An OB van is a broadcast facility that has to work wherever it parks, and its network has to do something studio infrastructure never does: deliver guaranteed-feeling connectivity from a different unknown location every week. I have written elsewhere about the bonding fundamentals for broadcast feeds; this piece is about the van itself, and how we architect the vehicle as a system.
From SNG to IP
Traditional outside broadcast was built on satellite news gathering and leased lines: reliable, and priced accordingly, with rigid booking windows and one-way transport. The industry's move to IP contribution changes the economics and unlocks remote production, but it swaps a guaranteed circuit for the public internet, which is best-effort by design. The engineering job is to assemble guaranteed-feeling transport out of non-guaranteed parts.
What makes modern OB harder than the SNG era is that the link is no longer one-way. Remote production needs bi-directional connectivity for tally, intercom, camera control and return video, all of it latency-sensitive, all of it running alongside the contribution feed. A one-way satellite booking never had to solve that; the van's IP architecture does.
The transport layer, briefly
The broadcast is won or lost in transport, not in the encoder. SpeedFusion bonds every available path into one packet-level tunnel, so a carrier dropping out costs capacity rather than the session; WAN Smoothing duplicates packets across paths to hold jitter down for RTP and UDP streams that cannot tolerate retransmission; FEC reconstructs lost packets at the receiver without a round trip. Budget 10 to 25 percent overhead for those protections depending on how hostile the environment is, and tune FEC and smoothing per location rather than shipping one static profile; a rural single-mast site and a saturated stadium need different settings from the same van.
The hybrid WAN mix
The strongest van architectures in 2026 mix three transport families. Multi-carrier cellular is the workhorse: SIMs across EE, Vodafone, O2 and Three so one operator's congested mast costs you a contributor, not the feed. LEO satellite, Starlink in practice, adds capacity that does not care about the stadium effect, though it brings its own personality: brief interruptions during satellite handovers and sensitivity to obstruction, which is precisely why it belongs inside the bonded tunnel rather than alone. And where the venue offers fixed-line or venue fibre, we pull it into the bond too; it is usually the lowest-latency path on site.
The stadium effect deserves its own paragraph because it is the failure mode that catches crews out. Coverage checked clean on the recce; then 40,000 phones arrive and uplink capacity evaporates. The defences are carrier diversity, external high-gain MIMO antennas mounted clear of the van's bodywork, and satellite capacity that congestion cannot touch. If your bandwidth plan assumes rig-day RF conditions, it is wrong by kick-off.
Design principles for the van build
Treat the antenna system as broadcast infrastructure, not an accessory: roof-mounted MIMO arrays with short, low-loss feeder runs, because every metre of cheap coax is signal you paid for and threw away. Keep the routing and modem hardware serviceable mid-event; racks get hot in a van in July, and thermal shutdown during a live is an avoidable failure. Separate production traffic from crew connectivity with VLANs so a runner's phone backup cannot compete with the programme feed. And design the headend properly: the van dials out to a FusionHub or data-centre endpoint with a fixed address, which means ordinary SIMs in the field and one controlled hand-off point into the gallery.
Fleet visibility ties it together. InControl 2 gives an engineering manager every van on one screen: per-modem signal, data consumption per SIM, GPS position, and the ability to reconfigure a unit that is three hundred miles away. On multi-van operations, the historical telemetry is quietly valuable; it tells you which venues, and which corners of which venues, have RF problems before you park there again.
Scoping a van properly
We scope from the programme backwards: contribution bitrate and codec, return and comms requirements, worst-case venue RF, then the transport mix and hardware to carry it with headroom. That order matters, because a van specified from a hardware catalogue tends to be perfect for the demo and undersized for the derby. We design and build these architectures, configure the SpeedFusion topology end to end, and train broadcast engineers to own them, because the crew that understands its own transport layer is the crew that stays calm at thirty seconds to air.
The short version
Mix carrier-diverse cellular with LEO and venue fibre, bond it all into one tunnel, engineer the antennas and thermals like they matter, keep production traffic segregated, and watch the whole fleet from one pane of glass. Do that and the van stops being the risky part of the production chain. If you are building or refitting an OB vehicle, get in touch for a scoping conversation.
Frequently asked questions
Can an IP-based van really replace an SNG truck?
For most productions, yes, and at a fraction of the transport cost. The exceptions are venues with no usable cellular or sky view at all, which is rarer every year and mitigated further by carrying LEO capacity in the bond.
Is Starlink reliable enough for live contribution?
Inside a bonded tunnel, yes; it contributes serious capacity and ignores cellular congestion. On its own, its handover behaviour and obstruction sensitivity make it a gamble no broadcast engineer should take with a live feed.
How much bandwidth headroom should a van carry?
Enough that the bonded capacity comfortably exceeds peak contribution plus comms plus the 10 to 25 percent protection overhead, in the worst RF you expect, not the best. If the numbers only work on the recce-day signal report, they do not work.
What about camera control and intercom over the same link?
That bi-directional traffic is exactly why the tunnel architecture matters: it needs a persistent session and low jitter in both directions. We prioritise comms and control traffic within the tunnel so a bandwidth spike in the programme feed cannot starve the intercom.
Can multiple vans share one architecture?
Yes, and they should. A common headend, a common configuration template and fleet-wide InControl 2 visibility mean any engineer can walk into any van and know exactly how it behaves.