October 25, 2026 · Technical Guide
Ask ten buyers what a MIL-DTL-38999 is made of and nine will say aluminium. The shell is the part you can see, the plating is the part you can specify, and the contact is the part that gets cross-referenced against a competitor. The insert — the dielectric block that holds every contact at its exact position and insulates it from its neighbours and from the shell — is usually compressed into one word on the drawing and never discussed again.
That word decides four things at once: the connector’s temperature class, its arc path at altitude, whether a contact still sits where it was pressed after a hundred thermal cycles, and whether the part can go into a vacuum. It is also the part most often changed without anybody noticing, because a material substitution leaves no mark on the outside of the connector, no change to the part number, and nothing that a room-temperature dimensional check will catch.
This guide is about that word. It is not a datasheet: every number below is a typical published range for a material family, and the grade — filler type, filler loading, molecular weight, moulding conditions — moves all of them. Take the actual values from the compounder’s datasheet and the UL yellow card for the exact grade you are buying, not from a family average and not from this page.
An insert is not a passive spacer. It is a structural, electrical and sealing component at the same time, and the four duties pull the material in different directions — which is why no single polymer wins on all four.
| Job | What the material must hold | How it fails when the material is wrong |
|---|---|---|
| Contact positioning | Cavity spacing and alignment tight enough that mating force stays inside the coupling envelope over the full temperature range | Warped or creeping insert closes the gap between pin and socket bore; mating force climbs, contacts deflect on entry, and the coupling ring feels tight before it is seated |
| Dielectric isolation | Contact-to-contact and contact-to-shell withstand at the specified test voltage — and at altitude, where the air gives up first | Tracking along a contaminated surface, partial discharge in a void, or a flashover path where the insert meets the shell. Appears as a dielectric-withstanding-voltage failure, not as a short |
| Contact retention | Cavity walls and shoulders that hold the retention clip at the specified pull-out load, hot and cold, without yielding | Contact pushback on mating, retention loss under vibration, or a cracked cavity wall after a re-termination — the field failure that reads as “the pin moved” |
| Sealing interface | Flat, dimensionally stable faces for the interfacial seal, and stable cavities for the rear grommet or individual seals | Creep or thermal set lets the seal relax; ingress appears months later as an insulation-resistance drop, and is blamed on the grommet rather than on the insert behind it |
Notice what is not on the list: strength. Inserts almost never fail because the polymer was too weak in bulk. They fail at a surface (tracking), at an interface (seal relaxation), at a thin section (a cavity wall), or at a weld line — all places a datasheet does not measure.
Four unrelated measurements all get called the temperature rating, and confusing them is the most expensive mistake in insert selection.
The practical consequence: a connector’s headline temperature range is set by the weakest of its insert, its seal elastomer and its plating system — and on a standard circular the top of the range is usually the silicone seal, while the bottom is usually the elastomer’s brittle point rather than the plastic.
| Material family | Typical continuous-use band | What actually sets the ceiling |
|---|---|---|
| Glass-filled epoxy (GFRP) | ~175 °C to 200 °C | Oxidative ageing of the resin; brittle well before the temperature limit, and cracking on contact insertion is the more common complaint |
| PPS, glass-filled | ~200 °C to 220 °C | Inherently flame-retardant chemistry with good dimensional hold; long-term embrittlement and low elongation |
| LCP, glass/mineral filled | ~240 °C to 280 °C | Highest of the thermoplastics used in inserts; weld-line weakness and notch sensitivity become the limit before temperature does |
| PEI (polyetherimide) | ~150 °C to 180 °C | Strength and stiffness are excellent, but solvent stress-cracking and moisture uptake cap it well below PPS |
| PEEK | ~240 °C to 260 °C | Chemical resistance plus temperature, at a cost that usually rules it out of volume circulars |
| Silicone / fluorosilicone (seals) | ~−55 °C to +200 °C | Compression set and reversion at the top; low-temperature stiffening at the bottom |
| Neoprene grommet (legacy) | ~−55 °C to +100 °C | Compression set and hardening; the reason a “200 °C” connector with a neoprene grommet is not a 200 °C connector |
Which polymer appears in which connector is mostly a story about cavity count and cavity wall thickness. As contact density rises the walls between cavities get thinner and the flow path gets longer, and the polymer has to fill a mould without warping — that, more than any specification limit, is what pushed inserts from thermoset to thermoplastic.
| Material | Where you meet it | Real strength | Real weakness |
|---|---|---|---|
| Glass-filled epoxy | Legacy and current MIL-spec circulars, compression- or transfer-moulded inserts, high-contact-count shells | Excellent dimensional stability, low CTE, good dielectric behaviour, understood by every moulder | Brittle: a cavity wall cracks rather than yields, and a cracked insert is a scrap part, not a repair |
| PPS, glass-filled | Volume circular connectors, industrial and automotive-grade families | Inherent V-0 without halogenated additives, good chemical resistance, stable cavities | Low elongation — thin walls are notch-sensitive and weld lines are the weakest plane |
| LCP | High-density and miniature inserts, thin-wall cavities, fine-pitch geometries | Very low melt viscosity fills 0.4–0.6 mm walls, low warpage, high temperature, low outgassing | Anisotropic: properties differ along and across flow, so a weld line can be a fraction of the bulk strength |
| PEI | Some industrial and instrument connectors; parts needing stiffness | High strength and modulus, good dielectric strength | Attacked by chlorinated solvents, ketones and strong bases — a cleaning-agent problem as much as a service problem |
| PEEK | Downhole, subsea and high-reliability parts where chemical exposure drives the choice | Broad chemical resistance at temperature, low outgassing, high purity | Cost, and it is harder to mould into thin high-density walls than LCP |
| DAP / melamine | Older MIL-C-era connectors still in service | Good electricals, dimensionally stable, cheap | Brittle, moisture-sensitive, and a reformulation candidate wherever a restricted substance is involved |
| Glass-to-metal | Hermetic connectors and headers | Satisfies a leak-rate requirement no polymer can — the seal is the joint itself, not an elastomer | Not a polymer at all: a matched or compression glass seal in a metal shell, and a different qualification route |
Tracking is a surface failure. Under voltage a wet contaminated surface develops a conducting carbon path, the path heats, it widens, and the insert becomes a resistor and then a short. The material property that resists this is the comparative tracking index, measured to IEC 60112 and reported in UL 746A as a performance level class.
| PLC | CTI band | What it means in a connector |
|---|---|---|
| PLC 0 | ≥ 600 V | Rare in insert materials; where it exists it buys creepage margin in dirty, wet installations |
| PLC 1 | 400–599 V | Comfortable margin for standard circulars at low altitude |
| PLC 2 | 250–399 V | The usual band for glass-filled engineering thermoplastics |
| PLC 3 | 175–249 V | Workable, but the creepage path now has to be designed deliberately rather than assumed |
| PLC 4 | 100–174 V | Poor: a contaminated surface can track at working voltage, not just at test voltage |
| PLC 5 | < 100 V | Not an insert material for anything above signal level |
Two things make CTI matter more in a connector than in a housing. First, the insert is where the whole creepage and clearance budget is spent — there is no room to add distance, because contact spacing is fixed by the insert arrangement. Second, altitude attacks the air and leaves the surface carrying the stress: at 70,000 ft a given gap breaks down at a fraction of its sea-level voltage, so a design that was air-insulated at sea level becomes surface-insulated in the flight regime, and the surface is where contamination and condensation live. The derating mechanics are in our temperature and altitude derating guide; CTI is why derating a low-CTI insert is not simply a matter of subtracting a factor.
Insulation resistance is the measurement that finds a wet insert, and it is the one buyers most often treat as a formality. The mechanism is undramatic: the polymer absorbs a small percentage of water, ionic contamination dissolves in that water, and the surface and bulk leakage paths lose resistance. Nothing looks wrong. The part passes continuity and contact resistance, and fails insulation resistance — or worse, passes it cold and fails it warm, because diffusion is faster at temperature.
| Material | Typical water absorption, 24 h immersion (ASTM D570) | Practical reading |
|---|---|---|
| PPS, glass-filled | ~0.02–0.05 % | Low enough that insulation resistance is stable in damp heat |
| LCP | ~0.02–0.04 % | Lowest of the group — a real advantage in condensing environments |
| Glass-filled epoxy | ~0.1–0.5 % | Acceptable, but the absorbed water has to be driven out, not cleaned out |
| PEI | ~0.2–0.3 % | Higher, and coupled with its solvent sensitivity |
| PEEK | ~0.1–0.2 % | Low for its temperature class |
The operational lesson is about recovery. A connector that fails insulation resistance after a humidity or salt-fog exposure can often be restored by drying, because the water is the conductor and the ions are still there: clean it, dry it, and the resistance comes back. Skip the drying and the next reading is worse than the first. This is why insulation resistance has to be read as a sequence rather than a single number, and why a “failed” insert should be dried and re-tested before it is scrapped.
Every cavity in an insert is a tolerance stack-up: cavity position in the mould, mould shrinkage, post-moulding shrinkage, thermal expansion in service, and the deflection of the cavity wall under contact insertion force. The first three are the moulder’s problem. The last two are the material’s.
| Material | Typical CTE (ppm/K) | Consequence against a 23 ppm/K aluminium shell |
|---|---|---|
| Aluminium shell (reference) | ~23 | The baseline every insert is fighting |
| Glass-filled epoxy | ~15–25 | Close to the shell — the reason the thermoset generation has such a good thermal-cycling record |
| PPS, glass-filled | ~20–30 | Reasonable match; anisotropy with flow direction is modest compared with LCP |
| LCP | ~5–20 along flow, ~30–60 across | The mismatch is not one number: a cavity measured along the flow direction moves differently from one across it |
| Unfilled thermoplastics | ~50–70 | Not suitable for a high-density insert; included only to show why fillers exist |
Two failure modes follow directly. Warpage is differential shrinkage, and its signature is a connector that mates easily on one side of the shell and fights on the other — the pin field is not parallel to the socket field. The weld line is the plane where two flow fronts met during moulding: mechanically the weakest plane in the part, and very often located exactly where two cavities are closest together, in the thinnest wall, taking the highest insertion load. A material that fills beautifully and welds badly produces an insert that passes every dimensional check and cracks on the third re-termination. This is why insert failures are so often written off as “tool abuse” in the field: the tool may have been correct, but the wall it was pushing against was a weld line.
In a vacuum or a sealed optical enclosure, the insert becomes a contamination source. Volatiles leave the polymer and re-deposit on the coldest surface in the assembly — frequently an optical face, a relay contact or a thermal radiator. The screen is ASTM E595, and the two numbers that matter are total mass loss and collected volatile condensable materials.
| Screen | Limit usually applied | Why it bites in a connector |
|---|---|---|
| Total mass loss (TML) | ≤ 1.0 % | Catches absorbed water and low-molecular-weight residue, including moulding aids and unreacted resin |
| Collected volatiles (CVCM) | ≤ 0.1 % | The number that actually matters: what condensed on the cold surface |
| Water vapour regained (WVR) | Reported alongside the other two | Separates water from solvents, so a wet part is not mistaken for a dirty one |
Glass- and mineral-filled thermosets, LCP and PPS generally clear these limits as moulded. Silicone is the classic offender on TML, because its volatile fraction is high unless the part is post-cured — and silicone is also present in the same connector as the seal, so a vacuum application has to screen the assembly, not just the insert. The same logic drives material choice in hermetic and vacuum applications, where the sealing medium is glass rather than an elastomer and the outgassing argument changes completely.
“V-0” is a material rating at a stated thickness under UL 94, and it is not a part rating. A burner test on a bar says nothing about a sixty-cavity insert with thin walls, voids and a metal shell acting as a heat sink. The claim that matters depends on where the connector is installed:
Inherently flame-retardant chemistries such as PPS and LCP avoid the halogenated additive packages older formulations relied on — which is also why a legacy insert formulation is a standing candidate for reformulation under restricted-substance rules. A reformulation is a material change, and a material change is a qualification event, not a paperwork event. Our obsolescence guide reaches the same conclusion from the procurement side.
Chemical-resistance tables are among the least trustworthy documents in the connector business, because a “resistant” rating is usually a room-temperature, short-duration, unstressed opinion. What destroys an insert is a fluid plus a temperature plus a stress: environmental stress cracking is a mechanical failure triggered by a chemical, and it happens well below any published attack threshold.
| Fluid family | What it does to a polymer | What to ask the compounder |
|---|---|---|
| Jet fuel / kerosene | Swelling and extraction of low-molecular-weight fractions; softened walls, reduced retention load | Volume swell and retention-load change after immersion at the service temperature, not at 23 °C |
| Phosphate-ester hydraulic fluid | Aggressive to many polar polymers; can attack gaskets and inserts at the same time | Whether the grade appears on an aviation hydraulic-fluid compatibility list at all |
| Mineral oils and greases | Generally mild swelling; the risk is the additive package rather than the base oil | Compatibility with the finished grease, including additives and thickener |
| Glycol de-icing and coolants | Absorption and hydrolysis at temperature; plasticisation and creep of thin walls | Hydrolytic stability under sustained immersion, not a splash test |
| Ketones and chlorinated solvents | Dissolution or stress cracking, especially in PEI; also a cleaning-agent hazard | An explicit attack list — and a check on what your own cleaning process uses |
| Strong alkalis and steam | Hydrolysis, especially in polyesters and some polyetherimides | Continuous temperature under alkaline or steam exposure, which is much lower than the dry rating |
The rule that survives all of this: for any critical fluid exposure a coupon test in the real fluid, at the real temperature, under the real stress beats any table — and the insert should be judged after immersion by retention load and dielectric withstand, not by appearance. A swollen insert that looks fine and has lost a third of its cavity grip is a field failure that has not happened yet.
The insert and the seal are chosen by different criteria and fail on different clocks, and the connector’s rating is the lower of the two. Silicone and fluorosilicone dominate because they hold elasticity from roughly −55 °C to +200 °C, a wider band than any insert needs. Fluorosilicone buys fuel and oil resistance at the cost of tear strength and price; plain silicone is better mechanically and worse chemically. Neoprene, still found in older and lower-cost connectors, is the usual reason a part is sold as a 200 °C connector while its grommet quietly limits it to about 100 °C.
The mechanism to watch is compression set: an elastomer held under compression at temperature gradually stops pushing back, the interfacial seal relaxes, and ingress begins without anything visibly cracking. That is why the insert is asked to be dimensionally stable in the same test — if the insert creeps, it takes the seal’s compression away with it, and the seal is blamed. Our shelf-life and elastomer ageing guide covers the storage half of the same problem.
This is the failure this article exists to prevent. “PPS” is a chemistry, not a grade. Two moulders can both buy “40 % glass-filled PPS” from two compounders and get materials with different filler size distribution, different coupling agents, different molecular weight and different crystallisation behaviour — and therefore different flow, different shrinkage, different weld-line strength and different cavity-wall toughness.
What that looks like in practice: the new insert moulds cleanly, every drawing dimension is in tolerance, a first-article inspection at room temperature passes, and the connector ships. Then mating force is a little higher in one quadrant, dielectric withstand margin is thinner than it was, and a re-termination cracks a cavity wall. Nothing in the inspection record explains it, because the thing that changed — weld-line strength in a 0.5 mm wall — is not a dimension.
The controls are organisational rather than technical:
Inserts are selected per family rather than per catalogue page, because the cavity geometry — not the polymer name — is what reliability actually depends on. The J599 Series III range uses high-density moulded inserts matched to the MIL-DTL-38999 shell and contact system; the ZH83723 bayonet and threaded families and the MIL-DTL-26500-class aerospace circulars each carry their own insert construction and temperature class; and the specialty and miniature families push the same materials into thinner walls and smaller pitches, where the weld line becomes the design driver.
Where a drawing names only a generic polymer, we will ask for the grade, the wall thickness and the fluid list before quoting — because those three inputs, not the acronym, decide whether the part still holds its contacts at the end of its life. The acceptance basis behind each construction, including the electrical tests every unit is screened against, is published on our quality page; the accessory and contact side of the same system sits in the M85049 backshell and M39029 contact ranges.
Specifying an insert by material name and need the grade-level data behind it?
Send the connector family, the shell size and arrangement, the working temperature, the fluid exposure and the altitude. We will come back with the insert material and grade we would build it in, the CTI, RTI and outgassing data for that grade, and the seal elastomer that sets the other end of your temperature range.
Email: info@airoadcon.com | WeChat: 18991927716