PSN-L Email List Message
Subject: Re: STM 8 (LVDT subject)
From: ChrisAtUpw@.......
Date: Sat, 16 Nov 2002 14:29:47 EST
In a message dated 15/11/02, meredithlamb@............. writes:
> >> Agreed, but there are several distance transducer approaches whic=
h=20
>> can be successful. To my mind, there is a distinct shortage of amateur=20
>> 'off the shelf' sensor designs capable of giving near professional=20
>> resolution and stability. There is only one LVDT Kit sensor available wit=
h=20
>> a PCB, to my knowledge, which can give about 7 nano metres resolution at=20
>>=20
> Presume its the Italian, Nuova Elettronica unit (?), you've been working=20
>=20
Hi Meredith,
I have done some experiments modifying / improving my Nuova=20
Elettronica LVDT board, reducing the two pole detector filter to 10 Hz, whic=
h=20
is more suitable for seismic use, changing the output opamp to a low noise=20
type and increasing the time constant of the high pass filter, along with=20
several other small changes. NE seemed to have used a circuit designed by=20
Philips for wider bandwidth commercial LVDT applications. I have also been=20
working on other magnetic sensor systems using the Philips NE5521 LVDT chip,=
=20
including modified transformers, somewhat like Sean's, some using ferrite=20
components and lately, a capacitative system. I have a full write up nearly=20
finished.=20
> Perhaps the greatest curiosity item I wonder about their transformer part,=
=20
> is whether its really a precision wound unit like mentioned as a=20
> (professional) necessity as on other web sites? It appears to be much lik=
e=20
> other general looking transformers. Without any access I'd guess its=20
>=20
If you want to measure several inches of movement and require a linea=
r=20
response, you may need a precision linear winding. If you only want a few mm=
=20
and the field of the sensor rod can be sensed over a physically much larger=20
sensor winding, you just need the windings to stay in a fixed position. Ther=
e=20
will of course be thermal expansion and changes of resistance with=20
temperature, but these effects are very small over the 10 Hz bandwidth and=20
the 5 to 30 sec of the high pass filter time constant. =20
> Perhaps the biggest thing I "seem" to see, is that Seans original LRDT=20
> transformers are in likelyhood kind of a crude nonlinear wire wrapping;=20
> whereas, your LVDT, is a painstaking relatively precision wire layered=20
>=20
With Sean's variable reluctance transformer sensor, the magnetic=20
circuit is defined by the position of the flat iron armature plate and the=20
iron polepieces. In a LVDT, the interaction is between the position of a=20
magnetised armature and the varying distance it moves inside the sensor coil=
,=20
so the coil needs to be quite uniformly wound. For a description of an LVDT,=
=20
see http=
://www.msiusa.com/schaevitz/pdf/lvdt/LVDT_Intro.pdf It is not too=20
difficult to make even windings, so long as you do not use extremely fine=20
wire, do not try to wind too fast, slow up right at the end of each layer=20
and, most importantly, use very thin paper, like cigarette or tissue paper,=20
for interleaving each layer. This makes it easy to spot and correct any=20
overwinding or incorrectly spaced turns. It also makes winding easier and=20
helps to prevent shorted turns in between layers. I made two matched 40,000=20
turn precision differential transformer coils this way, for measuring=20
paramagnetic susceptibility. A very small plastic tube makes a good wire=20
guide. It is also possible to just layer wind two LVDT secondaries end to en=
d=20
on a sensor tube and use a ferrite or powder iron cored choke to provide the=
=20
magnetic excitation. This avoids having to wind a central primary magnetisin=
g=20
section in between, or over the top of the secondaries.
=20
Yet.....according to Seans website, his LRDT is sensitive to a incredible=20
0.1nm. I've=20
> zero experience but it makes me wonder further about coils themselves=20
> between=20
> the two. It would seem the LRDT is much simpler to make and more sensitiv=
e=20
>=20
Coupled with the very high sensitivity of Sean's transducer, the=20
movement range is small, perhaps +/-0.5 mm in a 2 mm total gap, and the=20
output is only linear for a limited range around the zero position. This is=20
just fine for a feedback seismometer, where you try to hold the armature=20
position constant. The out of range signal increases dramatically before=20
saturating, which can be bad news if the seismometer feedback system gets ou=
t=20
of lock.=20
> All this is leading up to the question of whether a fairly linear wound (b=
ut=20
> not exactly precisely layered) coil "could" somewhat work simarily (with=20
> less sensitivity) with the same general ferrite core type setup. Such=20
>=20
If you have an E core transformer type winding, the magnetic flux is=20
linked strongly to the winding, so great winding precision is not required,=20
only a good physical stability. The magnetic field interactions are=20
controlled by the positions and separations of the iron laminations, so thes=
e=20
need to be firmly fixed. These E core types can be either modified audio=20
transformers similar to Sean's, ferrite cup transformers with DIY windings,=20=
a=20
linear differential reluctance system of the magslip type, or two physically=
=20
large coils pairs bridged by a ferrite rod, like the NE type.
J
> ust some (unexperienced) thoughts. I think the Nuova item costs around $1=
03=20
> US dollars (PCB and LVDT) plus whatever shipping/handling. Sometimes I=20
> think that a homebrew LVDT with perhaps a Phillips or Analog IC (and other=
=20
> parts) might be a cheaper alternative......but probably more costly for an=
y=20
> R&
The two Analogue Devices LVDT chips have a relatively high noise floo=
r=20
and are not suitable for our use. The Philips NE5521 device is about 1/3 the=
=20
price and you can get right down to the intrinsic opamp noise level. =20
The LX1358 Kit price is =E2=82=AC 51.65 + Carriage. Looking up today'=
s=20
exchange rate, this is just US $51.23 !! There will be a charge on your=20
Credit Card for the currency conversion and the exchange rate will be lower=20
than the International Banking Rate quoted. Air Mail packets are reasonably=20
priced and may take 10 days to arrive. Be warned that International Carriers=
=20
may have a minimum charge of near $50 for any packet! They seem to be only=20
relatively reasonably priced within the USA.=20
=20
With a LX1358, I get ~7 nano metre resolution over a +/-6 mm=20
accurately linear range, with 65 mV / micron output. The output is 10% down=20
at +/-12 mm and 20% down at +/-15 mm, since the ''out of range'' response is=
=20
of the usual S shape. This wide range was designed for Lehman type systems,=20
but the sensor will work equally well on simple pendulums.
Regards,
Chris Chapman
In a message dated 15/11/=
02, meredithlamb@............. writes:
Agreed, but there are sever=
al distance transducer approaches which can be successful. To my mind, there=
is a distinct shortage of amateur 'off the shelf' sensor designs capable of=
giving near professional resolution and stability. There is only one LVDT K=
it sensor available with a PCB, to my knowledge, which can give about 7 nano=
metres resolution at 10 Hz over +/-6 mm, with a max range of +/-12 mm.
Presume its the Italian, Nuova Elettronica unit (?), you've been working=
with / on over time?
Hi Meredith,
I have done some experiments modify=
ing / improving my Nuova Elettronica LVDT board, reducing the two pole detec=
tor filter to 10 Hz, which is more suitable for seismic use, changing=20=
the output opamp to a low noise type and increasing the time constant of the=
high pass filter, along with several other small changes. NE seemed to have=
used a circuit designed by Philips for wider bandwidth commercial LVDT appl=
ications. I have also been working on other magnetic sensor systems using th=
e Philips NE5521 LVDT chip, including modified transformers, somewhat like S=
ean's, some using ferrite components and lately, a capacitative system. I ha=
ve a full write up nearly finished.=20
Perhaps the greatest curios=
ity item I wonder about their transformer part, is whether its really a prec=
ision wound unit like mentioned as a (professional) necessity as on other we=
b sites? It appears to be much like other general looking transformers=
.. Without any access I'd guess its somewhat linearly wound but no where near=
precision.
If you want to measure several inch=
es of movement and require a linear response, you may need a precision linea=
r winding. If you only want a few mm and the field of the sensor rod can be=20=
sensed over a physically much larger sensor winding, you just need the windi=
ngs to stay in a fixed position. There will of course be thermal expansion a=
nd changes of resistance with temperature, but these effects are very small=20=
over the 10 Hz bandwidth and the 5 to 30 sec of the high pass filter time co=
nstant.
Perhaps the biggest thing I=
"seem" to see, is that Seans original LRDT transformers are in likelyhood k=
ind of a crude nonlinear wire wrapping; whereas, your LVDT, is a painstaking=
relatively precision wire layered device.
With Sean's variable reluctance tra=
nsformer sensor, the magnetic circuit is defined by the position of the flat=
iron armature plate and the iron polepieces. In a LVDT, the interaction is=20=
between the position of a magnetised armature and the varying distance it mo=
ves inside the sensor coil, so the coil needs to be quite uniformly wound. F=
or a description of an LVDT, see http://www.msiusa.com/schaevitz/pdf/lvdt/LVDT_Intro=
..pdf It is not too difficult to make even windings, so long as you=
do not use extremely fine wire, do not try to wind too fast, slow up right=20=
at the end of each layer and, most importantly, use very thin paper, like ci=
garette or tissue paper, for interleaving each layer. This makes it easy to=20=
spot and correct any overwinding or incorrectly spaced turns. It also makes=20=
winding easier and helps to prevent shorted turns in between layers. I made=20=
two matched 40,000 turn precision differential transformer coils this way, f=
or measuring paramagnetic susceptibility. A very small plastic tube makes a=20=
good wire guide. It is also possible to just layer wind two LVDT secondaries=
end to end on a sensor tube and use a ferrite or powder iron cored choke to=
provide the magnetic excitation. This avoids having to wind a central prima=
ry magnetising section in between, or over the top of the secondaries.
=20
Yet.....according to Seans website, his LRDT is sensitive to a incredib=
le 0.1nm. I've=20
zero experience but it make=
s me wonder further about coils themselves between=20
the two. It would seem the LRDT is much simpler to make and more s=
ensitive overall.
Coupled with the very high sensitiv=
ity of Sean's transducer, the movement range is small, perhaps +/-0.5 mm in=20=
a 2 mm total gap, and the output is only linear for a limited range around t=
he zero position. This is just fine for a feedback seismometer, where you tr=
y to hold the armature position constant. The out of range signal increases=20=
dramatically before saturating, which can be bad news if the seismometer fee=
dback system gets out of lock.=20
All this is leading up to t=
he question of whether a fairly linear wound (but not exactly precisely laye=
red) coil "could" somewhat work simarily (with less sensitivity) with the sa=
me general ferrite core type setup. Such coils are rather scarce I'am=20=
sure.
If you have an E core transformer t=
ype winding, the magnetic flux is linked strongly to the winding, so great w=
inding precision is not required, only a good physical stability. The magnet=
ic field interactions are controlled by the positions and separations of the=
iron laminations, so these need to be firmly fixed. These E core types can=20=
be either modified audio transformers similar to Sean's, ferrite cup transfo=
rmers with DIY windings, a linear differential reluctance system of the mags=
lip type, or two physically large coils pairs bridged by a ferrite rod, like=
the NE type.
J
ust some (unexperienced) th=
oughts. I think the Nuova item costs around $103 US dollars (PCB and L=
VDT) plus whatever shipping/handling. Sometimes I think that a homebre=
w LVDT with perhaps a Phillips or Analog IC (and other parts) might be a che=
aper alternative......but probably more costly for any R&D overall.
The two Analogue Devices LVDT chips=
have a relatively high noise floor and are not suitable for our use. The Ph=
ilips NE5521 device is about 1/3 the price and you can get right down to the=
intrinsic opamp noise level.
The LX1358 Kit price is =E2=82=AC 5=
1.65 + Carriage. Looking up today's exchange rate, this is just US $51.23=
!! There will be a charge on your Credit Card for the currency conversi=
on and the exchange rate will be lower than the International Banking Rate q=
uoted. Air Mail packets are reasonably priced and may take 10 days to arrive=
.. Be warned that International Carriers may have a minimum charge of near $5=
0 for any packet! They seem to be only relatively reasonably priced within t=
he USA.=20
=20
With a LX1358, I get ~7 nano metre=20=
resolution over a +/-6 mm accurately linear range, with 65 mV / micron outpu=
t. The output is 10% down at +/-12 mm and 20% down at +/-15 mm, since the ''=
out of range'' response is of the usual S shape. This wide range was designe=
d for Lehman type systems, but the sensor will work equally well on simple p=
endulums.
Regards,
Chris Chapman
[ Top ]
[ Back ]
[ Home Page ]