KitGuru plans to cut through the subjective with noise and try to establish a baseline for all of the future work we do in the Lab. This article will lay down the foundation for what we believe is true. As ever, we will listen to you, The Reader, as well as the manufacturers before making any changes to our belief system. When KitGuru gives you a definitive dBA level for a product – what exactly do we mean?
Simple words that are amazingly hard to pin down.
Also, the perception angle is tough. Everyone knows that youngsters have the best hearing, so why is it they listen to pile-driving music all day long, while the near-deaf older folk we know & love – complain bitterly about each and every noise in the world?
We then move onto things like ear size and frequency sensitivity. In general, a larger ear should channel more sound to your receptors. Which means that everyone has a few decibels different from their neighbour in terms of how they hear certain sounds at specific frequencies. In KitGuru's youth, we were playing with friends in a studio and managed to knock the drummer clear off his chair with a well-placed, ultra-high guitar note coming out of the cranked Marshall Stack next to the poor guy's head. His ears were slightly larger than normal and we hit the right note on the Strat. Very funny.
Then there is the testing methodology.
For those folk that believe noise from a graphic card should be measured with the PC on the desk, side open and the test instrument a few inches from the fan – would you also suggest that the noise from a rifle should be measured by sticking your tester down the end of the barrel? Doesn't sound like the smartest method to us.
KitGuru’s Basic Principals
Most desktop PCs are towers and most of these are located under a desk. Gamers/modders might love the artwork they have commissioned so much that they want the tower on top of their desks – but in most cases they will be blowing things up in a first person shooter with either headphones on or huge speakers blasting away – so we’re not going to worry about them too much. That means we’re testing noise levels for people with tower PCs under their desks.
Distance & Measuring Point
You are going to be a metre away from most of the fans – with the side panel and part of a desk between you and the source of the noise. We’re going with 1m away from the front of the PC with the meter lined up dead centre.
The Digital Sound Level Noise Decibel Meter Style 2 is a really useful piece of kit. Affordable, with a resolution of 0.1db when measuring from 30 to 130 dBA. It’s commonly used to check if a work place conforms with regulations and to examine the acoustic properties of auditoriums, studios and home theatres.
We will say things like 37dBA and mean anywhere from 36.5 to 37.4 dBA. The idea is not to be precise, but to give readers an idea of where a given product sits in comparison to the rest of the world that you experience around you. Worth noting how we will use language to describe noise:-
No moving parts. Let’s not mess around, if it makes any noise, then it’s not silent.
Doesn’t add anything to the overall noise from a PC. If it’s quieter than a small office, then you’re unlikely to hear it. Putting a quiet component into a silent media centre will definitely add to the noise.
A decent rig might have a £100-200 CPU clocked up by several hundred MHz. The cooler on that CPU will need to move more air and you’re also likely to have a more powerful graphics card. All of these will add some noise. Components that are ‘average’ will not add extra noise to this kind of rig. Components at the low-end of this range will be almost inaudible in a small office. Toward the high-end of the range, you'd just about hear them in a large office – if you were standing near the PC.
Even in an overclocked rig, THIS is the component that you most notice. For most of us, it will be the cheap and nasty sub £15 DVD burner that powers up to full speed when we boot. Components which operate at this kind of level all the time (GTX480 owners – you know what we mean) are noisy. Doesn’t mean ‘do not buy them’, because that component might be crucial to your needs. Just be aware that it’s noisy and not likely to be getting better.
Jokes aside, if a component hits over the noise generated by cars whizzing by on a motorway or a hairdryer at the side of your head – then you need to believe it will somehow save your life before you start to use it. KitGuru cannot imagine a piece of computer kit in this zone, but we'll definitely report on it if we find it!
We will use a simple graph. The product being tested will be lined up alongside other numbers in a similar range so you can get a picture of what to expect. We’re not guaranteeing the result in any way, simply trying to give you some guidance. If you really want to know how loud something is before buying it, then pop into a PC specialist store like YOYOTech and ask them if they have a demo model fired up so you can see/hear for yourself.
The Reference Platform
We have built a passively cooled system into a Lian-Li chassis. No case fans. Intel SSD for boot etc and a heat-pipe based PSU. The mainboard also has no fans. When we say you cannot tell if it is on or off, then we mean middle of a moonlit night with your head sellotaped to the side of the chassis can't tell. It isn't quite DIN45635, but it's as close as we can get without entering a psychologically dangerous place.
We will fire the KitGuru Reference Platform up with the new component in place. Once booted, we will get it to loop on a full set of 3DMark Vantage. After 10 minutes, we will make our measurement 1m from the front of the KitGuru Reference Platform and repeat the test 5 minutes later. As long as the 2 numbers are very close, we will record the result and add the new component to our database.
That's enough background. Here's the chart that we will be using to show you where a given component sits. It will change with time, and we're open to suggestions, but this is the starting point:-
KitGuru says: We're not normally bothered by the noise that a component generates, but that's just how life is when you're testing in a lab. Cases are rarely shut and there are so many tests running simultaneously that the National Grid must be rubbing its collective hands at the size of KitGuru's electric bill. But it is important that you know what noise a component generates – especially if you have a specific use in mind. Home Theatre PC (HTPC) is the most obvious example, but there are other cases where you need a machine to be quiet. In a recording studio, for example. We'll deliver the numbers, you decide what's reasonable.
Let us know your thoughts over here in the KitGuru forum.