Tweaks

To change the amp's gain, you can change whichever of the two resistors is most convenient. You should pick values somewhere in the 1 KΩ to 100 KΩ range. Lower is better, for a couple of reasons. First, high values will increase your circuit's suspectibility to stray RF and electromagnetic noise. Second, higher resistor values are inherently noiser. (You can explore this phenomenon with my op-amp noise calculator.) If you go much below 1 KΩ, the current wasted through the feedback loop will become significant.

The default gain of 11 is fine to start with. It won't let you use the full sweep of the volume knob unless you have very inefficient headphones or a very weak source, but it does provide good, stable operation of the op-amp. As you lower the gain, you make it more and more likely that the op-amp will become unstable. Indeed, some op-amps won't work reliably at low gains. The OPA13x series chips are "unity gain stable", meaning that you can take them down to a gain of 1, if your circuit is built soundly. (You get a gain of 1 by connnecting the output straight to the inverting input.)

I have used a gain as low as 2 with inefficient Sennheiser 500 series and AKG K401 headphones, when the source is strong enough. Similarly, a gain of 2 or 3 with efficient headphones should work fine even with weak sources. A gain of 5 or 6 is practical for almost all combinations of sources and headphones.

For your first amp, build it with a gain of 11 at first. Then once it's working correctly, try out different feedback resistor values, checking carefully for audible distortions with each change. Once you get to a gain level you like and it's stable, box it up.

Before I go into more details, please read all the information on noise in the companion article Basic Troubleshooting for Headphone Amplifiers. The solutions in that article, if they work, are superior. Adding R5 should only be done as a last resort.

Having tried everything else to get rid of noise, try putting 10 Ω to 100 Ω in the R5 position. Higher values will be more effective, but because lower values have fewer bad side effects, your incentive is to use the lowest value you can get away with.

The easiest way to solve this problem is to use a "closed circuit" DC power jack. These add a switch and a second circuit to a regular power jack. This second circuit is normally closed; thus the name. When you stick the DC power plug into the jack, the second circuit opens. If you connect the batteries through the second circuit, they get cut out of the circuit when you plug the wall supply in. I show how to wire a closed-circuit DC power jack in the companion article, How to Wire Panel Components.

Unfortunately, the switch inside some closed-circuit jacks is rather delicate. The DC power plug can snag on the switch while you're inserting it and bend it. If you're lucky, you'll just bend the switch contact a little bit, so that you can bend it back with a small screwdriver. If you're unlucky or ham-handed, you'll bend the switch contact completely out of shape and be forced to replace the jack. To some extent, you can solve this problem by paying particular attention to the DC power plugs you use: the better kind for this application have chamfered leading edges, so they're less likely to snag than the blunt kind. The high road is to avoid closed circuit jacks if you don't absolutely have to use them.

Batteries Plus Wall Power: Clever Method

It's possible to have our cake and eat it, too. We can use a plain, switchless DC power jack and still have both battery and wall power. The easiest way to do this is to use a diode OR bridge:

Diode OR Bridge

The name comes from the fact that it allows only one power source or the other to run the circuit, never both at the same time. This prevents the wall supply from trying to charge the battery when both are connected.

The diodes can be any generic type, such as the common 1N4001. The bar on the diode symbol in the schematic represents the diode's cathode, and the corresponding terminal on the physical diode is marked with a stripe. So to build this circuit, you simply connect the two cathodes together with the V+ line to the amplifier, connect the positive side of each power source to one of the other diode legs (the anodes), and the negative side of each power supply to V- in the CMoy amp.

There are a few restrictions with this method. First, because a typical diode has a forward voltage drop of around 0.7V (the "diode drop"), this can prevent you from getting the full use out of your batteries. (Whether this is the case depends on matters discussed in the companion article, Op-Amp Working Voltage Considerations.) The second restriction is that the wall power supply must be at least one diode drop higher than the voltage of a full battery, or else it will not reliably cut out the battery. Beware that fresh alkaline batteries have a higher terminal voltage than they are rated for; a fresh "9V" battery actually measures more like 9.6V, for instance. Rechargeables similarly have higher terminal voltages than their nominal voltage when fresh out of the charger.

The way the diode OR bridge works is actually quite simple. A diode only conducts current when its anode (positive) voltage is at least one diode drop above its cathode (negative) voltage. Since the cathodes are tied together, connecting the wall supply pulls the common cathode up so it's less than a diode drop below the battery voltage, so only the wall side diode conducts. (In practice, you usually overdesign it so the common cathode is actually higher than the battery voltage.) When the wall supply is removed, the common cathode point falls back down to one diode drop below the battery voltage, so the circuit still gets power.

Let's work a concrete example. We have 2×9V for our battery supply, which we'll conservatively estimate at 19.2V when fresh. We'll use standard 1N4001 silicon diodes, which have a diode drop of 0.7V at the low currents you'll see with a CMoy amp. And finally, we'll use a 24V wall power supply. So, with only the batteries connected, the common cathode point is 0.7V less than 19.2V, or 18.5V. When we connect the wall supply, the wall side anode goes to 24V, which raises the common cathode to 23.3V. The battery can never run the circuit while the wall supply is connected, because the battery side anode is always at least 4.8V lower than it has to be for the battery side diode to conduct.

The next most important spec is whether it's regulated or not. You can count on an unregulated supply to put a lot of noise on the amp's power rails, which may get through to the output in audible amounts. (See my article Op-Amp Power Supply Quality Considerations for further details.) Regulation fixes this problem. The two most common types of regulation are linear and switch-mode.

Switch-mode regulation also puts out a fair bit of noise, though of a different type than you get from an unregulated supply. Whether or not this is audible is a complicated issue. See the above-linked article for more discussion on this topic.

Linear-regulated power supplies are best for high-performance audio. Unfortunately, there aren't that many commercial linears suitable for audio DIY on the market. The most readily available ones are a few models in the WM series (not all of them!) from Elpac Power Systems, carried by Digi-Key, Mouser, Newark and Allied Electronics. By far, the most popular is the WM080 (24V/330mA), Newark part number 16F126.

It's possible to DIY your own linear power supply, with a bit of work. I offer a few different power supply boards which are designed with headphone amp circuits in mind.

If you want to pick a different power supply from the ones I recommend above, it should supply at least 12 VDC, and no more than 36 VDC. (See the companion article Op-Amp Working Voltage Considerations for further information on that topic.) The common voltages are 30, 24, 18, 15, and 12, in that order of preference. Check the power supply voltage specs in your op-amp's datasheet to be sure it can accept the voltage you choose! Most op-amps work fine from 30V, but some more modern chips have much lower voltage tolerances.

The amount of current your supply can put out really doesn't matter for this application — even the crummiest little wall wart can supply 100 mA, whereas your CMoy amp shouldn't ever draw more than about 20 mA.

You might think you could actually get away with a 10V cap for our example, because the virtual ground divider is supposed to divide the power supply voltage by 2, putting only 9V across each power supply cap. The first problem with this logic is that the divider in the CMoy will be imbalanced in actual use. If you to use 10V power supply caps with our 18V example power supply, it's very likely that one of the two caps will see more than 10V and be damaged. The second problem stems from the fact that you should never optimize something like voltage tolerance assuming that the circuit is working properly. Let's say you're using 10V caps with a 15V power supply instead. In this case, it's quite unlikely that you will get more than 10V across one side of the supply in normal operation. But, what happens if something goes wrong? There are several ways the virtual ground divider could fail, putting all 15V across one side of the power supply. Bye-bye power cap.

Bottom line, it's prudent to make your power supply caps' voltage tolerance at least equal to your rail-to-rail power supply voltage in this amp.

Capacitance Rating

The caps' values will change the way the amp performs. For the power caps, 220 µF is adequate, but bigger ones will provide a bigger current reserve, which can be useful in handling high instantaneous loads, like big drum hits. I've tried 470 µF caps in my CMoy amps before, and it does improve the bass handling significantly and they're not all that much bigger than 220s.

As for the input caps (C2), a larger cap will improve bass handling, but it's by a different mechanism than with larger power caps. See the companion article, Input Capacitors for Headphone Amps for a full discussion. Bottom line, the default 0.1 µF is a bit on the low side. Try 0.22 µF, 0.47 µF, or even 1.0 µF instead.

With film caps, the smallest voltage tolerance you'll likely be able to find is 50V. Although even 50V is way too high, don't worry about it: just get the lowest-voltage caps you can find. If you go with polypropylene caps, you probably can't find caps with less than 200V tolerance. If that's the lowest voltage tolerance you can find, go ahead and use them. Just don't think that getting the 800V versions will do anything more than waste board space and money.

Capacitor Type and Quality

There are many types of capacitors, because we have so many reasons for using capacitors. No single capacitor type works for all situations.

For the CMoy power supply, electrolytics are the best choice. Only they have the high capacitance values and low cost needed for this task. Different brands of capacitors can sound different. Many companies make good electrolytics: Panasonic, Nichicon, ELNA, Rubycon, Sanyo...

Electrolytics do have a downside, which is that their value tolerance is very bad, and they have high distortion. Neither problem matters in a power supply, but it would be foolish to use these for a CMoy's input caps (C2). For this job, plastic film caps are best. Good brands are Panasonic, Vishay, Wima...

Not all capacitor lines from the above listed companies are equal. Do some searching online for advice if you want to use different capacitors than the one I recommend in the parts lists. Here is a typical thread. Also, there is some advice in the companion article, Input Capacitors for Headphone Amps.

Before you go deviating from caps recommended to you, especially if you're swapping them into an existing layout, check their datasheet to see how large they are. If you're staying with the same product line and just getting a different value, the replacement could be larger in diameter, or height, or both. Or, you might get lucky and find another cap in the same line that's the same size as the one it's replacing. (It happens!) If you're switching to a different manufacturer or product line, it's doubly important to check the datasheet because it's common for a higher quality cap to be larger than a lower quality cap, all else being equal.

In the standard CMoy power supply, the two 4.7 KΩ resistors create a "virtual ground". This circuit works adequately in many situations, but it has weaknesses. See the full discussion in this article, which also shows improved virtual ground circuits.

This space intentionally left blank. :)