Ultra Precision Reference LTZ1000

[Kleinstein]

Measuring a LTZ1000 referenz with an 34465 or similar meter is mainly measuring noise and drift of the meter internal LM399 reference. Even with a better meter it is a combination of both references, so something like two LTZ1000 circuits.

[Dr. Frank]

So I run my tests with my heated LTZ1000 (non A - no compensation).
Results are opposite to what I expected, than more I approach the upper limit of the heater, than more TC increase in a non linear way.
I have to think at this.

...

Well, maybe you should have read before, what I have measured recently..
The T.C. seems to rise as soon as you approach the oven temperature by some margin.

It is also quite convenient to draw the output deviation over the temperature, then you can see directly, what's going on, and you may draw a box around the curve for determining the boxed T.C.

Frank

[TiN]

Indeed, output/temp graphs might be easier to grasp:

[kj7e]

  There is no real curve or slope, its seems to average right down the middle, no non-linearity seen.  So over a 29.6 deg span that comes to ~0.01ppm? It appears to be simply flat across any temp I would ever expect this box to be used in. 

Hello,

seems to be only noise (perhaps by thermal EMF).
Did you cover your inputs to the DMM with a cloth?

with best regards

Andreas

I did not, the test was done just as the photo above shows. The real problem is as Kleinstein points out, trying to measure a LTZ1000 reference with an LM399 based meter.  At this point I can only confirm my references seem to be performing well and possibly far better than the meter I have to measure them with.  At least the drift I was able to see before installing the 300K TC resistor appears to be gone now.  It's as good as I can get it, going to have to send it off for any further evaluations at this point.

[Andreas]

Hello,

let´s do the math:

In 10V range a 1 ppm drift of the LM399 reference will give you 1 ppm drift (7uV) of the 7V reading at constant input.
In 100mV range a 1 ppm drift of the LM399 reference at 20 mV input will give you 1 ppm drift (20nV) of the 20mV reading.

So even when you loose more than 1 ppm due to the scaling resistors in the 100mV range, the relative measurement will be better than the absolute measurement in 10V range.

with best regards

Andreas

[beanflying]

This thread is truly epic, I think I have digested and understood 'some' of the first 10 pages 

Not sure why but I am also having flashbacks to the most boring Uni lecturer I had to tolerate in 2 hours sessions for Probability and Sadistics 

[Zucca]

This thread is truly epic, I think I have digested and understood 'some' of the first 10 pages 

I think some of us need an "Ultra Precision Reference LTZ1000 for beginner..." you could probably make 10 YT Video on this.

[d-smes]

Thermographs of the FX reference, LTZ1000A with 1K/15K setpoint, powered by +11VDC. Large size clickable. 

@TiN-  Very nice thermographs!   Is the board laying horizontal?  If so, I'm surprised at the 3K temperature difference surrounding the LTZ pins.  Had you expected this?  On your thermograph of the back side, frame IR000740, it looks like three pins are cooler than the rest.  Is this due to different amounts of solder, thicker traces on these pins, different emissivity, or something else?

[Andreas]

  Had you expected this?

Hello,

if you look at the chip photo: there are different numbers of bond wires connected to some pins.
(different thermal resistance).

With best regards

Andreas

[branadic]

I'm the opinion that because of different emission factors of the different materials the results shouldn't be overrated. For a quality statement some coating with known emission coefficient is essential.

-branadic-

[branadic]

All I wanted to say is, take it only as a nice colorful picture but nothing more, even though TiN has marked several points with temperatures. These temperatures are not valid or reallity unless each point of the measurement was corrected for its individual emission coefficient or the complete board coated with a coating of known emission coefficent. There are a lot of people out there and I'm sure also on this board that don't know the difficulties on thermogaphics and take that pictures as representing reallity. No, it isn't!

-branadic-

[Kleinstein]

Much of the thermal gradients would be stable and thus would not change that much. It is mainly the part of the thermal gradients that does change, that could be trouble. Sources for this is the changing heater power of the  LTZ and the heater transistor and to a smaller degree the diode. Another factor it heat loss from convection, if the instrument is tilted.
Temperature differences between the critical pins (mainly  pins 3 , 7 and 6) can add thermal EMF, that can add to the final TC.
The tricky part is finding a good balance between thermal gradients that would favor thin long lines and low copper resistance that would favor thick copper.

With the thermal camera reading emissivity can be a factor - especially readings on metallic surface are not real. So the small gradients close to the hot LTZ1000 might not be all real, but could include environmental radiation and also heat flow by convection if the board is tilted. It still is nice to look at the thermals.

From below the temperature around the LTZ looks really nice and even. At least one of the two slightly colder pins is the rather thick ground connection and thus not a surprise. The other could be due to the lines going there too. Things will change with additional cover anyway. The more telling thermal picture would be from below, with the cap on the top side mounted.

[MisterDiodes]

Just as helpful thought, not bashing TiN's work at all:

The photos are an estimate of what's going on, but not close to reality unless you have a proper constant-emissive coating designed for real thermal testing.  We don't for an instant take the temperature annotation to mean anything for an absolute measure of temperature.

That being said what you do see here that on such a large board you get big temperature gradients from one end to the other.  Hopefully they don't change too much but that depends on enclosure and environment.  On a much smaller Vref module you'll get a board hopefully in equilibrium at about one temperature around the Vref components (and reacts much less to changing environment), and that's normally what you want for production.

On the other hand, the other thing we don't know about this board is how it works in an enclosure - you really have to look at the entire final system as a whole.  It may be that the temperature evens out inside the enclosure - an IR photo of the -bare- board could be misleading. What counts is how does it work in the final enclosure system, and we'll see that on the final TC testing of the finished system - and it could be that everything works out just fine and within LTZ specs.

TiN:  Where is the air-draft cover on the LTZ top and bottom?  Or will this be part of the enclosure system??

[branadic]

That being said what you do see here that on such a large board you get big temperature gradients from one end to the other

That is a typical misinterpretation I was warning about. Your interpretation implies a constant emission coefficient of the solder mask along the complete board, but this is not the case and depends on several aspects such as thickness of the solder mask, copper density underneath, the compound iself, reflections of the surrounding, ...

-branadic-

[Edwin G. Pettis]

Regardless of the thermal photo, you don't need a thermal photo to know that this board is going to have a significant thermal gradient from one end to the other, that is not an interpretation due to the image.  Small board is better than large board in nearly all cases.

[zhtoor]

Regardless of the thermal photo, you don't need a thermal photo to know that this board is going to have a significant thermal gradient from one end to the other, that is not an interpretation due to the image.  Small board is better than large board in nearly all cases.

maybe somebody tries this to reduce thermal gradients across the ref board.
ie; 3-dimensional pcb construction.

[TiN]

Where is the air-draft cover on the LTZ top and bottom?  Or will this be part of the enclosure system??

They will be installed for final tests. All these above are indeed just colorful pictures to get some rough idea of things, not aiming for any kind of accurate thermal test, can't agree more

I still need to assemble two remaining refs with 10V stage. There was no intention to make this board isothermal, however to reduce hotspot area board will be thermally coupled to metal base of the enclosure (except LTZ1000 area, of course).  Kelvin connection for 7V voltage output is taken at inner layer traces, not surface trace.

[3roomlab]

It may be that the temperature evens out inside the enclosure

im curious, do you purposely design your boards to maximise thermal "even-ness" (excluding the LTZ area)?

[TiN]

It's easier to try keep temperature same across the parts to limit the root cause for issues, instead of battling with temperature compensation and other wizardry to fix the consequences.

[cellularmitosis]

Does anyone speak French?  Is there a way to automatically translate French captions into English?

https://youtu.be/SVXdSIp9stc

[plesa]

Does anyone speak French?  Is there a way to automatically translate French captions into English?

https://youtu.be/SVXdSIp9stc

Yes, enable subtitles and in settings select auto-translate to English. It is not 100%.

[kj7e]

Does anyone speak French?  Is there a way to automatically translate French captions into English?

https://youtu.be/SVXdSIp9stc

Yes, enable subtitles and in settings select auto-translate to English. It is not 100%.

I hope its not 100%, otherwise we have all been doing it wrong.

[hwj-d]

Yes, that at precisely 01:12:53 -54 it's my telling too, that's where i'm totaly understand him ... 

[beanflying]

I don't speak or read french but it made more sense untranslated 

Somewhere in the first 10 minutes the auto translate dropped the "F" bomb 

[Grapsus]

Does anyone speak French?  Is there a way to automatically translate French captions into English?

https://youtu.be/SVXdSIp9stc

I do, feel free to ask if you don't trust the automatic translation at some point. Also Jipihorn, the guy who made this video is a member of this forum...