EU - CalClub

[TiN]

Maybe we start separate thread on this? As where did you get 55ppm/K figure? I recently stress-tested LTZ1000A ref (chamber temp outside of oven setpoint) and got much smaller TC, around 3ppm/K. If the real swing would be near +/-155C that would mean LTZ die will be very much dead already.

I will get to it, but all my meters are currently busy characterizing FXes, need to get these shipped to EEVBlog members already, including members of EU Club. 

EDIT: Wrong terminology used, stuff I measured was NOT the unheated-zener tempco.

[Dr. Frank]

Maybe we start separate thread on this? As where did you get 55ppm/K figure? I recently stress-tested LTZ1000A ref (chamber temp outside of oven setpoint) and got much smaller TC, around 3ppm/K. If the real swing would be near +/-155C that would mean LTZ die will be very much dead already.

I will get to it, but all my meters are currently busy characterizing FXes, need to get these shipped to EEVBlog members already, including members of EU Club. 

Hello Illya,

measurement technique and T.C. determination are here: https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg1336575/#msg1336575

That ~ +55ppm/K was also confirmed by other volt-nuts, but I can't find the link at the moment.

I doubt that you correctly measured these 3ppm/K.

Anyhow, something else happens in the 7000s cycling process, than only the change of the oven temperature. To find out, you really need to measure directly this Ube of the reference element. Maybe your 3ppm/K measurement was done at the Ube of the transistor of the oven control circuit, instead ?!

Frank

[SvanGool]

...
Hot shipping is the alternative method, with the known disadvantages and shortcomings.
...

Hot shipping under constant temperature, generically, also for portable resistor-references, is a brain project of mine. The advantages, assuming that one can also take care of humidity stabilization (e.g. using silica gel method):

• Much lesser temperature fluctuations and therefor much lesser TC dependability
• Reproducibility of accuracy in different environmental conditions, with supplied cables and connectors and when load conditions are met

The disadvantages, I can think off:

• Power need, will require a substantial battery pack (preferably non Li-Ion because of shipping limitations)
• Weight and size
• Is it allowed to ship a "hot" device?
• Needs an accurate oven circuitry and isolation

Ideally this would be a 23°C device, but then you need heating and cooling. Otherwise a 45°C should do for a heating-only device. It should be possible to have a PPMish reproducible accuracy and a PPMish long term stability.
I also know that a lot of people will classify this as "cannot be done as diy, look at what is needed for a 732A", but I didn't hear very strong arguments yet of not achieving this by putting an LTZ1000xxx and a few hermetic VPGs in such a device.

Any suggestions/comments/laughs?

[Dr. Frank]

...
Hot shipping is the alternative method, with the known disadvantages and shortcomings.
...

Hot shipping under constant temperature, generically, also for portable resistor-references, is a brain project of mine. The advantages, assuming that one can also take care of humidity stabilization (e.g. using silica gel method):

• Much lesser temperature fluctuations and therefor much lesser TC dependability
• Reproducibility of accuracy in different environmental conditions, with supplied cables and connectors and when load conditions are met

The disadvantages, I can think off:

• Power need, will require a substantial battery pack (preferably non Li-Ion because of shipping limitations)
• Weight and size
• Is it allowed to ship a "hot" device?
• Needs an accurate oven circuitry and isolation

Ideally this would be a 23°C device, but then you need heating and cooling. Otherwise a 45°C should do for a heating-only device. It should be possible to have a PPMish reproducible accuracy and a PPMish long term stability.
I also know that a lot of people will classify this as "cannot be done as diy, look at what is needed for a 732A", but I didn't hear very strong arguments yet of not achieving this by putting an LTZ1000xxx and a few hermetic VPGs in such a device.

Any suggestions/comments/laughs?

we (e61_phil and I) already did successful tests on hot shipping on his 'bread box' reference assembly.
Very massive and heavy, but very long lasting battery.

Unlike the 732A/B, it is not necessary at all to build another oven around the whole PCB.
Just to continuously heat the LTZ1000 oven, and of course power the whole reference circuit (~25mA @ 12V) is enough.
The other components of the reference circuit do not have an impact on the hysteresis, and if trimmed to near zero T.C., also do not influence the absolute output voltage of the LTZ1000 over temperature.

Everything else is complete over-engineering.  In another use case, you would also intend to temperature stabilize the 10V upscaling resistor divider.

I'd like to refer to the LTZ1000 thread, where < 1ppm stabilities (over time, temperature, noise) have been experimentally demonstrated already, by many 'amateur' volt-nuts, despite the (unproven) statements of a few naysayers.

Frank

[Berni]

12mA at 12V is actually not that bad. But im guessing this is heavily dependent on the ambient temperature and how well insulated the thing is. Gonna take heck of a lot more power to keep it warm at -10°C ambient compared to 30°C ambient wrapped in a bunch of insulation.

But going off that number that is 150mW of power. So running it for 3 weeks would consume 75Wh of energy. With the typical Li-Ion energy density of 200Wh/kg this would take a 380g battery. Safer types such as LiFePO or Ni-MH have about half the density at about 100Wh/h making the required battery about 750g. So yeah lugging a battery along would indeed add quite a bit of shipping cost, but can still be kept under 1kg.

Oh and what about any potential package x-ray scans? Since high energy radiation is known to have effects on semiconductors, would that upset its stability?

[babysitter]

Besides investing in isolation of the box, how about non-ohmic, non-peltier heating? Adding chicken to the package (Blue peacock bomb, anyone?) might be overengineering, too.

But how about slow exothermic chemical processes? Think of the hand warmers, but slower. Calcium oxide getting only very little water over shipping time? Or actually biological processes? It shouldn't emit anything nasty, of course, no one wants the shipped artifact full of rust. Dang, lets control the exothermic process by electronic sensing - logging might be a good idea anyway.

BR
Hendrik

[hwj-d]

Hmm ..   

very interesting positions, what an eu-calclub reference that can be sent all around should have in terms of properties. 

It seems to be fixed, there is not only one maybe two reference(s) with one transmitter and one receiver like usa cal club in the moment. No, there are many references with many transmitters and many receivers, that shouldn't be not that problem (many to many). It gets more problematic when many transmitters and many receivers also have many different ideas about what the respective reference should do for their respective interests ('nay' sayer vs 'beginner' vs real heavy voltnuts, but not in wintertime, but keep it small, but heavy isolation, and more...).

My concluson (and proposal) is:
 do what you can do in the real moment, but
   keep it simple!

 

[Echo88]

@babysitter: You could use Paraffin which uses its latent heat to warm the ref to ~40°C, depending on used Paraffin-type.

[FriedLogic]

My concluson (and proposal) is:
 do what you can do in the real moment, but
   keep it simple!

That would be my thought too. There are all sorts of options for a reference which can be put in the boot of a car and driven around, but a small ref to mail around is another thing.

(and thanks for starting this up!)

[Berni]

Hmm interesting idea with wax. It is indeed well known for this property.

Lets crunch the numbers on that Wikipedia claims wax has about 200 J/g of latent heat. Converting back to the same units gets us 55 Wh/kg. This is a pretty impressive figure when you consider its simplicity and essentially infinite charge/discharge cycles. But its still only half as good as a Ni-MH battery while having the disadvantage that you can't concentrate all of the heat into the desired place like you can using electrical sources. This spread out heat source means more heat loss trough the same insulation.

There is another way to go about it. Increase the performance of the insulation instead. There are practical limits to how much insulation you can wrap around it before its a impractically large ball of fluff. However another insulation technology is a vacuum thermos, providing incredibly high amount of insulation without being physically large. This does add about 500g of weight due to the inside often being glass. But from some info online it appears a reasonable thermos will let 1 liter of 100°C water drop to 85°C over 5 hours in freezing ambient. So water has 4.2 J/g/K of heat capacity meaning the thermos lost 63 kJ of energy or about 12.6 kJ/h, this works out to 3.5W. Taking a more reasonable delta T of half that gets is to about 1.7W. So yeah keeping anything physically large warm is quite difficult. Tho having it fit inside a thermos might be a good form of secondary insulation to ease the work of the insulation around the hot reference (At the cost of significant extra shipping weight however).

[AG7CK]

Maybe we start separate thread on this? As where did you get 55ppm/K figure? I recently stress-tested LTZ1000A ref (chamber temp outside of oven setpoint) and got much smaller TC, around 3ppm/K. If the real swing would be near +/-155C that would mean LTZ die will be very much dead already.

I will get to it, but all my meters are currently busy characterizing FXes, need to get these shipped to EEVBlog members already, including members of EU Club. 

Hello Illya,

measurement technique and T.C. determination are here: https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/msg1336575/#msg1336575

That ~ +55ppm/K was also confirmed by other volt-nuts, but I can't find the link at the moment.

I doubt that you correctly measured these 3ppm/K.

Anyhow, something else happens in the 7000s cycling process, than only the change of the oven temperature. To find out, you really need to measure directly this Ube of the reference element. Maybe your 3ppm/K measurement was done at the Ube of the transistor of the oven control circuit, instead ?!

Frank

A TC of +50ppm or so has been mentioned many times on the forum. Even if I have never measured or seen published any value, there is a compelling reason to believe that it is a reasonable number.

In several Datron schematics there is a series resistor added on top of the LTZ avalanche diode. Typical value afaik is 21 ohm or so.

Suppose that the naked LTZ has output 7.0 volt and a TC of +50ppm. One degree temperature rise will then rise the ref output with 0.00035 volt.

Assume also that the temperature compensating Vbe of the LTZ1000 in-reference transistor falls with -0.002 volt, and finally assume that the current determining resistor (that is in parallel with the base-emitter diode of the above mentioned transistor) is the standard 120 ohm.

A one degree rise in temperature will then reduce the current in the avalanche diode
with -0.002volt/120ohm= -0.000166(6666....) ampere. This fall in current will give a fall in voltage over a 21 ohm resistor equal to -0.0001667amperex21ohm= -0.00035 volt or so (ignoring numeric rounding). This incremental voltage fall thus cancels the voltage rise from the supposed LTZ1000 TC of +50ppm.

Hence the addition of the resistor in the Datron LTZ1000 reference circuits strongly suggests that the raw TC is +50 or so ppm.

[AG7CK]

Maybe we start separate thread on this? As where did you get 55ppm/K figure? I recently stress-tested LTZ1000A ref (chamber temp outside of oven setpoint) and got much smaller TC, around 3ppm/K. If the real swing would be near +/-155C that would mean LTZ die will be very much dead already.

I will get to it, but all my meters are currently busy characterizing FXes, need to get these shipped to EEVBlog members already, including members of EU Club. 

A measured or calculated TC of +3ppm/K seems imo erroneous. Please document your findings.

[IanJ]

Hi all,

Forgive the ignorance but what is the EU - CalClub.........I saw the thread for the US one but have never followed it so not really sure whats going on.

Ian.

[Dr. Frank]

There ist a deeper solid state physics reason/law, why ALL the LTZ1000 chips have about this T.C.
Any silicon diode with CC has quite exactly -2mV/°C.
The T.C. of zener diodes is always around zero @ 6.2V zener voltage, like 1N821.
It is predictably more and more positive at higher zener voltages.

Fluke has chosen and specified a Reference Amplifier voltage, so that the sum of both T.C.s nearly cancel, and that is about 6.9 V. The LTFLU is explicitly specified, that T.C. can be trimmed to zero by slight change of collector current.
As the LTZ 1000 has a zener voltage of more than 7.0V, its sum T.C. is  definitely higher, and predictibly at these +55ppm/K @ the 7.118V, which I measured.
Therefore, for physics reasons, the 3ppm/K must be wrong, or TiN has got an LTZ1000 with about 6.9V reference voltage.

If LT creates a new version of the LTZ1000, they should change the diffusion for a 6.9V ref. voltage.

Frank


 

[hwj-d]

Hi all,

Forgive the ignorance but what is the EU - CalClub.........I saw the thread for the US one but have never followed it so not really sure whats going on.

Ian.

Hi Ian,
that's a legitimate question, because we've not yet defined this more precisely.

I'll start from the back. Shipping costs, and -handling (extra costs), customs duty, duration, stress to the Ref's makes it unwieldy to send them regularly abroad/overseas and back. So we want to do this within europe.

To help someone with a relative precise and (edit: ourself-) calibrated voltage and resistance standard as adjustment of its own measurement devices is the closer reason.

The ref's themselves should be continuously logged (ftp) and, depending on which hands they run through, also freshly rated.

[The Soulman]

Hi all,

Forgive the ignorance but what is the EU - CalClub.........I saw the thread for the US one but have never followed it so not really sure whats going on.

Ian.

Just a group of people that send out a artifact/reference to each other to be able to compare each other volts and ohms,
the usa cal club has one reference (ltz1000 and vishay vhp(?)) that is shipped from one member to the other.

The discussion here is to start something similar but different in the eu, suggested are multiple artifacts and shipped in star pattern.
So the artifact returns to its "home base" after each go, to check for drift and deviation etc.

What that artifact should be and should be capable of is now getting discussed here.

imho multiple small, light weight and robust references should be used and perhaps a small (flat) form factor so it can be shipped as a letter, cheaper and more convenient.
Also 10V and 10K Ohm should suffice, people should do their ratio's at home, or get a 3458a to do it for them. 

[YetAnotherTechie]

Hello,
I see you guys are considering shipping the references hot. Would a drop of 5º of the internal heater be acceptable? what about 10º? Sure it's not instantly settled at the destination, but knowing the audience, i figure people would leave it to settle for many hours/days, after receiving it.
Is there an amount of temperature drop whose drift that can be absorbed in the usual settling time at the destination? every degree less would save a considerable amount of power, and battery.

[bsdphk]

One note about shipping hot:  Remember that delivery vehicles can become very hot during summer:  Do not over insulate if the thermal regulation timeconstant is large.

[nnills]

There ist a deeper solid state physics reason/law, why ALL the LTZ1000 chips have about this T.C.
Any silicon diode with CC has quite exactly -2mV/°C.
The T.C. of zener diodes is always around zero @ 6.2V zener voltage, like 1N821.
It is predictably more and more positive at higher zener voltages.

Fluke has chosen and specified a Reference Amplifier voltage, so that the sum of both T.C.s nearly cancel, and that is about 6.9 V. The LTFLU is explicitly specified, that T.C. can be trimmed to zero by slight change of collector current.
As the LTZ 1000 has a zener voltage of more than 7.0V, its sum T.C. is  definitely higher, and predictibly at these +55ppm/K @ the 7.118V, which I measured.
Therefore, for physics reasons, the 3ppm/K must be wrong, or TiN has got an LTZ1000 with about 6.9V reference voltage.

If LT creates a new version of the LTZ1000, they should change the diffusion for a 6.9V ref. voltage.

Frank

If the vref = 6.9V the zener should be 6.2V(assuming a LTZ1000 circuit). If the zener has no tc. at 6.2V why put the transistor(diode) in series(again assuming a LTZ1000 circuit)? Would it not create a -2mv/K tc.?

[Dr. Frank]

If the vref = 6.9V the zener should be 6.2V(assuming a LTZ1000 circuit). If the zener has no tc. at 6.2V why put the transistor(diode) in series(again assuming a LTZ1000 circuit)? Would it not create a -2mv/K tc.?

Well, your assumptions are not correct.

At first, a silicon diode, or a Ube diode have about 0.5V forward voltage at small currents of 20..200µA where the LTFLU is specified, but at room temperature of 25°C only.
At the intended 45°C oven temperature, the forward voltage is about 0.45V only, so the zener must have about 6.45 V, which gives a pronounced positive T.C.

Don't forget, that these -2mv/K are related to a 14 times smaller voltage than the zener, and to simply calculate the necessary T.C. of latter, it still is +2mV/6.45V*K = +310ppm/K !!

So to re-formulate your wrongly derived last question, why not simply use an 1N821 type zener with small T.C.?

The manufacturing variations of these 1n821 series (or of any zener diodes), are very big, so you can only get low T.C. by selection, which range from +/- 100ppm/K for the 1N821 down to +/- 5ppm/K for the 1N829. The yield for near zero T.C. is extremely low..
In any case, you have no chance to trim the T.C. externally to exact zero. This is required for non-ovenized applications (The LTFLU is specified for 25..45°C application temperature.)

Therefore, you need such a Reference Amplifier, which can be trimmed to exact zero T.C. by variation of collector / Ube diode current @ 3mA zener current.

The other intention is to quite simply amplify the reference voltage to 10 or 15V, inside the different FLUKE reference applications.

Frank

[branadic]

Unfortunately, the rare owners of this device, like TiN, did not yet measure the real oven temperature, and also not the cycling profile.
Otherwise, we also could design this scheme into our existing references... as far as it's really working.

That is a bit understatement, but yes, I'm guilty of not showing W7000 love.. I still hope to find time for complete reverse-engineering, so I don't have it closed up still, to do proper measurements.
But I did perform measurement of power on conditioning cycles year ago.





I'm not very happy with it, my last try to measure it's tempco (10V) provided wierd results that made no sense. And switching charger/isolated power supply emit nasty squeak which I can hear across the room. Don't like switching power supply next to reference board even single bit.

Used setup:



[/offtopic]

I pretty much appreciate tear down pictures. This unit is really interesting and with the upgraded battery pack with 1700mAh instead of 850mAh and thus 30h instead of previously 16h together with it's low power consumption <1W this is the perfect candidate by now for hot shipping. So Illya, you might want to put the teardown in the upper place of the priority list?
I'd also like to know more about the recovery methode, especially the details not covered by the patent or the paper. Would be great. A live stream on that is also welcome

-branadic-

[TiN]

My apologies for mixing up tempco of unheated LTZ zener vs out-of-oven margin LTZ tempco output. These are obviously not the same things. 

[FriedLogic]

Hi,
Did you notice this happening repeatedly or with different references? I'm just wondering if this is a fundamental issue, or one which can be fixed or reduced by things like component selection, circuit design, processing and usage.

I observed that effect on 4 of my LTZ1000 references experimentally, 2 of them where sent in winter, when outside temperatures were below zero °C.
For the other 2 references, I cycled them between -20°C and +80°C with decreasing temperature swing, and could bring both back to original value.

You've been busy!

Have you ever tried cooling them down to a low temperature before each power up? That sort of temperature cycling is not ideal, but at least it would happen regularly and the starting point would be similar each time.

[The Soulman]

Before another endless discussion about ltz1000 circuitry and compensation starts..

What would the potential members of the EU calclub would like to see in terms of specs and features?

10V and 10K or more than that?
sub ppm/K tempco or do we keep our labs close to the 20 dgr. C mark?
Noise?
Size/weight, small box that could be mailed more frequently or big and heavy with integrated battery's to ship it "hot"?

[Berni]

Well i for one don't have any sort of air conditioning in the lab so its never 20°C. Especially when summer comes around and night time temperatures start to rise above 20.

As for the hot shipped reference i guess most of the argument for that was the drift after being frozen in winter. Perhaps do both and have people opt into the hot shipped or cold shipped mailing chain. Could perhaps give a bit of insight if hot shipping makes a diference when both compare again at home base. That is if it never hangs up in the mail somewhere and its batteries run dry.