VG 96918 Connectors: The Souriau 847 & 848 Energy-Supply Series, Decoded

Most connector standards describe how to get a signal from one box to another. VG 96918 describes how to get energy from one vehicle to another — and the difference shows up in three places that a signal connector never has to think about: the earth contact has to connect before anything else, the connector has to be able to open under load, and the insulator has to hold 440 V while it is wet and dirty.

The standard is realised in Europe as the Souriau 847 and 848 series: a reverse-bayonet circular built for military ground equipment, trailers, shelters, artillery and heavy weapons, rated to 63 A and 440 V three-phase, sealed to IP68, and carrying a pilot contact so the vehicle knows the cable is seated before the contactor closes. It is one of the most specified power interfaces in European land systems and one of the least documented in English.

This guide covers what the family actually is, why each of its mechanical features exists, how the three layouts map to the standardised energy-supply classes, and — at the level most buyers need — how to read the order code field by field, including the three-digit segment that most listings print as an opaque number.

What VG 96918 is, and what it is not

The VG prefix marks a German defence equipment standard. The family splits cleanly by what it connects, and knowing the split is what stops a buyer ordering the wrong circular:

So VG 96918 is the odd one out: it does not inherit a US shell library. Its three housings are its own, its insert is built around a protective conductor, three phases and a pilot contact rather than around a contact count, and its ratings are expressed as classes of energy supply rather than as pin counts. The coupling, however, is inherited: the 847/848 uses a reverse bayonet derived from VG 95234, which is why the connector feels like a 5015 in the hand while being a completely different part.

Two series carry the same standard:

That table is the whole commercial argument for 848. Cadmium is a superb connector finish — sacrificial, low contact resistance, forgiving on threads — and it is also the finish that restricted-substance legislation has been pushing out of European programmes for two decades. The aluminium and zinc-nickel version exists because a programme could no longer write cadmium on the drawing, not because cadmium stopped working. When a legacy drawing still calls for 847, the justification has to be a real one — a cadmium-plated mating half, a repair of existing equipment, a temperature case — rather than habit.

The three layouts, and the energy classes behind them

The three numbers in the family — 25, 48 and 58 — are layouts, not shell diameters in millimetres and not contact counts. What each one carries:

Apparent power is arithmetic, not a catalogue figure: S = √3 × U × I for the three-phase layouts, S = U × I for layout 25. Real load depends on power factor.

Two things follow from this table that catch people out.

Layout 48 and layout 58 have the same contact set. The upgrade is not more pins; it is bigger contacts in the same insert, which is why 25 A and 63 A live in what looks like the same family. If you size a system by counting contacts you will conclude that 48 and 58 are interchangeable, and they are not.

The layouts correspond to the standardised energy-supply classes of VDE 0620-1 — single-phase 250 V / 16 A, three-phase 500 V / 25 A, and three-phase 500 V / 63 A. That is the point of the standard: a connector that claims to implement one of those classes can be plugged into any equipment built to the same class, across nations and manufacturers. It also means the class, not the connector, is what a programme should specify — and the connector is then chosen to serve it.

Note the collision of numbering schemes inside one datasheet: the electrical tables refer to shell sizes 2, 4 and 5 while the order codes use layouts 25, 48 and 58. They are the same three parts. This is a reliable source of quoting errors, and it is worth writing both numbers on the RFQ.

Why a power connector needs a pre-leading earth contact

On a vehicle, the chassis is both the return path and the reference for everything else. A connector shell is bonded to that chassis through the panel, the harness and sometimes a braid. If a phase contact made before the protective conductor, the shell and the cable screen would briefly sit at phase potential, and the only path back would be whatever incidental impedance the installation happens to offer — a bearing, a paint film, a fastener.

So the 847/848 sequences its contacts. The protective conductor is arranged to make first and break last, and the pilot contact is arranged to make last and break first. In service the order is: earth, then neutral, then phases, then pilot on connection; and the reverse on disconnection, with the pilot opening first so the equipment has already dropped the load before the power contacts separate.

Two practical consequences:

• The earth contact is a power contact, not a signal pin. It has to carry the full fault current for the duration of the protective device’s clearing time. A connector whose earth path is a shell-to-shell contact through a plated coupling ring is a different, and generally weaker, proposition.
• You cannot substitute a same-size signal circular. Two connectors can both be 63 A and IP68 and still differ on whether the earth leads, whether there is a pilot at all, and whether anything is rated to break. Those are the three questions to ask about a power interface, and none of them appear in a current rating.

Breaking capacity: the part of the rating nobody reads

A continuous current rating says how much current the contact pair can carry without exceeding its temperature rise. A breaking capacity says how much current it can interrupt. They are separate declarations, and a connector can be excellent at the first and useless at the second.

Alternating current helps. At 50 Hz the current crosses zero a hundred times a second, and an arc drawn between separating contacts extinguishes at the next crossing once the gap can hold off the recovery voltage. Direct current does not help at all: 500 VDC has no zero crossing, so the arc sustains until the gap is long enough and the plasma cool enough to stop conducting.

The energy involved is not small. At 63 A and 500 V, a 10 ms interruption deposits roughly 500 × 63 × 0.01 ≈ 315 J in the arc — into silver-plated brass contacts and a neoprene insert. On a battery-fed DC bus the source impedance is low, so the fault current is limited mostly by the cable, which means the available current at the connector can be very much higher than the load current. This is the reason the 848 is specified with breaking capacity as a headline feature rather than as a footnote: it is a connector designed to be pulled apart by someone who has to.

The material choices follow. Contacts are machined brass with silver plating rather than stamped and formed: there is more metal mass to absorb the arc, and more wall thickness where the arc root sits. Silver is chosen for the lowest film resistance at the contact interface — and it is worth remembering that silver tarnishes in sulphur-bearing air, which is why contact resistance drifts over years in an industrial or coastal environment rather than staying at its first-article value.

Insulation coordination at 440 V and pollution degree 3

The working voltage is 260/440 V three-phase, so the line-to-neutral stress is around 254 V — undramatic on its own. What makes it a design problem is the declared pollution degree 3 to DIN EN 60664-1 / VDE 0110-1: conductive contamination is expected. A connector on the outside of a trailer, under a shelter, or on a weapon carriage lives in condensation, dust, salt and road spray, and the surfaces between contacts are not clean.

Pollution degree is not a sealing claim. An IP68 connector can be pollution degree 3 at the same time, because the contamination that matters accumulates on the internal surfaces between the contacts — through the grommet holes, around the contact barrels, in the creepage path between phases — and because the mated interface is only sealed while it is mated.

The declared withstanding voltage is the margin that pays for all of this:

Hipot is a bench test, not a working condition, and the ratio between the two numbers is the allowance for contamination, altitude and transients. Take the actual creepage and clearance distances from the standard for pollution degree 3 at your altitude — they are a function of the material group, and the material group is a property of the grade of neoprene in the insert, not of the word “neoprene”. That is a question worth putting to a supplier in writing.

Reverse bayonet: why this coupling and not a thread

A reverse bayonet inverts the familiar arrangement: the coupling ring with the ramps sits on the half that would normally carry the exposed contacts, so the contacts are recessed behind the coupling and the ring itself takes the knocks. The 847/848 inherits this from VG 95234, and for a power connector on ground equipment the case for it is straightforward.

• It is faster to mate and to break. A quarter turn against several turns of thread matters when the person doing it is wearing gloves, in the dark, and needs the cable off now.
• It resists being walked off by vibration, because the ramps are detented rather than relying on friction alone.
• The coupling ring protects the interface from scooping damage when the two halves are brought together off-axis.

The trade is real too. A thread can be torqued, and torque is measurable, recordable and inspectable. A bayonet gives you no torque value: mating verification becomes visual (full engagement of the witness marks) plus a pull check, and the anti-decoupling feature has to be inspected rather than measured. On a power connector a partially engaged bayonet is worse than no connector at all, because it is a controlled arc gap with a human hand near it. If your work instructions rely on torque values to prove a joint, this family will not give them to you.

The pilot contact is an interlock, not a signal

The pilot contact is a low-current contact in the same insert as the phases. Its function is to tell the equipment that the connector is fully seated. Because it makes last and breaks first, a circuit closed through it can be trusted to mean “mated”, and the contactor that applies power can be interlocked on it.

This is the same idea that the electric vehicle industry later named HVIL, and the three design rules carry over unchanged:

• Supervise for both open and short. A simple series loop reads healthy when it is shorted, so a chafed pilot wire that shorts to the shell will hold the interlock closed while the connector is unmated. The detection circuit has to see both failure directions.
• Break-before-make has to be mechanical. The sequencing is a property of contact length in the insert. A software delay cannot substitute for it, because software is not what is pulling the connector apart.
• The pilot must drop the load, not light a lamp. If the pilot circuit only drives an indicator, the interlock is decoration.

For a designer, the important consequence is that the interlock requirement propagates all the way back into the order code: only the three-phase layouts carry a pilot, and only two of the four insert versions include it. That is a system decision made by one digit, and it cannot be recovered downstream because there is no contact to wire to.

The order code, field by field

The manufacturer’s ordering format is positional, and it is the single most useful thing to understand about this family:

84 [plating] [layout] [shell style] [contact type] [backshell] [orientation] [insert] [specification]

Take 84758F500N002 apart:

Now the same reading on 84748P147N302: series 847, layout 48, style P (a cable-connecting receptacle), contact type 1 (crimp pin), backshell 47 (straight, shield termination and heat-shrink sleeving), orientation N, insert 3 (grommet and pilot), specification 02. A 25 A three-phase cable receptacle, pin contacts, shielded, with a working interlock.

The trap is the three-digit segment. Many listings — including some distributor pages and some catalogue extracts — print the middle as a single opaque “code segment”: 100, 147, 500, 547. It is not opaque. It is contact type (one digit) followed by backshell (two digits), which is why exactly four combinations occur in this catalogue:

Read as one number, 500 tells a buyer nothing and two buyers can order “the same” part and receive the two halves of one mated pair. Read as two fields, it answers the two questions that actually matter: which half am I buying, and does it come with a backshell. A pin half mated to a pin half is a delivery dispute, not a connector.

A second trap sits in the same position: the letter after the segment is the orientation, not a finish code. The finish is carried by the two-digit specification suffix at the end — 02, 03, 09 or 08. Mistaking the letter for a finish is an expensive kind of wrong, because the finish decision changes the salt-spray life of the part while the orientation decision merely decides whether it will key into the receptacle on your panel.

Shell styles: which half of the pair you are buying

Two of those columns do the real work, and neither is obvious from a photograph.

Smooth holes versus threaded holes decides how the connector is fixed to the panel: a through-bolt with a nut behind it, or a screw into a threaded insert in the flange. On thin sheet metal the threaded flange is usually the wrong answer, because the insert becomes the load path and the insert is what strips.

“With thread for backfitting” decides whether a backshell can be screwed onto the rear of the receptacle later. Buy the without-thread version (A, L, R) and the backshell stops being an accessory and becomes an engineering problem: you are left with a smooth rear face, a cable to terminate, and no interface. This is the most common specification mistake in the family, and it is invisible until the harness is being built.

Style F is the plug and style P is a receptacle that terminates a cable — useful where the fixed side of the installation is a plug on a bulkhead and the cable side needs the receptacle geometry. Choosing between them is a question about which half your maintenance procedure expects to be replaceable.

Insert versions: grommet and pilot in one digit

The grommet is what makes a crimp-contact connector sealable at the rear: it is a resilient wafer with a hole per contact, the contact is pushed through it from behind with an insertion tool, and the hole closes around the wire. Without it the rear of the insert is open, which is fine if a backshell and a boot close the volume and unacceptable if they do not.

The pilot is the interlock contact described above. So insert version is not a cosmetic choice: it decides whether the connector can implement a mating-sense circuit at all.

A warning worth stating plainly: published availability tables for the four insert codes do not agree with each other, and the restrictions differ by layout. Some catalogue tables restrict code 0 to layouts 25 and 48, code 2 to layout 48 only, and code 3 to layouts 48 and 58; other listings, including some of our own product pages, carry combinations outside those restrictions. Treat the insert table above as the meaning of the codes and confirm availability for your specific layout against the current catalogue entry rather than against any summary, including this one.

Finishes, cadmium, and what the salt-spray number really says

The published figures are 48 hours of salt spray for the cadmium-plated steel shell and 500 hours for the black zinc nickel aluminium shell. A tenfold difference in a test that is already a harsh one, and it is not simply a claim that one plating house is better than another.

Cadmium protects steel sacrificially: it is consumed while it protects, so its life is a function of coating thickness and of how aggressive the environment is. Zinc nickel protects as a barrier with a secondary sacrificial role, and it holds up far longer in chloride environments — which is why it became the standard answer when cadmium had to go. What it does not do is match cadmium’s low, stable contact resistance on threads and on shell-to-shell interfaces, and that difference is why legacy drawings are sometimes genuinely hard to convert.

Three practical rules for specifying a finish in this family:

• Do not mix finishes across a mated pair and call it equivalent. A cadmium-plated shell mated to a zinc-nickel shell creates a galvanic couple at the coupling interface, and the couple is what corrodes. If the mating half is fixed by an existing installation, that constraint may be the reason to stay on 847.
• Aluminium shell plus steel panel plus salt means the fastener becomes the anode, not the shell. The finish on the screw and on the panel cutout matters as much as the finish on the connector.
• Ask what the 500 hours was run to. Salt-spray hours without a test method and a pass criterion are marketing. The useful answer names the method, the concentration and what constituted failure.

Sealing and the environmental envelope

The IP68 figure deserves the same scepticism as the salt-spray figure. It is declared for the mated connector, and it is a system property: grommet plus interfacial seal plus backshell gland. It survives only if the cable jacket outside diameter falls inside the gland range and the grommet holes match the contact size actually used. A connector sealed to IP68 on a cable whose jacket is too small is a funnel.

The endurance number is worth reading in the same spirit. Five hundred cycles is far more than a vehicle interface will ever see, which means the part that fails first in service is not the contact but the elastomer — the grommet taking a compression set, the interfacial seal hardening, the boot cracking. That is a shelf-life and environment question rather than a mating-count question.

Specifying a VG 96918 connector: the checklist

Items 3, 5 and 6 are the ones that turn into delivery disputes, because each of them is decided by a single digit that looks like housekeeping.

Where this maps to the Airoadcon catalogue

Our Souriau 847 series listing carries the family across all three housings and all eight styles, with the order code, insert version and heat-shrink molded part published per model — for example 84758F500N002 (layout 58, plug, socket contacts, no backshell, insert 0), 84748P147N302 (layout 48, cable receptacle, pin contacts, shield-termination backshell, insert 3 with grommet and pilot), 84725B500N102 (layout 25, square flange with backfitting thread, grommet, no pilot) and 84725D100N102 (layout 25, jam nut, pin contacts, grommet).

The cross-reference logic for this family is style plus size plus insert version — not size alone. Two parts with the same layout and different shell styles are not alternatives; they mount differently. And a part with the same size and style but insert 1 instead of insert 3 has no pilot contact, which changes the system rather than the connector.

Where the requirement sits next door, we would point you at the VG 95234 family for the 5015-dimensioned reverse bayonet that this coupling came from, the Souriau 8525 series for the hermetic, EN 3646-qualified case where the requirement is a pressure boundary rather than a power interface, and the ESC10 / ESC11 range for the aerospace engine side, which uses the same machined-crimp contact philosophy in a completely different shell. For the signal side of the same vehicle, MIL-DTL-5015 remains the volume answer.

If you are holding a drawing that says VG 96918 with no order code, send the layouts, the load class and the interlock requirement and we will confirm the code rather than quote from the drawing number.

Frequently asked questions

Is VG 96918 the same as MIL-DTL-5015?

No, and they do not intermate. The two share a coupling principle, not a shell system: the 847/848 reverse bayonet is derived from VG 95234, which is the German military rendition of the MIL-DTL-5015 dimensional family, so the locking action feels familiar. But 847/848 uses three energy-supply housings (25, 48, 58) with a contact set built around a protective conductor, phases and a pilot contact, while 5015 uses the 10SL to 36 shell series with its own insert library. Treat them as two different connectors that happen to share a coupling style.

What does the 847 to 848 change actually buy?

The shell material and the finish, and through them the corrosion budget. 847 is a steel shell with olive-green or white cadmium plating and a 48-hour salt-spray figure. 848 is an aluminium shell with black zinc nickel plating, a 500-hour salt-spray figure and RoHS conformity. That is a tenfold difference in salt-spray hours, which is the number that decides whether a connector on a vehicle exterior survives a coastal deployment.

Can I substitute an 847 for an 848?

They cross-reference at housing-size and style level, so mechanically the substitution looks trivial. Treat it as an engineering change rather than a line-item swap: the finish change also changes the shell alloy from steel to aluminium, which changes the galvanic couple with the mating half and with the panel, and it changes what the threaded inserts and the backshell interface are made of. Do a first-article check on the mated pair, not just on the part.

In the order code, what does the three-digit segment such as 500 or 147 actually mean?

It is two fields written without a separator: a one-digit contact type followed by a two-digit backshell type. 500 is contact type 5 (crimp socket) plus backshell 00 (none). 147 is contact type 1 (crimp pin) plus backshell 47 (straight, for shield termination and heat-shrink sleeving). 100 and 547 complete the four combinations that appear in this catalogue. Read as a single opaque number it tells you nothing; read as two fields it tells you which half of the mated pair you are buying and whether a backshell is included.

Which digit decides whether I can implement an interlock?

The insert digit. Insert versions 2 and 3 carry the pilot contact that a mating-sense interlock needs; 0 and 1 do not. Because the pilot is part of the three-phase insert, an interlock requirement also fixes the layout at 48 or 58 before you have chosen a shell style. Get this decision wrong and no amount of downstream wiring recovers it, because there is no contact to wire to.

Does a 63 A rating mean the connector can break 63 A?

No. Continuous current rating and breaking capacity are separate declarations, and a connector can carry 63 A happily while being rated to make and break nothing at all. On AC the arc is helped out by the current crossing zero; on a 500 V DC bus there is no zero crossing and the arc has to be stretched and cooled by the contact geometry. If the application disconnects under load, ask for the breaking capacity in writing rather than assuming the current rating covers it.

What crimps the contacts?

The M22520/1-01 crimp tool with the positioner matched to the contact size. The contacts are machined brass with silver plating, and the insert carries protective-conductor, phase and pilot contacts, so the contact list is not one part number repeated N times. Order the contact set against the insert, not against the shell.