Simple DC Adapter Power Supply

You need a power supply for a project, but only have a DC adapter available, so you can't use my AC power adapter trick (Project 05).  This little project came about because a reader had just this problem, and didn't know what he could do.

This idea will work best with DC adapters having 12V DC or more - lower voltages will work (but not with all opamps!), but the dynamic range will be very limited.  For example - using a 12V DC supply you will get ±6V, allowing a maximum signal level of about 2.8V RMS (allowing for losses in the opamps).  While this is still more than enough for most applications, it is generally better to have as much headroom as possible.

The supply here will not be suitable for low level circuits (such as phono preamps), as these really do need lots of headroom, and hum might be a problem, since the standard transformer based plug-pack power supply is often not regulated.  Most of those available now are switchmode, and while these are regulated, they are also somewhat noisy.  Whether (or not) this gets through to the audio can be a lottery.  For high level applications, such as the surround decoder or a crossover filter, this circuit will allow you to use a supply you already have, saving a few dollars, pfennings, shekels (etc.).  The added capacitance should keep hum levels low, but you can increase the 1,000µF caps if you want to ensure lower ripple.

There are two versions, one is ultra-simple, and is fine where there is no load imbalance - which is the majority of opamp circuits.  Where there is a different current from +ve and -ve supplies, you will need to use the second circuit, which allows up to 30mA current imbalance between the supplies.  In either case, if the external supply is switchmode (which most modern ones are), there is limited filtering to get rid of high frequency noise.  It may be necessary to add extra filtering if noise is a problem.

NOTE: The external power supply used must not be used to power other equipment along with the circuitry attached to this project.  While multiple PCBs can be powered from the splitter, they must all use the same supply voltage (e.g. ±6V).  Attempting to power circuitry with different supply needs (such as a single +12V supply) as well as the adapter shown here may lead to the power supply being overloaded or short circuited, and all devices powered will likely malfunction and/or be damaged.

Circuits don't come much simpler than this.  The input DC is given a voltage divider to establish a 'virtual/ artificial earth', and this is used as the 0V reference for the unit to be powered.  In its simplest form, it uses a pair of resistors and two additional filter caps to make sure that any hum is within the capability of opamps to reject, and to allow for transient current with higher signal levels.  This also provides a pair of properly decoupled supply rails to power the circuit.  As mentioned above, you can increase the 1,000µF caps to get less hum, but it is not likely to cause a problem - opamps have very high power supply rejection, so even quite high hum levels on the supply will not be audible at the output.

Figure 1 - Simplest Form of the Circuit

There will be instances where the currents from each supply will be unequal.  Where this is the case, the resistor divider is not sufficient, and the +ve and -ve voltages will be unequal.  By using a cheap opamp (such as a µA741), a DC imbalance between supplies of up to about 15mA will not cause a problem.  However, we can do better with a dual opamp (which will cost the same or less anyway), and increase the capability for up to about 30mA of difference between the two supplies.  Using higher resistances for the divider means the caps can be smaller, and the caps at the outputs help to remove any noise as well.

Figure 2 - Dual Opamp Buffered Supply

The 10 Ohm resistors allow the opamps to share the load properly.  Without these, one half of the opamp (that with the higher gain) will do all the work, and the circuit will not work nearly as well.  Opamp noise or other sonic characteristics are of no consequence, since their influence is completely swamped by the capacitance, so the circuit looks like a regular DC supply to the following circuitry.

Very simple, but works very well.  I have used this technique many times in different designs, and it is extremely effective.  I found that the circuit will keep the supply balanced with as much as 40mA of unbalanced load (i.e. with up to 40mA from +ve or -ve to ground), however, this is pushing the opamp right to its limits.

Note that there is no reverse polarity protection, since the diode voltage drop will reduce the limited voltage even further.  If your supply is high enough in voltage (or the diode drop is not a problem for you), use a 1N4004 in series with the positive supply lead.  You can use a Schottky diode for reduced voltage drop if preferred. Also, if a LED power indicator is to be used, make sure that it (and its series resistor) go between supply rails, and not to the artificial earth.  This is important ! If connected between one supply rail and the artificial earth, the supply voltages will be unbalanced.

There are a number of issues if you decide that a 'supply-splitter' is required for a power amplifier.  The main problem is SOA (safe operating area), because the splitter booster transistors will have half the supply voltage across them, along with the peak speaker current.  For example, a 48V supply can be split to become ±24V, but with a power amp driving 8 ohms, the peak current is 3A (it's actually a bit less, but as a 'thought experiment' we'll use 3A).  3A at 24V is a peak dissipation of 72W.  There's also a problem with linearity - if the splitter isn't perfectly linear, it will add distortion to the signal.  The output capacitors need to be quite large to ensure that the audio is carried by them, not the transistors.

Arranging the transistor network isn't as straightforward as you'd think, as it really needs to be almost a complete power amplifier to be successful.  The solution is deceptively simple.  By using a simple buffer and increasing the output capacitance, the majority of the current is passed via the capacitors, and not the transistors.  Provided the caps are big enough, the current handled by the supply splitter transistors is minimal.  I showed a pair of 1mF (1,000µF) caps below, but more is better.  Just make sure that the power supply can tolerate a large capacitive load.  Some switchmode supplies have limited peak current capability, and will be unable to charge the capacitors.

Ideally, if you only have a single supply available, use of a BTL (bridge-tied load) amplifier is far a better proposition.  A BTL design has the advantage of very little ground current, as the load current flows through the amp's transistors from positive to negative (and vice versa), and the only ground current if for the amplifier's reference voltages.  The basics are shown below ...

Figure 3 - Transistor Buffered Supply

The transistors don't do very much, but they do allow for offset currents.  With the BD139/140 transistors you can have an offset current of up to around 50mA or so.  If you need (or expect) more, TIP35/36 transistors are recommended, with lower value emitter resistors.  The two diodes are 1N4004 or similar, and all 1k resistors should be 1W.  A heatsink is essential for Q1 and Q2, and the maximum recommended input voltage with the scheme as shown is around 48V.  A BTL amplifier with a total supply voltage of 48V can deliver around 120W into 8Ω, so the comparatively low voltage isn't really a limiting factor for most systems.

With a single-ended amp (i.e. not BTL), the two capacitors across the output handle the majority of the audio signal.  If either transistor is forced into a non-conducting state there's a risk of distortion because the transistors are operated open-loop and have no linearising feedback.  If you use such an amplifier you should use the second set of transistors/ emitter resistors shown (TIP35/36, 1Ω), and the transistor current will be around 60mA (it could be up to 100mA, and the transistors will dissipate 2.4W even with no signal.

For anyone who wants the minimum of fuss, the TLE2426 'rail splitter' IC is another solution.  TI describes it as a precision virtual ground.  It's available in either a 3-lead TO92 package, DIP or SMD, with the latter two providing a terminal for noise reduction.  The IC has a maximum voltage of 40V (±20V maximum) and can handle an offset current of ±20mA, with a quiescent current of no more than 400µA.  It's stable with a capacitive load, but there is an unstable region that requires that the output capacitance should be greater than 100nF.  My recommendation would be a minimum of 10µF, which ensure stability with any load imbalance.  If you want to know more, please see the datasheet (available from the TI website).

This is a reasonably priced device (they should be less than AU$5.00 each), but a pair of opamps connected as shown in Fig. 2 is just as good, and will generally be cheaper.  The advantage of the opamp circuit is that you don't need to order an additional part, and can use whatever you have available.

Main Index Projects Index