Hello Russell
Batteries generate noise just like other components. The only way to stop this would be to reduce the temperature to absolute zero. Fortunately the noise is relatively low at room temperatures and often noticeably lower than typical regulator chips. The noise generated by batteries is wideband. I haven’t done any tests with respect to noise levels verses current flow but it is possible that increased current flow would increase current noise due to the increased chemical activity. The internal impedance of the batteries will dictate the absolute level of the noise at the output terminals and if the impedance rises with frequency so will the noise voltage. Adding a parallel capacitor would tend to reduce this noise by reducing the impedance at high frequencies.
In my search for power supply perfection, I conducted some performance trials on 12 A/Hr SLA batteries (medium capacity) in the early 1990's, using sine wave, square wave and pulsed loads. Most audio and digital circuits present a varying load to the power supply. The noise specs of a battery may well measure very well under static conditions, but add a varying load, and things change. Unfortunately, I didn’t do any dynamic noise testing at the time. My notes were left with a now extinct company I was consulting for at that time so I am unable to refer to them. The results, with the batteries I tested, were generally favourable for low frequency operation compared to typical mains supplies with three terminal regulators. Batteries rely on a chemical reaction to provide power and their transient response, at higher frequencies, like typical three terminal regulators (typical transient response of 5 microseconds with noticeable overshoot), is quite poor, so no real gains here. This is very relevant with digital loads. Batteries do remove mains born interference from the supply equation and typical three terminal regulated mains supplies are poor in this respect due to the regulators poor supply rejection (typically 70 dB at low frequencies but falling rapidly with frequency often starting barely above 1 KHz and being virtually non existent at 1 MHz). These regulated supplies also have poor impedance performance. Lack of mains interference, reasonably low noise and their better impedance characteristics are the reasons why batteries have been generally preferred against these general mains supplies.
Another point to consider is load transient induced noise on the battery terminals. If you have a power amplifier drawing a 5 amp transient through say 0.003 ohm (Red Top impedance specification) this will generate a signal related voltage fluctuation of 0.015 volts. This will interfere with the wanted signal (subject to the supply rejection ratio of the circuit under power, which is often not good, particularly at higher frequencies) and become part of the noise floor, which rides on the wanted signal. A busy signal with lots of different frequencies happening at once can really clutter up the noise floor, masking very low level information (you know, the bit that gives you all the subtle clues about how the instruments are played, the song is sung and the acoustic space is portrayed).
I can’t advise you on the relative merits of various battery types, as I have never conducted a specific test. I chose the SLA battery for the trials because I was after a low impedance supply and this type fitted the physical and electrical requirements at the time.
Paul Hynes
