Claude designs a 1 ohm standard resistor, how would you do it?

[aronake]

I had some reasoning back and forth with Claude on how to build a 1 ohm standard resistor.

Claude come with quite many good ideas. Like process on how to "dry" oil with sieves, include sieves in the box to continue to keep oil dry, what epoxy to use, calculations on thermal mass for different material, physical build etc. Claude quite frequently search eevblog and xdevs to figure things out.... Its quite impressive, but some human guidance still needed which I provided.

The build idea is basically to build a giant DIY 1 ohm Vishay VHA518-11. 15 well dried out 15.2 ohm S102K + some trimming resistors kept in a sealed cast aluminum box filled to 80% with medical grade paraffin oil. Then put this in another cast aluminum box with some insulation in-between and terminals on this box.

I have scanned 45 15.2 ohm resistors for TC, so that is done. Some of the other components I have and other are on the way from taobao.

Any "human" though on the design?

There are lots of 10K standard resistor builds presented here on EEVblog, but not many 1 ohm. So if someone have builds to share, that is also of interest.

Below from the Claude generated document. More in the attachment

Executive summary:

Purpose and Performance Target

This document describes the design and build of a homebuilt 1Ω precision resistance standard based on a parallel array of 15 Vishay S102K bulk metal foil resistors. The design targets sub-0.1 ppm/°C temperature coefficient and long-term stability in the ±0.5 ppm/year range — performance comparable to a commercial working standard such as the Fluke 742A or ESI SR104. The primary application is as a transfer standard for precision resistance calibration using an HP 3458A multimeter.

Build Philosophy and Key Principles

The design is guided by several metrological principles applied consistently throughout:
•   All-copper measurement path: bus bars, feedthrough rods, connecting cables, and solder (CdSn) are copper or copper-compatible throughout. Dissimilar metal junctions in the measurement path generate thermal EMF errors; eliminating them removes this error source entirely.
•   Kelvin 4-terminal connection: current and voltage paths are separated at the resistor field. Voltage sense taps are at the midpoint of each bus bar, not at the current entry points, ensuring bus bar resistance is excluded from the measurement.
•   Parallel resistor array: 15 individual S102K units in parallel. Individual unit TC values partially cancel, improving net TC by approximately −15 ≈4× over a single unit. Individual drift contributions also partially cancel, improving long-term stability.
•   Pre-selection and iterative trimming: all 40 available S102K units are characterised for TC before assembly. The 15 with the best cancelling TC combination are selected. After assembly, TC is verified by thermal cycling and individual resistors are swapped if needed. Two additional trim slots allow a higher-value resistor to be added to bring the final value to just above 1.000Ω.
•   Thermal isolation: a sealed oil-filled inner box (VT7) is surrounded by XPS foam insulation inside a sealed outer box (VT20). The effective thermal time constant of approximately 3 hours makes short-term bench temperature fluctuations essentially invisible to the resistor array.
•   Moisture exclusion: both boxes are hermetically sealed diecast aluminium with nitrile O-rings, neutral-cure RTV, and Araldite 2011 epoxy on all terminal penetrations. 3Å molecular sieve desiccant inside both boxes provides active moisture scavenging. The inner box has a nitrogen headspace. The outer box acts as a first moisture barrier before the inner seals.
•   Isothermal design: a copper base slab (100×53×5mm) under the resistor array provides an isothermal mounting plane. Combined with oil immersion, temperature differences across the array are minimised, which both improves TC measurement accuracy and reduces thermal EMF at solder joints.
•   Ratio measurement method: resistance is not measured directly on the 3458A ohms function. Instead, a stable voltage source drives current through the new standard and a Fluke 742A 1Ω calibrated reference in series. The 3458A measures the voltage across each resistor alternately. The ratio V_x / V_ref equals R_x / R_ref, so R_x = R_ref × (V_x / V_ref). This method cancels current source uncertainty and leverages the excellent linearity of the 3458A rather than its absolute accuracy.

Key Specifications
Parameter                                       Target / Value   Basis
Nominal resistance                       1.000 000 Ω           15× S102K parallel + trim resistor(s)
Temperature coefficient               < 0.05 ppm/°C   Pre-selection + iterative TC trimming
Long-term stability                      ±0.5 ppm/year   After 6-month stabilisation period
Thermal time constant              ~3 hours           XPS insulated double-box construction
Measurement method                      Ratio vs Fluke 742A 1Ω   3458A measures V_x/V_ref, R_x = R_ref × ratio
Measurement terminals              4-terminal Kelvin + GND   M4 copper rods through inner box wall, epoxy sealed
Bar dimensions                              85 × 15 × 2mm copper, 1.81mm gap   Parallel bars: 3.81mm lead pitch, holes 1mm from inner edge
Resistor slots                              17: 15 S102K + 2 trim   4mm hole-to-hole centres, single row
Temperature sensor                      Omega 44006 NTC 10kΩ β=3694K   In oil, 3.5mm mono jack on outer box
Layout                                      Inner box on one side, terminals on other   Short Cu cables cross insulation gap to outer binding posts

[aronake]

i have 14 pieces of 10 ohm VHP101, VHP202Z etc, but even if choosing the 10 with lowest TC, i would end up with positive, TC so until I have found more 10 ohm sealed vishay resistors with low TC, building something similar with these will have to wait. Same with 10 ohm VPR221, have around 13 of these, but still too high net TC. I think 10 pieces of 10 ohm VPR221 would have been an interesting build.

[miro123]

Seems known procedure to build resistor standard.I want to mention mistakes that I made building it on this way.
 - 80% filed and hermetically sealed. - My sealing was  that standard AL case silicone rubber o-ring.  outside i sealed wit RVM - result after 2 years some minuscule leak appeared. in 3-th year leak becomes visible. I assume temperature variation causes some pressure.
My goal was to keep it for at least 20 years. Oil change for 1 ohm will not be an issue but in my case 10K.
Question.
DOe you need to measure hydroelectrical constant. Many modern DMM dont need it. E.g my Keythleys has option to  measure the 4wire sense voltage without current

[aronake]

Great input!

Claude previously warned against silicone seal, and recommended nitril in the design document that should be quite a bit more oil resistant, slower aging and let through less water. I now pushed Claude a bit more on this topic, recommendation is Vitron/FKM/Fluoroelastomer, which is better in most ways. Then apparently there is FFKM (Kalrez, Perlast) which is very expensive. I now placed an order for some FKM.


Claude comments

Material comparison for your application
Nitrile (NBR/Buna-N) — what you currently plan

Good mineral oil resistance
Water vapour permeability: moderate — better than silicone but not the lowest
Long-term compression set: moderate — will slowly lose clamping force over years
Practical service life in static seal at room temperature: 10–20 years before noticeable relaxation

FKM (Viton) — the recommended upgrade

Excellent resistance to oils, considered very reliable in hydrocarbon service. Offers broader chemical and temperature resistance than nitrile. IET Labs
Water vapour permeability: significantly lower than nitrile — roughly 3–5× better
Hardness remains practically constant during ageing, unlike HNBR and EPDM which harden significantly — meaning FKM maintains its sealing contact force over decades IET Labs
Compression set resistance: excellent — best of any common elastomer at room temperature
Practical service life: essentially indefinite at room temperature in static application
Cost: 3–5× more than nitrile — but O-rings are inexpensive regardless IET Labs

FFKM (Kalrez, Perlast) — overkill but ultimate choice

Near-zero gas permeability
Essentially zero compression set
Used in semiconductor and pharmaceutical hermetic sealing
Very expensive — a single O-ring costs $50–200 depending on size
Genuinely worth considering for the inner box lid where seal integrity matters most over decades

EPDM — not suitable here

EPDM resists swelling in water-based fluids but is not suitable for hydrocarbon oils. It swells in paraffin oil — eliminate from consideration. IET Labs

Silicone — already discussed, avoid

High water vapour permeability — worst of the common options for moisture exclusion

[miro123]

Thanks for the feedback.
I wonder how your Claude chat session compares Paraffin vs silicone oil filling. It is worth mentioning that my10k an 2k resistors are filled with clean silicone oil HV electrically certified. High viscosity 320cT@40C

[aronake]

Thanks for the feedback.
I wonder how your Claude chat session compares Paraffin vs silicone oil filling. It is worth mentioning that my10k an 2k resistors are filled with clean silicone oil HV electrically certified. High viscosity 320cT@40C

Cut and paste below on Claude's view on liquid paraffin vs silicon oil.

Executive summary, both are fine and similar. As I plan to put some sieves in my resistor for long term extra dry oil, liquid paraffin seems to have a slight advantage.

"Liquid paraffin" goes under many different names depending on use and some very slight variations: white oil, Mineral oil, Transformer oil (typical lower purity). The more common name for metrology use would probably be white oil or mineral oil. For pharmaceutical purposed often Liquid paraffin, and it seems that is typically the purest kind, so i just happened to continue to use that expression.

Flukes "bath guide":
https://media.fluke.com/df74e46c-7444-4c40-b21e-b3190159b463_original%20file.pdf

and "how to select a calibration bath fluid" guide
https://media.fluke.com/4682a306-784f-4664-984c-b2e6009311e5_original%20file.pdf

give some more limited insight. It seems silicon oil is aviable in more different versions for wider temperature coverage.

------

how would you are using medical grade liquid Paraffin vs silicone oil?

9:23 PMClaude responded: Good comparison to make before committing to the fill medium.Good comparison to make before committing to the fill medium.
Medical grade liquid paraffin — your current planSearched the webSearched the webGood question — worth comparing properly before you fill the box.
Key properties comparison
PropertyMedical liquid paraffinSilicone oil (PDMS)Thermal conductivity~0.13–0.15 W/m·K0.10–0.16 W/m·K depending on viscosity Hammond Mfg.Specific heat~2.1 J/g·°C~1.5 J/g·°C (roughly 1/3 that of water)Density~0.85 g/cm³~0.93–0.96 g/cm³Volumetric thermal mass1.79 J/cm³·°C~1.4 J/cm³·°CViscosity stability with temperatureModerate variationVery low viscosity-temperature coefficient — changes very little with temperature Don AudioChemical stabilityGood at room temperatureExcellent — very high thermal and oxidative stabilityMoisture absorptionLowVery lowCompatibility with copperExcellentExcellentCompatibility with epoxyGoodGoodCompatibility with FKM gasketExcellentExcellentOdourOdourless (medical grade)OdourlessLong term stabilityGoodExcellent

Thermal conductivity — closer than expected
Silicone oil thermal conductivity is between 0.10 and 0.15 W/m·K for viscosities below 50 cSt, approximately 0.16 W/m·K above 100 cSt. This is actually slightly lower than mineral oil at comparable viscosities — the opposite of what many people assume. Silicone oil thermal conductivity is roughly one quarter that of water, and nearly the same as benzene and toluene. Hammond Mfg.Stompbox Layout
So the thermal conductivity advantage goes slightly to paraffin oil, not silicone.

Thermal mass — paraffin oil wins clearly
Silicone oil has lower specific heat (~1.5 J/g·°C) and slightly higher density than paraffin — giving volumetric thermal mass of roughly 1.4 J/cm³·°C vs paraffin oil at 1.79 J/cm³·°C. Paraffin oil has about 28% more thermal mass per unit volume — meaningful for buffering short-term temperature fluctuations.

Where silicone oil is genuinely better
Silicone fluids can maintain chemical stability at temperatures exceeding 200°C — mineral oils tend to oxidise and degrade at elevated temperatures. For your room temperature application this advantage is irrelevant. Don Audio
Silicone fluids have a low viscosity-temperature coefficient — viscosity changes very little with temperature variations. This means convective circulation inside your box is more consistent across the 15–30°C cycling range. A minor but real advantage during thermal cycling. Don Audio
Silicone oil is also cleaner to handle — less tendency to leave residue, easier to clean up if spilled.

Moisture absorption
Both are very low. Silicone oil is slightly more hydrophobic than paraffin oil — marginally better at resisting moisture absorption from residual headspace humidity. The difference is small and your molecular sieve handles this regardless.

One concern with silicone oil — molecular sieve interaction
Silicone oil (PDMS — polydimethylsiloxane) contains silicon-oxygen bonds. Some grades contain trace low-molecular-weight cyclic siloxanes (D4, D5) which can potentially interact with zeolite molecular sieve pores. This is well documented in industrial contexts where silicone contamination of molecular sieve beds reduces capacity over time. For your application the quantities are tiny and pharmaceutical grade silicone oil has minimal cyclic siloxane content — probably not a practical concern but worth noting.

What commercial resistance standards use
The ESI SR104, Fluke 742A, and Ohm-Labs standards all use mineral oil — typically transformer oil or similar. This is the established choice in the metrology community. Silicone oil is used in some temperature bath applications but is less common in sealed resistance standards specifically.

Honest recommendation
Medical grade liquid paraffin is the better choice for your build for these reasons:

Higher volumetric thermal mass — better short-term temperature buffering
Equal or slightly better thermal conductivity
Established use in commercial resistance standards
No cyclic siloxane concern with molecular sieve
Easier to source in pharmaceutical grade with guaranteed purity
Less expensive

Silicone oil's main advantages — high temperature stability and viscosity consistency — are largely irrelevant at room temperature in a sealed static application. Its lower thermal mass is a genuine disadvantage for your purpose.
Stick with medical grade liquid paraffin BP/USP as planned.