Why do our radios buzz, hiss, or lose range near LED lighting and VFDs, and how do we fix it?

I get called to this problem more than almost any other, and it usually arrives described as a coverage failure: "the radios worked fine last year, now they're noisy and the range has collapsed in half the building." Nine times out of ten it is not a coverage failure at all. Nobody moved a repeater or lengthened a cable run. What changed is the electrical noise floor of the building, and once you understand that, the fix is straightforward and often cheap. Let me explain what is actually happening.

Your radio has two enemies: weak signal and loud noise

Reliable two-way radio voice needs a certain signal-to-noise ratio at the receiver, not just a certain signal strength. On a UHF DMR portable I design to a level of around −95 dBm at the handset, and the receiver's own sensitivity floor sits near −119 dBm for 12 dB SINAD. That roughly 24 dB gap is your working headroom. It only holds if the ambient radio noise in the band is close to the receiver's internal floor.

LED drivers, variable-frequency drives (VFDs), escalator and lift motor controllers, and cheap switch-mode power supplies all raise that ambient floor. They use fast switching — tens of kHz up into the MHz — and every switching edge throws off harmonics that smear energy right across 400–470 MHz. When that broadband hash lands on your receiver, it lifts the noise floor from, say, −119 dBm to −105 or −100 dBm. You have just thrown away 15 to 19 dB of your headroom without touching the transmitter. A signal that was comfortably readable at −90 dBm is now sitting almost on top of the noise, so you hear the hiss, the buzz, the flutter, and the "range" appears to have halved. The signal did not weaken. The noise got loud.

Why it seems to appear overnight

This is why a system that passed acceptance testing degrades a year later with no RF changes. A tenant refits with hundreds of cheap LED panels. Facilities swaps the escalator drive for a modern VFD unit. A new chiller plant goes in with unfiltered inverters. None of it touches your radio system, and all of it raises your noise floor. The interference is also loudest exactly where the offending equipment lives — plant rooms, escalator wells, retail units, carpark lighting — which is why the dead zones feel random and localised rather than following the coverage map.

How I diagnose it on site

I do not guess. I put a spectrum analyser or a service monitor on the band and look at the noise floor with the suspect equipment switched off, then on. If the floor jumps 10–15 dB when the escalator VFD or a lighting circuit energises, that is your culprit, and the trace usually shows the characteristic comb of switching harmonics. I also walk the affected area with a portable on a known channel and watch the received signal strength and SINAD, because a desense problem shows a healthy RSSI with poor audio quality — that combination is the fingerprint. Genuine coverage holes show low RSSI and poor audio together; desense shows adequate RSSI and poor audio. Distinguishing the two decides whether you need filtering or more antennas, and the two remedies are completely different.

What actually fixes it

The order of attack is: kill the noise at source, then protect the receiver, and only then touch the RF system.

At source, the biggest wins come from replacing non-compliant emitters. Reputable LED fittings certified to EN 55015 and drivers meeting EN 61000 emission limits are dramatically quieter than the unbranded product that usually causes this. Where you cannot replace, ferrite chokes on the DC leads and lighting flex, proper bonding and screening of VFD motor cables, and the shielded cable and line filters the drive manufacturer already specifies (and installers routinely skip) recover most of the lost margin. A VFD installed to its own EMC instructions is rarely a radio problem; one installed without the screened cable and filter almost always is.

At the receiver, physical separation matters more than people expect — moving a DAS antenna two or three metres away from an escalator drive or a lighting transformer can recover several dB, because near-field noise falls off fast. Band-pass filtering on the repeater or BDA receive path rejects out-of-band energy, though it does nothing for in-band hash. If the noise sits right in your channel, only fixing the source works.

Only after that would I look at the RF system itself, and usually it does not need changing. Adding antennas to fight a noise problem is throwing money at the wrong layer — you raise the signal, the noise rises with it in the same area, and the ratio barely improves.

One regulatory note for Singapore: if the interference is severe and traces to equipment radiating well beyond permitted limits, that is an IMDA matter, and equipment sold here is meant to meet emission standards. Documented spectrum evidence is what turns "our radios are noisy" into an actionable complaint.

How Suneast helps

At Suneast we treat noise-floor problems as an engineering diagnosis, not a sales opportunity to sell more radios. We carry the test gear to measure your actual band noise with the suspect plant on and off, identify the offending LED, VFD or motor-drive source, and give you a written remediation plan — source filtering, antenna relocation, or receiver band-pass filtering — with the margin numbers to back it. Where a genuine coverage gap is found alongside the interference, we design the in-building DAS or repeater fix to the same standard we apply on every Singapore building. The goal is to restore the headroom you paid for, at the lowest-cost layer that solves it.