Calibratory D-105 DC Precision Voltage Reference Standard

[IanB]

In the early history of science, most research was done by wealthy people of independent means. It's not all about making a profit. The search for knowledge is a worthy goal in itself.

[TunerSandwich]

In the early history of science, most research was done by wealthy people of independent means. It's not all about making a profit. The search for knowledge is a worthy goal in itself.

of course....but you can't assign an economic value in that case...like "poor mans standard".....it implies that it's somehow a replacement for a more expensive tool......I don't see how this point can be so corrupted. 

we can't just replace a fluke 732B with NIST cal cert with a $7 IC....and some unpaid time.....at that point why bother getting it certified?  That would be like taking a device in for EMC that you did zero pre-compliance work on....it would just be foolish and a waste of money.....

I mean is anyone here seriously suggesting that ANY of these devices are a replacement for a NIST traceable cert, on a device that can actually prove it's stability over time?  Using a set of resources none of us here can afford?  I mean for gods sake Fluke has an in house JJA to compare it's devices against.....how can we ever do something equivalent for the "sake of knowledge".  If you are wealthy enough to invest millions of dollars into duplicating the Fluke 732B, so that you can have ONE device in you r personal lab....then more power to you.....but that is not the real world

Also these are monolithic IC's.....how are we to do anything to better them?  We aren't going to stumble on some radical discovery, using someone elses "chip",  I suggested earlier on that a discreet ref might actually lend more insight into discover and personal education, because you can change individual component values and seek net results.  That is what research is all about...and that is fantastic, but I don't see how soldering an IC onto a PCB and following some app note = anything other than the predicted results, by the manufacturer of the monolithic IC 

There isn't going to be some magical AH HAAAA moment where we best the predicted uncertainty by many orders of magnitude.....and a Josephson array already bests a 732B by many orders of magnitude....so what groundbreaking thing is it we are chasing? 

[TunerSandwich]

"Has been so for a week or 2 .. rock solid."


Explain to me how this (see above) is a replacement for this (see below)

http://us.flukecal.com/products/electrical-calibration/electrical-standards/732b-direct-voltage-standard-0?quicktabs_product_details=4

All i see is a statement and a static picture.....even if someone here was the most honest guy in the world....do you think a picture and some totally UN-vetted statement is a trustworthy benchmark, in comparison to the millions of research dollars spent by Fluke to generate the data in the link I posted?  c'mon.....

Everyone is entitled to their own opinions and can spend their time and money as they see fit.....but I don't see any flood of industry professionals, who rely on data to CYA rushing to buy eBay DiY ref over the Fluke hardware.....it's nowhere near the same league.....

[timb]

If I make a device that can hold a stable voltage to 1ppm or less for 7 days an output 1mv to 100v, that had great value as a transfer standard for hobbyists. That could be nulled to a 732B and sent around anywhere in the world. You could use it to calibrate your hobbyist lab meters. That has business potential.

As for gathering data on monolithic ICs, that can actually prove to be useful as most IC manufacturers only gather initial data on 100 chips over 1000 hours. That's what goes into the datasheet. If we've got people gathering data on batches of chips for 10,000 hours, that's producing real data that could be very useful to others. That's science.

At any rate, no one is trying to replicate the 732B. I think the goal is to make a standard good enough to validate and maybe even calibrate up to 6.5 digit meters.


Sent from my Smartphone

[TunerSandwich]

If I make a device that can hold a stable voltage to 1ppm or less for 7 days an output 1mv to 100v, that had great value as a transfer standard for hobbyists. That could be nulled to a 732B and sent around anywhere in the world. You could use it to calibrate your hobbyist lab meters. That has business potential.

As for gathering data on monolithic ICs, that can actually prove to be useful as most IC manufacturers only gather initial data on 100 chips over 1000 hours. That's what goes into the datasheet. If we've got people gathering data on batches of chips for 10,000 hours, that's producing real data that could be very useful to others. That's science.

At any rate, no one is trying to replicate the 732B. I think the goal is to make a standard good enough to validate and maybe even calibrate up to 6.5 digit meters.


Sent from my Smartphone

The gentleman above was comparing his DiY standard to a 732B....

also where is the proof of holding such calibration?  graphs, control to measure against....test methodology, ambient conditions....tes equipment certificates of cal etc etc etc

I am not saying I doubt you or what any of us here can or can't do....I can make a standard that holds as well.....but how would anyone accept that data, when it doesn't conform to industry specifications?

Also that statement about only characterizing a couple IC's isn't always the case....Ti actually has a massive, and in detail report of long term testing on the REF102....showing exactly how and when chips were chosen, ambient conditions...PSU injected error...uncertainty against a standard cell (null) etc....

I just don't understand what "breakthrough" anyone is going to have.....basically the ideal scenario (assuming you don't have the build quality of Awesome14's "ref") is that it follows the typical...min or max results of the datasheet....

not at all the case if you build your own discreet ref.....a nice current stabilized zener and stabilized precision amplifier to follow.....

I mean if the devices aren't to sell and make profit from, and all you learn is that it conforms to the datasheet (monolithic IC)....I don't get it? 

Now implementing various forms of heater control...that is a different story.  However that is far beyond the scope of the manufacturer, and you will still have pretty large obstacles to overcome to calibrate each unit....as the drift in each IC will be different.....so you would have to hand cal each device, burn in for 1k hrs min, then datalog the drift long term, before you could accept it as a usable standard.....

[TunerSandwich]

I mean I can make my meter show a number also....what does it prove?  Where is the datalog...what were the conditions etc etc etc.....

If no datalog at appropriate intervals....did someone actually watch the meter for 7 days or two whole weeks?  can they prove it? 

The majority of commercial standards can prove their claims.....if they can't they would have never received a NIST cal cert.....that's the point of the traceable reference....anything else is just pissing in the wind and hoping none blows back on you.....how could one derive a known level of uncertainty across their entire test reference system? 

here is a picture.....for the hell of it lets say it's been there "for a week or two" (a week or so?  which one is it...a week, or two....?) ROCK SOLID!!!

[paulie]

A poor man can't afford 2 3458A's + a few 732B's......

show me how you can amortize the cost of equipment + resources into a sub $100 MSRP, and come out ahead?  You can't....

If you are doing this for purely academic purposes....then you are once again not a poor man, because you have no financial gain...and only spend resources.....so again not an excercise for a poor man.

As often the case a collection of random non-sequitur, out of context misquotes, and generally lame nonsense. To assume academic efforts are not possible on a budget is absurd. Equally absurd to assume everybody must have multiple 8 1/2 digit lab meters and at least a hundred grand invested in a wall full of similar gear. Estimated at least 9 out of 10 of these forum users are not involved with mars rover or developing heart and lung machine.

http://dictionary.reference.com/browse/hobby

And to assume a poor man cannot pursue academic research for eventual profit just as senseless.

those DMM's and baseline ref aren't "cheap"

Actually they can be.

http://www.ebay.com/itm/AD584KH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-Simpl-/181219600361?pt=LH_DefaultDomain_0&hash=item2a3187a7e9

$8 but very stable and comes with calibration data from 6 1/2 digit lab meter so an excellent transfer standard for most setups.

http://www.ebay.com/itm/161247229877?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

5 bucks on Ali. 0.1mv resolution. More than adequate for 99% of hobby AND professional voltage measurement.

I mean I can make my meter show a number also....what does it prove?  Where is the datalog...what were the conditions etc etc etc.....

If no datalog at appropriate intervals....did someone actually watch the meter for 7 days or two whole weeks?  can they prove it? 

Again you don't have to spend more than a new car to arrange data collection. A one or two dollar MCU perfectly able to not only convert meter readings to serial for PC logging but actually store weeks or months internally depending on sample rate.

[JohnnyBerg]

A poor man can't afford 2 3458A's + a few 732B's......

show me how you can amortize the cost of equipment + resources into a sub $100 MSRP, and come out ahead?  You can't....

If you are doing this for purely academic purposes....then you are once again not a poor man, because you have no financial gain...and only spend resources.....so again not an excercise for a poor man.

As often the case a collection of random non-sequitur, out of context misquotes, and generally lame nonsense.

Hear, hear! 

Englisch is not my native language. If it were, I would have given the same reply.

[timb]

I mean I can make my meter show a number also....what does it prove?  Where is the datalog...what were the conditions etc etc etc.....

If no datalog at appropriate intervals....did someone actually watch the meter for 7 days or two whole weeks?  can they prove it? 

The majority of commercial standards can prove their claims.....if they can't they would have never received a NIST cal cert.....that's the point of the traceable reference....anything else is just pissing in the wind and hoping none blows back on you.....how could one derive a known level of uncertainty across their entire test reference system? 

here is a picture.....for the hell of it lets say it's been there "for a week or two" (a week or so?  which one is it...a week, or two....?) ROCK SOLID!!!

I might post a picture of my meter, but that doesn't mean I'm not collecting long term data.

I've got a nice GPIB setup with an NI GPIB to Serial converter hooked to a BeagleBone Black doing the logging via some Python tools. I've also got a +-1% Honeywell HumidIcon sensor hooked to the BBB collecting RH and temp data that can be plotted with any of the the data.

My HP and Tek meters have valid NIST cal certs from last year. My Keithley 197A and 2000 (not shown) don't. However the 2000 reads to one digit of my HP and the 197A reads spot on to my Tek, so I consider them in calibration.

I have a Keithley 195A that is a bit out, but I was thinking of swapping the reference in it before getting a calibration.

Oh, and my humidity sensor has been calibrated to an NIST standard to within 5%. (They've got these cards in sealed plastic bags you can open and they show the RH by the color of various dots changing. The cards are NIST certified. Each lasts 6 months.)

So yeah, I log my data to traceable standards. I just haven't processed any data yet, but rest assured it's being collected.






Sent from my Smartphone

[alhoop]

Man! This is like reading a good book and just when I thought it couldn't get any better, it does.

As to Awesom14 building his circuit on perf board or whatever , what is wrong with that. As to  Jim Williams(a hero to some on here),
I bet he never laid out any of his circuits for production. If Awesome14 put his circuit on a pretty PCB it might not work.
I also knew my boy Tim couldn't go four posts without incurring the Wrath of God.

Al

[TunerSandwich]

A poor man can't afford 2 3458A's + a few 732B's......

show me how you can amortize the cost of equipment + resources into a sub $100 MSRP, and come out ahead?  You can't....

If you are doing this for purely academic purposes....then you are once again not a poor man, because you have no financial gain...and only spend resources.....so again not an excercise for a poor man.

As often the case a collection of random non-sequitur, out of context misquotes, and generally lame nonsense. To assume academic efforts are not possible on a budget is absurd. Equally absurd to assume everybody must have multiple 8 1/2 digit lab meters and at least a hundred grand invested in a wall full of similar gear. Estimated at least 9 out of 10 of these forum users are not involved with mars rover or developing heart and lung machine.

http://dictionary.reference.com/browse/hobby

And to assume a poor man cannot pursue academic research for eventual profit just as senseless.

those DMM's and baseline ref aren't "cheap"

Actually they can be.

http://www.ebay.com/itm/AD584KH-4-Channel-2-5v-7-5v-5v-10v-High-Precision-Voltage-Reference-Module-Simpl-/181219600361?pt=LH_DefaultDomain_0&hash=item2a3187a7e9

$8 but very stable and comes with calibration data from 6 1/2 digit lab meter so an excellent transfer standard for most setups.

http://www.ebay.com/itm/161247229877?_trksid=p2057872.m2749.l2649&ssPageName=STRK%3AMEBIDX%3AIT

5 bucks on Ali. 0.1mv resolution. More than adequate for 99% of hobby AND professional voltage measurement.

I mean I can make my meter show a number also....what does it prove?  Where is the datalog...what were the conditions etc etc etc.....

If no datalog at appropriate intervals....did someone actually watch the meter for 7 days or two whole weeks?  can they prove it? 

Again you don't have to spend more than a new car to arrange data collection. A one or two dollar MCU perfectly able to not only convert meter readings to serial for PC logging but actually store weeks or months internally depending on sample rate.

Those suggestions are so laughable it's hardly worth the time to respond.....if you can't do the simple math to reach a conclusion that you won't be able to even verify these "cheap standards"....then best of luck....

It's very hard to take anything you say seriously with these suggestions....and then your general superiority complex...that you somehow know better than the industry at large, well.....like I said best of luck....feel free to waste your time on these products.....although you could have learned the same exact lessons (with some accredited level of accuracy) by simply reading a few NIST reports about calibration.....

all you will wind up learning with tools like these, is that you can't trust the tools....but it's your time and your dime   

Not one of these devices comes with ANY form of trustworthy traceability.....it's just Awesome14 all over again.....I suspect something about conspiracy and angels will come out next....good grief....

no 6 1/2 digit "lab meter" is going to verify stability to PPM range.  The least significant digit is not trustworthy and @ 10V your least significant digit is 1ppm....

and YOUR suggestion for measuring vREF is this....?    why even bother speculating about "ppm stability".....there is simply ZERO data....

[TunerSandwich]

Man! This is like reading a good book and just when I thought it couldn't get any better, it does.

As to Awesom14 building his circuit on perf board or whatever , what is wrong with that. As to  Jim Williams(a hero to some on here),
I bet he never laid out any of his circuits for production. If Awesome14 put his circuit on a pretty PCB it might not work.
I also knew my boy Tim couldn't go four posts without incurring the Wrath of God.

Al

His "ref" doesn't work either way....PCB or not....it can't be verified and as soon as you open it it no longer powers up....as demonstrated....

meh best of luck gentlemen....you are wasting massive resources chasing readily available information, thinking that somehow you will make some breakthrough.....it reminds me of the over-unity endeavor....with a bit of reading, you would soon realize that the arguments posed here can't hold water.....you are essentially refuting decades of documented knowledge, saying that with zero documentation you will somehow best the efforts of giants in this industry....

there isn't even academic value here, because you are simply duplicating extensive test results, spanning decades....using far more resources than anyone here has in their "hobby lab"....it's nonsense....then as soon as someone points anything of relevant value out, you all join in some pat each-other on the back circle-jerk to dispute decades of research and documentation....with these notions your efforts will most certainly remain in the "hobby realm".....especially seeing that you are ignoring the simple math that shows you can't verify ppm stability @ 10V on any of these 6 1/2 digit meters....nor can you publish such documentation w/o a concise breakdown of your measurement methodologies....all the piles of gear in the land aren't going to change that fundamentally flawed methodology....and to even defend anything Awesome14 posted here only furthers the nonsense....


yet NOT ONE PIECE of relevant data is presented....

I will leave you folks to it though....you obviously get some delight in admiring each-others nonsense claims...sooner or later though (after you piss away your time and resources) you will have a very rude awakening that the datasheets were right after all....such a pity to see all the time wasted

[timb]

Uh, no one here is claiming what you're claiming they're claiming... You're going off on a massive tirade about, fuck, I really don't know, but it's evident we're all wrong and you're brilliant.

If you want relevant data, go take a look at the LTZ1000 thread, or any other countless numbers of threads here on the forums.

You seem to be missing the point of a hobby grade voltage standard. It's not to replace a proper cal lab or Fluke gear. It's to have a "good enough" reference to sanity check your < 5.5 digit meters. Or perhaps you're making some boards with ADCs or DACs on them and you want to trim out the references, again, this "hobby voltage standard" would be perfect for that.

So yeah, don't be such a fucktard is what I'm trying to say here. Just because you're shortsighted doesn't mean others are.


Sent from my Tablet

[HighVoltage]

yet NOT ONE PIECE of relevant data is presented....

I will leave you folks to it though....you obviously get some delight in admiring each-others nonsense claims...sooner or later though (after you piss away your time and resources) you will have a very rude awakening that the datasheets were right after all....such a pity to see all the time wasted

I can not believe you would say that. You seem to be a smart person and sometimes you are going off in a wrong tangent.

As I have said before, if I need lab grade calibration and verification and transfer, I go to my official calibration lab here in my town and they use their Fluke 732B and an Agilent 3458A and a Fluke 8508A.

But for some purpose I do need 5.000 V as a reference in some projects and some of these good quality "hobby" references are doing their work perfectly.
And if I take 3 different meters in my lab, a 34401A (calibrated) a 34461A (calibrated) and a 34410A (calibrated) and all of them show 5.0000xV with the x floating 1 digit up or 1 digit down, and the difference between the meters is 1 or 2 digits off in the last place what is wrong with that?
It has fulfilled my requirements and expectations and I am getting a job done.

[timb]

Oh, and TunerSandwich, you state 6.5 digit meters can't verify 1ppm at 10V? I don't know what sort of shitty meter you've got, but mine certainly goes down to 10uV at 10V (10.0000X).


Sent from my Tablet

[engineer_in_shorts]

yet NOT ONE PIECE of relevant data is presented....

I will leave you folks to it though....you obviously get some delight in admiring each-others nonsense claims...sooner or later though (after you piss away your time and resources) you will have a very rude awakening that the datasheets were right after all....such a pity to see all the time wasted

I can not believe you would say that. You seem to be a smart person and sometimes you are going off in a wrong tangent.

As I have said before, if I need lab grade calibration and verification and transfer, I go to my official calibration lab here in my town and they use their Fluke 732B and an Agilent 3458A and a Fluke 8508A.

But for some purpose I do need 5.000 V as a reference in some projects and some of these good quality "hobby" references are doing their work perfectly.
And if I take 3 different meters in my lab, a 34401A (calibrated) a 34461A (calibrated) and a 34410A (calibrated) and all of them show 5.0000xV with the x floating 1 digit up or 1 digit down, and the difference between the meters is 1 or 2 digits off in the last place what is wrong with that?
It has fulfilled my requirements and expectations and I am getting a job done.

Well, that's the spec of the 2ppm reference swamped (by floating 1 digit noise) that you are trying to measure.
I think I would struggle to verify the performance of these hobby standards, with either my trusty fluke 8508A or my dusty hp 3458A. The techniques used, other sources of error and uncertainty guard banding makes this a very error prone 'experiment'.  Forget trying to do it on a crappy 6.5 digit whatever. 

[timb]

If you're only going for 5.5 digit precision it's more than adequate. That's the point!


Sent from my Tablet

[JohnnyBerg]

Well, that's the spec of the 2ppm reference swamped (by floating 1 digit noise) that you are trying to measure.
I think I would struggle to verify the performance of these hobby standards, with either my trusty fluke 8508A or my dusty hp 3458A. The techniques used, other sources of error and uncertainty guard banding makes this a very error prone 'experiment'.  Forget trying to do it on a crappy 6.5 digit whatever.

If the PCB and circuit  is designed well, and one uses a well known reference IC, verifying performance boils down to verifying if the reference IC meets its spec.

For example, the REF5025A has a max T.C. of 8ppm, and typically 3ppm. Changing the operation temperature 10°, would result in a 30 ~ 80ppm change in output. No problem for a 6,5 digit meter.

Different story of course if one try's to make things better .. but that's more for the daredevils among us, not my league 

[HighVoltage]

Well, that's the spec of the 2ppm reference swamped (by floating 1 digit noise) that you are trying to measure.
I think I would struggle to verify the performance of these hobby standards, with either my trusty fluke 8508A or my dusty hp 3458A. The techniques used, other sources of error and uncertainty guard banding makes this a very error prone 'experiment'.  Forget trying to do it on a crappy 6.5 digit whatever.

So, you consider a 34401A, a 34461A and a 34410A crappy 6.5 digit whatever meter?
Really?

If I am looking for 5.5 digit precision and my 34461A shows 5.00002 V and over several thousand measurements the last digit will not even move and it is confirmed by a 34401A and a 34410A, how is that not verified?

 

[nctnico]

Oh, and TunerSandwich, you state 6.5 digit meters can't verify 1ppm at 10V? I don't know what sort of shitty meter you've got, but mine certainly goes down to 10uV at 10V (10.0000X).

IMHO what you are forgetting is long term stability. I have been looking at voltages references myself for an upcoming project. I think that the best you can expect is a long term stability of 20ppm per year with a very well designed and thermally controlled production circuit. Power cycles, noise, thermal EMF, mechanical stress all add to long term INstability. A reference may be accurate to 1ppm today but how about next month and next year?

There is a difference in resolution an accuracy!

[JohnnyBerg]

I think that the best you can expect is a long term stability of 20ppm per year with a very well design and thermally controlled production circuit.

Most manufacturers specify a 1000 hour stability, after a burn in of 1000 hour. Doing 20ppm/year will be very challenging IMHO.
 
You have to outsmart the specification from the manufacturer. And then comes the problem of proof into play. As long as you are staying in spec. with the device, you simply refer to the datasheet. As you outsmart the manufacturer, you have to come up with some proof of yourself. This can be a long, costly process. Wen using the reference internal, this does not have to be a problem. But when selling a reference, it is.

If course, there are a lot of people out there, who promise a lot but in the end cannot deliver 

[engineer_in_shorts]

Well, that's the spec of the 2ppm reference swamped (by floating 1 digit noise) that you are trying to measure.
I think I would struggle to verify the performance of these hobby standards, with either my trusty fluke 8508A or my dusty hp 3458A. The techniques used, other sources of error and uncertainty guard banding makes this a very error prone 'experiment'.  Forget trying to do it on a crappy 6.5 digit whatever.

So, you consider a 34401A, a 34461A and a 34410A crappy 6.5 digit whatever meter?
Really?

If I am looking for 5.5 digit precision and my 34461A shows 5.00002 V and over several thousand measurements the last digit will not even move and it is confirmed by a 34401A and a 34410A, how is that not verified?
 

What time period? ... as per TunerSandwich posts look at the accuracy specs:

34401A specs *after 1 hr warmup

5V is in the 10V range and therefore within 24hrs the relative accuracy is 15ppm plus 2ppm.

Have you made an uncertainty budget ?

P.S yes, for those kind of measurements 6.5 digit is crap, my fluke on that range is around 0.5ppm within 24hrs (relative).

[HighVoltage]

5V is in the 10V range and therefore within 24hrs the relative accuracy is 15ppm plus 2ppm.

Have you made an uncertainty budget ?

Do I really need to worry about 15ppm + 2ppm, when I require only 5.0000V on a 6 1/2 digit multimeter?
Here is my setup:

[Dr. Frank]

5V is in the 10V range and therefore within 24hrs the relative accuracy is 15ppm plus 2ppm.

Have you made an uncertainty budget ?

Your calculation is not correct, either!

uncertainty of 5V on 10V range:
15ppm of  reading (5V)  => 75µV
plus 4 ppm of range (10V) => plus 40µV

=> 115µV / 5V = 23ppm

To be exact, you also have to add the A/D non linearity, which is 2ppm of range => 20µV => gives additionally 4ppm, for a  Grand Total of 27ppm.
And then also some ppm for the temperature coefficient, which I'd like to omit now, because it's already included, if you stay within +/- 1°C.

On a 6 1/2 digit DMM, 27 ppm of 5V or 135µV uncertainty on 10V range are visible on the last two digits: 4.99987 .. 5.00014 V.

That's not so good, more a 4 4/5 digit result, than a 5 1/2 digit one.

It's mainly caused by the mediocre stability specification of the LM399H reference plus the assorted amplifier-resistors.

Frank

[engineer_in_shorts]

5V is in the 10V range and therefore within 24hrs the relative accuracy is 15ppm plus 2ppm.

Have you made an uncertainty budget ?

Do I really need to worry about 15ppm + 2ppm, when I require only 5.0000V on a 6 1/2 digit multimeter?
Here is my setup:

As TunerSandwich said, a single picture proves nothing in any terms.  Temper Coef? Drift over time? etc etc
OK, leave that setup and take a picture this time tomorrow (24hrs) - without ANY adjustments.  Note the temperature when photos was taken too.