Ultra Precision Reference LTZ1000

[splin]

“O Metrologists," imo said, "the task I have designated you to perform is this. I want you to tell me" ......... "The Answer."

      The Answer?     The Answer to what?

"LTZ1000s!" urged imo. "The ultimate question of 80s/90s LTZ1000s compared to post merger ADI/LT LTZ1000s and everything!"


      <long pause>


      Tricky


"But can you do it?"


      <long pause>


      Yes.......... We can do it.

"There is an answer?" said imo with breathless excitement.

      Yes.   LTZ1000s and Everything. There is an answer. But, We'll have to think about it.

“How long?”.

      Seven and a half million minutes

"Seven and a half million minutes...!” imo cried. "That's nearly 14.27 years!

      Yes, We said we’d have to think about it, and take rather a lot of measurements didn’t we?

[SilverSolder]

Hopefully there have been advances in manufacturing and materials over 30-40 years?

[Kjelt]

Hopefully there have been advances in manufacturing and materials over 30-40 years?

Yes!................. but not on precision ultra low noise zenerdiodes 

[SilverSolder]

Perhaps it is time to switch to AC? 

[dr.diesel]

Hopefully there have been advances in manufacturing and materials over 30-40 years?

Yes!................. but not on precision ultra low noise zenerdiodes 

Perhaps no need/market for it?  Or maybe companies that require that performance have rolled their own, Intel maybe as an example?

[mrflibble]

      Yes.   LTZ1000s and Everything. There is an answer. But, We'll have to think about it.

“How long?”.

      Seven and a half million minutes

"Seven and a half million minutes...!” imo cried. "That's nearly 14.27 years!

      Yes, We said we’d have to think about it, and take rather a lot of measurements didn’t we?

I'm surprised ArthurDent hasn't made an appearance yet. 

[dietert1]

Kleinstein wrote an answer about reference noise filtering and i did a little experiment to get some numbers.

Took an 18 KOhm resistor to charge a 650 uF film capacitor we had to 10 V by our AD587 reference and then left it connected with a 3.3 MOhm resistor over night, watching with HP 3456A. Such capacitors are currently offered at ebay for 44 € per unit. After connection of the 3.3 MOhm resistor the voltage dropped slowly, in total about 70 or 80 ppm. But in the morning the voltage drop recovered and ended up as only 23 ppm.
It means that HP 3456A input parallel with that capacitor exhibits a current of 70 pA at 10 V, corresponding to 143 GOhm. Probably this is the input current of the HP3456A, so with a good OpAmp this may be much lower.

Regards, Dieter

PS: I can imagine that the capacitor may produce unwanted effects under temperature change but that's another step.

[exe]

What about this circuit? (values on the schematic are arbitrary).


I saw it in AoE. The idea is upper capacitor has zero DC voltage accross it, so leakage doesn't matter. As a result, you can use  even electrolytic cap (so it is claimed).

[dietert1]

No, you want the capacitor to serve as a reference during the filter time constant. Sorry, but i guess the dirt effects will be much worse with electrolytic caps. Probably you won't even reach a time constant of 35 minutes due to self discharge.
Now i remember we have some gold supercaps that i could try as well. But then the problem will be charge balancing to reach 10 V. As Kleinstein wrote, it's not something simple.

Regards, Dieter

[SilverSolder]

Kleinstein wrote an answer about reference noise filtering and i did a little experiment to get some numbers.

Took an 18 KOhm resistor to charge a 650 uF film capacitor we had to 10 V by our AD587 reference and then left it connected with a 3.3 MOhm resistor over night, watching with HP 3456A. Such capacitors are currently offered at ebay for 44 € per unit. After connection of the 3.3 MOhm resistor the voltage dropped slowly, in total about 70 or 80 ppm. But in the morning the voltage drop recovered and ended up as only 23 ppm.
It means that HP 3456A input parallel with that capacitor exhibits a current of 70 pA at 10 V, corresponding to 143 GOhm. Probably this is the input current of the HP3456A, so with a good OpAmp this may be much lower.

Regards, Dieter

PS: I can imagine that the capacitor may produce unwanted effects under temperature change but that's another step.

Interesting experiment.

You could also compensate the 70pA DMM current with some kind of current source, to balance things out.   The enormous time constant would likely "eat" any noise produced by the current source.

[3roomlab]

Kleinstein wrote an answer about reference noise filtering and i did a little experiment to get some numbers.

Took an 18 KOhm resistor to charge a 650 uF film capacitor we had to 10 V by our AD587 reference and then left it connected with a 3.3 MOhm resistor over night, watching with HP 3456A. Such capacitors are currently offered at ebay for 44 € per unit. After connection of the 3.3 MOhm resistor the voltage dropped slowly, in total about 70 or 80 ppm. But in the morning the voltage drop recovered and ended up as only 23 ppm.
It means that HP 3456A input parallel with that capacitor exhibits a current of 70 pA at 10 V, corresponding to 143 GOhm. Probably this is the input current of the HP3456A, so with a good OpAmp this may be much lower.

Regards, Dieter

PS: I can imagine that the capacitor may produce unwanted effects under temperature change but that's another step.

in some older posting by mr blackdog, he did a guarded capacitor, ie capacitor sitting on a guard voltage which sits on another capacitor to gnd. the guard rail provides all the leakage the gnded capacitor needs, while the AC goes thru 2 capacitors to ground, the precision rail is suppose to see zero leakage to the guard rail because the potential is nearly the same.
it is complex, but it looks like you can make as big a filter as you want as long as you can wait for the charge up.

[dietert1]

Probably this is the circuit exe showed above with two capacitors, the lower one meant to be the helper (guard). Still there is one critical capacitor that needs to have reference quality up to a certain time scale. It is the one which holds the main voltage and the other one is more or less decoration, like the electrical drive in a hybrid SUV. Maybe i am wrong, but somebody needs to show the numbers.

Regards, Dieter

[Kleinstein]

Leakage is just one problem with a low pass filter. Another is dielectric absorption, that may requires hours to weeks (with electrolytic caps) for settling.  Capacitors also react to temperature changes and possibly also air pressure changes, which can add low frequency noise. Instead of using two $44 capacitors, I would prefer a second reference chip.

It also depends a lot in how the reference is used. Most DMMs act as a low pass filter. Filtering would in this case not help very much (it still can with some DMMs).

The 2 capacitor circuit helps against leakage, but only to a limited extend against dielectric absorption and one needs 2 caps of twice the size.

[iMo]

Fyi - I've seen somewhere Mickle(T) did a measurement of the current coming out of the DMM's inputs (tens pA till nA as a short pulses pos/neg during AZ etc., about 10 DMMs, as I can remember). Thus the direct measurement of the voltage at the capacitor might be somehow influenced by that current..
PS: re above 2 caps filter - do the electrolytic (and tantalum) capacitors need a "minimum polarization voltage" in order to work properly??

[dietert1]

Yes, somebody should write a clear statement that building analog storage devices with time constants of minutes or hours in 2019 is complete retro. Let the DVM run, take averages and use those numbers. For example i plotted the Allan deviation of my AD4522 preamplifier and found that it has a broad minimum of 3 nV at about 30 to 60 seconds. Nobody would do this exercise in the analog domain.

Regards, Dieter

[SilverSolder]

Of course analog filters with long time constants are retro - but they are still cool!

[TiN]

Wonder what that might be...



6-layers of ceramic substrate...
Price?  If you have to ask, you can't afford it ... 

[martinr33]

3458a ovenized quad reference board?

Does it support the Pickering patent?

[TiN]

What patent? Oven oscillator at startup? No, I am little sceptical that it helps much anyway.

[martinr33]

Yes - the Pickering patent claims to significantly reduce LTZ1000 drift after a power off period.

https://patents.google.com/patent/US5369245A/en

"An electronic component is conditioned to remove distortion in a temperature response characteristic due to a temperature hysteresis effect, by subjecting the component to a controlled temperature variation prior to operation of the component at a given temperature. "

[mimmus78]

Wonder what that might be...



6-layers of ceramic substrate...
Price?  If you have to ask, you can't afford it ... 

I guess it's just one more iteration of this other ...

Inviato dal mio ONEPLUS A5010 utilizzando Tapatalk

[TiN]

I'm bit lost how your board assembles together? How are the results? I see PCB made in May 2019?
And unconnected copper pads bugging me 

[mimmus78]

They assemble one o top of the other.

The board have pin 1 and pin 2 swapped on both LTZ1000 ... Kicad played some bad trick when it lost the symbols reconnecting it in reverse order when I reassigned it. Anyway I managed to not notice the problem despite the many times I have checked the schematics. I had to bodge some fix that fortunately seems to have no much impact on thermal design.

The board had also some filtering if you wanted to use autozero opamp for the LTZ reference. This seems to cause more problems so I just reverted to original LTZ circuit bypassing filtering.

Once fixed board I had to build a TEK mini thermal chamber to properly test TC ... and this has taken me some more time.

Averaged reference seems pretty stable and also TC looks ok ... you'll never know how it works until you do not put inside the 3458a.
Then the project was suspended because I was busy moving to a new house ... things are moving slow now.

The upper board hosts the two LTZs.
The bottom board hosts connection to 3458a, averaging/buffering circuit, and virtual ground generator for second zener.

There are some unconnected copper pour (both sides) just aiming to improve temperature equalization around the board, don't know how it improves ...

Virtual ground is used for second zener, referencing pin 7 of first LTZ.

I used only one divider shared between the two LTZ for temperature setting, "powered" from the averaged/buffered reference. Sensing of the averaged buffered out reference is done at the 3458a connector pin, so what it goes to the divider can suffer from some copper TC. I made some raw calculation and the effect should be very minimal.

PS: Routing the board trying to keep EMF balanced was a hell of work ... I can see why you selected ceramic substrate.

[feedback.loop]

Pins 1 and 2 are the heater. Does it matter which way around they are connected?

[MiDi]

It is very important to have the correct polarity and both pins have to be >= the voltage at pin 4.
The substrate diodes are drawn in the block diagram in Datasheet.