Calibratory D-105 DC Precision Voltage Reference Standard

[zhtoor]

OK, I took some advice from that IT guy who summed up the thread. Why does my standard outperform the raw regulator IC? It's all in the datasheet, if anyone actually read the whole thing. Long-term drift: the IC is aged 1-2 years. That brings the drift way down. We age the entire assembly, so any other components that experience long-term drift benefit.

Why is the temp drift lower than the raw IC? A 100kohm NGT is wired into the trim circuit. When the temp rises, the resistance drops, and the output gain is decreased. The IC drifts up with temperature, so it evens out.

But that's not enough. Because the thermistor has to be a bit different for each individual IC, because temp drift is not constant between specimens. The manufacturer rates temp drift over a huge range. I'm only concerned with 17-26C. The drift is fairly flat there, and so is the NGT.

A pot used as a rheostat in parallel with the thermistor adjusts the resistance delta of the thermistor with temperature, which is all we really care about (how much does it change). It is the quantity of change that determines how much the TC compensates the output gain.

The absolute resistance in the TC circuit is not critical. By adjusting the rheostat, the resistance delta of the NGT decreases or increases with temp, giving less or more tempco. The copper heat pipe keeps the delta temp similar between the IC and the NGT.

Without it, calibration of the tempco would not be possible. The device would oscillate because of the tempco. The exact thickness of the heat pipe is critical. If it's too thin, it doesn't transfer enough heat. If it's too thick, it acts as a heat sink, which throws off the thermodynamic balance of the entire assembly.

Unlike that other guy who made the standards and then quit, I have a much faster way to test the tempco. Then, I only require the environmental chamber to test the final result.

Why is it so cheap? I am not ambitious. I make 1500% profit on the parts, after eBay and Paypal fees. I don't need to sell them for more! I'd feel guilty!

How can you do what no one else seems to be able to do for the price? I persist, I believe in myself, and I apply scientific principles in new ways. That's it. And, I received much welcome advice from a very intelligent MSEE, without whom the entire project would have failed.

No one is going to get a surprising result unless they try something new.

Why is the soldering so bad? Because it's made with a perf board. There are no solder pads. With traces I get stray capacitance, which causes oscillation.

How can you make such exaggerated claims? A: How would anyone know my claims are exaggerated if they haven't tested the device?

Why are you such an idiot? Well, I truly know nothing. But my IQ is a standard deviation of 4 from the mean. I don't consider myself intelligent. But compared to other humans I guess I am.

How can you show your face on eevblog? I have nothing to be ashamed of. I can show my face anywhere. I'd prefer to make a product that works right with bad soldering than one that has neat solder joints and doesn't work.

Why are your long-term drift figures so unbelievable? I just age the ICs for a long time.

How can you claim +-2ppm accuracy? I make the claims that are justified by experimental evidence. I haven't been able to get +-1ppm accuracy. So, I stick with +-2ppm.

Do you have any new specs? Yes, thank you for asking. Typical standard deviation of the output of 250 samples at 100PLC is 500nV. That means 99% of the readings are within 1.5uV of each other.

If you are willing to spend 800.00USD on a calibration from Fluke Calibration, why do scam everyone? I don't think anyone who intends to scam actually requires certified instruments. 

Why did you allow everyone to call you a liar, and only after that post the cal cert? To piss you off!

Why don't you just leave us to our prejudices? Because ignorant is no way to go through life. I see a lot of technical skills on this board, but that's about where it ends. A man is measured by the kindness he displays to others.

Why is your design so stable? Mostly because I put it in a box. That prevents changes in thermal EMF caused by air currents across the IC pins. Read the datasheet!

Why should we accept you here? You have the mind of a child! I'm not asking for acceptance. I could never hope to attain unto the examples set by the other members on this board. I'd be happy if you just based your criticisms on experimental evidence rather than idle speculation. I mean, you use experimental evidence in every other area.

Are you mentally ill? I'm a creative genius. Mental illness seems to go along with that quite frequently. My prose  have been compared to those of James Joyce and Virginia Woolf, two authors who were considered by many to be mad.

hello,

keep up the good work. even if it is not good enough for anybody or everybody.

welcome to the board,
we are all brothers/sisters here, and share a common parent; ie; Adam & Eve.

and if i have understood correctly,
the best advice from the mighty prophet called Jesus / Isa (may peace be upon him)
is to forgive and "Let him who is without sin cast the first stone".

best regards.

zhtoor/pakistan.

[zhtoor]

and by the way,

using a perfboard probably cuts down on warping of the board due to thermal gradients,
which can easily be simulated by using finite element thermal analysis,
sort of like drilling *MANY* holes in the board, instead of drilling them around the LTZ1000 in
some published pcb designs.

regards.

[Awesome14]

and by the way,

using a perfboard probably cuts down on warping of the board due to thermal gradients,
which can easily be simulated by using finite element thermal analysis,
sort of like drilling *MANY* holes in the board, instead of drilling them around the LTZ1000 in
some published pcb designs.

regards.

Yes, and allow me to complete the principle. The holes in the perf board relieve thermal stress which would otherwise induce changes in the output, due to hysteresis in the regulator IC. Actually, regarding the LTZ1000, there is an engineering note that advises the pins be held in rubber grommets fit into the PCB, and that the connections to the IC pins be made with 30 gauge bare copper wire coiled to reduce stiffness.

So, thank you for your observation and for your vote of confidence.

[Awesome14]

My Dad actually told me that over the phone some time ago. I'll ask him where he got the information, and post it if I can find out.

[2N3055]

......................

LOL... Epic! 

[Awesome14]

Hello Awesome14,
                            actually it was me - in person - that revised latest datasheet of the LTZ1000A - and all comments were approved, typos removed and errors modified by engineers from LTC and this became DS LTZ1000 rev. E 11/15-2015.

Slot holes are effective for certain footprints (e.g. LTC6655 LS8) - some LTC videos prove this, and I discussed this also with a leading, eminent specialist in buried zeners - and his conclusion was that after many years monitoring slot holes on the LTZ1000 - they did not see any statistical effect and they stopped using slot holes more than one decade ago ... - those boards w/o slot-holes were shipped to CERN and other leading institutions.

Certain people that did not understand the way the LTZ1000(A) works and should be implemented - copied the black magic over and over ... - also the chinese do ... - very funny for insiders to see this happen

During all my observations I never spotted Fluke, Keysight, Keithley using slot-holes - Last couple of years I have invested many hundreds of hours in reading many articles, app. notes, from LTC a.m.o. ... about LTZ1000A - and if such design note from LTC would exist - I would probably have known this - and after more than 25 years as professional design engineer - I do not believe in VooDoo solutions ...

So I leave it up to you to proof - that I am wrong - by showing such design note from LTC discussing the effect of slot holes to improve the stability of the LTZ1000(A), i said in particular case for LTZ1000(A) ...

Good luck !

cheers
butterfly

This ap note applies to all LT references, although the example of slotting is applied to a different IC. The paper doesn't specifically exclude the LTZ1000. But the latest data sheet might be newer. Then LT should correct AN-82. http://cds.linear.com/docs/en/application-note/an82f.pdf

[Jay_Diddy_B]

This ap note applies to all LT references, although the example of slotting is applied to a different IC. The paper doesn't specifically exclude the LTZ1000. But the latest data sheet might be newer. Then LT should correct AN-82. http://cds.linear.com/docs/en/application-note/an82f.pdf

AN-82 says that metal packages TO-5 and TO-46 are largely immune to board stress. Since the LTZ1000(A) is a metal package, it is excluded. Here it is:



Regards,
Jay_Diddy_B

[zhtoor]

so the board stess "voo-doo" *does* apply to dip packages then, as used in calibratory d-105?

[2N3055]

so the board stess "voo-doo" *does* apply to dip packages then, as used in calibratory d-105?

Plastic DIP and SMD packages..
And it's not voodoo..
Plastic packages are affected by mechanical influences, and also by humidity absorption.
Also they have rather large hysteresis on temperature and humidity changes, from both mechanisms..
Well known physics phenomena..
That is why you want metallic or ceramic hermetic case to minimize both...

All that is well discussed here on EEVBLOG, on Internet, and in numerous app notes...

And we all do prototypes and one off circuits on perfboard. But you shouldn't solder with a hammer..

This is workmanship to aspire to..
http://elm-chan.org/docs/wire/wiring_e.html

[Cerebus]

Since the LTZ1000(A) is a metal package, it is excluded. Here it is:

Specifically for the LTZ1000(A), the die attach material is unique to the part and was developed both to thermally insulate the die from the case and to reduce stress transmission between the die and the case. This is specifically the 'A' part, not the 'non-A' part, which uses a more conventional die attach.

[Zucca]

This thread is better than Game of Thrones now. Thanks butterfly.

[lars]

Here is two screen shoots from excel with about two years of measurements of two D-105, two SVR-T and one VREF10. As reference I have an REF102CM (metal case) with temperature compensation that has been powered for 17 years. The VREF10 is compensated with the value 1.3ppm/C I got from the manufacturer. The others are not compensated. The room temperature has varied between 19 and 29 °C.

In the beginning I did a humidity test as can be seen in the graphs. Both D-105 and SVR874 were in a box with a large amount of silica gel prepared to get about 80-85%RH. As can be seen from the quick drop the box was not very tight. The VREF10 was in a plastic bag with a silica gel bag.

The D-105 and SVR874 were continuously powered about the first year. The last year it has been in a shelf in a box between measurements so now with a humidity of 35%RH in the room but above 50%RH most of the last two months the references probably have acquired humidity. This can be clearly seen on the D-105's as they dropped about 5ppm the last week when powered in the about 35%RH environment.

The VREF10 has only been powered about 15 minutes at every measurements. The original batteries are still ok!

Lars

[timb]

Wow, the graphs for the two D-105s are basically identical, and pretty bad. It literally jumps all over the place, following the humidity curve, which is what you don’t want to see! The VREF10 is climbing, but that looks like natural settling over time.

[Gyro]

It's just what I'd expect to see from an epoxy packaged reference (as seen in the teardown photos) in response to humidity. I just wonder why it doen't appear up in the specified performance. 


P.S. Thanks for the continuing SVR-T data Lars, it gives me a warm and fuzzy feeling.

[martinr33]

The star of the show here is definitely the SVR device, where both units are closely tracking 0 ppm. This figure also gives some confidence in the reference.

The Calibratory devices are obviously vulnerable to humidity, but it looks like they would be not so bad with some better sealing. Or even a bag of silica gel inside the package.

The SVRs have the advantage of a ceramic package,plus a lot of careful tuning to compensate for temperature.

The VREF10 part is interestingly independent of humidity, maybe because it is in a plastic bag with silica gel.  Plus, it is temperature compensated. If there is correlation between temperature and humidity, then the VREF10 is getting an unfair advantage over the other plastic part devices.

All in all, some solid work here! Thanks for having the foresight...

[Macbeth]

Hello VoltNuts - I spoke today with a main responsible of our national accredited Lab (CE approval) and apparently to export a DC source to europe in the way it is presented by Awesome14 - the manufacturer and export/import company - both are fully liable to comply with CE legislation (CE compliance, RoHS, etc. ... ).

So where are the documents that grant such export license to europe ... ?

best regards
Butterfly

Just stick a crappy "CE" or the "China Export" mark on it.

CE is worthless shit, much like everything else dreamt up by unelected Brussels bureaucrats. You know the HP 3458A is now illegal for sale within the EU?

As much as I dislike Awesome14's insane claims, we really do not want political zealots to decide who gets to sell voltage refs. Let the free market decide. We also have traceable cal certs for a reason.

This also applies to any dickhead in charge of a lab that decides to use Awesome14's D-105 instead of a Fluke 732 for example. Instant job loss. 

[Echo88]

Did you hear they want to ban asbestos?! *cough* Those bastards! *cough*! Next time they might take away our lightbulbs, those dickheads!

[zhtoor]

Hello Awesome14 (Tom K.) - I am still waiting on you to show us that particular design note ... or went your B.S. generator out of service lately ... ?

http://www.atrixnet.com/bs-generator.html

Hello Awesome14,
                            actually it was me - in person - that revised latest datasheet of the LTZ1000A - and all comments were approved, typos removed and errors modified by engineers from LTC and this became DS LTZ1000 rev. E 11/15-2015.
...
So I leave it up to you to proof - that I am wrong - by showing such design note from LTC discussing the effect of slot holes to improve the stability of the LTZ1000(A), i said in particular case for LTZ1000(A) ...

Good luck !

cheers
butterfly

respectfully sir,

lets not start a flame war again...

best regards.

[Awesome14]

I asked my Dad where he read the ap note on the LTZ1000, and he didn't remember. He only mentioned it in passing over the telephone when I was working on a LTZ1000 reference. But I did mock up a LTZ1000 reference and push on the IC a bit, and the voltage output changes.

Now, for Lars graphs; Good, solid work. I'm happy to have the info. If the unit was designed to operate at over 50% RH, I'd have a problem. But the standard is designed to be operated in a controlled environment not to exceed 50% RH, and with a temp between 17-26C.

I haven't done any testing at RH above 50%, because I work in a standards lab with a controlled environment (RH 30-40%). I wasn't aware there was a market for high-humidity, low-cost standards. But, in any event, the unit is not designed to operate above 50% RH or above 26C.

[Cerebus]

But the standard is designed to be operated in a controlled environment not to exceed 50% RH, and with a temp between 17-26C.

This is the first time, as far as I am aware, that you have specified this humidity limitation and, having just checked your ebay sale offer, there is no mention of that limitation there. I find that rather odd, to not mention a basic, critical design parameter. Heck, even my TV comes with that information clearly supplied.

You say:

I haven't done any testing at RH above 50%, because I work in a standards lab with a controlled environment (RH 30-40%). I wasn't aware there was a market for high-humidity, low-cost standards. But, in any event, the unit is not designed to operate above 50% RH or above 26C.

Didn't it ever occur to you that: 1) Buyers of cheap voltage references aren't likely to be people running standards labs with highly controlled environments (if you can afford the air-con you can afford the 732), 2) that 50% RH is quite average for much of the temperate world, indeed it's most definitely not a "high-humidity" environment.

[beanflying]

Wow was on my watch list and considering buying one. - unwatched for bad attitude to peer review alone.

I wasn't going to post up my veroboard MAX6350 reference in the projects section due to my "average soldering" but I think I will 

[Andreas]

From my friend volt nut I have the indication of around 0.3-0.4 ppm/% rH for the REF102 in plastic package.
Time constant of epoxy packages with humidity is around 3-7 days reaction time.

So my suggestion would be to do verification measurements in february / march when there is low humidity and in august / september with high humidity.

From a 5 V Reference (LT1027CN, 2 samples) I get a humidity coefficient of around 0.5ppm/%rH

Hmm,

thats what we knew already 2.5 years before. (plastic package = humidity sensitive with individual sensitivity and time constant)

Anyway if the calibration is done at 30% rH and travels 1 week to a customer with 50% rH it will be already 10 ppm off.
In my lab with 40%rH in winter and up to 70% in summer even more.

with best regards

Andreas

[ManateeMafia]

Has anyone else wondered if Awesome14 is also keen101? This appearance of the 3458A reference board makes me wonder even more.

[saturnin]

Hello Tom K. (awesome14) - Oooh it was you working on this LTZ1000 reference ... ? We observed this ad with your great soldering artwork on Flebay - "used briefly for a prototype" 

https://www.ebay.com/itm/HP-3458A-Voltage-Reference-Board-REV-A-03458-66509-LTZ1000/263337867308?_trkparms=aid%3D111001%26algo%3DREC.SEED%26ao%3D1%26asc%3D49452%26meid%3Dfa21f476378b41c3b41e43bb691f0928%26pid%3D100675%26rk%3D1%26rkt%3D15%26sd%3D263337867308&_trksid=p2481888.c100675.m4236&_trkparms=pageci%253Af8c65513-cfbf-11e7-8aec-74dbd18045d9%257Cparentrq%253Ae55276b115f0ab6ac789a120fffb06cc%257Ciid%253A1

A true craftsman at work ... 

Wow! $288 for abused 3458A reference module with very questionable performance. Wouldn't it be at least fair to explicitly mention that some components were desoldered and put back (or replaced)?!

[MK]

Hi, that reference still has solder splashes on it! does not inspire confidence, they could move later at any time.