Home / FAQ
FAQ
The questions that land in our inbox week after week, sorted by what they are really asking.
01 / Choosing an interface
Which one, and why.
-
help_outlineAre U.FL and IPEX MHF the same thing?
Close enough to fool you, different enough to bite. U.FL is Hirose's trademark on a 2.0 mm connector. I-PEX MHF I is the same footprint wearing another badge, and the two mate without complaint, as do Amphenol AMC and Molex UMCC. Grief starts further down the family: MHF II, III and 4 shrink step by step and turn their backs on U.FL. MHF 4 keeps the family name and nothing besides.
Our cross-reference grades every pairing fully interchangeable, conditional or do-not-mix, built from parts we mated on a bench rather than from datasheet promises.
-
help_outlineMMCX or MCX - what separates them?
MMCX runs around 45% smaller with a snap-on detent you feel seat, which wins it the job whenever board area is the thing squeezing you. MCX is bigger, more forgiving of rough handling, standardized under CECC 22000, and the one of the pair you find in 75 Ω without going on safari.
Both reach 6 GHz and around 500 mating cycles. With space and impedance out of the argument, the tiebreaker is whatever the rest of the design already stocks.
-
help_outlineHow do I know if a design needs 50 or 75 Ω?
Wireless runs 50 Ω top to bottom: Wi-Fi, cellular, GPS, LoRa, every box on your bench. 75 Ω belongs to video and broadcast. DVB tuners, CATV, SDI. A module datasheet that stays quiet on the subject means 50 Ω.
Here is the trap. 50 and 75 Ω MCX parts mate physically and look identical sitting in the tray. Mix them and you have bought return loss that eats an afternoon before you find it, which is why 75 Ω parts wear a colored band.
-
help_outlineWhat is the lowest stack height currently achievable?
MHF 4 and 4L get down to 1.2 mm mated on 0.81 mm coax. That is the floor in a production connector as things stand. MHF III sits at 1.6 mm, U.FL at 2.5 mm.
Before you sign up for 1.2 mm, stare at the cycle budget for a minute. Thirty insertions is the rating, and nobody drew these parts expecting them to be unplugged in service.
02 / Handling & assembly
Getting them on and off.
-
buildWhy do ultra-miniature connectors have such low cycle ratings?
They hold on by interference fit. No latch, no thread, no separate retention feature anywhere: the shell grips the receptacle ring directly, and each insertion works that grip a little looser. Thirty cycles is where most manufacturers stop promising a retention force.
Nobody reads this number. A jumper pulled once at manufacturing test and again during rework has already spent a real slice of its life.
-
buildWhy does a snap-on joint fail if it is pulled sideways?
The detent was drawn to release along the axis. Pull at an angle and you lever the plug into the receptacle wall, deforming the collar and the ground path instead of letting go clean. It reseats afterwards, looks perfect, but has surrendered half its retention force.
Same family of damage: popping a connector off by hauling on the cable, which dumps the whole load into the termination.
-
buildDoes torque really matter that much on SMA?
Yes, and in both directions. Leave it loose, and the joint walks around with temperature. Crank it down and the dielectric deforms for good, so the part underperforms every time you touch it afterwards.
0.6 N·m is the figure. SMA is the only interface in this reference carrying a defined torque at all, which is exactly why it holds its numbers at 18 GHz.
-
buildWhy does bend radius keep coming up?
Micro-coax has almost no mechanical margin to spend. Bend it hard near the termination and you distort the dielectric, shove local impedance around, and funnel every joule of vibration energy into the weakest spot on the assembly.
Give it a service loop, support the run. A few millimeters of space buys you most of the mechanical failure modes we write about, gone.
03 / Performance & measurement
Reading the numbers.
-
show_chartWhy do two datasheets disagree about the same interface?
Each one measured on its own fixture, with its own calibration, at whichever frequency flatters the part, on a sample somebody picked by hand. Nothing dishonest in any of that. Nothing comparable across vendors either.
We run controlled comparisons wherever published figures fight each other, and we print the sample size and the method every time.
-
show_chartIs the mean insertion loss the number I should design against?
Not if you ship volume. Means quote beautifully and build badly, because a production run lives out in the tail of the distribution and never at its center. Where we know the spread, design against worst case.
The gap opens up high. In our U.FL against MHF 4 comparison, the means sat 0.21 dB apart at 8 GHz while the worst cases sat 0.27 dB apart.
-
show_chartHow much does an adapter really cost me?
Roughly 0.05 dB for a well-built in-series SMA bulkhead, north of 0.3 dB for a pigtail transition, both at 6 GHz. On pigtails the cable does most of the damage while the connectors collect the blame.
The transition matrix tabulates sixteen combinations with measured figures. We would argue the mechanical bill hurts more: a second joint working itself loose while nobody watches.
-
show_chartWhat changes over temperature?
The cable, mostly. Dielectric properties shift as temperature moves, and repeated cycling leaves small permanent changes in loss once materials settle where they wanted to be all along. Thin micro-coax drifts further than RG-316.
Connector side, the classic is a threaded joint nobody torqued properly, backing off as things expand and contract.
Not answered here?
Ask, and it turns into a page.
Repeat questions become reference pages around here. Send the module datasheet, the constraint boxing you in, or the pairing you cannot find written down anywhere.