Glass · Acoustics
Why the two panes in an acoustic unit must not match
Every pane goes acoustically transparent at one frequency, and that frequency is set by its thickness. Two identical panes therefore fail at exactly the same place, and the unit gets one deep hole instead of two shallow ones. Making them different costs nothing in glass and is the first thing to fix.
The two things going wrong, and they are not the same thing
Almost every confused conversation about acoustic glazing is these two effects being mistaken for each other. They sit at opposite ends of the range and they are fixed by opposite means.
The coincidence dip, high up. A pane of glass carries bending waves. At one frequency the bending wave in the glass travels at the same speed as the sound wave arriving through the air, the two stay in step, and the pane hands on almost everything it receives — it radiates about as much as it takes in. Reduction collapses across a band around that frequency. Thickness sets where it lands: stiffer, thicker glass has its coincidence lower down.
| Float glass | Coincidence | Which matters because |
|---|---|---|
| 4 mm | 3,200 Hz | above most speech and traffic, so it is the least harmful place for it |
| 6 mm | 2,100 Hz | consonants, and the edge of what makes speech intelligible through a wall |
| 8 mm | 1,600 Hz | squarely in speech |
| 10 mm | 1,250 Hz | speech and the top of traffic noise |
| 12 mm | 1,050 Hz | the thicker the glass, the worse-placed its weak point |
Approximate: near enough to 12,800 divided by the thickness in millimetres. Take the tested curve for the actual product from the maker rather than these figures — they are here to show the direction, which is the part that decides the specification.
The mass-spring-mass resonance, low down. The two panes are two masses and the trapped gas between them is a spring. Like any such system it has a resonance, and at that frequency the unit performs worse than a single sheet of the same total glass. For an ordinary 4/16/4 it sits somewhere in the low hundreds of hertz — which is exactly where a lorry is. Widening the cavity softens the spring and pushes the resonance down out of the way. Thickening the glass makes the masses heavier, which also pushes it down.
The four decisions, in the order they are worth making
Ordered by what they buy against what they cost, which is not the order people usually try them in.
-
Make the two panes different — this one is free
4 mm against 6 mm rather than 4 against 4. 6 against a 6.8 mm laminate rather than 6 against 6. The two coincidence dips move apart, and where one pane has gone transparent the other is still doing its job. The glass costs the same, the unit weighs about the same, and nothing else in the specification changes.
Do this before anything else. A symmetrical acoustic unit is a specification mistake, not a budget choice.
-
Make one of them an acoustic laminate
A laminate is two plies bonded to an interlayer, and the interlayer shears as the pane bends. That shearing turns bending energy into heat, which is precisely what damps the coincidence dip — the pane no longer has a clean resonance to hand the sound over at. Acoustic-grade PVB is formulated softer than the standard structural kind for exactly this, so ask for it by name rather than assuming any laminate does it.
It is the largest single improvement available in a sealed unit, and it brings the safety and the security of laminated glass with it.
The biggest gain per euro, and the one that also does something for you when somebody puts a brick through it.
-
Take the widest cavity the frame will hold
Straight at the mass-spring-mass resonance. Wider cavity, softer spring, lower resonance, out of the range where traffic sits. Note that the best cavity for sound and the best for heat are not the same number: thermal performance peaks around 16 mm of argon and gets slightly worse beyond it as convection starts in the gap, while acoustic performance keeps improving. If both matter you are trading, and it is worth knowing you are trading rather than assuming one number serves both.
Real, and free in glass terms — but the frame has to accept the unit, so it is a decision that has to happen early.
-
Then stop looking at the glass
The frame, the seals, the reveal and any trickle vent decide more of the finished result than the last few decibels of glass do. Sound follows the leak: a small gap around a frame will undo a unit you paid a great deal for, and a vent is a designed hole. The installed performance of an opening is not the tested performance of the pane in it, and the gap between them is where the complaints come from.
Where the money is usually lost. A superb unit in a mediocre opening is a mediocre window.
The question people actually ask
Same total glass, same cavity, three ways of arranging it. This is the comparison that comes up whenever somebody is specifying laminated both sides.
| Make-up | What it does |
|---|---|
| 3+3 / 15 / 3+3 | Symmetrical. Both leaves are the same laminate, so both dips are in the same place — the fault this whole page is about. The interlayers damp it, so it is far from a disaster, but it is leaving the free improvement on the table. |
| 3+4 / 15 / 3+4 | Each laminate is internally asymmetric, which is worth having, but the two leaves are still identical to each other. It fixes the smaller of the two symmetries and not the one that governs. |
| 3+3 / 15 / 4+4 | The two leaves differ. Different masses, different stiffnesses, dips in different places, and both laminated so both are damped. Of the three, this is the one to build, and it costs no more than the first. |
If only one leaf can be laminated, put it on the noisy side and make the other a different thickness of monolithic. And whatever the arrangement, ask for the tested figure for that exact make-up in that exact cavity: acoustic performance is not something you can add up from the components, which is why every manufacturer publishes tested curves rather than a formula.
The number on the brochure is not the number you get
Acoustic glass is sold on Rw, a single figure condensed from the whole tested curve. Underneath EN ISO 717-1 there are two corrections that reweight that curve for what you are actually listening to: C for higher-frequency sources such as voices, and Ctr for low-frequency ones — road traffic, and the bass through a wall.
They are always negative, and Ctr is the larger of the two. The same unit can be quoted at Rw 38 dB and deliver Rw+Ctr of 32. Neither number is dishonest; they answer different questions. But a window facing a road is a Ctr problem, and Ctr is the number a brochure is least likely to put in the headline.