5V precision referene

[alm]

The LTZ1000 is also used in commercial voltage references (eg. Fluke 732B).

No it is not !. i got proof !. keep your eyes peeled....

Please check your sources before declaring a post false four months later. The Fluke 732A and early 732B models used a Motorola reference, and the later Fluke 732B models used the LTZ 1000 (obviously a highly selected and burned in version). Here is a document from someone who might know a thing or two about Fluke voltage references that supports this.

[fmaimon]

I can't find any datasheet on it, but this guy has some info: http://www.eefocus.com/lymexbg2vo/blog/10-01/183124_69352.html

It's in chinese, so google translate to the rescue: http://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http%3A%2F%2Fwww.eefocus.com%2Flymexbg2vo%2Fblog%2F10-01%2F183124_69352.html

[Dr. Frank]

The LTZ1000 is also used in commercial voltage references (eg. Fluke 732B).

No it is not !. i got proof !. keep your eyes peeled....

Please check your sources before declaring a post false four months later. The Fluke 732A and early 732B models used a Motorola reference, and the later Fluke 732B models used the LTZ 1000 (obviously a highly selected and burned in version). Here is a document from someone who might know a thing or two about Fluke voltage references that supports this.

Sorry alm,

Fluke never used the LTZ1000  in any of their own designed calibrators or references.

Fluke always has used custom specific 'Reference Amplifiers' like the Motorola SZA263 or its (exact) replacement "LTFLU", from Linear Technology. 
Their characteristic is, that the zener diode is connected to the emitter of the transistor in the same package.
That special Fluke reference schematic was already used for the very early calibrators, as 332B, 335B/D, where an equivalent TI Reference  was used. The advantage is, that you can amplify the ~ 7V volt of the Reference amplifier directly to 10,000V with the same OpAmp.
 
In the article from Deaver, a "Linear Technology Reference" is mentioned, but that's simply the "LT-FLU".
Reason for that change is the Part Termination of Motorolas SZA 263. See the 732B teardown of free_electron.
Deaver nowhere mentions explicitly an LTZ1000 like type.

The use of the LTZ1000 instead of the SZA263 would have implied a complete redesign of the 732B, because the LTZ1000 is not compatible at all with the overall Fluke reference schematic.

Only when Fluke acquired Metron Designs Ltd., and the 7001 reference, invented  by famous John R. Pickering, it was the first LTZ1000 based reference with the brand of Fluke on it.

Btw.: Fluke just terminated the 7001 reference system.
That's a pity, because obvioulsy no big oven and no steady powering was required for those devices.

Frank

[amyk]

I can't find any datasheet on it, but this guy has some info: http://www.eefocus.com/lymexbg2vo/blog/10-01/183124_69352.html

It's in chinese, so google translate to the rescue: http://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http%3A%2F%2Fwww.eefocus.com%2Flymexbg2vo%2Fblog%2F10-01%2F183124_69352.html

There's a bunch more info here too, also in Chinese.

(Any idea why the Chinese seem to know so much more about these references? Maybe they get a lot of "scrapped" but working equipment sent there for disposal...)

[Neganur]

(Any idea why the Chinese seem to know so much more about these references? Maybe they get a lot of "scrapped" but working equipment sent there for disposal...)

Without having too much of an idea about how people live their lives in either regions, the sheer amount of people^* in China lets me imagine, that there also are more people who are interested in electronics.

^*: world population 7x10⁹ (2011), USA 0.32x10⁹ (2012), Europe 0.74x10⁹ (2012), China 1.344x10⁹ (2011)

[alm]

Fluke never used the LTZ1000  in any of their own designed calibrators or references.

You appear to be correct, I must have been confused with the later 7000 series that they acquired through Wavetek. I couldn't find any authoritative source for the reference used in the 732B (although I did find a bunch of non-authoritative volt-nuts posts claiming it is the LTZ1000), but the schematic clearly shows a zener diode connected to the emitter  of a transistor, unlike the LTZ1000.

[amspire]

I can't find any datasheet on it, but this guy has some info: http://www.eefocus.com/lymexbg2vo/blog/10-01/183124_69352.html

It's in chinese, so google translate to the rescue: http://translate.google.com/translate?hl=en&sl=auto&tl=en&u=http%3A%2F%2Fwww.eefocus.com%2Flymexbg2vo%2Fblog%2F10-01%2F183124_69352.html

This page and the pages it links to are really great. Thanks.

Even has information about the LTFLU ic.

Temp Stability: 5ppm/C
0.3 ppm/ 1000 hours
1ppm /  year
RMS noise:  0.06 ppm

The 5ppm may not look that great but that would be an average coefficient over a range of temperatures. You can adjust the temperature coefficient by adjusting the zener current, so Fluke can get better then that .

Once it is put in an oven that regulates to perhaps 0.01 degC, and you adjust the current to zero temp coefficient at the oven temperature, the end result could easily be below 0.01ppm/C.

The thing about adjusting the zener current for zero temp coefficient is it is hard to adjust for zero temp coefficient over a range of temperatures, but you can precisely adjust for zero temp coefficient at one precise temperature.

[nukie]

Its very sad that we have to reference a Chinese website for an American made item. That's just show where the future of engineering is heading. I was very frustrated when I first started off studying precision voltage reference, the lack of information in the western. Without this Chinese 'nut' we would have lost great wealth of information.

[amspire]

What seems to be clear is that making great buried zener references is still as hard as ever, and IC companies would prefer not to know about it.

It wouldn't surprise me if it comes down to trail and error tweaking - make a wafer. Test the devices. Tweak the parameters a bit and make another wafer. Continue until you have reached the correct performance.

My guess is we will continue to see an improvement in other reference technology - like bandgap references - until they get good enough for secondary references. There is no incentive to even try and make better zener references.

If a compact way can be invented to cool an IC substrate down to 4 deg K, then companies like Fluke may start using Josephson Junction references instead of zener-based references. Is there a lower temperature limit for the Peltier/Seebeck effect?

[ecat]

I don't know if there is a theoretical limit to commercial Peltier devices but iirc there is a practical limit in that the efficiency is inversely proportional to the temperature difference between the hot and cold sides of the device. So a delta T of 300degC requires either a massively powerful device (?) or a hideously inefficient stack of ever more powerful devices.

That new fangled laser cooling looks interesting but I have no details.

[Dr. Frank]

...
My guess is we will continue to see an improvement in other reference technology - like bandgap references - until they get good enough for secondary references. There is no incentive to even try and make better zener references.

Nope, Bandgap references will never beat Buried Zener devices regarding long term stability:

The Buried Zener structure is diffused deeply and safely below the surface and occupies a relatively big volume / area in the chip. Therefore, the diffusion and decay processes in the silicon crystal (depending basically on temperature, Arrhenius law) do not lead to such high drift rates over time, compared to the bandgap structures, latter ones being relatively small and situated vulnerable at the surface of the chip.
2nd reason: Oxygen and humidity do not deteriorate the active silicon structure  of the Buried Zener Devices, because of their hermetically sealed package.

The ageing behaviour of Buried Zener devices can also not be improved significantly by a better design, because the dominating decay process in silicon follows a natural law.


Only if bandgap references are ovenized, like buried zener devices, they may reach the same low T.C.
That can not be achieved by better trimming of the bandgap refs, due to yield problems and quick worsening of trimmed T.C. over time due to above described deterioration of the chips structure.

If a compact way can be invented to cool an IC substrate down to 4 deg K, then companies like Fluke may start using Josephson Junction references instead of zener-based references. Is there a lower temperature limit for the Peltier/Seebeck effect?

Sorry, that will not happen:

JJ arrays consist of Superconductors (e.g. Niobium structures), and can definitely not be made from silicon technology.

Commercially available JJ arrays are produced by NIST, PTB and some specialized university institutes or one or two private companies only.

You definitely need liquid Helium(4) to achieve 4.2K. No cooling element can do that.

Even if you would construct a miniature Helium liquefier, you also would need a Helium dewar enclosure to set up your experiment.
It's much easier to simply rent a dewar can filled with 50l of liquid Helium and dip your experiment inside.

That all is already commercially available; complete metrology standards based on a JJ array @ 4.2K cost around 200k. (Fluke owns several of them)


Recently, physicists demonstrated, that High Tc Superconductors (from YBaCuO) can be structured to make a JJ, and to get around 10^-8  uncertainty for the 'Josephson-Volt', at ~60K (pumped liquid nitrogen). That would be much easier regarding the low temperatures, but the YBCO - JJ is still far away from being available and affordable.

Frank

[amspire]

I don't know if there is a theoretical limit to commercial Peltier devices but iirc there is a practical limit in that the efficiency is inversely proportional to the temperature difference between the hot and cold sides of the device. So a delta T of 300degC requires either a massively powerful device (?) or a hideously inefficient stack of ever more powerful devices.

It would have to be a stack, but at the inner level, you could be looking at cooling a chip a small as 1mm square. Each device has to be two or three times the power of the previous device and probably six or more layers would be needed. I did a quick search. The current commercial  Peltier devices cannot get down to 100K, but I did see a patent for a barium copper oxide superconductor Peltier device that can cool to a stable 65K. I know, it is a long way from 65K to 4K, but they are gradually getting closer. There has been a lot of work on high temperature Josephson Junctions for computers, and even back in 1995, they had junctions that worked up to 70K. So perhaps it is possible.

I think the laser cooling you mentioned is only good for cooling Helium gas that is already extremely cold to start with.

[amspire]

JJ arrays consist of Superconductors (e.g. Niobium structures), and can definitely not be made from silicon technology.

Agreed. I was never considering using silicon for a Josephson Junction.

Commercially available JJ arrays are produced by NIST, PTB and some specialized university institutes or one or two private companies only.

Recently, physicists demonstrated, that High Tc Superconductors (from YBaCuO) can be structured to make a JJ, and to get around 10^-8  uncertainty for the 'Josephson-Volt', at ~60K (pumped liquid nitrogen). That would be much easier regarding the low temperatures, but the YBCO - JJ is still far away from being available and affordable.

That is just about in the range of Peltier cooling, and so that is promising. Of course devices built in a lab will be uncommercial, but once they know how to make these high temperature junctions, they will be much cheaper. The main motivation for development of high temperature Josephson Junctions is computing and very high frequency processing. When developed commercially, you will be down to fractions of a cent per junction.

So combine a 70K high temp junction array with an achievable Peltier cooling down to 65K and problem solved!

I know, it is easy to say.

Richard.

[branadic]

What about LT1236LS8?

[casinada]

There is some new competition to the Pentaref from voltagestandard.com

I wonder if anybody got any of these:

http://www.ebay.com/itm/New-AD584LH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/221210908532?pt=LH_DefaultDomain_0&hash=item3381329774
http://www.ebay.com/itm/NEW-AD584LH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/221208115533?pt=LH_DefaultDomain_0&hash=item338107f94d
A search on Taobao found:
http://item.taobao.com/item.htm?spm=a230r.1.14.20.ZEFK2r&id=12553765969
75Y = $12 US dollars

I don't know if they come with a certificate of calibration. 

[bingo600]

There is some new competition to the Pentaref from voltagestandard.com

I wonder if anybody got any of these:

http://www.ebay.com/itm/New-AD584LH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/221210908532?pt=LH_DefaultDomain_0&hash=item3381329774
http://www.ebay.com/itm/NEW-AD584LH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/221208115533?pt=LH_DefaultDomain_0&hash=item338107f94d
A search on Taobao found:
http://item.taobao.com/item.htm?spm=a230r.1.14.20.ZEFK2r&id=12553765969
75Y = $12 US dollars

I don't know if they come with a certificate of calibration. 

I'd not trust it ...
Just the fact thait it switches the 3 voltages automatically to the plugs via a MCU , would make me want to see the the circuit

I'll keep on trusting my Geller 10v

/Bingo

[ecat]

There is some new competition to the Pentaref from voltagestandard.com

I wonder if anybody got any of these:

http://www.ebay.com/itm/New-AD584LH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/221210908532?pt=LH_DefaultDomain_0&hash=item3381329774
http://www.ebay.com/itm/NEW-AD584LH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-/221208115533?pt=LH_DefaultDomain_0&hash=item338107f94d
A search on Taobao found:
http://item.taobao.com/item.htm?spm=a230r.1.14.20.ZEFK2r&id=12553765969
75Y = $12 US dollars

I don't know if they come with a certificate of calibration. 

No calibration that I can see, so just an out the box AD584 with 10V +/-5mV specs (assuming it is the L version)  sitting right beside a boost converter.