Precision Resistors - General Thread

[dietert1]

I think one can do the "10" or even higher on the most significant digit, but the other digits should remain 0 .. 9. So the maximum resistance would be 10 999.99 Ohms. Anyway one can use it that way to avoid confusion.
Maybe the idea was to provide a built in means of checking the lower decades at calibration time.

Regards, Dieter

[2N3055]

I think one can do the "10" or even higher on the most significant digit, but the other digits should remain 0 .. 9. So the maximum resistance would be 10 999.99 Ohms. Anyway one can use it that way to avoid confusion.
Maybe the idea was to provide a built in means of checking the lower decades at calibration time.

Regards, Dieter

That was the idea I think.

[leighcorrigall]

Perhaps you have another 10 kOhm resistor with a known value that you can check your DMM against.

Sorry about the delay. Here are the results.

Keithley 2000 (uncalibrated)
Vishay Y4078: 1.00065 kΩ @ 23.4 °C on 2021-09-24
C5-60T: 10.00711 kΩ @ 23.5 °C on 2021-09-24
Vishay Y4078: 10.00680 kΩ @ 23.4 °C on 2021-09-24

Keithley 6500 (NIST calibrated traceable within 1 month of measurements)
Vishay Y4078: 1.00000 kΩ @ 28 °C on 2021-04-13
Vishay Y4078: 9.99997 kΩ @ 28 °C on 2021-04-13

With compensation,
C5-60T: 10.00027 kΩ @ 23.4 °C

I must have made a calculation error last time I measured it. 

The Keithley 2000 seems to be OK. Measurement error could explain why the resistor is out. Until I perform a direct measurement with a calibrated instrument, I won't know.

[TiN]

Your 6500 is not NIST calibrated, it it calibrated traceable to NIST. Small wording change but huge practical (and money) difference

[guenthert]

In any case, you do not have the opportunity to buy that unobtanium things from our domestic market.... example like this:

      Huh, a IPv6 address in the wild.  More so, an unroutable IPv4 address embedded in an IPv6 address.
 How did that happen?  Might that just be a configuration issue on their firewall?  What happens when you use an IPv4 address?

[leighcorrigall]

Perhaps you have another 10 kOhm resistor with a known value that you can check your DMM against.

...

Keithley 2000 (uncalibrated)
Vishay Y4078: 1.00065 kΩ @ 23.4 °C on 2021-09-24
C5-60T: 10.00711 kΩ @ 23.5 °C on 2021-09-24
Vishay Y4078: 10.00680 kΩ @ 23.4 °C on 2021-09-24

Keithley 6500 (NIST calibrated traceable within 1 month of measurements)
Vishay Y4078: 1.00000 kΩ @ 28 °C on 2021-04-13
Vishay Y4078: 9.99997 kΩ @ 28 °C on 2021-04-13

With compensation,
C5-60T: 10.00027 kΩ @ 23.4 °C

...

Until I perform a direct measurement with a calibrated instrument, I won't know.

Results of a freshly calibrated NIST traceable Advantest R6581T (ACAL = ON, NPLC = 100, MEAN = 100):
10,000.065 ± 0.003 kΩ (n = 7) on the 100 kΩ range @ 25.89 °C, 64.94 % RH
10,000.1614 ± 0.0003 kΩ (n = 10) on the 10 kΩ range @ 25.92 °C, 64.51 % RH

Therefore, the resistor is within a 0.002 % tolerance.

Thank you for the resistor, shodan@micron. 

[leighcorrigall]

Does anyone have suggestions for a multi- or single-stranded wire that can be used to connect a resistance standard to a binding post? Many distributors carry hookup wire that is tinned or alloyed. I would prefer copper or silver wire.

I would like to mimic what Fluke or EIT has done with their standards:

https://www.eevblog.com/forum/metrology/teardown-standard-resistors/msg891672/#msg891672
https://www.eevblog.com/forum/metrology/teardown-standard-resistors/msg891673/#msg891673

[bdunham7]

I stripped down an old 4-wire POTS (phone) cable that I removed from the attic and got 4 nice pure copper color coded 24AWG wires that I twist up and use for voltage references and the like--just bare wires, no terminals.

[3roomlab]

Multipole PTFE cable is not easily available and even more difficult to find if shielded type is required.

Even though the polyethylene insulation sets some limitations, some people have succesfully used CAT6 cable as a replacement. It is dirt cheap, available from every electronics store and has very good performance if not used in extreme conditions.






INSULATION RESISTANCE

Measurements were made with a HP 4329A High resistance meter.

The setup represents the "worst case". Resistance was measured between the plain color wires of all pairs connected together and the white striped wires connected together. Cable shield braid connected to the meter Guard to avoid noise pickup. The cable length was 15 meters.

R ~ 200 Tohm @ U=10V
R ~ 20 Tohm @ U=100V
R ~ 2 Tohm @ U=1000V

High enough for most metrology applications. For example 2 Tohm in parallel with 10 kohm causes an error of 0.005 ppm.

The voltage dependence was remarkable. The 10V and 100V were stable, but the 1000V seemed to be too much for the insulation and the readings became "noisy". The sample was a cheap no-name shielded CAT6 cable ("worst case" again). It would be interesting to see if others get similar results.

For comparison a same style PVC insulated cable measured only ~3 Gohm. It was connected the same way as the CAT6, but the length was only 2 meters. No visible voltage dependence. The previous 10 kohm example would give 3 ppm error.


CHEMICALS

Several years in mineral oil bath didn't affect the cable in any way (only the polyethylene was in the oil, not the PVC jacket).


TEMPERATURE

Not suitable for use at elevated temperature.

https://www.eevblog.com/forum/metrology/replacing-teflon-cable-with-cat6-in-metrology/msg887437/?topicseen#msg887437

[DavidKo]

Look for the oxygen free speaker cable. It is cheap, but check the isolation. I have made a cable with bananas and there is resistance between wires  , you can always split it to two single wires and use air as insulator.

[Alex Nikitin]

I use CAT6 cable where I need a screen and where the wire resistance is not a major issue (voltage measurements) however it is not very flexible and needs careful handling. For mid-low resistance measurements and high currents I build my own test leads from silicone rubber isolated wires and Staubli plugs as on the photo. I've done some isolation tests and silicone rubber is pretty good, leakage current measured in a fixture as on the photo below is <0.5pA at 100V (a plain mains grade PVC isolated wire measured in the same way leaks about 100pA at 100V). The Staubli plugs are made from gold plated "copper alloy" (somewhere I've read "beryllium copper") unlike a bronze used for the majority of banana plugs and test leads like on the photo easily give me offsets below 1uV without special measures when used to connect something like SR1010 and KVD. I used two different makes of silicone rubber isolated wires, one Chinese make from Amazon (1.5mm2 , 400 strands of tinned copper, 600V 20A rated) which is surprisingly good and comes in various colours. 0.9m of this wire is about 10mOhm terminated. Another is a double-insulated wire from RS (1mm2, 250 strands of plain copper, 1000V 20A rated) which comes pre-cut at 1.2m and at that length measures about 25mOhm terminated. Both wires are very flexible and nice to use.

Cheers

Alex

[TimFox]

Are the units with extremely high resistance values, such as 1000 megohm, three-terminal TEE/WYE networks?
If not, what is the resistance material?

[bsw_m]

Some glass encapsulated microwire resistors.
С5-66  333.3MOhm ±0.02% TCR ?
МРХ    100MOhm -0.1% TCR: ?
РМП-7 10MOhm ±0.02% TCR 10 ppm
С5-60  1MOhm ±0.01% TCR 5ppm

[bsw_m]

Also good sapphire insulators for DIY high-resistance standards. 
The large insulator has a diameter of 10mm, a thickness of 4mm and a hole diameter of 2mm.

[bobAk]

The paper version is not custom.  You forgot something that must be still on this version for the resistor to be preferred.

[bobAk]

That's a coincidence.  My experience suggests otherwise.  And “best of the best” sounds odd to mrh, albeit with a paper label.  However, let me remind you once again - a paper label is not a custom, you forgot to indicate something additionally.  How many units have been measured?

[bobAk]

Can I see the results, for "paper label"?  With the separation of mvsg(мвсг) or mrh, for presentability.

[bsw_m]

I'm measure about 120pcs 30+ years old 100MOhm 0.1% MRH resistors without paper label.
All resistors have -0.02%.....-0.06% error from nominal value. Except one which was faulty.

[bobAk]

You can send me the result in pm if you are afraid of people in black.  Offtopic.  How does mrh hurt me?  Especially if you haven't even figured it out mrh or mvsg.  Let's talk seriously without the "pertuberant period word game".

[1audio]

I have a question relating to trimming precision resistors. I have 10 JRL 250K .01% resistors. They are all within tolerance (at least as far as I can tell with my limited resources). I want to build a variation on a SR1010MT divider with them. ESI used this circuit to make the resistors adjustable. What pitfalls and aspects should I be aware of both for the resistors and the trimmers (scaling to the 250K level I'm working with). My goal is to trim all the resistors to be as close as I can to each other.

[Conrad Hoffman]

In my experience those schemes work well and have few pitfalls. You can easily work out the sensitivity for each part to be sure what you use is good enough. The pot is often the limiting factor and the smaller range it covers, the better. I also have some JRL resistors but don't know how good they really are.

[leighcorrigall]

If you are serious about trimming, you could try the following Vishay products:

http://www.vishaypg.com/foil-resistors/trimmers/

You should consider the 'through the wiper TCR' and stability factor in your design.

Through the wiper TCR examples:
Accutrim 1240/1260: 10 ppm/°C
Accutrim 1285G: 5 ppm/°C

If TCR must be minimized, then the trimmer needs to be small because 5 ppm/°C can contribute a lot depending on the configuration. There are trade-offs when designing any circuit.

Keep in mind that vibrations can potentially knock trimmer wipers. If a trimmer is necessary, you might want to get one with the maximum number of turns per ohm to minimize the potential risk of offset due to vibrations.

You could also go with a more permanent solution by implementing a resistor network/array instead of a trimmer and 'trim' by various combinations of series and parallel resistances. I have a 4 x 350 Ω example in a past post: https://www.eevblog.com/forum/metrology/precision-resistors-general-thread/msg3664213/#msg3664213

Good luck

[Conrad Hoffman]

Another way to trim, used by Julie in some of their products, is to get a length of manganin wire and put a short piece in series with all the resistors to get them up to the value of the high one. You can't adjust it with a screwdriver but once it's done it isn't going to move much. 

[1audio]

I actually have a small supply of the Vishay foil trimmers. Not sure how many and what value. The Manganin wire idea is quite interesting and would probably work OK. I will need to find a source of around the right resistance.  On my Keithley the 250K resistors are all within about 40 Ohms which is inside the .01% original tolerance. Given that these predate zip codes (1963) its safe to say they are well aged!.

[TiN]

Little report on testing of 1000 Ω Guildline 9330 resistor. This resistor was acquired from second-hand market. There was no known history other than calibration sticker label on resistor body from Cal-Matrix Metrology, Inc. which does not seem to have a web-site but located in Burlington, ON, Canada. Last known value from sticker is 1000.0250 Ω measured at somewhat strange +24.0 °C with unknown uncertainty.

Resistor arrived in decent shape without any damages, just slightly dirty. There was no oil residues that could suggest use in bath.

Other than already mentioned value and calibration date May 13, 2015 there is no much other useful information from the stickers.

So we will start measurements on this resistance standard and collect history starting from clean slate. Before that can happen binding posts and brass nuts were carefully cleaned with soft abrasive cloth and cleaned with 99% IPA.

Resistor then was connected to measurement setup with Measurements International 6010B DCC bridge to perform performance tests.

RAW Datalogs and results for 1000 Ω Guildline 9330, S/N 45793

For the initial measurements at xDevs Lab in January 2022 this unknown resistor was placed in an air bath at nominal 23.0 °C ±0.1 °C. For better accuracy DUT temperature was measured with separate Omega capsule RTD and Fluke 1529 thermometer readout. This RTD was installed inside Guildline 9330 thermal well. The average temperature of the resistor was measured to be 22.992 °C and uncertainty of this measurement estimated ±0.18 °C. Standard resistor used in first test was thermally stabilized Fluke 935-10kΩ S/N 001, placed in lab ambient environment. Fluke 935 was running on battery during the measurement period.

Ratio between Fluke SL935-10k (316 µA) and MM Guildline 9330-1k (3.16 mA)

Ratio between Fluke SL935-10k (316 µA) and MM Guildline 9330-1k (3.16 mA) in reverse

Ratio between xDevs.com's ESI SR104 (316 µA) and MM Guildline 9330-1k (3.16 mA)

Ratio between T2 ESI SR104 (316 µA) and MM Guildline 9330-1k (3.16 mA)

Ratio between Fluke 5720A/H1 9999.9769 Ω (316 µA) and MM Guildline 9330-1k (3.16 mA)

Uncertainty and transfer analysis with Sandia PSL's open-source toolkit

Run, date	Rx	Ix	Rs	Is	Reversal time	Result	SDEV, ppm	ppm from average	Deviation from sticker
FSL935 FWD, JAN.9.2022	9999.9757 Ω ±0.3 ppm	0.316 mA	G9330-1k	3.16 mA	20 s	1000.03476 Ω	±0.04	+0.06 ppm	+9.76 ppm
SR104 forward, JAN.10.2022	10000.00128 Ω ±0.13 ppm	0.316 mA	G9330-1k	3.16 mA	20 s	1000.03477 Ω	±0.03	+0.07 ppm	+9.77 ppm
SR104 T2, JAN.12.2022	10000.0013 Ω ±0.13 ppm	0.316 mA	G9330-1k	3.16 mA	20 s	1000.03463 Ω	±0.04	-0.07 ppm	+9.63 ppm
Averaged	N/A	0.316 mA				1000.03470 Ω ±0.18 ppm			+9.72 ppm
Final 27.JAN.2022	N/A	0.316 mA				1000.03477 Ω ±0.22 ppm			+9.77 ppm

Based on these measurements we can estimate long-term drift of this resistance standard. To do so, one can take previous known calibration value from date May 13, 2015 which was 1000.0250 Ω with unknown uncertainty. That was 2452 days from today, 27.JAN.2022. Today measured value is 1000.03477 Ω with uncertainty ±0.22 ppm (95%).

Deviation between two values = ( (1000.03477 Ω / 1000.0250 Ω) - 1 ) * 1e6 = 9.77 ppm of change.

Making assumption of linear steady drift (which may not be true, but we don't have anything else to work with just two points history) we can calculate daily drift = 9.77 ppm / 2452 days equals 0.003985 ppm/day.

By multiplying this into 1 year = 0.003985 * 365 days = 1.454 ppm/year. This is still well within Guildline 9330 specification of ±3.5 ppm/year.

Temperature coefficient was also measured using two active Peltier air bath chambers.

RAW Datalog page with tempco measurement experiment



More details and internal design of Guildline 9330's also available on my review/teardown article.