A small active speaker is a Fundamentally Good Idea for a small listening triangle and small rooms. It’s equally right for mix engineers, media producers, gamers, home workers and audiophiles – with one caveat . . . .
You might reasonably expect (demand, even) that speakers primarily designed for near-field listening should have low distortion at short range. Drivers should integrate seamlessly at a distance of 1m or less. Imaging and detail projection should behave like headphones not PA speakers. As work tools, they should be an accurate window onto a source. And, it goes without saying, they should be silent when not intentionally making noise.
Sadly, commonly, they’re not. They hiss. Some users find thhhhhhisssssss irritating. As noted elsewhere, one of the blind spots of pro audio culture is noise – amps to 11 and all that; never mind the quality feel the volume. The workaday fix in a studio is ‘turn it up’; drown the noise in more noise – you’ve probably got 20 good years before the tinnitus kicks in.
But how can it be acceptable, in a tool purposely crafted (in both senses) for close listening, to swamp fine-grained information in hash? What impact is intermodulation having on the driver’s ability to render high frequency information? Why go to great lengths to engineer the perfect listening room, but bury the recording’s acoustic cues in self-noise? How did it come to this? Is new-fangled Class D to blame? And is there a fix?
First, money. The tight margin scheme of the pro industry offers great value to customers: a studio monitor is a hugely smarter buy than anything recommended in What HiFi. Second, space. Small, self-contained speakers are desirable – they sell – and small, active speakers need small amplifiers with small power supplies. In combination, these market forces squeeze component choices. Studio monitors are built to a price point. Low-noise amplifiers tend not to be good, small and cheap: pick any two.
Monitors hiss because that the operational noise floor is significantly lower than the sensitivity of the high frequency driver. It doesn’t matter whether the tweeter amp is Class A/B, D or H: a low-mass tweeter invariably has better efficiency than a rigid low frequency driver and so is more vulnerable to self-noise.
For instance, the datasheet of a given ‘chip amp’ might show, in an idealised implementation, a SNR of 90dB for a given (or maximum) wattage. However, the operational noise floor of the amplifier and crossover is a sum of all components, including DSP stages which are notoriously noisy and increasingly common. If you want to complain about newfangledness, don’t blame Class D – if anything, Class A/B amps are worse in this respect. Instead, fall back on the more familiar canard of noisy digital: “in the good old days, crossovers were done properly with wound wire and EQ was rugs on the floor.”
It’s not unusual to find monitors that exhibit an array of subtle squeaks, pulse and crackles when idling. Active PA speakers are a worst case for self-noise: combining very efficient drivers, very powerful amplifiers in a very small space with very little need to consider requirements of near-field listening.
The question often arises on forums: “Which monitors hiss, and how badly?” Manufacturers don’t publish this data, or like to comment on it. Therefore we’ve compiled a standardised chart of measurements that compare a few popular models – see ‘Hiss Ranker‘ for details.
Unfortunately there are no easy fixes: self-noise is baked into the design. However, there are fixes. A leading university purchased from us a number of monitors to use in psycho-acoustic research that required flat response to 30KHz. The brand scored quite highly on the Hiss Ranker, but self-noise was still a deal-breaker. Their solution was to switch out the on-board amps for higher quality, lower noise equivalents. Silence restored.
In some instances, the problem is exacerbated by noisy mains and bad earthing, which multiply the interaction of random fluctuations and 50Hz hum into the mix. Isolation transformers and AC regenerators can help. Earth-lifting or isolating devices will remove hum, but these tend only, at best, to ameliorate tangential issues: they don’t address the core issue: the noise floor of the amp + DSP/XO.
With most monitors, optimising the gain structure reduces hiss. In fact, most monitors increase hiss above a baseline when deploying gain. Whatever your monitor’s method- a digital volume control, a -10/+4dB switch or input sensitivity selector, the aim is the same: ramp down the monitor; ramp up the input.
Noise is sometimes introduced by conductively attached interfaces and computers and poor cabling. The solution is low-noise power supplies, good quality D-A devices and balanced (XLR), not single-ended (RCA), cabling. A USB filter on the PC output has helped in some cases.
Occasionally, we see poor board design or other internal layout issues where convenience is placed above quality that don’t improve self-noise, but very rarely is the solution as simple as extra screening or a dab of absorption material. The design fumble is usually a tacit admission that the amp has bigger problems than its signal path.
The fact that passive speakers rarely hiss as badly points to one solution in the case of monitors with highly configurable DSP. Almost invariably, active crossovers are compensating for a HF driver’s massively higher efficiency, often exacerbated by a waveguide. Performing this ‘padding’ in the analog domain (like a passive monitor) will reduce tweeter hiss. But only if you’re prepared, and there is physical space, to modify the tweeter – and, providing you can digitally add back whatever the mod take away.
For instance, a 4 ohm resistor and an 8 ohm resistor in parallel will drop an 8 ohm’s tweeter’s sensitivity – and the hiss – by 6dB. If you check your monitor on the Hiss Ranker, you’ll note that well behaved speakers have measurable peaks of 6-8 dB at 180mm. A 6dB reduction would therefore render them practically silent. However, this mod won’t work for all monitors as a retrofit.
Why such a simple expedient isn’t standard practice is beyond me. Also, there are low-noise chip amps available for only a few quid more . Would the market tolerate a 15% price increase to obtain silent running? Maybe. Personally, I’d like to be given the option of a Special Edition low-noise Kali IN8, for instance: it would be worth an extra £100 on the sticker. The problem in 2021 is not ‘too much digital trickery’ in speaker design, it’s ‘not enough digital trickery’, even though DSP plays an inevitable part in increasing the noise floor. If a monitor is ground-up designed for a digital crossover, an SE version with quieter amps is only a firmware upgrade away from being identical to the base design. But a monitor with an analog XO and a noisy amp is a dead end.
Unfortunately, until technology improves, if hiss bugs you, and you don’t fancy gutting your monitors to upgrade their innards, the only choice is to buy the least objectionable option, or crank the volume.