How much would you pay for a assembled 10V reference?

[The Soulman]

Evidently, people did not read my post.  I NEVER mentioned 0.05ppm.

I have asked, "what is this for, and what do you expect it to do for you?"

STILL no answer...   

Sorry about that, the original plan was to have a little reference that could be shipped back and forth to other forum members to
import "The Volt" back to my lab, however on the very first return trip DHL lost my package (that also included my resistance standards  ).

So I'm looking at building about five (never put all your eggs in the same basket..) new ones for my personal use, and design a pcb
and probably out-source machining of the housing, just to speed up the process.
And as I'm doing that I could also build twenty or fifty in a single batch.
If there is a need for such type of thing and what people would be willing to pay for it also influences the design and parts used a bit.
 
5½ digit of accuracy (=10ppm/=0,001%)
is fine for me.

I hope that clarifies it a bit.

[The Soulman]

I understand. I think what you could do is find someone to help or work with you to do this. You could design/build and someone else could calibrate and ship. You do need to at least make sure the lt1021s you'd be using are stable.

Aging as a assembled product would be 1000 hours minimum I'd also gather data in that time and weed out any drifters
or ones with unreasonably high temp-co.
Also at this moment I'm not committed to use the lt1021, if people have better idea's please let me know.
It also appears that lt1021 H(ermetic) versions are EOL, not sure it's because there is a new one coming or because they (LT/AD) are phasing out all all 10V v-refs in favor of low noise 2,048V ones. 

[The Soulman]

While it's true that I don't really know how close my ICs and my 3456A are, they come from different sources and agree amazingly well.  I also got hold of a 3455A that stopped working but until that happened it agreed with the 3456A to the last digit.

So I think that, statistically, I am pretty much spot on with this gear.  And if I am off a few microvolts I would neither know nor care.  Certainly more accurate than any other instrument I or my friends have.

Exactly, I was pulling your leg a bit.

In the end it all comes down to the confidence that you have in your measurements, if your meter agrees with a number of external
independent references the confidence is good.
Now if one of those references gets measured by a another meter (from a fellow volt-nut) and that agrees as well
confidence is even higher.
Do that more frequently and add a couple of high-spec and or traceable meters in the mix, confidence is optimal. 

[exe]

I'd be interested in a product like DMMCheck, but in Europe . So, not just 10V, but also resistance, capacitance, and ac voltage. Probably I want too much .

As of calibration voltage, it's tough question. My understanding is, 10V is used to calibrate "old" gear. In my personal projects I use 3.3V reference as this is the voltage needed for the stm32 mcu.

[guenthert]

I sense some confusion.  If you
* just want to "import" the volt to your lab, then why not participate in one of the existing "calibration clubs"?
* build/sell a device for such "calibration clubs" (tiny market), such a device would be a "transfer standard" and as such needs to be
  -- low noise to allow determining the DC value with great confidence
  -- ruggedized (as shippers and members *will* drop them, *cough*, *cough*)
  -- protected input (overvoltage, reversed polarity) and output (short circuit)
  -- small and light to reduce shipping costs
  -- withstand temperature extremes while showing only minimal hysteresis
  -- should use a standard voltage (so that it can be compared easily), that'll be 10V these days and
  -- standard connectors 4mm "banana" 5 way binding posts pretty please with possibility to screw in bare copper wires
  -- not too expensive, as it might get lost during shipping
 it does *not* need to be
  -- long term stable, as it is "re-calibrated" (or current value recorded) every few days (few weeks at the most), when returning to home base

* If, on the other hand, you want build/sell a home reference (much larger market, but apparently not large enough for e.g. Mr. Geller's reference to be economically viable), you want it to be first and foremost long term stable (which is time consuming and hence expensive to determine and demonstrate)

[Magnificent Bastard]

I sense some confusion.  If you
* just want to "import" the volt to your lab, then why not participate in one of the existing "calibration clubs"?
* build/sell a device for such "calibration clubs" (tiny market), such a device would be a "transfer standard" and as such needs to be
  -- low noise to allow determining the DC value with great confidence
  -- ruggedized (as shippers and members *will* drop them, *cough*, *cough*)
  -- protected input (overvoltage, reversed polarity) and output (short circuit)
  -- small and light to reduce shipping costs
  -- withstand temperature extremes while showing only minimal hysteresis
  -- should use a standard voltage (so that it can be compared easily), that'll be 10V these days and
  -- standard connectors 4mm "banana" 5 way binding posts pretty please with possibility to screw in bare copper wires
  -- not too expensive, as it might get lost during shipping
 it does *not* need to be
  -- long term stable, as it is "re-calibrated" (or current value recorded) every few days (few weeks at the most), when returning to home base

* If, on the other hand, you want build/sell a home reference (much larger market, but apparently not large enough for e.g. Mr. Geller's reference to be economically viable), you want it to be first and foremost long term stable (which is time consuming and hence expensive to determine and demonstrate)

It should not require batteries during shipping, which entails a whole bunch of problems.  In today's world, things that show batteries inside on an X-Ray, will get flagged for further investigation-- and may get lost or destroyed as a consequence of that.  It should not be affected by the input voltage much (with an input range of say, 12V to 16V)-- and it should not be affected by humidity or barometric pressure.

What this means is that every reference that is NOT in a hermetically sealed package is NOT acceptable.  Unfortunately, almost all hermetically sealed references are being phased out by the idiotic bean counters at the various semiconductor companies, with the possible exception of the LM399 and the LTZ1000.

The (heated) LM399 has lower mechanical hysteresis than even the LTZ1000A, let alone the LTZ1000.  The LTZ1000(A) references are better off if they are continuously powered.

So, this absolutely SCREAMS to use the LM399.  You *could* use 4 of them in parallel to get the low frequency noise down, without getting too ridiculous on the final price of the reference.  I once talked to my Linear Tech rep about the LM399.  I asked what the big DMM manufacturers were doing to sort these into acceptable parts.  He would not say what they were doing, but he did tell me that only about 1 out of 10 references ever made it into the 6-1/2+ digit meters.  They are selecting for low 1/f noise (1Hz and below), low popcorn noise, low temperature drift, and low temporal drift.  Some of these have to be monitored for over a year to find the "shining jewels".  Probably, they will need some kind of artificial aging schedule before the drift testing is done.  This could easily turn a $7 part into a $70 part (plus labor and fixturing).

Now, the LM399 has very poor initial accuracy, but very good stability.  There are some very old DMMs that have this reference in them, that have a surprisingly low annual drift rate (less than 1ppm), but this is only after 10 years or so of power-on time.  Using the LM399 would mean that you need to have a highly adjustable 7V to 10V boost circuit that does not drift over time, and this (in turn) indicates a PWM scheme.  This also provides the opportunity to trim the reference for temperature and barometric pressure related drift.  There is already work being done on this on the LM399 thread here:

https://www.eevblog.com/forum/metrology/lm399-based-10-v-reference/

This could be initially adjusted and calibrated at a Cal-Lab (ENI Labs maybe?)-- and then calibrated each year just after the Cal-Lab gets their 732B calibrated at Fluke for minimum uncertainty.  (Here, "calibrated" means "accurately measure" and "adjust" means "change the absolute output value closer to 10V".)

This reference could then be circulated around through the Cal-Club, so users could calibrate and/or adjust their in-house 10V standards, DMMs, and calibrators.

[The Soulman]

I'd be interested in a product like DMMCheck, but in Europe . So, not just 10V, but also resistance, capacitance, and ac voltage. Probably I want too much .

As of calibration voltage, it's tough question. My understanding is, 10V is used to calibrate "old" gear. In my personal projects I use 3.3V reference as this is the voltage needed for the stm32 mcu.

Resistance could be added in the same device or perhaps a separate apparatus, capacitance and AC are a different animal
but maybe, if this seems viable, in the distant future.

What voltage to use is a easy as it is difficult,

10V seems a tradition perhaps because a particular expensive dmm over-ranges at 12V..
7V is the native voltage of a buried zener, and the best to use for long-term stability, but apart from that a completely oddball value
that is unpractical to use with a meter that isn't limited to 8.000 or 10.000 counts (or the same with more zero's)  because of
the lost amount of resolution.
3,5V might be the ticked and useful for the home hobbyist to check his/hers 4000 or 6000 count (or 40.000 etc.) dmm
The 3,5V could be derived from the 7V zener voltage with a switched capacitor divider that has less of a an impact on stability than
a "simple" resistor divider, but the added components also add cost and might push it out of the price range for that type of user, i.e.
 a "normal" (not a volt-nut) hobbyist..

 

[The Soulman]

I sense some confusion.  If you
* just want to "import" the volt to your lab, then why not participate in one of the existing "calibration clubs"?
* build/sell a device for such "calibration clubs" (tiny market), such a device would be a "transfer standard" and as such needs to be
  -- low noise to allow determining the DC value with great confidence
  -- ruggedized (as shippers and members *will* drop them, *cough*, *cough*)
  -- protected input (overvoltage, reversed polarity) and output (short circuit)
  -- small and light to reduce shipping costs
  -- withstand temperature extremes while showing only minimal hysteresis
  -- should use a standard voltage (so that it can be compared easily), that'll be 10V these days and
  -- standard connectors 4mm "banana" 5 way binding posts pretty please with possibility to screw in bare copper wires
  -- not too expensive, as it might get lost during shipping
 it does *not* need to be
  -- long term stable, as it is "re-calibrated" (or current value recorded) every few days (few weeks at the most), when returning to home base

Yes all that, for voltage see my previous post, long term stability is use-full for any type of standard including a transfer standard,
as it helps to build confidence.

* If, on the other hand, you want build/sell a home reference (much larger market, but apparently not large enough for e.g. Mr. Geller's reference to be economically viable), you want it to be first and foremost long term stable (which is time consuming and hence expensive to determine and demonstrate)

Well I wan't it to be both, it could be the twentieth standard for a incurable volt-nut and a "gold-reference" for a
electronics hobbyist to check his 40.000 count 0,1% hand-held dmm once a week.

[The Soulman]

It should not require batteries during shipping, which entails a whole bunch of problems.  In today's world, things that show batteries inside on an X-Ray, will get flagged for further investigation-- and may get lost or destroyed as a consequence of that.  It should not be affected by the input voltage much (with an input range of say, 12V to 16V)-- and it should not be affected by humidity or barometric pressure.

What this means is that every reference that is NOT in a hermetically sealed package is NOT acceptable.  Unfortunately, almost all hermetically sealed references are being phased out by the idiotic bean counters at the various semiconductor companies, with the possible exception of the LM399 and the LTZ1000.

The (heated) LM399 has lower mechanical hysteresis than even the LTZ1000A, let alone the LTZ1000.  The LTZ1000(A) references are better off if they are continuously powered.

So, this absolutely SCREAMS to use the LM399.  You *could* use 4 of them in parallel to get the low frequency noise down, without getting too ridiculous on the final price of the reference.  I once talked to my Linear Tech rep about the LM399.  I asked what the big DMM manufacturers were doing to sort these into acceptable parts.  He would not say what they were doing, but he did tell me that only about 1 out of 10 references ever made it into the 6-1/2+ digit meters.  They are selecting for low 1/f noise (1Hz and below), low popcorn noise, low temperature drift, and low temporal drift.  Some of these have to be monitored for over a year to find the "shining jewels".  Probably, they will need some kind of artificial aging schedule before the drift testing is done.  This could easily turn a $7 part into a $70 part (plus labor and fixturing).

Now, the LM399 has very poor initial accuracy, but very good stability.  There are some very old DMMs that have this reference in them, that have a surprisingly low annual drift rate (less than 1ppm), but this is only after 10 years or so of power-on time.  Using the LM399 would mean that you need to have a highly adjustable 7V to 10V boost circuit that does not drift over time, and this (in turn) indicates a PWM scheme.  This also provides the opportunity to trim the reference for temperature and barometric pressure related drift.  There is already work being done on this on the LM399 thread here:

https://www.eevblog.com/forum/metrology/lm399-based-10-v-reference/

This could be initially adjusted and calibrated at a Cal-Lab (ENI Labs maybe?)-- and then calibrated each year just after the Cal-Lab gets their 732B calibrated at Fluke for minimum uncertainty.  (Here, "calibrated" means "accurately measure" and "adjust" means "change the absolute output value closer to 10V".)

This reference could then be circulated around through the Cal-Club, so users could calibrate and/or adjust their in-house 10V standards, DMMs, and calibrators.

Totally agree on the hermetic enclosure, as mentioned LT1021-H are eol, but at this moments remaining stocks are plenty full..
LTZ1000 are overkill for what I'm attempting here, LM399 could be it but a boost stage to 10V hurts stability and adds a lot of cost
for good parts (VHP?).
PWM is another option also adds cost thru complexity.
As mentioned above, a switched capacitor divider could be used for a non-traditional but useful 3,5V output.

[Magnificent Bastard]

{snip!}
PWM is another option also adds cost thru complexity.
{snip!}

Uhmmm.... no.  PWM is arguably cheaper and more stable than ANY resistor-boost scheme of any kind.  The "complexity" is in the firmware, which is not that bad, can be done by an expert here on this forum, and should only take 2 weeks or so.

[alex-sh]

I think that's a tough market your trying to break into there buddy considering what is already on offer for a very reasonable $65!
I hope it works out for you but I think it will be a very hard sell.

(I assume you are referring to Doug Malone's voltage reference)

It's not the type of device I would buy only because of the housing and connectors.
And him being in the US shipping costs to the EU might be prohibitive, also he is out of stock at the moment just like
the last couple of times I've check.

I have ordered from Doug and shipping to Europe is very reasonable.
Most important is that I can return the reference for re-calibration.

[alex-sh]

I will start the bids. I would pay a tenner for a 10 volt ref, and sick squid for a 6.8 volt ref. This is The Queens GBP here, not that EURO/USD rubbish. 

That sick squid sounds tempting but I'm looking for something around/below the 50 euro area, all depends
on the specs/features people come up with.

If it is just me I'd go for the LT1021DMH-10 again, to minimize long-term drift leave the trim pin not connected, so no trimming and/or temperature compensation.
Instead something like a lm35 temperature sensor on the same board close to the v-ref to keep track of temperature and made available on a separate output.
Power input would be a common barrel jack, that would accept 15 to 30 volts dc, current consumption below 10mA.

Calibration is the main reason for me to have started this whole journey, I currently don't own anything traceable or even very accurate by specs.

At Euro 50, I'd be looking at LM399 board with an excellent long-term stability LT1021 cannot match.
A few of these boards were up for sale on this forum at Euro 50.

[alex-sh]

LM399 could be it but a boost stage to 10V hurts stability and adds a lot of cost
for good parts (VHP?).

Really? And what costs does it add? A couple of 1% resistors for a correct op-amp gain? A buffer op-amp is needed anyway.

[The Soulman]

The lm399 has more noise (by typical specification)

LM399 could be it but a boost stage to 10V hurts stability and adds a lot of cost
for good parts (VHP?).

Really? And what costs does it add? A couple of 1% resistors for a correct op-amp gain? A buffer op-amp is needed anyway.

Op-amp is indeed required anyway for lm399, and the absolute values of the gain resistors aren't important but the stability of the ratio is,
generic resistors aren't good enough for that.
If you think it is:

https://www.ebay.nl/itm/LM399H-10V-Precision-voltage-Reference-Calibrated-LM399/254200638725?hash=item3b2f89ed05:g:RtsAAOSwPA1cOaqw