Part Selection Guide

Passive Parts

To understand the issues involved in deciding whether to add C1 or to put a jumper across this position, read my article Input Capacitors for Headphone Amps.

Many types of capacitors will fit here, up to 0.6" pin pitch. You should have no problem finding polypropylene types with large enough values to fit here.

Optional? Yes, jumper across it.

Largest Part Size: 17.5mm × 9mm

Next, you need to know your power supply voltage. It's best to use caps with a voltage rating that's higher than your power supply's maximum output voltage, but no higher. For instance, if you have a 30V supply, 25V caps could be damaged by the power supply, 35V caps are good, and 50V caps are wasteful. (For more on this topic, read my article Op-Amp Working Voltage Considerations.)

Optional? C2, no, C3 yes. (Or vice versa, if you want to be difficult. :) ) Do not jumper.

Largest Part Size: 12.5mm diameter

Op Amps for Everyone by Ron Mancini et al., 2083 KB PDF

High Speed Amplifier Techniques by Jim Williams, 5223 KB PDF

Optional? Yes, do not jumper.

Largest Part Size: 4mm × 12mm

These are the feedback resistors, which set the amplifier topology and the gain. Click here for details about how to pick values for these parts.

Optional? In some circumstances, some of these parts are optional, and it's a mixture of jumpering and leaving parts out entirely. Follow the link for details.

If you bias the op-amp into class A with JFET cascodes, you should also add R10. JFETs have an input capacitance, and you don't want a capacitor across the output of the op-amp. R10 is a kind of "insulation" between the two, so they don't affect each other as much.

The exact value of this resistor is not critical, but it does have to fall within a certain range. If it's too low, it doesn't do its job very well; 100 Ω is probably the smallest value that will provide appreciable benefit. If it's too high, it will cause the cascode to fall out of regulation during parts of the swing of the op-amp's output, negating its advantages; 1 KΩ is about the largest you can get away with. Since you need 1K's for R1 and 100's for R7, it's simplest to just double up on some of those.

Optional? Yes. Populate it if you bias the op-amp into class A. Leave it empty otherwise.

	I = V/R
	I = (15 - 1.8) / 4700
	I = 13.2 / 4700
	I = 0.0028
	I = 2.8mA

Setting this circuit up correctly requires some work. The main thing you must figure out is the minimum working voltage for your op-amp. Follow the link for the detailed procedure.

Now, if you use two or more different kinds of headphones, the working voltage will be different for each one. In this case, this LED cutoff circuit is only useful if the batteries are fully drained before any of your headphones begins sounding bad. If any of your headphones start sounding bad before the batteries are fully drained, the LED will still be on when it's time to change the batteries with your voltage-hungry headphones, or it will turn off while there is still usable battery capacity with your voltage-efficient headphones. In either case, you're probably best off not using this feature.

Once you know your minimum working voltage, you need to look up the voltage drop of your LED. This voltage plus the zener voltage tells you the point that the LED will be completely dark. Unfortunately, this circuit doesn't have a perfectly sharp cutoff point. Between the cutoff voltage v and about v + 0.5V, the LED brightens rapidly compared to an 0.5V power supply swing when using RLED. Beyond this point, it brightens less rapidly until about v + 1.5V where you get full brightness. This is a lot better than the behavior with RLED, but it's imperfect. Such is life.

This circuit works best if you're using rechargeable batteries and a high voltage. The advantage of rechargeables in this situation is that at the end of their usable life, they drop off in voltage quickly, unlike an alkaline battery. This ensures that the cutoff feature works over a matter of minutes instead of an hour or more. The advantage of having a high voltage is that your batteries will be depleted before your op-amp starts sounding bad due to clipping. For example, you might find that the minimum working voltage for your amp with your most difficult headphones is 12V, but if you've got 16 AA-style rechargeables powering the circuit, the batteries will be unusable before they hit 12V. With rechargeable AA's, they stay at 1.2V for most of their working life, so the normal voltage for your amp is 14.4V. If you set this circuit up so that the LED turns off at oh, say, 13V, the LED should turn off as the batteries are giving up the very last of their juice. I'm making these numbers up, so you'll need to experiment to find the actual numbers.

When picking parts for this circuit, you should know that the holes in the board are the standard size for resistor leads. (29 mils?) There are many zeners with rather thick legs, apparently for high power applications. The zeners that work best are in the DO-35 package. For CRDs, I recommend the 1N5283-1N5314 series, which are also available in the DO-35 package.

Optional? Yes, do not jumper.

If you're using AC power, picking the proper diode amperage rating is easy — the power supply will have a maximum amperage spec, and you simply pick a diode based on that. (But see below for a better alternative.)

Batteries are a different matter. A shorted 9V battery will initially put out 4-5A and will take about a minute to ramp down to below 1 ampere. If you use a 1A diode like a 1N400x or a 1N5817-9, you will at least make it get hot enough to burn you, and you may make it fail. If it fails "open", the crowbar will fail to protect your amplifier from the reverse voltage. If you use a 3A diode like a 1N5820-2, that may be safe enough, since a 9V battery obviously drops from 5A down to 3A in much less time than it takes to get from 5A down to 1A. I was able to get a 1N5821 warm, but not hot. If you're using batteries in parallel, remember that the current increases for each additional battery. Batteries in series have a current rating only as high as a single battery.

Another thing you have to pay attention to is that some Schottkys have a lower reverse voltage spec than their forward voltage spec. It's reverse voltage you care most about, since that's how the diode will be used virtually all the time. For example, the 1N5817 has a 20V forward voltage tolerance, but only a 14V reverse voltage tolerance. If you were using a pair of 9V batteries in series, this diode would likely fail in normal operation, since it's reverse biased except when the batteries are plugged in backwards. You'd need to step up to the 1N5818, which has a 21V reverse voltage tolerance.

If you use the crowbar, you might consider also adding a fuse to the amplifier. If you add, say, a 500 mA fuse, it will only blow if the current in the amp gets extraordinarily high, which should only happen when the amp has been crowbarred. (If you have buffers stacked high on your amp outputs, a high load or an output short could also blow the fuse, but that's probably a good thing for a headphone amp.)

Optional? Yes, do not jumper.

The crowbar method works best for battery-powered amps. For wall-powered amps, a better alternative is to put a diode in series with one of the power supply wires. This is the purpose of D2.

The disadvantage of D2 is that it will drop the power supply voltage by 0.6 to 0.9V, depending on the diode type. That matters with batteries because it means your battery lifetime is shortened. For wall-powered amps this small voltage drop won't matter.

D2 has a few advantages: 1) the diode's current rating can be much lower than if you'd used D1 for power supply reversal protection; and 2) if the power is connected backwards the circuit is simply broken, rather than short-circuited.

If you use D2, you should connect the leg farthest from the power supply wire pads to the pad closest to C2-. That is, you should cross the power switch pads with the diode. Then the power switch must go inline with one of the power supply leads; you can't use the power switch pads on the board when you use D2.

Optional? Yes, do not jumper.

A power supply voltage somewhere in the 9 to 24V range will serve you best. More voltage will probably hurt more than it helps, and lower voltage will require very careful part choices to make a workable amp. For the full ugly details on how to measure and calculate your way to the ideal power supply voltage level for your situation, see my article Op-Amp Working Voltage Considerations.

If you're going to use batteries, it's simplest to put them all in series. This gives you a single supply. It's possible to make a dual supply with batteries, but there are dangers in this; do a search in the archives on Headwize or Head-Fi, it's been discussed before. A third option is to put two or more batteries in parallel, for increased battery life.

If you're going to use a wall power supply (a.k.a. "wall wart", "AC/DC transformer"), you also need to take care of the type of power supply you use. The two words you need to look for are regulation and isolation.

An unregulated power supply's output voltage will fluctuate as the wall voltage fluctuates. Also, unregulated power supplies tend to be cheap all around, so they'll have a lot of ripple and noise on their outputs. The META42 design doesn't do much to eliminate noise and ripple (N+R) on the power rails; the power caps and the op-amps and buffers will reject some of the N+R, but they won't get rid of all of it. (See my Op-Amp Power Supply Quality Considerations article for more info.) If there's enough N+R on the rails, it can get into the amp's output in audible levels.

A regulated power supply's output will not fluctuate as the wall power fluctuates, and they generally have much better N+R behavior than cheap unregulated wall warts. Most power supplies that are simply called "regulated" are in fact switching power supplies. These can have more noise than unregulated supplies, because of the way they work. The better type of regulation (for audio, anyway) is linear regulation. If it doesn't say "linear regulated", assume it's a switcher.

Your power supply also needs to be isolated, which means that the output leads are not directly connected to any of the input leads. It's common for non-isolated single-voltage supplies to tie their V- on the DC output side to the earth ground connection on the AC side. Since the META42 uses a "virtual ground", tying V- to earth ground can cause problems, since this places the virtual ground several volts above earth ground. If you plug the amp into a source component that uses earth ground for its outputs, the amp's virtual ground will sit there fighting against the source's true earth ground. The amp probably won't win that fight.

If you use an isolated supply, the amp's virtual ground can "float" to whatever level is required by the source, and V+ and V- will float right along with the virtual ground. Any supply with a transformer directly between the AC side and the DC output (i.e. most linears and unregulated supplies) is isolated. There are ways to make a switching power supply isolated, but check to be sure: most switchers are not isolated, in my experience. Because it's uncommon, if you have a switching power supply that doesn't say that it's isolated, assume that it isn't.

Bottom line: use linear power supplies if you can afford them, or follow an unregulated supply with a linear regulation stage, or use a really good, isolated switching power supply. Or use batteries, and avoid this whole rigamarole. :)

There are better, more balanced chips in Analog Devices' line, such as the AD843, the AD825, and the AD8610. These are all single-channel chips, so you need two of them and a BrownDog adapter: single DIP-8 to dual DIP-8 for the AD843 and single SO-8 to dual DIP-8 for the AD825 and the AD8610. BrownDog also has a simple SO-8 to DIP-8 adapter that you can use for converting SO-8-only dual-channel op-amps like the AD8512 to DIP-8, but you don't strictly need this one since the META42 v2 board has an SO-8 pad for dual-channel chips on the bottom of the board. You would use this adapter if you wanted to socket the chip using a DIP-8 socket.

The AD8620 deserves special mention. This is the dual version of the AD8610 mentioned above, whicn means that it can be soldered directly to the META42 board, or you can use the simpler SO-8 to DIP-8 BrownDog adapter if you want to socket it. The AD8620 has power specs nearly as good as AD823's and it sounds better. It's quite a bit more expensive, though: $12 for the AD8620 chip vs. $6.77 (Newark) for the AD823. It's also harder to find, so I'm offering it along with the META42 boards for your convenience.

You can get Analog Devices chips through Newark, RS Components, or direct from Analog's web store.

Burr-Brown also makes some nice chips. A cheap and workable chip for the design is the OPA2132PA. A better-sounding chip with the same sonic signature (and a higher price tag!) is the OPA627AP. Both of these chips are available in higher grades, but I personally can't tell them apart from the lower grade chips, sonically. You should also consider the OPA637, which is the same as the OPA627 but with an internal limiting capacitor removed which allows it to sound better, but it requires a higher gain to be stable. This is not a problem if you use the Jung multiloop topology; I typically use an inner loop gain of 200 and an outer loop gain of more like 5 with this chip. Burr-Brown chips are available from DigiKey and RS Components.

For more details about op-amps, see the companion article, Notes on Audio Op-Amps.

Choosing a Volume Control

First you need to decide which model you're going to use. The default choice is the cheap but good and small Panasonic EVJ-C20 series pot. This part is an excellent value, it works well, and you can get it quickly from DigiKey. The main alternative is the big, nice and expensive ALPS Blue Velvet pot (RK27112). This part is expensive, getting it requires dealing with small-fry distributors, and you have to pay special attention to grounding when using it. Your third choice is to use something else entirely. There are two other footprints on the board for Bourns 51 and Noble XVB93 pots, but neither of these are readily available at this time; however, they may become available in the future, and in the meantime, these are common pin layouts so you may find that your pot of choice fits these footprints already. If you can't board-mount your pot for some reason, see the pot wire pad labeling section of this document for info on how to wire your pot to the board.

There's a new and strange fourth choice now. Apparently some company has built a stepped attenuator into a package similar to that of the ALPS Blue Velvet. It's marked as an ALPS pot, but the compnay claims not to know about this part (labeled RH2702) and the web site doesn't have any information on it. One guess is that it's made by a Chinese company called Soundwell. I've measured one, and it's got excellent channel matching. It's a bit tough to turn the knob, and there's a click in between settings unless you turn it quickly, but it fits the board and it's inexpensive. You can buy them for $6 each with a minimum order of 4 by contacting Wai Kei Leung of Wai Yip Electronics, Hong Kong. You should read the Head-Fi thread about these parts before buying.

Next you need to decide on a pot value. The maximum reasonable range is 10K to 100K. If you use a pot lower than 10K, you put too much load on the source, which can cause it to sound bad. 10K is high enough that you shouldn't have any problems with source loading, though some people say there are exceptionally weak sources that will have problems even driving a 10K load. 50K is said to be a well-established input impedance. It's a good middle ground, and so if you can find the pot you want in this value, it's probably as close to ideal as you can get. 100K and above have their own problems. If the input impedance is too high, you can get noise and such. The only reason to go this high is if the pot you want is only available in these higher values.

By the way, if you go with a Panasonic EVJ series pot, make sure you get the C20 type, which is the horizontal mount version. (i.e. For horizontal boards if the shaft is to be horizontal.) There's also a Y10 version for vertical boards and horizontal shafts. That part might be used on a stereo receiver where there's a large PCB behind the front panel. Also, be sure to get the "D" taper. There are several other tapers, and since the curves aren't given in the datasheet, I can't say in what situations these others would be useful. I've only used the D curve ones, and so those are the only ones I can recommend.

Sources for ALPS Blue Velvet pots:

Angela Instruments — USA. Rather expensive, but reliable and quick to deliver.

Percy Audio — USA. He sells the motorized version only, and has a reputation for being slow to deliver.

RS Components — Basically worldwide, except USA. (Part# 236-9604 for 50 KΩ version) Reasonable price, and a reputation for great service.

Audiograde — UK. Cheaper than Angela, but it's just one guy with a personal web site. The guy's honest, but slow to bill customers and ship product.

THLAudio — Taiwan. Their price is quite low, but probably the slowest shipping option for most people. Also, they make you order via fax, which is a pain.

Octave Electronics — Malaysia. Very slow to respond, at least to quotes for people in the US. Prices in the midrange of others I've received.