DMM Volts/Ohm Calibration Club - anyone interested?

[Dr. Frank]

Like that Datron 4910 we might get from that German company you linked in the other thread (I believe it was you).

They are called "Singer Pharmacy", as they are totally overprized, and they often have very old stuff, w/o advising you, about ser. no. / date codes and real condition.
The instruments work all right, though, because they have a bunch of skilled technicians in their cave.
Sometimes, they sell instruments as "scrap" for a reasonable price, and these turn out sometimes to work fine, as my Fluke 5200A for a few hundred €uro.

Over 20 years ago, I've already bought several special RF tubes for our Aachen Physics II Institute directly from Helmut Singer.. he's quite a rustic guy..
 

That module you designed is fantastic!  I love it.  I am definitely trying to get into the design side of things. 

Please check out our Super LTZ thread, https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/, there are many other designs also, which have some advantages over the original schematic from LT, with less engineering effort and cost... Look for Andreas PCB, for example.

Do any of you here see a market for 10.5 digit multimeters?  I've been thinking about trying to design one in my off time.  First though, I was going to try to design a new way to get a more accurate voltage reference.  This is probably a few years in the making, however lol.... As I delve into it, it could very well be a thing that's not feasible.  I will definitely be looking to some of the great information I've found on this website when I get the go ahead.

This has already been discussed in several threads.
An analogue DMM with more than 8 1/2 digits is useless at this time, due to physical linearity and noise limitations.
As an analogue voltage reference, there is currently no other more stable device available than using zener diodes (reference amplifiers).
Drift is on the order of a few parts in 10^7/year, noise on the same order, again limiting useful uncertainty, and the resolution to about 8 digits.
See DMM noise comparison thread for that.

I've seen a concept, where a JJA as an A/D and reference is combined directly with a DMM, so to maybe have order of 10^-9 of noise/stability/uncertainty..
For physical reasons, as soon as you go from cryogenic/quantum world to real 300K analogue world, you will be limited to these 10^-9.

So there's not only no market for 9..10 digits analogue DMM, these simply do not exist.

Frank

[CalMachine]

They are called "Singer Pharmacy", as they are totally overprized, and they often have very old stuff, w/o advising you, about ser. no. / date codes and real condition.
The instruments work all right, though, because they have a bunch of skilled technicians in their cave.
Sometimes, they sell instruments as "scrap" for a reasonable price, and these turn out sometimes to work fine, as my Fluke 5200A for a few hundred €uro.

Over 20 years ago, I've already bought several special RF tubes for our Aachen Physics II Institute directly from Helmut Singer.. he's quite a rustic guy..

Thanks for the information!  I'll see what kind of information I can get about the unit before we make the purchase.
 

Please check out our Super LTZ thread, https://www.eevblog.com/forum/metrology/ultra-precision-reference-ltz1000/, there are many other designs also, which have some advantages over the original schematic from LT, with less engineering effort and cost... Look for Andreas PCB, for example.

I will definitely do that.  There has been some amazing work done by you all so far.  Very well done.

This has already been discussed in several threads.
An analogue DMM with more than 8 1/2 digits is useless at this time, due to physical linearity and noise limitations.
As an analogue voltage reference, there is currently no other more stable device available than using zener diodes (reference amplifiers).
Drift is on the order of a few parts in 10^7/year, noise on the same order, again limiting useful uncertainty, and the resolution to about 8 digits.
See DMM noise comparison thread for that.

I've seen a concept, where a JJA as an A/D and reference is combined directly with a DMM, so to maybe have order of 10^-9 of noise/stability/uncertainty..
For physical reasons, as soon as you go from cryogenic/quantum world to real 300K analogue world, you will be limited to these 10^-9.

So there's not only no market for 9..10 digits analogue DMM, these simply do not exist.

Frank

Hmmm well that's some food for thought.  You have some great insight!  I'm definitely going to take some time to digest this and see whats going on.  One of the things I wanted to incorporate was a specific resistance range to measure the Von Klitzing constant.   Maybe my eyes are too big for my mouth lol

And yeah, I realize that 8.5 DMMs are the longest scale out there right now... that's why I was maybe wanting to develop a 9.5 or even 10.5 DMM, but it sounds like it's going to take some serious design improvements along with maybe some quantum references.

[dr.diesel]

Even the big guns would struggle with a new true low noise/PPM 8.5 digit meter, 9.5 digit might be doable in theory today, but 10.5 IMO would take the collaboration of all involved and likely cost > $1B and 10+ years of serious development.  10.5 would take guts not even invented yet.

In any case, I'm struggling to think of even an uber niche need for such a 10.5 digit meter?  Aside form a few guys here sales would likely be very low, 

[enut11]

@calmachine
Hi Calmachine. Your offered services will be of much value to many (mainly local USA) voltnuts as well as those at the top of their game regardless of location.
However, we need to keep this thread on topic and find a practical way to help voltnuts, located in different parts of the world, check their out-of-cal 6.5 to 8.5 DMMs. A lot of these enthusiasts have home-made voltage references of different grades. So, for many, a stable reading using a calibrated DMM (eg HP3458A) would suffice.

From my pespective, this can be successful done provided each country has an affordable service such as yours. In reality, I think we have to find a way to help ourselves because the practicalities of international roving standards has yet to be resolved.

[TiN]

enut11

My personal 5c that given cost numbers from CalMachine for measurement/check it would be hard to beat, even including shipping to and from USA. Would you spend 300$ to design a box with reference and resistor, when you can just send your meter altogether and get it calibrated for similar sum? Unless you have more than few meters, this is viable way too. So I'd say this discussion is very well on-topic. There are still no specs/requirements defined and no-one seem to have large enough loads of time required to actually build first roving box to make this Club talk practical.

Surely CalMachine can start separate thread, called "Hobby-level calibrations service" or something like that if he desire to keep all eggs in one barn. 

[enut11]

@TiN
Yes it would be tempting to take up Calachine's offer from Australia but for the risks involved with international shipping. My HP3456A, purchased working from the USA, arrived with several errors. I found 2 dislodged PCBs, most likely due to rough handling. Hence my suspicion of this working for me - would I be confident of receiving a calibrated meter for my $300?

[e61_phil]

And yeah, I realize that 8.5 DMMs are the longest scale out there right now... that's why I was maybe wanting to develop a 9.5 or even 10.5 DMM, but it sounds like it's going to take some serious design improvements along with maybe some quantum references.

If you want more resultion only, you can connect a 3458A via GPIB and you will get more than 8.5 digits... I'm still interested in how do you want to achieve (with DIY methods) an ADC which is at least as linear as the 3458As one?

[TiN]

enut11

You already know the answer - only one way to find out, pack the box well, talk with CalMachine and sent it. We all can learn from outcome, before sending ours 3458A to him .
Of course your mileage will vary, but ALL my gear comes internationally, mostly from US eBay. Only once I had problem, with 4263B, broken front panel plastic due to horrid packaging with just tightly fit single layer cardboard box.

That's also reason why I usually never question shipping cost, which could be sometimes 30-100% of unit cost. Bright example was Agilent 4142B. Unit itself - 220$. Shipping of it : 250$ due to 29.9kg weight and size (4U frame 750mm deep).

And again, if you pay 100$ for shipping each way, and 100$ for cal of your 3456A, you will get much better bang per buck than spending same 300$ for building 10v, 10k reference box (which you still need pay for ship to someone for calibration and pay for it to come back). And that leaves you without current ranges except 1mA, without high/low voltage and without ACV/ACI and makes you need MFC/sources to transfer your reference values into same 3456A. To me choice is obvious

[Dr. Frank]

I found this AC Quantum Voltmeter, if somebody needs a 10 digit DVM, here it is:

http://www.supracon.com/en/ac_quantum_voltmeter.html

That's a cooperation of PTB and the companies esz and Supracon, latter offers JJA systems, also.

That system is able to measure DCV and ACV to about +/-10V with 0.1nV/V (10digits) uncertainty for DCV.. JJA serves as reference and D/A, obviously.

Frank



Frank

[mimmus78]

Ahhh!!! I just brought a 3458a ... it will never ends!

Inviato dal mio Nexus 6P utilizzando Tapatalk

[dr.diesel]

it will never ends!

Dang technology!   

I took the post about 9.5/10.5 digit meters as in something in a reasonable form-factor, not considering stuff with multiple kWs of consumption that arrives on a semi and requires a forklift to unload!

You pretty much have the best money can buy at the moment.

[mimmus78]

Form factor, consumption ... does it really matter!?

Inviato dal mio Nexus 6P utilizzando Tapatalk

[CalMachine]

And yeah, I realize that 8.5 DMMs are the longest scale out there right now... that's why I was maybe wanting to develop a 9.5 or even 10.5 DMM, but it sounds like it's going to take some serious design improvements along with maybe some quantum references.

If you want more resultion only, you can connect a 3458A via GPIB and you will get more than 8.5 digits... I'm still interested in how do you want to achieve (with DIY methods) an ADC which is at least as linear as the 3458As one?

I am not sure... this is in the infancy stage of just an idea that I have.

I found this AC Quantum Voltmeter, if somebody needs a 10 digit DVM, here it is:

http://www.supracon.com/en/ac_quantum_voltmeter.html

That's a cooperation of PTB and the companies esz and Supracon, latter offers JJA systems, also.

That system is able to measure DCV and ACV to about +/-10V with 0.1nV/V (10digits) uncertainty for DCV.. JJA serves as reference and D/A, obviously.

Frank

This is amazing!  This is the type of stuff I'm shooting for.  Maybe I need to go back to school for a physics degree!

it will never ends!

Dang technology!   

I took the post about 9.5/10.5 digit meters as in something in a reasonable form-factor, not considering stuff with multiple kWs of consumption that arrives on a semi and requires a forklift to unload!

You pretty much have the best money can buy at the moment.

Yeah the form factor I would try to design the unit around would be, like, the size of a Fluke 5700 or Datron 4808/4708

[Theboel]

Hi,

One think I like to do if I have access to lab standard is made a history of my own standard in control environment for a long time with let say 90 days interval let say in 2 years long with this work I can made a prediction about drift and accuracy in longer term.
   

[technix]

I wonder how much would this cost. I have a LM399 and an ADR03B chips to calibrate. Maybe I will calibrate my UT61D multimeter too.

[Theboel]

I wonder how much would this cost. I have a LM399 and an ADR03B chips to calibrate. Maybe I will calibrate my UT61D multimeter too.

I don't know how Your country tax policy, but from experience annually send my VNA calibration kit to overseas lab the shipment and tax are killing factor compare to my NMI cal

[VK5RC]

There is 'a bit' of an 'Aussie Cal Club' going - thanks to enut11 et al.

Thanks to TiN and minnimus78 and others for the run of TiN's KX boards (see xdevs.com/article/kx-ref for excellent background on the development)
After a bit of a struggle getting some good Vishay resistors into Australia - [as opposed to Ultohm Plus/Mr Pettis, who was so helpful and a delight to deal with, thanks again]  I have built three LTZ reference boards.
One has an LTZ1000CH with Ultohm Plus resistors (Mr Pettis)
One has an LTZ1000ACH with Ultohm resistors
One has an LTX1000ACH with mostly Vishay Precisions ( and one Mr Pettis Res - VK5RC stuffed up his order with Vishay!!!!!)
Currently they are 'running in', I aim to put ~400hours or so on them.
I boxed them in a cast metal box with some foam insulation which is then placed inside a plastic box for a bit more thermal insulation. They are powered by a linear PSU using a TPS7A49 low noise regulator putting out 15V, this is mounted in its own metal case nearby but 'out in the open' - to minimise heat transfer.
The PSU receives 18V from another linear PSU LM7xx series.
The resistors and LTZ are mounted just a little off the PCB so I could get a decent set of haemostat forceps in there during soldering to minimise heat stress. So far early drift (by simple measurement) has only been around 1 ppm.
Still undecided re logging strategy, my Linux/Rasp Pi/Python abilities are terrible, IanJ's excel macro/82357A or RF scientific/Prologix are possibilities.
With thanks to EEVblog Metrology forum contributors for helping me get this far
Robert

[TiN]

What is your next step in plan? If you wish to perform a transfer using one of modules with cross-shipping (I'd assume Australia-Taiwan be fairly cheap), you can include RPI3 with it and I can set it all up for you, so it will be as easy to start logging as entering one command ./make_me_logging.sh 

[VK5RC]

TiN,  that is a great  offer,  I would be delighted to accept,  I would want to be sure the unit I send is reasonably stable,   along with a RPi3,  this might take me a week or two.
I am reasonably happy my 3458 is moderately stable: It mainly lives switched off but if I warm everything up properly,  and read either,  1 an old separate 3458 voltage reference or,  2 my fluke 731b,  over the last 2-3 months,  I am getting readings consistent to around 2 to 5 uV.  Over many hours the results are better. 
Regards Rob

[TiN]

Sure, I'd expect to get reference to settle first 200+ hours. Then similar would need to be happen here, log for week or two to make sure it's all good and stable.
As bonus I'd be able to run tempco curve for you

[enut11]

Hi VK5RC. Great implementation of LTZ1000 refs. Now that I have successfully built a couple of LM399 refs I am looking forward to getting stuck into an LTZ project. Just need to source a PCB. BTW, any idea of the total cost of one completed ref?
enut11

[VK5RC]

Hi enut11,
The Vishays were crazy in price AUD $80+ EACH ! (Vishay sent me to an Asian distributor) The Ultrohm Plus resistors despite being prob very close in quality (I am not qualified to say) but were a fraction of the price, I recall about USD$5-10 each and Ultrohm was so much easier to deal with. Mr Pettis is a contributor to this forum - in the future it will definitely be Ultrohm for me.

The LTZ1000ACH I could only get from Digikey and were ~ AUD$80 ! I think the LTZ1000CH were a bit over half price.
The other components are pretty cheap. TiN on his KX reference article has a list of components and the price he could get them for - getting the Vishays in Australia was quite a bit more than TiN could in Taiwan.

Getting an old (but hopefully not buggered) 3458 reference board starts to look not bad value (often around USD240). The one I got is OK (I think!).

I am tempted later to try and make a 'rough and ready/good value' reference using an LTZ1000CH, a few well placed good resistors (some are more critical - I recall the 120R) on veroboard ! Just to see what difference it makes! HiHi.

I think I saw your 399s on another post - nice units!

Robert

[Kleinstein]

The vertical mounting of the PWW resistors is not such a good idea: this way the temperature at both ends of the resistor can be a little different and thus thermal EMF could be an issue. Ideally the resistors should be mounted horizontally with wires of equal length. So the PWW resistors should get a slightly different layout. The 70 K resistors are already less critical than the others - so I don't see a reason for the metal canned ones here.

The way the board is mounted looks strange - looks like possible mechanical bending the the board. This is usually something one wants to avoid. So ideally more like one fixed point and maybe 3 more with a little room for free movement.
Having a full metal enclosure is a good idea.

[Alex Nikitin]

My "1010B" reference box is available now in the UK for any EEV forum member who needs a quick check with a better than 50ppm uncertainty - please PM me if interested. The reference provides 10V with a low output impedance, 1V with 1K output impedance (so the value will be off by -0.01% on 10M load), 10K and 1K resistance (hermetic bulk foil). The stability for 3 weeks running (~500 hours) is very good, the temperature coefficient is about +1ppm/C for the voltage and less than 3ppm/C for the resistance. I will send this unit out with two conditions - the time to return should be less that ten days and the unit should be sent back by Special Delivery. I would prefer to get the reference back to my place every time as it would allow me to check it before sending it out again. 10v output is adjusted for 10.00000V at 23C, the actual measured values for 1V output, 10K and 1K resistance are marked on the reference case. The power supply could be from 16 to 26V DC, a couple of 9V batteries in series will do nicely, as the current consumption is low (~4mA) and a warm-up time from cold to <5ppm of a final value is about 10min . I can send a simple 24V switcher with the reference if required (I use a 24V stabilised linear supply with it, but it is quite heavy).
 
Cheers

Alex

[VK5RC]

Doing a few experiments to see if I really understand my vol-nut set-up ( the answer is I don't fully - yet ) I think I still have a few - I think - possible ground loop - gremlins.
 
Today was quite hot day (outside Min 18C max 33C)- 3458 warmed up overnight - mine is usually about 16C above ambient - below are some graphs of the measured voltage of an (TiNs KX design PCB - Thanks again to TiN for design and minimus78 for the PCB run) with Ultrohm (Mr Pettis' ) resistors and LTZ100ACH reference.
I am currently using a Prologix GPIB USB interface and 'RFScientific' program on a PC. Only one ACAL DCV at the beginning - clearly against the protocol - where it should be done every degree C change.
The bottom graph is the measured voltage which altered over the day (and room temperature) The airconditioning (reverse cycle hole in the wall type) was switched on between 1300 and 1500hrs. I was pleasantly surprised it didn't horrendously effect the voltage (via loops etc). It is on the same run/phase from the distribution board. The room warmed up from about 21 to 27C by 1300 then cooled to about 21 by the aircon. The 3458s internal temperature reflected this and I suspect probably also so did the references temp but the voltage of the MAX166 only indicated an internal temp change of the reference by about 2 or 3 degrees ( I forgot to write the result down ).
The top is a scattergram of the 3458 self measured Internal temp vs measured voltage of LTZ reference. To me it probably shows both the 3458 and the external LTZ1000s temp coefficient - poss other effects e.g. lead to terminal thermal gradients. etc. Hysteresis I think is also evident. Overall reasonably happy with 8uV / 7V  over 5C temp change so about 0.2ppm/C which I think is in spec - mine is a very old 3458. SN 28*******
Robert