20
"gravitometer" device from Scientific American - it has been discussed=20
before on this list:
http://www.scientificamerican.com/article.c=
fm?id=3Ddetecting-extraterrestria

Basically it's a force-feedback vertical with permanent-magnet=20
suspension and temperature regulation. I did some research on temperature=
=20
coefficients of different permanent magnet types (it turns out the=20
ceramic magnets used in the device have the WORST coefficient), and=20
found Alnico (I forget if it's "Alnico 5" or "Alnico 8") to have the=20
lowest of standard available types. But even more, I found that it's=20
possible to make permanent magnets with (near) zero temperature=20
coefficient, and indeed very large ones are used in some particle=20
accelerators. But as one might expect, it's a specialty item and surely=20
very expensive, and I didn't find a source selling any such magnets.=20
Perhaps for professional-grade instruments this may be worth looking into.

Here's an abstract on such a magnet:
http://ieeexplore.i=
eee.org/xpl/login.jsp?tp=3D&arnumber=3D5087293&url=3Dhttp%3A%2F%2Fi=
eeexplore.ieee.org%2Fiel5%2F4915369%2F5087117%2F05087293.pdf%3Farnumber%3D5=
087293
Setting the temperature at a point where the temperature coefficient=20
flattens out (where the integral is zero) would provide the least change=20
with temperature and the best stability of the field.=20


    An interest=
ing idea for magnet design, but even Alnico 5 has a tempco of -200 ppm. 
Which is still a VERY LONG WAY from the ~0.01 ppm stability required and yo=
u do 

need a cylindrical design. Stabilising the temperature of something the siz=
e of a magnet 

to milli degree levels could be 'quite challenging' - particularly with a '=
hot coil' inside ! 

Maybe you could use a permanent magnet boosted by thermally controlled curr=
ent 

through a coil ? 


But what about electromagnets? The magnetic field of an electromagnet is=20
determined by the number of turns (and their geometry), and the current=20
through the coil. A coil for both the "stator" (fixed element used=
 to=20
provide force opposing gravity) and the "rotor" (the moving arm) would=20
provide a force proportional to the product of the currents in the=20
stator and rotor coils. Both currents would be regulated, and the other=20
modulated by the position sense signal to maintain a fixed position.

Probably the biggest disadvantage would be the heat produced by the=20
coils. This could be stabilized by putting the mechanism at the top of a=20
vessel to reduce thermal currents. Very low frequency waves would cause=20
the coil dissipation to change slightly with the force-feedback current=20
changing to keep the arm in place.    
<=
/font>
    Maybe use a LVDT or a LCDT precision position senso=
r. Filter out the very long 

period to DC signal to provide the position feedback and differentiate the =
higher 

frequency signal to provide the dynamic damping ? 

    Sorry, but I think that this would 'the hard way to try to do it' !! 


    Regards, 



    Chris Chapman =