Enter for the full results page · Esc to close
Multimode reach is not a loss calculation — it is a modal-bandwidth limit fixed by the standard. Pick the standard and the fibre grade and this gives you the specified distance, the channel loss budget and the connector headroom that distance allows.
Multimode reach is a bandwidth limit, not a loss limit. This is the thing that catches people. On single-mode you can trade connectors for distance, because the limit is the power budget. On multimode the limit is modal bandwidth — different light paths through the core arrive at different times, and past a certain distance the pulses overlap no matter how clean your connectors are. That is why the standards publish a distance in metres rather than a budget in decibels, and why 100 m is 100 m.
What OM3, OM4 and OM5 actually differ by. OM3 is specified at 2000 MHz·km effective modal bandwidth at 850 nm, OM4 at 4700. That is the whole difference and it is why OM4 goes further. OM5 has the same 850 nm bandwidth as OM4 — it does not go further on a single-wavelength standard — but it is additionally specified out to 953 nm, so it supports short-wavelength WDM standards that put four wavelengths down one fibre pair. If you are not using SWDM or BiDi, OM5 buys you nothing over OM4.
Connector count is the other half. Every mated pair costs roughly 0.5 dB on multimode, and the budgets are tight — 1.8 dB for 400GBASE-SR8. Three mated pairs is 1.5 dB, leaving almost nothing for the fibre. This is exactly why parallel-optics data centres favour a single trunk with a cassette or harness at each end rather than several patch panels in series.
When to stop and go single-mode. If the link is over about 100 m and you are at 100G or above, single-mode parallel optics are usually the better answer, and often the cheaper one once you count the structured cabling. DR4 reaches 500 m and FR4 reaches 2 km on ordinary G.652.D. Check the standard lookup for what each variant needs.
Not on a standard single-wavelength SR link — both are 100 m for 100GBASE-SR4 and both are 100 m for 400GBASE-SR8. OM5's advantage appears only on wideband standards: 100G SWDM4 goes 150 m on OM5 against 100 m on OM4, and 400GBASE-SR4.2 BiDi likewise. Buy OM5 because you intend to use SWDM or BiDi, not because you want more metres.
Physically yes, and the link will usually work — but the channel is then limited by its worst segment. A 2 m OM3 cord at each end of a 90 m OM4 trunk barely matters; 20 m of OM3 in a 100 m channel does. If the design depends on OM4 distances, keep the whole channel OM4.
Because multimode is a short-reach technology using cheap VCSELs with modest launch power, and because the standard reserves part of the budget for modal and chromatic penalties rather than for your connectors. 1.8 dB across a whole channel is what you get at 400G — which is three or four mated pairs and nothing else.
For 1G, yes — 275 m. For 10G it manages 33 m. For anything at 25G or above it is not specified at all, and no amount of good practice will change that. Legacy 62.5/125 plant is a replacement project, not an upgrade path.
Pick a standard and get the wavelength plan, fibre type, connector, reach and planning loss budget
Open tool →Total channel loss, power budget and pass/fail margin for single-mode and multimode links
Open tool →Turn a switch port count into the trunks, cassettes and harnesses you actually have to buy
Open tool →Transceivers, MPO assemblies, WDM and patch cords — built and tested to the figures you just calculated.