High Precision Resistance Measurement - What would be best to use?

[Chipguy]

So yes, I would not try to claim anything unless there is proven meter with known absolute accuracy. 3458A on it's own does not produce any accurate readings, unless it is calibrated and proven stable against known reference standard.

One of the possible plans is to have a 3458A for that purpose.
Calibrating the resistance to Keysight specs can only be done by the PTB here and costs around 1000 EUR.
I did not find another lab in Germany that could meet the Keysight specs. I fact the one I talked to does indeed send their Datron 127x to the PTB for calibration as well.

[Edwin G. Pettis]

Most interesting, I have been talking to Try about those two resistors (he doesn't mention he has bought quite a few other resistors from me without complaint).  I originally tried to talk him out of ordering ±10 PPM resistors and told him why it would have been better to go with ±25 PPM instead.  I'm a little surprised at such a big fuss, I only charged Try $26.38 each for those resistors and frankly, that is an heck of a bargain (apparently he doesn't).  I rather doubt you could order 10 PPM resistors from anybody else for that price or even close to it, just how much is Vishay charging these days for even ±0.005%.  As with any precision resistor, it is going to drift and the first year is always the biggest shift, even a 5 PPM drift or a bit less could drop a 10 PPM resistor out of tolerance even if it was initially within the specified 10 PPM.  That was one of the reasons I tried to convince Try to go for 25 PPM plus lower cost.  The order was originally placed November 6, 2016 and likely shipped in early December (I'd have to check the records for the exact date) so these resistors are bit older than a year now, not nine months.

Given the unknown state of the instruments and the fact that a Vishay VHP 10K does not qualify as a resistance standard (and what was it measured on?), such tight PPM measurement uncertainty and accuracy is questionable for measuring a resistor that accurately.  There is the possibility that the resistors have moved out of the initial ±10 PPM, it would be silly to claim otherwise, however without the measurements being done on a properly calibrated instrument, the claimed measurements themselves are questionable at that accuracy.  For that level of accuracy, a direct comparison to a calibrated SR-104 is the appropriate method.

Since I have been busy working, I have not had the time to reply to Try's last email (last Wednesday) yet.  Does he want to send the resistors back to me?  Does he want a partial refund for the difference between 10 PPM and 25 PPM pricing?  At this point I don't know what he wants, I was rather surprised to see his posting here before trying to reach a solution with me.  Given the accuracy requirements here, while I accept that Try believes the resistors are "way out of spec", unless they are measured on a known accurate instrument, I really cannot accept his figures at face value in this case.  I am more than willing to try and find a solution but when splitting PPM hairs, I really need something better to go on.

[ap]

One of the possible plans is to have a 3458A for that purpose.
Calibrating the resistance to Keysight specs can only be done by the PTB here and costs around 1000 EUR.
I did not find another lab in Germany that could meet the Keysight specs. I fact the one I talked to does indeed send their Datron 127x to the PTB for calibration as well.

What exactly do you mean / refer to by Keysight resistor spec, and what is so special about it that you claim only PTB is able to calibrate it?

[Chipguy]

What exactly do you mean / refer to by Keysight resistor spec, and what is so special about it that you claim only PTB is able to calibrate it?

Simple as that: All the other labs I called are using something like: HP 3458A, Fluke 8xxx or Datron 127x for resistance calibration.
You see the problem?

Those are good enough to calibrate a Keysight 34401A for instance. But not good enough to calibrate another 8.5 digit multimeter to full spec.
The PTB however has some other means to do that. I have no idea what they use though (yet).

[Henrik_V]

This?
https://www.ptb.de/cms/en/ptb/fachabteilungen/abt2/fb-26/ag-262/the-quantum-hall-resistance.html
plus lots of stable artifacts in 24/7 stirred baths, AC-bridges..    ...    ...    skilled operators

[Chipguy]

This?
https://www.ptb.de/cms/en/ptb/fachabteilungen/abt2/fb-26/ag-262/the-quantum-hall-resistance.html
plus lots of stable artifacts in 24/7 stirred baths, AC-bridges..    ...    ...    skilled operators

Hallo Henrik,
yes, perhaps this. At this time I really don't know.
They should have diagrams in high res for download somewhere. Those show the tracability chain and/or rough diagrams of how they calibrate stuff.
I know that because we got a lot of them on the walls in our calibration lab at work.

[Chipguy]

I think that is what I mean:

https://www.ptb.de/cms/fileadmin/internet/fachabteilungen/abteilung_2/Kal-Messm-Abt2_V12_2017.pdf

[ap]

Simple as that: All the other labs I called are using something like: HP 3458A, Fluke 8xxx or Datron 127x for resistance calibration.
You see the problem?

Those are good enough to calibrate a Keysight 34401A for instance. But not good enough to calibrate another 8.5 digit multimeter to full spec.
The PTB however has some other means to do that. I have no idea what they use though (yet).

Well, then you called the wrong companies. There are several companies out there that have resistance standards that are specified (DAkkS) to sub ppm level over most of the range and low digit ppm in the 1 and 10Mohm ranges and which are very well positioned to calibrate a 3458A with very good TURs. Go look into the DAkkS data base and you will find them.
The meters you mention are indeed (in conjunction with a transfer decade or so) not suitable to calibrate the resistance ranges of a 3458A. But one would normally use a 5720/5730/4808 for calibration anyhow, if one would not want to work manually with resistance standards (which is more time consuming and costly than just connecting a calibrator). See also Fluke white paper on how to calibrate a 3458A with the 5720/30. Using low uncertainty dedicated standards is certainly preferrable though from a TUR perspective.

[branadic]

... a couple were damaged in shipping and 1 was slightly off spec....

Doh, this increases confidence to a maximum. If they got broken by shipping, this looks like a very sensitve component and construction leading in unreliable components. If they get broken by shipping, how reliable will they be in their final application?
But as I've learned, some resistor manufacturer are the opinion, that it is out of their scope what happens after the components have left their factory. This is one possible perception, with the chance of bad reputation. Another opinion is to improve construction and processes, so that shipping will not harm the component. Even though we are talking about a comparable cheap precision component this shouldn't imply poor reliability but more sustainability. My opinion.

-branadic-

[Echo88]

You do know that some package drivers are able to break the most ruggedized and careful packed stuff right? Since we didnt see any pictures of the damages all that remains are your opinions.

[branadic]

Even then it's up to the manufacturer that the package content is save during transport. I did already understand that you as well as MisterDiodes are fanboys of E.G.P., nobody could overlook that

-branadic-

[MisterDiodes]

... a couple were damaged in shipping and 1 was slightly off spec....

Doh, this increases confidence to a maximum. If they got broken by shipping, this looks like a very sensitve component and construction leading in unreliable components. If they get broken by shipping, how reliable will they be in their final application?

-branadic-

Let's see you build hundreds of a high precision anything on a production basis and see if you hit 100% after your product pass thru the post office.

The shipping damage was cause by the box getting jammed on a post office conveyor and crushing the box - not that it was an unreliable component.  We're guessing the out-of-spec resistor probably was part of the crushing process also since the lead was kinked hard.  Even then it was out only .013% on a .01% resistor - Edwin fixed that one too, and we didn't expect that.  That's what I mean by "slightly off spec". He wanted to make sure that we got within .01%.  OK, cool!

The important part is that even with correcting shipping damage, we wound up with 100% of what we ordered a week or two later.  Does that make me a fan??  Of course!

All the other many shipments we've received are perfectly fine.

We've got hundreds of Edwin's  PWW's (and other mfrs including GR, Reidon, Vishay) running in the field for years in test equipment literally running 24/7/365 and let's see...exactly zero have ever needed to be replaced.  That's what I can tell you.   As with ANY high precision resistor, you treat it with much care until it is safely mounted on a board, and don't overstress / crimp / kink / pull hard on the leads.  Like you would treat ANY. Precision. Component.  Regardless of where it comes from. 

As long as it's not overloaded or crushed it will probably outlast your lifetime.

Weren't you the one that was complaining about one of Edwin's resistors, and Edwin offered to replace it for you at no cost, regardless of mishandling?   Did you get your replacement?

Do other manufacturers do that for a hobbyist?

[MisterDiodes]

And by the way....How many other resistor manufacturers spend ANY time participating in this forum, explaining processes, answering questions, patiently offering free advice and help - even to those who are NOT customers?

I don't know of any... Maybe there are but they sure don't make it known.

[Echo88]

@Branadick: Its much simpler: i dislike you because of your attitude and pissing on the leg of multiple people. You know: whining about TiN not publishing his equations for his resistor-TCR-measurement, whining about Edwin who hasnt written a book which contains all his relevant resistor manufacturing knowledge. A resistor broke during shipment, DOH, clearly a major disaster which could have prevented by Captain Branadick.

I remember fondly the case, in which you showed your project here with attached eagle files and when people asked for schematic-pictures instead of eagle-files (because not everyone has eagle or wants to install it) and you basically said: fuck all those who dont have eagle to view it. Thats how you are and thats how i see you.

[Edwin G. Pettis]

An update on this resistor situation, first the person who originally bought the resistors has not contacted me to complain or ask for replacements.  While there is apparent reason to say that two resistors are broken, the root cause has not been determined, the fact that they were intact when originally shipped is the only solidly known fact as yet and I believe they were also intact when the buyer received them since there was no communication about bad parts from him.  The X-rays of the two resistors were not revealing in any manner, they did not show what had happened.  While Branadic contacted me earlier to complain a lot, he has not contacted me about the offer I made here on the blog.

Branadic seems intent to only complain and throw innuendo out, making unsupported remarks, resorting to silly claims about 'safe' shipping containers (that by the way, would be cost prohibitive compared to the cost of the resistors, that should be obvious).  The boxes I use now are rated for 200lbs and are hardly considered flimsy.

The point everyone seems to be making here, is either accept my offer (actually, the original buyer should ask since he bought them) or shut up and stop making an ass of yourself.  Get over it, it is only a couple of resistors, you are not making any brownie points here.

[branadic]

and you basically said: fuck all those who dont have eagle to view it.

@ Echo88
If you quote me, please quote me correct

-branadic-

[Chipguy]

Oh dear... what  have I done   
I made people argue... over nothing          
How did I manage that     

[branadic]

Get over it, it is only a couple of resistors, you are not making any brownie points here.

I'm the opinion, that a serious manufactor would backorder failing parts, analyze them to find out what was causing the failure, to improve manufacturing processes. But the statement reflects the main attitude. People are paying for functional parts, that's the point.

-branadic-

[Edwin G. Pettis]

Since this is getting way off this thread......last time!

The resistors are in Germany, I'm in America, now how do you think I'm going to be able to analyse those resistors from thousands of miles away?  What do you mean 'backorder' failing parts?  If you want to send them back for me to examine, feel free to do so, I'll be glad to post my findings and rest assured they will be accurate as possible.  Again, I think that the person who ordered them originally really should be making any requests, not somebody who did not purchase them.  As I said before, I received no complaint from the original purchaser which tells me the resistors were okay when they were received, by inference, that says something happened to those resistors after they were delivered....not before as you seem to imply.  It looks like you just want to complain for the sake of complaining.

Evidence has been presented by other customers that there is nothing wrong with my manufacturing processes except in someone's mind.  You did not respond to my earlier offer, you still haven't responded and that tells me that you really aren't interested in it, just in whining about it.

Enough of this silliness, you obviously aren't interested in any of this so stop it.  Unless I hear directly from the original purchaser in this matter, I consider this to be closed and done with, you have no standing.

[anishkgt]

Well a diy Version such as the milli ohm meter by scullcom (https://www.barbouri.com/2016/10/09/milliohm-meter-version-1-5/) would be a cheaper choice for just resistance measurement.


Www.Georgehobby.wordpress.com

Equipments: DSO104z, Hakko FX888D, Brymen BM869s

[amspire]

Since this is getting way off this thread......last time!

The resistors are in Germany, I'm in America, now how do you think I'm going to be able to analyse those resistors from thousands of miles away? 

Edwin, I think you are totally correct. If someone has an issue with your resistors, they should be talking to you directly. On a forum, discussions about issues with a few resistors is just annoying noise because it does not help the rest of us in any useful way. If it is such a big an issue, I do not know why they have not been posted back to you so you can analyse them.

I really appreciate your posts. Not everyone agrees with everything you say, but that is exactly what this forum is for. People have different experiences, different ways of working and they come to different conclusions. What you posts do tend to include is lots of information and that is gold.

Richard

[kutte]

and you basically said: fuck all those who dont have eagle to view it.

@ Echo88
If you quote me, please quote me correct

-branadic-

the correct quotes:
https://www.eevblog.com/forum/projects/oshw-24bit-adc-measurement-system-for-voltage-references/msg973338/#msg973338
https://www.eevblog.com/forum/projects/oshw-24bit-adc-measurement-system-for-voltage-references/msg973639/#msg973639
Kutte

[retroware]

I've been trying to use a frankenstein ESI 242 (ESI 240c, ESI RS925 - the one without trimmers, a Fluke 5440 and a Keithley 155) in an attempt to accurately calibrate the relative resistance difference of each resistor in a 1K per step SR1010. My end goal is to use the SR1010 as a calibrated decade divider.

I've been having no problems getting short term (~1min) stable .1 ppm resolution readings. But, it takes about 10~15 minutes to check all 12 resistors. As a check, I go back and spot check a couple of the resistors after measuring all 12 and the reading has drifted about 1 ppm, usually in the same direction but not by the same amount.

My first guess is TC, my lab is in a house basement with typical air conditioning and the SR1010 is specified at 5ppm/C. I'd have no problem believing that I have at least a .5C swing in temperature but I need to get an accurate thermometer to be sure.

My question is short of building a carefully temperature controlled room is there any way to get tighter readings? Are there other causes for this drift I should be looking out for? I'm using twisted 20awg solid copper wire for my interconnects and I adjust for lead and yoke variations after every resistor. Or is this about the best  I should expect given my setup? I'd be particularly interested in hearing from others that own a ESI 240.

Also, given that I'm not going to upgrade my environment, is there some way to factor TC drift into my calibration short of characterizing the drift of each individual resistor? Or am I going to have to read through the 32 page forum topic on measuring TC of precision resistors? 

[ap]

I like your Frankenstein statement 
The 1010 is not really built for very precise voltage dividing. Also, the 242 temperature drift adds. So the 1ppm is not a surprise. The 1010 is intended e.g for 1:100 ohms transfers and is pretty good at that, as temporary drifts are compensated out (see theory section of 1010 in user manual). For stable voltage divisions, there are better solutions (720A...). What you can do is to apply the error propagation methods of GUM, to calculate the temperature drift related uncertainty. It would go a little far to go into details how to calculate. In a nutshell, use the divider formula, calculate the weighted drift impacts of each of the contributors and combine this. You could assume a worst case scenario, all resistors on each side of the divider drifting 0.5x 5ppm up, the others down, but thats pessimistic. RSS would be more realistic.

[amspire]

The cheapest way to get accurate readings is to buy some PT100 (100 ohms) or PT1000 (1K) Platinum temperature sensors.

For example, you can get +/- 0.1 degC accurate ones from Mouser for about $6. Sensors are available down to at least 0.03% accuracy.

Rt = R0 * (1 + A* t + B*t2 + C*(t-100)* t3)

Where:

Rt is the resistance at temperature t, R0 is the resistance at 0 °C, and
A= 3.9083 E-3
B = -5.775 E-7
C = -4.183 E -12 (below 0 °C), or
C = 0 (above 0 °C)

So at room temperatures, it is very close to 3.9083 ohms/degree for the PT1000. At 20 degC, the resistance is 1077.9 degC.

The thing is measuring produces heat so the sensor has to be in good thermal contact with the thing you are measuring. I guess you could do this in a temperature well by adding oil.

The PT1000 sounds better then the PT100 with the higher resistance, but the PT100 is less susceptible to noise. You can use 1V across the PT1000 (as long as it is in thermal contact) but the lower the voltage the better. If you have 6 digit meter, you could a 100K range to measure a PT1000 just to minimize the power dissipation.