physical explanation of the tilt effect?
I read somewhere that the LTZ1000 chip is mounted in a special way
to reduce thermal conductivity. That would sort of indicate some
kind of edge mount with the chip floating, and that again would
certainly expose it to gravitational effects.
In that case, it should only have three modes: up, down and sideways,
where the chip is vertical, but the rotation around the center
axis should have very little effect.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
Poul-Henning,
I use an LTZ1000. Its chip is mounted in a standard TO99 package, so your explanation does not fit.
The LTZ1000A is assembled with a thermally insulating die attach (ie epoxy to glue the die to the leadframe).
What all of us is missing is the correct physical explanation, including my assumption, which is derived from the long time constant of the change only.
I.e. if it would be a gravitational effect, please explain how the orientation of gravitational force acts on the circuit, or on the physical package and changes the output that slowly, stable after a minute only, and not instantaneously.
In the case of of an XTAL, one can explain this (known) orientation effect by the Coriolis force, but should act instantaneously. Gyro sensors use this effect for navigation puposes.
For the LTZ, I do not see a similar physical relation at the moment, do you?
A magnetic effect could be the interaction between the vector of the earth magnetic field and the direction of the current flow within the chip, ie the Hall Effect.
But this interaction would be instantaneously also.
A thermal effect could be induced by different heat convection flow from the warm LTZ to the surrounding solder joints.
I isolated the LTZ1000 additionally by a PS cap around it, so i really wonder...
Anyhow, if anybody finds "the missing link" between the drift effect and the physical property, it would be interesting to improve the assembly.
PS: I also have -perhaps - seen a certain drift effect in the same order of magnitude when rotating the 3458A, but as this measurement was too noisy, I cannot tell for sure, neither give any quantitative value.
Frank
In message 4C95DD5C.3080307@freenet.de, "Dr. Frank Stellmach" writes:
In the case of of an XTAL, one can explain this (known) orientation
effect by the Coriolis force[...]
It has nothing to do with the Coriolis force, it is the force of gravity
adding stress to the X-tal, no more, no less. The "2G turnover" effect
is instantaneous. See John R. Vigs Quartz Crystal Tutorial.
For the LTZ, I do not see a similar physical relation at the moment, do you?
I would personally suspect the effect is mainly thermal.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
Convection effects inside the package are likely to be position dependent, even if the case itself is kept at a ~constant temperature.
The exact amount of air (or nitrogen) that flows over the chip itself inside the package can redistribute the thermal gradient, however small of an effect that is.
Didier KO4BB
Sent from my Verizon Wireless BlackBerry
-----Original Message-----
From: "Dr. Frank Stellmach" drfrank.stellmach@freenet.de
Sender: volt-nuts-bounces@febo.com
Date: Sun, 19 Sep 2010 11:52:28
To: volt-nuts@febo.com
Reply-To: Discussion of precise voltage measurement volt-nuts@febo.com
Subject: [volt-nuts] physical explanation of the tilt effect?
I read somewhere that the LTZ1000 chip is mounted in a special way
to reduce thermal conductivity. That would sort of indicate some
kind of edge mount with the chip floating, and that again would
certainly expose it to gravitational effects.
In that case, it should only have three modes: up, down and sideways,
where the chip is vertical, but the rotation around the center
axis should have very little effect.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts | TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.
Poul-Henning,
I use an LTZ1000. Its chip is mounted in a standard TO99 package, so your explanation does not fit.
The LTZ1000A is assembled with a thermally insulating die attach (ie epoxy to glue the die to the leadframe).
What all of us is missing is the correct physical explanation, including my assumption, which is derived from the long time constant of the change only.
I.e. if it would be a gravitational effect, please explain how the orientation of gravitational force acts on the circuit, or on the physical package and changes the output that slowly, stable after a minute only, and not instantaneously.
In the case of of an XTAL, one can explain this (known) orientation effect by the Coriolis force, but should act instantaneously. Gyro sensors use this effect for navigation puposes.
For the LTZ, I do not see a similar physical relation at the moment, do you?
A magnetic effect could be the interaction between the vector of the earth magnetic field and the direction of the current flow within the chip, ie the Hall Effect.
But this interaction would be instantaneously also.
A thermal effect could be induced by different heat convection flow from the warm LTZ to the surrounding solder joints.
I isolated the LTZ1000 additionally by a PS cap around it, so i really wonder...
Anyhow, if anybody finds "the missing link" between the drift effect and the physical property, it would be interesting to improve the assembly.
PS: I also have -perhaps - seen a certain drift effect in the same order of magnitude when rotating the 3458A, but as this measurement was too noisy, I cannot tell for sure, neither give any quantitative value.
Frank
volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.
It definitely sounds like a convection effect. It could be internal or
external to the package, or both. Varying heating/cooling of external solder
joint may be the cause. Further from the package there is the
heating/cooling effect of the surrounding circuitry and the equipment
enclosure in general. Internally it might change the temperature sensed
within the package relative to the actual zener temperature. Another
possibility is due to the very high thermoelectric potential generated with
silicon as one conductor. The thermoelectric potential for silicon/copper is
about 400uV/C compared to about 2uV/C for lead/copper. The chip is probably
electrically connected with aluminium or gold, but the thermoelectric
potential to silicon will be similar.
Since the package is hermetically sealed it would be possible to distinguish
internal/external convection effects by
testing the reference board under high vacuum where external convection
would cease.
Alan
----- Original Message -----
From: "Dr. Frank Stellmach" drfrank.stellmach@freenet.de
To: volt-nuts@febo.com
Sent: Sunday, September 19, 2010 10:52 AM
Subject: [volt-nuts] physical explanation of the tilt effect?
I read somewhere that the LTZ1000 chip is mounted in a special way
to reduce thermal conductivity. That would sort of indicate some
kind of edge mount with the chip floating, and that again would
certainly expose it to gravitational effects.
In that case, it should only have three modes: up, down and sideways,
where the chip is vertical, but the rotation around the center
axis should have very little effect.
--
Poul-Henning Kamp | UNIX since Zilog Zeus 3.20
phk at FreeBSD.ORG
https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts |
TCP/IP since RFC 956
FreeBSD committer | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by
incompetence.
Poul-Henning,
I use an LTZ1000. Its chip is mounted in a standard TO99 package, so your
explanation does not fit.
The LTZ1000A is assembled with a thermally insulating die attach (ie epoxy
to glue the die to the leadframe).
What all of us is missing is the correct physical explanation, including
my assumption, which is derived from the long time constant of the change
only.
I.e. if it would be a gravitational effect, please explain how the
orientation of gravitational force acts on the circuit, or on the physical
package and changes the output that slowly, stable after a minute only,
and not instantaneously.
In the case of of an XTAL, one can explain this (known) orientation effect
by the Coriolis force, but should act instantaneously. Gyro sensors use
this effect for navigation puposes.
For the LTZ, I do not see a similar physical relation at the moment, do
you?
A magnetic effect could be the interaction between the vector of the earth
magnetic field and the direction of the current flow within the chip, ie
the Hall Effect.
But this interaction would be instantaneously also.
A thermal effect could be induced by different heat convection flow from
the warm LTZ to the surrounding solder joints.
I isolated the LTZ1000 additionally by a PS cap around it, so i really
wonder...
Anyhow, if anybody finds "the missing link" between the drift effect and
the physical property, it would be interesting to improve the assembly.
PS: I also have -perhaps - seen a certain drift effect in the same order
of magnitude when rotating the 3458A, but as this measurement was too
noisy, I cannot tell for sure, neither give any quantitative value.
Frank
volt-nuts mailing list -- volt-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/volt-nuts
and follow the instructions there.
A thermal effect could be induced by different heat convection flow from
the warm LTZ to the surrounding solder joints.
I isolated the LTZ1000 additionally by a PS cap around it, so i really
wonder...
Anyhow, if anybody finds "the missing link" between the drift effect and
the physical property, it would be interesting to
improve the assembly.
Hello,
on my LM399 where heater, sensor and regulator are on the same chip I
measured a different heater current depending on orientation. So I think
since heater, reference element and temperature sensing element are not
exactly at the same location and the temperature spreading between the 3
elements is not exact symmetrical there will be small temperature
differences depending on wether the heater is exactly below the temperature
sensor or exactly below the reference element.
By the way: on LM399 the heater current is a minimum when the device is
mounted upside down (with the pins in the upside direction).
with best regards
Andreas
Hello all,
when having a look at the link from Alan to the chinese collegues:
the photograpy from the opened LTZ and the (mirrored) photo of the chip
you can see the following.
chip photo:
-
Pin1 (heater at lower left) has 3 short bond wires:
should be the coolest place on the chip
especially for the LTZ1000A which is isolated from the case -
Pin 8,7 and 7 are on the corner lower right.
-
so the lower half of the chip is cooled by 6 bond wires
against the upper half with 4 bond wires. (Pin 2-5)
I am not a chip designer but when looking at the photo
the buried zener seems to be the circular thing in the middle
surrounded by the transistors.
The both transistors seem to be splitted each into 4
90 degrees parts T1 and T2 interleaved by 45 degrees.
On the other side the inner heater circles seem not to be
bonded (not active ?)
with best regards
Andreas