Q&A: Dealing With Wind

This post is part of a series based on questions received from readers and colleagues.

Questions may be edited for clarity, general applicability, or to protect the anonymity of the individual.

If you would like to ask a question, please email me via the Contact page.

Q: When EQ’ing my mains, I did have some trouble getting good coherence in the higher frequencies due to wind. How do you combat this when working outdoors? Would adjusting the averaging parameters help this, or is it just an occupational hazard of being outside?

I’ll start with a bit of the “why” and then we will talk about the “how.” Without taking too much of a tangent, it’s worth noting that the term “transfer function” has a more rigorous definition than we might realize. We use it to measure the tonality, timing, and phase of our sound systems, but in a stricter sense, a transfer function is a mathematical function that allows us to say what the output of a system would be, for a given input. It assumes that the relationship between input and output is fixed (fader goes up on the console, system output level goes up in the room) and time invariant (the time it takes signals to move through the system is unchanging). We can combine these two requirements together into one: LTI, or Linear Time-Invariant.

Basically, a transfer function measurement helps us study the relationship between input and output of a sound system, but the data produced by the analyzer is only fully valid (from a mathematical perspective) if that relationship isn’t changing over time. If the response of the system is not LTI, then the transfer function data no longer can give us everything we need to completely describe that system’s response.

Now let’s translate that into practical terms: When we introduce wind into the equation, we have a constantly changing set of conditions between loudspeaker and measurement microphone, which is part of the entire system response we’re measuring, and the analyzer is going to (accurately and correctly) report this changing system response with a trace that just won’t settle down.

Wind will change both the magnitude and phase traces, because it changes both the level and timing of the energy arriving at the microphone (it violates both the Linear and the Time-Invariant requirements). And what’s more, that means the measurement and reference signals are not observing a solid relationship over time (the relationship between them keeps changing) which means the coherence measurement is going to tank as well. That’s essentially the coherence trace’s job: “hey, you keep asking me to consider the relationship between these two signals, and every time I run the numbers, I’m getting a different answer.” Low coherence.

If your analyzer uses coherence blanking, it is going to blank out portions of the trace whose coherence data falls below a certain threshold. You can disable the coherence blanking function or set the threshold to 0 to get the trace back, but that still doesn’t solve the problem. You can also try locking your measurement delay finder on, telling the analyzer to continuously adjust measurement delay, which may stabilize things in a light breeze, but in a heavier wind, that’s also not likely to do much for you. A modern realtime dual-channel analyzer only needs a second or two to generate a stable trace, so if wind is intermittent, just “shoot the gaps”, but in a constant breeze, we’re probably out of luck.

Ultimately, this is not an issue with the measurement, it’s an issue with reality, so we can’t be too frustrated with the analyzer. Odds are, the bulk of the disruption will be limited to the top couple octaves, where periods are short and even small differences in arrival time correlate to rather large fractions of a cycle. In the lower half of the spectrum, we can expect a lot more stability as long as we have done a good job of eliminating wind noise from the measurement (you are using a high-quality windscreen, right?). In practical terms, the bottom half of the spectrum is really the first thing I need to focus on with a mains measurement, since that’s where I typically end up doing most of the work to hit whatever target curve we’re going for in a certain circumstance. If you’re working with a line array, the manufacturer’s prediction ecosystem probably offered a starting point for HF filtering, which is likely a whole lot better than nothing at all in the top half of the spectrum.

And let’s not forget, even in optimal circumstances, once we’re done our initial measurement pass, we’re going to move to listening as the next step anyways, so the most likely outcome of a very windy tuning session is that we move to the listening stage more quickly and are further away from where we want to be when we start, so might have to do a little more massaging than usual at this stage - but that’s the job, isn’t it? If winds are very high I may dispense with the measurement step entirely and move straight to music from my tuning playlist.

As a realist: how much time and energy do we want to devote to the top octave or two that’s just getting absolutely obliterated by winds? Concentrate your efforts on the stable portion of the spectrum since that’s really all anyone (FOH mixer or audience member) is going to be able to make sense of, anyway, and just be ready to tweak things as you go - as always.