Effect Of Increased Thermal Capacity Of LM399/LTZ1000

[zhtoor]

Dear friends,

what would be the effect of increasing the thermal mass of LM399 / LTZ1000 to the tempco and stability of the reference?
also what would be the effect on position-sensitivity of LM399 as unearthed by andreas?

best regards and comments awaited.

[mimmus78]

Dear friends,

what would be the effect of increasing the thermal mass of LM399 / LTZ1000 to the tempco and stability of the reference?
also what would be the effect on position-sensitivity of LM399 as unearthed by andreas?

best regards and comments awaited.

I will discover it very soon ...

Inviato dal mio Nexus 6P utilizzando Tapatalk

[Andreas]

I will discover it very soon ...

Inviato dal mio Nexus 6P utilizzando Tapatalk

Hmm,

better thermal isolation helps.

but how do you increase thermal capacity without reducing thermal resistance to environment?

with best regards

Andreas

[zhtoor]

hello,

thermal capacity can be increased by simply increasing the mass of the chip by mounting on a
metal object (like a heatsink) and removing the polysulfone cap in case of LM399.

the heater would have a tough time though.

regards.

[Andreas]

Hello,

I´d rather try to isolate the chip from the housing like in the A-Version of the LTZ.
The lower the heater power -> the lower the termal gradients across the chip.

with best regards

Andreas

[Gyro]

That Polysulfone shield doesn't strike me as terribly efficient, it's compromised by the requirements of volume production, it certainly feels quite warm to the touch in operation. I suspect that you could probably match its performance with a larger metal block mass using decent thickness expanded poyystyrene of similar (not sure if 90'C is too hot for polystyrene long term).

You'd probably want to roughly match the thermal resistance that the heater circuit is designed to see - compare heater currents before and after?

[mimmus78]

Well I have this crazy idea that is to mount LTZ1000 on a super small PCB that will host only itself and the three main resistors (almost but not in contact with LTZ).
This should keep everything well confined, and running almost at the same temperature of the LTZ (or anyway hotter than usual temperatures).
Can be this considered as thermal mass increase?

[zhtoor]

trying to ovenize the resistors for free using ltz1k heater 

[Kleinstein]

Using thermal coupling to the temperature stabilized reference would only give a partial stabilization of the resistors. At best you would end up at something like 90% of ref. temperature + 10% outside temperature. It should not be so difficult to build a separate temperature control that is better than a factor of 10 in keeping the temperature stable.

The resistors abound the LTZ1000 are not that critical that it really warrants temperature stabilization. It would be different thing for the 7 to 10 V amplifier based on resistors - these resistors would be about 100 times as sensitive.

I don't know the LM399 internal heater circuit, but I very much doubt is will include a slow time constant. Even the LTZ circuit has something like 1/10 s as the longest analog time constant. So it should not care very much about things happening usually much slower. A direct contact to the case of the LTZ (non A Version) could be problem.

[mimmus78]

Using thermal coupling to the temperature stabilized reference would only give a partial stabilization of the resistors. At best you would end up at something like 90% of ref. temperature + 10% outside temperature. It should not be so difficult to build a separate temperature control that is better than a factor of 10 in keeping the temperature stable.

The resistors abound the LTZ1000 are not that critical that it really warrants temperature stabilization. It would be different thing for the 7 to 10 V amplifier based on resistors - these resistors would be about 100 times as sensitive.

I don't know the LM399 internal heat circuit, but I very much doubt is will include a slow time constant. Even the LTZ circuit hat soemthing like 1/10 s as the longest analog time constant. So it should not care very much about things happening usually much slower. A direct contact to the case of the LTZ (non A Veresion) could be problem.

Kleinstein I think "hovenizing" resistors have others advantages like stabilization of humidity and I think also that keeping resistors always at the same temperature (even if higher) can improve long term stability.



Inviato dal mio Nexus 6P utilizzando Tapatalk

[mimmus78]

Using thermal coupling to the temperature stabilized reference would only give a partial stabilization of the resistors. At best you would end up at something like 90% of ref. temperature + 10% outside temperature. It should not be so difficult to build a separate temperature control that is better than a factor of 10 in keeping the temperature stable.

The resistors abound the LTZ1000 are not that critical that it really warrants temperature stabilization. It would be different thing for the 7 to 10 V amplifier based on resistors - these resistors would be about 100 times as sensitive.

I don't know the LM399 internal heat circuit, but I very much doubt is will include a slow time constant. Even the LTZ circuit hat soemthing like 1/10 s as the longest analog time constant. So it should not care very much about things happening usually much slower. A direct contact to the case of the LTZ (non A Veresion) could be problem.

Kleinstein I think "hovenizing" resistors have others advantages like stabilization of humidity and I think also that keeping resistors always at the same temperature (even if higher) can improve long term stability.



Inviato dal mio Nexus 6P utilizzando Tapatalk

I don't think increasing thermal mass of lm399 is ok. LM399 is pretty small, has it's own caps. Everything suggest me that oven control is calibrated for this precise work. LTZ1000 can be more forgiving as you can tune the controlling circuit, but you still have the small heater inside it.

Inviato dal mio Nexus 6P utilizzando Tapatalk

[montemcguire]

Increasing the thermal mass of a component under an already stable thermal servo system will not make the thermal system more accurate, it will just increase the thermal settling time, and increase the heater overshoot when the system is slewing, reaching set-point. The whole point of a thermal servo system is to eliminate external factors, so rather than adding thermal mass, I think it'd be better to add thermal isolation so that the servo system will not have to respond as greatly to external thermal inputs, either thermal sources or conduction paths to other devices.

It's good to think of what could be done to the basic IC to improve its deployment, but adding thermal mass is not something I'd consider along those lines.

[zhtoor]

thanks all,

what i had in mind was to establish an external ambient in the form of a heated metal ring around
lm399/ltz1000 so that the internal heater has to work less, moreover, the external ring can hold
precision setpoint resistors embedded in it for stability.

regards and comments required.

[Kleinstein]

Adding a ring and thus thermal mass to the part surrounding the LM399 will not have much direct effect on the thermal regulation. If can help to keep fast temperature changes away from the chip.  Anyway I don't think the thermal environment is that critical with the LM399, though there are reports on orientation dependence which is supposed to be due to thermal effects.

For the resistors keeping them hot / warm is a 2 sided thing. I helpf against temperature variations and it can reduce humidity effects. However aging will be higher. So it depends on the resistors used it help.

[branadic]

For the resistors keeping them hot / warm is a 2 sided thing. I helpf against temperature variations and it can reduce humidity effects. However aging will be higher. So it depends on the resistors used it help.

Well, in the Prema 5017 the LM399 is mounted upside-down (bottom side of the pcb) with leads as long a possible with the original part. A melf resistor for the zener current as well as a ceramic capacitor are assembled between the four leads . This way the resistor is thermally stabilized by radiated heat too.

-branadic-

[Kleinstein]

The LM399 has an internal circuit to set the actual zener current. So it has very low differential resistance and thus low sensitivity to changes of the current setting resistor. In the normal circuit the current is set by 3 resistors like I = Uz*R1*R2/R3. 

[zhtoor]

@klienstien, thanks,now imagine, the metal ring surrounding the lm399 holding the gain setting resistors for 6.95 -> 10v opamp cct,
the op-amp (say LT1007 etc.) *and* the zener current setting resistor, and the whole assembly now insulated from the environment
using a high-temp plastic insulator like polysulfone. maybe the heater would require some active current limiter, resulting in longer
thermal settling times, resulting in longer time to stabilize.

care to comment on expected performance?

best regards.

[Kleinstein]

Keeping the resistors of a 7 to 10 v amplifier at a stable temperature can help. However this should not be the high temperature of the LM399. So the right way would be to have the LM399 on it own, and only stabilize the 2 resistors of the 10 to 7 divider with a separate small temperature controller to a much lower temperature of maybe 40 C. As the resistors have lower power consumption, there oven can also work with a relatively low power level and good thermal insulation. 

The LT1007 might not such a good choice, because of rather high bias currents, so a more suitable would be an LT1001 or OPA177.