IPEX3 Connector Selection For Compact RF Boards
When a wireless module has to fit into a product smaller than the palm of a hand, the board level antenna interface becomes one of the largest components on the board. A threaded connector is mechanically excellent and far too bulky, so the design moves to a micro coaxial receptacle that is soldered to the surface and mated with an ultra thin cable. The IPEX3 connector, also known as MHF3, is the common choice in that space, and it brings its own set of layout and process rules.
This article covers what matters when such a connector is specified and placed: the footprint, the impedance through the transition, the coplanarity that decides reflow yield, and the cable that has to go with it. The connector is only one element of the radio frequency path, and the path is what has to work.
What The Micro Connector Is For
A receptacle of this class measures roughly 2.0 by 1.3 millimetres with a mated height under a millimetre, so it can sit beside the radio chip with almost no impact on the mechanical envelope. The body is a metal shell with a gold plated contact, and the base is a moulded insulator. The part is intended for a cable that stays inside the product rather than for an antenna that a user removes and refits.
The intended use is an internal connection: a video transmitter board, a data link module, or a wireless module whose antenna is mounted on the enclosure. Where the user has to attach and detach an antenna frequently, a threaded connector at the enclosure wall is the better engineering choice, with a short cable running to the micro receptacle inside.
Footprint And Pad Geometry
The signal pad and the shell ground pads have a defined geometry, and the supplier drawing should be followed rather than approximated. The signal pad is narrow, and the ground pads surround it on two or three sides. Copying a footprint from a similar part with a different body size is a common error, because the pad positions differ by a fraction of a millimetre while the electrical consequence is a discontinuity at the launch.
The ground pads should connect to a ground plane with several vias each, placed as close to the pad as the geometry allows. Those vias are the return path for the signal as it leaves the cable, and their inductance is part of the transition. A single thin trace from the shell pad to a distant plane defeats the purpose of the ground pads and shows up as a rise in the reflection at the higher part of the band.
Characteristic Impedance Through The Transition
These receptacles are specified for a 50 ohm system and operate from DC to about 6 GHz, which covers the common wireless bands used by consumer modules. The connector itself is close to 50 ohms, but the transition from the coaxial structure to the microstrip on the board is where the impedance usually deviates. The launch has to be designed as part of the transmission line rather than as a pad that a trace happens to reach.
The practical approach is to keep the trace at the signal pad at the same width that the impedance calculation gives for the stack up, and to avoid a step or a stub between the pad and the trace. Any tuning that the supplier drawing shows is there to compensate for the pad capacitance, so it should be reproduced rather than simplified. The broader routing considerations for high frequency traces apply from the connector all the way to the matching network.
Coplanarity And Reflow Yield
Coplanarity is the flatness of the solderable surfaces and is quoted as a maximum deviation, often 0.10 millimetres for this class of part. If the terminations are not coplanar, one pad lifts off the paste during reflow and the joint forms with a gap. The joint may pass a continuity test and behave intermittently at radio frequency, which is the worst possible failure mode because it is difficult to reproduce.
Stencil design follows the drawing, and over printing is a real risk with a part this small. Excess paste can bridge the signal pad to the shell ground and short the radio frequency path, which produces a link that does not work at all. Inspecting the joints under magnification after the first production run, and checking the parts per million of bridges, is worth the time before the design goes to volume. Standard pad and stencil practice is described in pad design standards, and the placement considerations in placement and pad positioning apply to a part of this size.
Cable Selection And Link Budget
The cable that mates with the receptacle is a micro coaxial type, commonly the 0.81 millimetre variant, and its loss has to be included in the link budget. A short run of thin cable looks negligible on a schematic and can contribute a measurable fraction of a decibel per centimetre at the higher bands, which matters when the antenna gain is fixed and the receiver sensitivity is specified.
The mating height and the bend radius of the cable are mechanical constraints that are easy to overlook. A cable routed with a tight bend loses more than its specification suggests, and a bend right at the connector loads the mating interface. Route the cable with a gentle radius, anchor it so that vibration does not work the connector, and keep the total length as short as the mechanical layout allows.
Mechanical Life And Environment
A micro receptacle has a lower mating cycle rating than a threaded connector, because the contact is small and the retention depends on the shell geometry. It is a service-free internal connection rather than a user interface. Where the design requires repeated mating during development, the cable should be handled with a dedicated extraction tool rather than pulled by the cable itself.
The operating temperature range of these parts covers the usual industrial window, but the connector has no sealing. In a humid or salt laden environment the protection has to come from the enclosure rather than from the component, and any conformal coating or potting applied to the assembly must not enter the mating interface, since contamination there changes the impedance and the contact resistance. Generation compatibility is the last check: the different generations of this connector look alike and are not interchangeable, so the cable part number and the board receptacle must be confirmed as the same generation before either is ordered.
Additional Considerations for This Build
Practical attention to SMT receptacle pays for itself here, because it decides whether the finished board behaves as the drawing intended. Where the requirement is not stated on the fabrication drawing or in the assembly notes, the shop has to assume a default, and that default is rarely the value the design was simulated with. Stating SMT receptacle explicitly, together with the tolerance that applies, removes the assumption and keeps the result predictable from batch to batch.
Deliberate attention to ultra thin coax pays for itself here, because it decides whether the finished board behaves as the drawing intended. Where the requirement is not stated on the fabrication drawing or in the assembly notes, the shop has to assume a default, and that default is rarely the value the design was simulated with. Stating ultra thin coax explicitly, together with the tolerance that applies, removes the assumption and keeps the result predictable from batch to batch.
FAQ
Can the shell ground pads be connected with a single trace? Technically yes, but the inductance of that trace degrades the transition. Connect each pad to the plane with its own vias placed as close to the pad as possible.
Why does a connector that works on the bench fail after assembly? The usual cause is a joint that is not coplanar or a bridge between the signal pad and the shell. Both are process issues that appear only after reflow, so inspection under magnification is the reliable check.
Is this connector suitable for an antenna that the user connects? No. It is intended for an internal cable. Where the user attaches the antenna, use a threaded connector at the enclosure and a short internal cable to the board.
