For typical music, the crest factor is much higher than for a sine wave, so headroom is required to have a healthy signal level without the peaks becoming distorted.
I’ve read so many people saying that a 9V power supply for guitar pedals means they can’t give you enough headroom, and the same for the 500-series 32V (±16V) supply. But surely there’s a way designers can ‘step up’ the voltage inside a device. After all, I’ve seen units running on 24V DC supplies or off USB ports or even 9V batteries that can give me 48V phantom power. So why can’t they all use a standard power supply and do that, to give higher voltages inside if they need them? Are significant trade-offs involved — and is there any reason to avoid units that do this?
Douglas Bannister
SOS Technical Editor Hugh Robjohns replies: The short answer is yes, voltage can be stepped up, and there are a variety of techniques available to achieve that depending on whether we’re talking AC or DC, the amount of increase needed, and the power consumption requirements. Today, a lot of equipment employs ‘DC-DC converter’ chips, which can synthesise almost any required DC voltage from any other with remarkable efficiency, but inevitably such devices add to the equipment build cost, consume power, and generate heat, all of which limit their practicality in some applications. Although relatively rare today, they can also cause interference in some cases, which isn’t what you want in a high-quality audio device!
While it could be argued that the ideal approach is always to generate the required power-rail voltages directly from a mains supply, that’s obviously not always possible. Modern DC-DC converters offer a mature and effective technology, which makes operation from batteries or low-voltage supplies very practical in many circumstances, and I see no reason at all to avoid such devices.
Some guitar-pedal manufacturers, such as Roger Mayer (his 456 analogue tape-machine simulator is pictured here), opt for higher voltage power supplies than the ‘standard’ 9V DC to allow much greater headroom without having to incorporate DC-DC converters inside the devices.
The headroom issue warrants a little closer investigation. The power-rail voltage for audio circuitry inherently sets the limit for the maximum signal voltage that can be processed; any signals that try to go higher will clearly become ‘clipped’. As audio signals swing above and below a nominal mid-line reference level, a 9V supply allows for a maximum voltage swing of ±4.5V, but because of the way active components work in audio circuitry we can’t usually access that whole range.
For the sake of argument, let’s assume a 9V-powered device supports an audio signal with a swing either side of the mid-line of up to ±4V. Converting that into a more meaningful signal level in terms of dBu values is slightly complicated because it depends on the signal’s ‘crest factor’ (the ratio of the peak value to the RMS value). If we’re talking sine-wave tones, then the crest factor is 3dB, but for typical music the peaks can easily be 12-20dB higher than the RMS value, and that’s why we need headroom in our audio equipment! (See diagram.)
So, a sinewave signal with a 4V RMS value equates to a signal level of just over +14dBu; the calculation is =20*log(4/0.775). However, the peak level of a sinewave signal is 3dB higher than its RMS value, and the peak level matters here because it’s going to be restricted by the power rails. So a ±4V working range supports a sine-wave signal amplitude of up to about +11dBu (14 minus 3dB), and for a really peaky signal (such as from a guitar!) the maximum supported signal level before clipping could be as low as -6dBu (14 minus 20dB).
The output from an electric guitar is generally classified as ‘instrument level’ which is specified as a nominal -20dBu, in which case we would appear to have around 14dB of headroom (20 minus 6) which isn’t bad but it’s not great, either, and with more efficient pickups and more percussive playing styles we could easily find ourselves with surprisingly large transient signals that could well start to clip. For this reason, some manufacturers prefer to employ higher power rails in their equipment to afford greater headroom margins.
Professional equipment generally supports a maximum interface level of +24dBu (measured with sine-wave tones), but again taking the crest factor into account, the peak voltage of a +24dBu tone would be just under 17.5V. From that, we can see that the equipment power supply must be at least ±18V, and in practice it’s often ±22V or even higher.
You mentioned the ±16V rails of the 500-series racks, so let’s consider the maximum signal levels that can be supported there. If we assume the maximum voltage swing in a device powered in a 500-series rack is ±15.5V, the highest sine-wave signal level that could pass without clipping would be around +23dBu, which is surely good enough for most situations. In the few cases where a higher output signal level was needed, one simple solution would be to use a step-up output transformer, providing some ‘free’ signal voltage gain!
So while having ±16V power rails is inherently slightly restrictive compared with ±22V rails, say, the practical impact probably isn’t that significant in most applications, can careful and sympathetic gain structuring will easily avoid any clipping issues anyway. One final observation along those lines: the Neve 1073 preamp/EQ module operates on a single +24V supply rail, allowing a maximum internal signal level of around +20dBu — but with transformer outputs! If it was good enough for Neve...