X-Ray Specs (Part 1)
Ben Duncan explains the language of the specification sheet
find out what the small print means
Introduction — by Dr. Spliff, M.Inst. Electronics Lexigrapologers (Hons). "Electricity was first harnessed in the 1830s for railway telegraphy (-..-...-), then for telephony (Mr Edison, I presume), and by 1884 a public power source, ye mains juice. At the turn of the century Mr Marconi proved transatlantic wireless telegraphy (-..-...- again, this time without wires) to be feasible. By 1915 it was already being used to aim big guns at people across the Flanders salient. In the 1920s wireless telegraphy became radio telephony and audio recording, and reproduction became tied in with wireless techniques and valves. By 1945, so many techniques had sprung from radio — from radar to hydrophony — that they were all lumped together under a separate technology known as electronics. Thus the lingo used by the high-tech amber priesthood has charted a curious course, from the utterings of deranged Victorian inventors, through the earthy cynicism of Edwardian power engineers and the jolly-good-show-spiffing speak of wireless chappies to the earnest seriousness of 1950s nuclear electronics when pure theory displaced mere human feelings.
This is their dictionary..."
Another word for volume or gain reduction. A fixed attenuator is a doubrie which gives a known reduction in signal level, eg: -20dB = 1/10th, or a 90% reduction. When an attenuator is variable, it's called a pot or volume control, and when it's frequency selective, it becomes an EQ boost/cut control.
The ordinary, original species of electronics — rarely fails altogether; uses lots of capacitors, transistors and op-amps. A signal is an electrical analogue of the original, acoustic sound at all stages in the signal path. (See digital)
The output end of a console; or the speaker stacks-cum-power amp racks in a big PA.
A symmetrical input or output, connected to a paired cable (= two-core). By devious means, balanced inputs can distinguish between the legit signal and any garbage or interference picked up en route, and can cancel out the latter. A facility known as common-mode rejection (CMR).
The ordinary species of transistor. Now superseded by the newer MOS-FET variety in high power applications.
Gentle cooling by natural air currents. Hot air rises, dragging in cool air beneath it. Thus an airflow is available without resorting to fans.
If you try to magnify a colour slide overmuch, the edges of the picture are clipped off. The same happens when you try to magnify a signal to a greater size than the equipment can cope with, but the results are much more dire. Synonym for overload.
Col. for the unexpected, aggressive sounds of severe overload.
(CMR or CMRR for short). A measure of how good a balanced input is when it comes to chucking out hum and RF nasties. Expressed in dB, it gives the ratio between the actual signal and the noise/interference remaining after the balanced input has done its stuff. A good figure is 100dB (1/100,000th); a poor one is 30dB (1/30th), ie the noise is only 30 times less than the signal. But every little helps. The CMR figure always falls off at high frequencies, but HF filtration hides this.
Usually a switch which partially mutes playback or the PA.
Like doing a wheelie, a PA or monitoring amp with a poor damping factor fails to 'take hold' of the speaker properly. The results are most evident as sogginess or overhang in the low bass and these FX are easily swamped out by common speaker defects.
Adverts by upmarket amp makers have frequently hyped the large damping factor figures that can be achieved in amps with bipolar power transistors. However, the fact remains that most critical listeners prefer the bass sound of Tube or MOS-FET amps, neither of which need have particularly spectacular damping factor figures. A damping factor of ten is the bare minimum and means the amplifier's source impedance is ten times greater than the speaker's nominal impedance. Therefore, by referring the measurement to a high impedance speaker (eg: 15 ohms), the figure of merit can be made to appear larger than it really is. To be meaningful, the damping factor should be quoted against an 8 (or 4) ohm speaker impedance, and over the full range of audible frequencies (20Hz to 20kHz), otherwise the maker is very likely covering up performance defects at high frequencies, which an honestly specified damping factor is apt to make embarrassingly obvious. Verdict: Damping factor 10 to 50: if it sounds tight at the bottom end, love it. Damping factor 50 to infinity: fine, but don't part with any extra cash unless it does something spectacular to the bass sound of your existing speakers.
Decibels are confusing because while essentially a dimensionless (!) ratiometric quality used to compare things in a relative sense, they are also widely used as units in their own right. So let's begin again:
As a ratio between any two things — a relative form of measurement without reference to absolute size. You can also think of the smaller dB quantities as a percentage. Eg: +6dB = twice the size. So a brown note is 6dB up on a fiver, or, 2dB = x1.26 bigger. That's to say that I may be 2dB taller than you or 126% of your stature. Decibel comes from deci, a tenth, and Bel., the name Bell Telephone labs gave to their conceptual baby in the 20's. The full sized Bel is inconveniently large for the real world; decibels and bels alike follow a logarithmic scale and you don't need very many of them to get into large numbers. For example, 1700dB encompasses 1084 which is reputed to be the number of atoms in this particular universe. But then we need a logarithmic scale for precisely this reason — for our hearing range of 152dB is just a convenient way of saying that the loudest sounds we can stand are only 39,810,717 times bigger than the quietest sounds perceivable.
One other point: log scales also make life easy because division comes by merely adding a minus sign. So -6dB = the opposite of double, that's half. -20dB is a tenth, -2dB is 1/127% = a 78% reduction. Now a big question for you: is it better to score -4dB at the casino or lose +0.5dB? Answers on a postcard to Dr. Bell, please.
When we say that 0dB (or 'X'dB) means so many volts or so many of any other quantity, we can begin to use dBs as units. When 'U' is added to dB, it means that 0dB = 776mV, or about ¾ of a volt. The reasons for choosing a stupid figure like 776 are all part of history. Having established the zero level (ie. at zero = 0dB), we can derive a whole range of voltages, both -dBU and therefore smaller than 0dBU, and +dBU, for the voltages above 776mV. Eg: -20dBU = 77.6mV (that's 776 divided by 10), while +20dBU = 7.76 volts (ie, multiplied by ten). dBU is the standard measurement unit in Europe and the UK. Verdict: the cool one.
Today, this can be taken to be pretty much identical to a figure in dBU. Strictly, 0dBm is 766mV applied across a 600ohm resistor. But today's sound engineers don't pfaff around driving expensive multitrack sessions into racks of steaming resistance coils — this one is a relic. However, it can still be seen at work today in certain UK broadcast stations and in the US. Verdict: Quaint laughter.
IF YOU HAVE NOT GONE DEAF, this one was explained in our recent examination of high sound levels. It's basically the same idea as dBU or dBm except that the reference level is set very low so -dBSPL figures don't arise much, and the +dBSPL figures tend to be much bigger. Second, it's all about what comes out of a speaker or goes up the mic, contrast dBu, which is an electrical signal level inside equipment.
Not necessarily a noise reduction unit. Putting any letter after dB is telling you it's a scale not a dB ratio.
A dummy part: "Do Fark All"
(or EIN). In mike amplifiers, excess noise or hiss is primarily down to the front end devices or the first stage of amplification. For practical measurements, though, it's handiest to measure at the output. The actual noise at the front end is then derived by dividing the output noise at the output by the overall amplification. Eg: if for 40dB (= x100) amplification the output noise is -70dBU then the EIN is 100 times less, that's -70, -40dB = -100dBU. This is a good figure for a mike amp, but not the best that's possible.
Either the mike and or line input amplifiers on console or the mike/desk/outboard gear of a complete PA system.
This figure tells us how smoothly and accurately any part of a sound recording or reproduction system reproduces the tonality of the music. Psychoacoustic research in the 50's suggested that a flat response across the full audible band wasn't appreciated by more than a few of the population. So who could they be?? Today, more subtle factors have emerged. In consequence, while the limited effective response of most speakers and tape machines would seem to make a nonsense of a better response in the associated electronics, this is not, in fact, the case.
At low frequencies below 100Hz, phase-shift becomes audible and to move the problem down out of ear's way, the frequency response needs to be extended to below 1Hz. On the other hand, above 20kHz (so that any HF garbage is massively attenuated) are no sonic disaster. An ideal response, attained by some digital recorders, is <1Hz (meaning "less than 1 Hertz") up to >20kHz, ("greater than or equal to 20kHz"), ±0.1 dB (meaning the response deviates no more than + or - 0.1 dB, or 1% within these frequency limits). At 40kHz, though, the response may be -60dB, to kill the hash from the digital processes. Budget gear may be quoted either cheap, cheerfully and absolutely uselessly sans les limites: "RESPONSE 20 TO 18KHZ". This means SFA and what is worse the 'z' should be in lower case! It's like the great bore of today boasting "Yar, the Romans built Ar'mine street in a rular straight course from Londinium to York... er... Ebuculiam... er" without mentioning that the straight line deviates ±25 miles between Lincoln and York.
Beware also of comparing frequency responses with different dB limits. Conventionally, the spec writers tell you the signal has fallen 'X' dB (Where 'X' is minus 0.1, 0.5, 1, 2 or 3dB) depending on the equipment's calibre) at the most impressive possible frequency. It follows that the response can be made to look wider by quoting it at -3dB limits, say, rather than -1dB limits. For example, '20Hz to 18kHz, -3dB' is actually less good than '30Hz to 20kHz, -1dB', although it appears to be the wider response. So when converted to the same frequencies, the second spec reads: "-2dB down at 20Hz, -1.5dB down at 20kHz". To sum up, the fewer dB by which the response is down at 20kHz or below, the better, but beyond this point it's cool for the response to drop out like a stone. Meanwhile, at the low end, a completely open response (to 1Hz) or below is good but infrequently encountered. A response which extends down to 40Hz, -1dB will be about -3dB down at 20Hz, and can be considered the minimum for a good kick sound.
Like resistance, this one tells us how much current will flow in a circuit when X volts are applied. The higher the impedance, the lower the current for any particular voltage. But there's a special caveat: impedance (as opposed to resistance) implies the inclusion of components whose resistance varies with frequency. So to be useful, people who write all about impedance should also tell us what frequency range their impedance applies to. This is particularly true for speakers; the eight ohms written on the back is often measured at 400Hz and may fall to a much lower value like 2½ohms at a higher frequency, causing your amp to crunch up.
Our normal way of scaling is linear, which means counting up in equal steps. Thus one, two, three, four are all steps of one. Logarithmic scaling is just as natural (at least in the natural world around us), yet seems anomalous, because it don't go "one, two, three, four". Essentially, a log (for short) scale goes up or down by equal ratios instead of equal steps. This means it can encompass a very large range of numbers. For example, if we count up with a constant ratio of two, the number doubles each time: one times two equals two, two times two equals four (ad nauseam). If you remember vainly trying to fold One Two Testing 12 times to reach the moon, you'll get the general picture. Log scales are relevant to the art of sounds, because our sensation of frequency, pitch, and sound level intervals all follow the logarithmic pattern. Octaves and decibels are musical concepts expressly designed to help us cope with these logarithmic relationships.
Part 2: Ben Duncan refreshes the alphabet after M.
Feature by Ben Duncan
mu:zines is the result of thousands of hours of effort, and will require many thousands more going forward to reach our goals of getting all this content online.
If you value this resource, you can support this project - it really helps!