Low-Level Signals and Contact Design
Most people assume more current means a harder connector problem. At the contact interface, the opposite can be true: a contact that happily carries ten amps may turn flaky at ten millivolts. The reasons are surface films, dry circuits, wetting current, plating, and fretting. We'll unpack those, then use one simple question to decide whether a contact system belongs anywhere near your signal.
It extends the plating table in §5.1 and the contact-resistance row in Reading Datasheets §6; the hobby-track version of the boundary is Power vs Signal.
1. The problem: films win at low level
Every non-noble contact metal grows a surface film in service — oxides, sulfides, contamination. Films are thin, but they are insulators, and whether your circuit works depends on whether metal actually touches metal through them.
At power levels, the circuit helps: enough voltage and current will break down and punch through a thin film, and the heat and force at the interface keep the conduction path open. At signal levels — millivolts of open-circuit voltage, microamps to milliamps of current — the circuit never breaks the film. Whatever conduction path exists is whatever the contact system created mechanically at mating, and whatever the environment has done to it since. That is the entire subject of this page: at low level, contact reliability must come from the metallurgy and mechanics of the contact, because the electricity contributes nothing.
This is why the failure signature is so characteristic: the interface "works when freshly plugged" (mating wipe scrapes a clean spot), then drifts intermittent over weeks or months as films and debris re-form — and often "fixes itself" when someone re-mates the connector, which just wipes a new spot and restarts the clock.
Same force, same geometry — the film decides. On tin the signal lives with whatever the oxide and debris allow; on gold, metal touches metal and stays that way.
2. "Dry circuit," defined
A dry circuit is one whose voltage and current are too low to alter the contact surface — too low to break down films electrically. The connector industry's standard measurement embodies the idea: the low-level contact resistance (LLCR) method, EIA-364-23, deliberately caps the measurement at approximately 20 mV open-circuit and 100 mA precisely so the measurement itself cannot break down films — it measures the interface as your signal will actually experience it, not as a power circuit would.12
The practical reading for a design engineer:
- If your signal lives at or below that class of levels — thermocouple and RTD millivolts, strain-gauge bridges, low-voltage logic, bus lines with milliamp drive, anything a datasheet would call "measurement" — you are designing a dry circuit, and the contact must be reliable without electrical help.
- When a manufacturer publishes contact resistance "per EIA-364-23" (or an equivalent low-level method), that is the number that speaks to signal duty. A contact resistance measured at rated current says much less about how the same interface behaves at millivolts.
3. Minimum wetting current
The relay and switch world — which has fought this exact battle for a century — names the other side of the boundary. A film-forming contact needs a certain minimum load to stay reliable: enough current to break through and keep clear the films that keep forming. Below that level, the contact drifts intermittent. Relay manufacturers publish this as a minimum switching capacity or minimum applicable load — a reference value, not a guarantee — and their microload guidance is blunt about the failure mechanism (oxide film on the contact surface) and the fix (gold-class contact surfaces, often redundant crossbar/bifurcated geometries, for microload duty).3
The same concept applied to connectors is often called minimum wetting current. Connector datasheets rarely state one — which is exactly the point:
- Tin-class contacts have a real minimum. Below it, nothing keeps the interface clear, and the only question is how long the mating wipe's clean spot lasts.
- Gold-class contacts effectively don't, in clean, correctly designed service — gold grows no insulating film, so there is nothing for a minimum current to break through.4
- If you find yourself asking "is there enough current on this line to keep the contact happy?", the robust answer is not to add current — it is to specify a contact system that doesn't need it.
4. Gold vs. tin is a mechanism choice, not a price tier
The §5.1 plating table gives the summary; here is the mechanism behind it.
Tin works by fracture. Tin oxidizes instantly, but the oxide is thin, hard, and brittle on top of a soft, ductile metal. Apply enough contact normal force and wiping action, and the oxide shell cracks; clean tin extrudes through the cracks and forms gas-tight metal-to-metal spots. That is a genuinely reliable mechanism — if the design maintains it. The classic tin guidelines (AMP's, later Tyco's, "Tin Commandments") spell out the conditions: high contact normal force (a 100-gram class figure, far above what fine signal contacts run), a mechanically stable mated interface that cannot micro-move, contact lubrication, adequate coating thickness, and no continuous high-temperature service.5
Gold works by staying noble. Gold grows no oxide or sulfide in normal environments, so a gold-to-gold interface conducts at any level, at low normal force, indefinitely — which is why it is the default for low-current signals, dry circuits, and mil-spec contacts (the 50 µin-class plating on mil contacts exists for this reason). Its enemies are mechanical and economic instead: it is soft, it wears, and it costs money. Gold flash (a very thin layer) buys the surface chemistry but not the wear life — flash wears through with cycling, exposing the nickel or base metal beneath, after which the interface is no longer a gold interface. Cycle-count expectations and plating thickness travel together; check both on the exact contact P/N.4
Never mate gold to tin. A mixed interface gets the worst of both: it fretts, tin transfers to the harder gold surface, and tin oxide builds up exactly where the gold was supposed to prevent it — the gold half is wasted and the joint behaves like a bad tin joint. This is the AMP/Tyco Golden Rules whitepaper's Rule 12, and it applies per mated pair: both halves, same plating class, chosen deliberately.46 In families where plating is selected per contact P/N (most of them), it is easy to violate by accident across two BOMs — the mating-pair check belongs in the ICD.
5. Fretting: the low-level killer
§1.6's failure table lists fretting corrosion first for a reason. The mechanism:
- Vibration, thermal cycling, or cable motion moves the mated contacts against each other by micrometers — far too little to notice, far too much for the interface.
- Each micro-slide exposes fresh metal, which (on tin and other film-formers) instantly oxidizes.
- The oxide debris doesn't leave. It accumulates in the contact zone, and the interface resistance climbs — steadily or intermittently — until the signal fails.
Two properties make fretting the characteristic low-level failure. It is invisible: a fretted connector looks perfect, and the damage is under the contact spot. And it is self-hiding: re-mating wipes the debris aside and the fault "goes away," which is how harnesses end up in the maintenance folklore of "reseat the connector every few months." A dry circuit cannot burn through the debris the way a power circuit partially can, so low-level signals see fretting first and worst.
Mitigations, in the order to prefer them: a gold-to-gold interface (no film to grow), a mated pair that cannot micro-move (positive latching or threaded/self-locking coupling, strain relief so cable loads never reach the contacts, board-to-board stacks that are mechanically supported), and — where the manufacturer specifies it — contact lubricant, which is a legitimate engineering material here and part of the tin guidelines, not a hack.5
6. When a power-oriented contact is the wrong home for a signal
The Anderson Powerpole page makes this argument for one family; here is the general version. Power-oriented contact systems are engineered around a different problem: carrying amps with low temperature rise. Their platings (tin, silver), normal forces, geometries, and published data all serve that lane. Pressing one into signal duty fails on three counts:
- Surface chemistry. Tin and silver are power platings; at dry-circuit levels their films never get broken down (silver's sulfide tarnish is its own version of the problem — see §5.1).
- No data. The documentation publishes current ratings and temperature rise, not low-level contact resistance. Nothing tells you how the interface behaves at millivolts, because the manufacturer never characterized it — you'd be designing on vibes.
- No signal mechanics. Power families frequently lack the latch, keying, shielding path, and fine-pitch density that signal interconnects are built around.
So the one-question test:
Does the manufacturer publish low-level / dry-circuit contact resistance (EIA-364-23-class data) — or otherwise state that this contact system is characterized for low-level signals? If yes, you're in the documented lane; design to that data. If no, your millivolt signal is an experiment the manufacturer never ran.
The legitimate mixed case is real and common: one connector carrying both power and signal contacts, each specified for its job — a mil-circular insert with gold-plated signal contacts beside larger power contacts, an M23 motor connector's separate feedback insert. What makes those work is that the signal contacts are signal contacts. The illegitimate version is "the power connector has spare poles, run the sense lines through them" — same shell, wrong lane.
Signal ≠ small power, in both directions: a 500 mA accessory feed through a power family is fine; an encoder pair through it is not. The mirror-image mistake — real power through signal contacts — is Power vs Signal's subject.
7. What to check on the datasheet
Adding the low-level lens to the §6 field guide:
| Check | Where it hides | Why it matters at low level |
|---|---|---|
| Plating material and thickness on the exact contact P/N | Contact drawing / ordering table, not the family page | Gold flash and 50 µin gold are different products with different lives (§5.1) |
| Contact resistance test method | Spec/qualification table footnotes | Low-level (EIA-364-23-class) data speaks to signal duty; rated-current data mostly doesn't1 |
| Cycle rating at that plating | Durability spec | Wear-through converts a gold interface into something else mid-life |
| Mating half's plating | The other BOM | Gold-to-tin is a per-pair defect no single datasheet will flag4 |
| Normal force / stability features | Application spec (latch, coupling, TPA) | Tin's mechanism needs force and zero micro-motion; low force + vibration = fretting5 |
| Any stated minimum load | Usually absent on connectors | Its absence on a power-oriented family is the lane test failing |
Source status
The film, dry-circuit, fretting, and gold-vs-tin explanations come from the EIA-364-23 method listings and manufacturer engineering documents listed below. The “lane test” and mitigation order are engineering judgment, tracked as heuristics in Source Notes. There is no universal magic number on this page: minimum loads, normal forces, plating thicknesses, and cycle lives belong to exact part numbers and their datasheets.