KIH #89 – Mind the gap

  • Hamburgers or humbuggers? Dark specs are specs which hifi manufacturers don’t publish because they offer no bragging rights; or specs that apply non-standard tolerances or definitions. With the first, we must be informed enough to ask directly. With the second, we must pay attention to the small print. In no particular sequence, here are a few dark, colourful and hazy specs for your amusement.

    Nominal speaker impedance. It’s typically 4Ω or 8Ω. To the average shopper it suggests that as long as their amps are happy into either, all is dusted. In reality, impedance spikes with a ported speaker could hit 50Ω with power-demanding low frequencies. If our transistor amp delivers 100wpc into 8Ω, it’ll tend to do just half that into twice the impedance. That means 50wpc/16Ω, 25wpc/32Ω, finally ~15 watts into 50Ω. That’s nearly 90% less than the nominal 100-watt rating. Each time music traverses where our speaker’s impedance spikes like a heart attack, the actual power which our amp can deliver shrinks dramatically. That’s nonlinear suboptimal power transfer. If an actual impedance plot isn’t published to show that our dream speaker hits serious impedance peaks, we should ask for the plot to make an informed decision about the kind of amplifier we will need. That’s extra vital when a nominal 4Ω spec could hide the fact that the speaker nearly hits 1Ω in two broader troughs. In the absence of a published plot, what we want now is to at least know the impedance minima and preferably, at what frequencies they occur. Low impedance at high frequencies is less power-demanding than low impedance in the bass.

    Damping factor. It’s the ratio of amplifier output Ω and speaker input Ω. As the former drops, damping factor increases. So far so easy. But we already know that speaker impedance is far from a fixed value. It’s very variable relative to frequency. The result of the basic damping factor equation thus depends on frequency. What’s more, it doesn’t include the impedance of the speaker cable and contact resistance of its connectors. That addition instantly lowers the damping factor figure. Whilst marketeers wield 4-figure DFs as guarantors of sound quality and amplifier control, such figures are theoretical, not actual. Amplifiers with variable damping factor like Gold Note’s also suggest that depending on speaker, higher output impedance can sound better. Fanciers of no-feedback SET amps and the speakers which love them never have high damping factors on their side but still can make good sound. So damping factor isn’t so much a dark but colourful spec. Its actual relevance is subject to both interpretation and circumstance.

    Linear phase. It’s often paraded as a mark of a speaker’s distinction. As soon as we request a frequency/phase plot however to see the advertised linear phase in action, makers routinely clam up. It’s happened to me many times. Despite protestations to the contrary, should we see the actual plot, it would reveal obviously non-linear phase. That’s why such plots are withheld. There are exceptions of course but because they’re quite rare, they tend to be paraded like show ponies. If an actual phase vs. frequency plot isn’t published, you can now predict why. It wouldn’t be very pretty.

    3rd-octave smoothing. This is a default way to make frequency response measurements look a lot better than reality. Another trick to make deviations look less severe is to change the ratio of the graph axes. By lengthening the horizontal and compressing the vertical, peaks and troughs shrink and reduce their jaggedness. Season with smoothing. Et voilà, very pretty text-book graphs. But be mindful. They’re virtually always doctored; and not by Doc Martin. If response graphs look too good to be true, they invariably aren’t raw but prettified measurements.

    Digital latency. If you’re of the time-alignment persuasion to favor speakers which respect it, modern DSP subs can throw us a major curve ball. That’s especially so for compensation which relies on FFT filters to create big processing delays. For a range of Norwegian subwoofers for example, DSP latency is 8ms. That equals the time delay caused by 2.7m of distance. Even if such a sub sat equidistant with the main speakers to look time-aligned, its signal still arrives at our ears 8 milliseconds late. If we stuck said sub into a front corner for maximum room gain whilst the speakers sat out in the room two meters in front of the wall, now its bass would arrive late by 4.7m or 14ms.

    To know this digital latency spec nearly always means that you must ask for it. It’s usually not published. Now you appreciate why. For our Dynaudio S18 sub, it’s 2.5ms. I offset that with the necessary 0.86cm closer to the chair. KEF’s KC62 incurs a just 0.5ms delay which equates to 16cm.

    Grimm’s SB1 pushes latency down to 38μs. That’s truly negligible relative to late/slow bass. Adding a subwoofer to a two-way speaker transforms it into a 3-way across three boxes. If your speaker sweats time alignment with 1st-order filters and a sloped baffle, bolting on a woofer that’s 2.7m behind the baffle is truly counterproductive.

    Bandwidth. KEF’s marketing department misfired when they published a 105dB and 11Hz spec for their micro KC62 sub in the same breath. On their own, both figures are true. In tandem, they’re not. 2 x 6.5” drivers can’t do 11Hz at 105dB. To be meaningful, subwoofer specs can’t just list an impressive cut-off frequency. They must specify at what SPL that target is met. A proper spec would say something like “30Hz at 100dBSPL, 20Hz at 87dBSPL, 17Hz at 65dBSPL”. Now the spec reflects reality, not wishful thinking.

    Bluetooth latency. This can intrude as a problem when a wireless subwoofer suffers broadcasting latency whilst being matched to wired passive loudspeakers which don’t. Now the wireless bass will be late regardless of placement. Wireless latency can intrude with a Bluetooth phonostage where a visible needle drop doesn’t create instant feedback but a delayed reaction. Bluetooth delay can interfere with lip-sync during video. Kii Audio’s popular active monitor for example suffers high latency from intense DSP processing. But the manufacturer doesn’t keep it a secret. They offer a low latency setting which doesn’t correct their speaker’s response to the same extreme extent but guarantees lip-sync compliance. If latency figures aren’t published, be sure to inquire so you know what to expect.

    Bluetooth resolution. CD or near-CD quality? Neither LDAC or aptX are lossless 1’411kbps. They lack the necessary bandwidth. Don’t believe propaganda to the contrary. It’s fake news.

    Bluetooth power. This is a dark spec if you suffer WiFi allergies like our household. It’s the case when for music signal transmission or tablet app code, a component includes Bluetooth. We might assume that if we don’t sync up Bluetooth sources or use the wireless tablet app, Bluetooth would be deactivated. Wrong. Routinely it never turns off. It broadcasts microwave radiation as soon as the device powers up. Whilst you won’t see a Wifi antenna or other indication that Bluetooth is live, it will pollute your entire environment. Should you be allergic, your brain will tell you soon enough. It could simply take a while to identify the culprit. If you spent money, you’ll really be mad. It’s like a vegetarian ordering a veggie burger but being served a classic hamburger. When I canvassed compact subs for our upstairs system, I specifically asked SVS whether their new DSP-based smart micro sub can defeat its Bluetooth. It cannot. It’s always on. Likewise for the iFi Aurora Bluetooth transmitter. This could be a growing trend. For ancient brains like ours, that could become one of the darkest hifi specs.

    Gapless. You’d expect gapless playback to be a digital streaming default by now but it’s not. If the specs don’t specifically say gapless, check to ensure that the device or player software you’re considering can do gapless.

    Android antics. Contributed by John Darko, these shenanigans are in play with Android users who must first install the UAPP aka USB Audio Player Pro app to enjoy native hi-res sample rate playback when their device outputs via USB. Otherwise, Android OS applies fixed 48kHz conversion regardless of file. Yet even this saver app which restores native-rate playback won’t accommodate offline content for Tidal or Qobuz. In short, Android users beware.

    Pure analogue. You’d think so with vinyl. But more often than not, modern records are cut from hi-res files. If you thought spinning black kept you away from digital, not so fast.

    DAC-direct drive. It sounds like the ideal spec. Why bother with analog preamps if you’ve only got digital sources with their own analog, digital or hybrid volume control? In practice, it’s a common observation that the ‘redundant’ extra circuitry of the analog preamp improves the sound. Why? Speculations include higher current delivery so better signal transmission across the interconnect and superior drive over the amp’s input stage; and better impedance matching.

    Amplifier input sensitivity. This is less a dark spec than it is hazy. It’s routinely published and not in any misleading fashion. Why include it? Because many consumers don’t add up the implications. If our digital source outputs an industry-standard 2Vrms—many do 4Vrms or more–and our amp’s input sensitivity is 0.3V for full output, consider how much signal you must cut to play your amp at the desired levels. That’s hardly ever full output but a mere fraction thereof.

    But it gets worse. Many add an active preamp with 12dB or more gain. It takes little math to realize that unless its volume taper is exceptionally gradual, a few clicks beyond mute will get too loud already. What’s more, channel tracking right at the start of particularly wiper pots isn’t very precise. Call it gain poisoning for a wakeup call. It’s audiophile bad cholesterol. Not only is it inefficient to first create something only throw it right away again. It also clogs up our signal-to-noise ratio. For the most gradual hence practical control over our volume and noise floor, we ideally want an amplifier input sensitivity that’s close to our source output. If we throw in an active preamp, it should be of only low voltage gain like 6dB; or apply no voltage gain at all, just pure current gain to make the signal transmission to our amplifier more robust.

    Fully balanced. When a preamp announced as such uses a 2-gang volume control, it isn’t fully but only partially balanced. It converts balanced input signal to single-ended for volume then back to balanced for the outputs. To maintain a twice-of-everything signal path with separate circuitry for each half cycle (+/- right channel, +/- left channel), the volume control must be a quad type. That is costlier but essential to avoid conversion. Not all that glitters balanced is balanced gold. Related to this are XLR i/o on a single-ended machine. Unless they’re de/re-symmetrized by balancing transformer, active driver or opamp, they’re not true but convenience XLR. Not all XLR are created equal. Even with true XLR, utility in the context of a single-ended circuit is sheer convenience, not any sonic advantage.

    Single-ended push/pull. That’s an outright oxymoron. Single-ended circuits use a single gain device or paralleled parts for both phase halves. Such transistors or tubes never turn off. Push/pull uses a single gain device or paralleled parts for each signal phase. These parts do switch on/off. The intended meaning behind the oxymoron is a minimalist push/pull circuit that uses just one gain device per phase (so two per channel).

    Zero negative feedback. In some quarters used as a spec of distinction and sonic superiority, we must distinguish between global feedback, local/nested feedback and the ubiquitous degeneration resistor. To truly be considered a circuit with zero negative feedback, all three types of feedback would have to be absent. They hardly ever are. The loosest definition of no NFB avoids the global feedback loop. A tighter definition includes both global and local feedback. Not all zero feedback means the same!

    Missing or non-standard tolerances. Calling an amplifier 100wpc but referencing it against a non- standard 6Ω for example is deceptive. In reality, the amp only make 75 watts into the standardized 8Ω. But if marketing insists that 100 watts is the magic number, this sleight of hand will make it so. An example for a missing tolerance returns us at a subwoofer’s cut-off frequency which lists without the SPL reference. It’s irrelevant. Should a speaker claim extension to 20Hz but the small print specifies a non-standard -10dB (or no tolerance at all), that extension figure can’t be used to compare to a speaker which specs at 35Hz -3dB to actually go lower louder.

    In-room speaker specs. This gets hazy again because manufacturers disagree on how much room gain to add. For their KC62, KEF applies a whopping 15dB over their anechoic measurements. What applies if we set up their subwoofer between the speakers out in free space? At what frequency do we measure room gain? This soft spec also affects speaker sensitivities when they’re appended with the ‘in-room response’ qualifier. That’s done to raise the published figure.

    In closing, remember that whenever we shop on spec comparisons, the specs must be based on the same standards and tolerances. Otherwise, we compare hamburgers to humbuggers to get buggered. Remember our dark specs. Just because they go unpublished doesn’t mean they’re unimportant. They go unpublished because they aren’t very pretty. For a shopper, that’s simply a bad reason not to know. Finally, there are colourful specs like damping factor and room gain. They describe real effects but it’s hard to discern where the line between theoretical and practical cuts.

    Avatar photo

    Written by Srajan

    Srajan is the owner and publisher of 6moons. He used to play clarinet at the conservatory. Later he worked in audio retail, then marketing for three different hifi manufacturers. Writing about hifi and music came next, then launching his own mag. Today he lives with his wife Ivette and Chai the Bengal cat in a tiny village overlooking the estuary of Ireland’s Shannon river at County Clare’s border with County Kerry. Srajan derives his income from the ad revenues of 6moons but contributes to Darko.Audio pro bono.

    Campfire Audio announces two new entry-level IEMs

    Rotel RA-1592MKII integrated amplifier review