ESP Logo
 Elliott Sound Products Project 109 

Portable Headphone Amplifier

© November 2004, Meraj Salek (edited by Rod Elliott)


The modern day dynamic headphone drivers are very efficient.  Just a few milliwatts are sufficient enough for reaching SPL that can easily render you with permanent ear damage.  Caution, therefore, is not just a recommendation, it is a necessity.

Headphones are by far the most affordable of all audiophile equipment.  The quality of reproduction and SPL offered by even moderate headphones can easily be regarded as a performance standard for the most desirable of loudspeakers.

Still, headphone listening is not as blissful as it might have been expected.  The headphone outputs of most commercial systems receive very little attention from the manufacturers.  This neglect manifests itself in the form of cheap quality sound and frustrations for the listener.  A dedicated headphone amplifier can easily cure these ailments.

In my case, it all started when I got myself a Sennheiser PMX 60 headphone.  When connected to my Sony portable, the sound left a lot to be desired.  As I increased the volume, the bass simply disappeared while the treble became a ringing in my ears with all the hostility of a raging gale.  I tried the 'phones with my IPAQ and this time the sound was even worse.

If you use Grado or any other low Z (≤ 32Ω) headphones, then this may very well be your song I'm singing.  The built-in headphone outputs of most systems, by their very design, cannot keep up with the high current appetite of a low Z headphone.

My design goals for this amp were quite straightforward:

The amplifier, as it now stands, sports three opamps per channel, one as the voltage gain stage and the rest as current amplifiers.  That's a total of three dual opamps for stereo.  There is also a crossfeed network sandwiched between these two active stages.


To locate and externalise sources of sounds, we use both of our ears.  The sound from a source on the right (say, the right speaker) is heard not only by the right ear, but also heard, delayed and attenuated, by the left ear.  The brain compares the delayed and attenuated sound with the original to deduce the exact location of the sound source.

Of course, this is some what an over simplification as reflections at the ear pinnae and from the walls of the listening area also contribute complex information important to the localisation process.  All the info from these sources is furthered by the movements of the head.

When listening to a headphone, all these sources of info are absent.  Transducers mounted directly on the ears cause the unnatural 'super-stereo effect', where one ear doesn't hear, in any form, what the other is hearing.  The perceived spaciousness, which doesn't occur in normal listening conditions, might be very impressive in the beginning but quickly fatigues the listener with headaches and occasionally, dizziness.

This is where a crossfeed comes in.  It is an acoustic simulator of the simplest from.  The crossfeed electronically mimics the inter-channel interactions of the real world by delaying and attenuating the signal from one channel and feeding it to the other.

The use of the crossfeed results in a realistically spacious sound stage where instrument locations seem more natural.  The perceived depth also lowers the listener fatigue considerably.

The crossfeed presented was originally designed by a Swedish audio engineer named Ingvar Ohman.  It was published in an article called "Den Lilla Stereo-kontrollboxen SP12" in the December 1994 issue of the "Musik och Ljudteknik" ("Music and Audio Technical Society") magazine.


The headphone amplifier circuit is shown in Fig.1.  As you can see, it is a very simple design requiring you to detach yourself from the wonderful world of weekend chores for just a few hours ... I promise!

fig 1
Figure 1 - Schematic of Headphone Amplifier

U1 is the gain stage, and as shown it has a gain of 4.  The gain can be adjusted by changing the value of R4 (L+R), 3.3k resistor.  A gain of more than 11 is not recommended.  Reduce R4 for higher gain and vice versa.  Gain is equal to ( R3 /R4 ) +1.

The Left channel is shown, along with the crossfeed parts for the Right channel.  SW1 bypasses the crossfeed network.  I have reconfigured the original crossfeed schematic so that now the 100k resistor always bridges the bypass switch and thereby reduces any 'crackle' or 'click' or whatever you may call them.  Don't omit these 100K resistors as they form a part of the crossfeed network and omitting them would bear undesirable results.  Note that R6 and R9 are indicated as 4.53k, however the use of 4.7k resistors will be perfectly adequate in practice.

U2 and U3 are paralleled as current-boosting amps.  This doubles the output current into the load as established by Burr-Brown's AB-051 application note: Double The Output Current To A Load With The OPA2604 Audio Opamp.

Meraj suggests that the value of the output resistors might require a little bit of experimenting for optimally matching the amp with your headphones.  However as shown, output impedance is close to zero, and changing R9 and R10 will not affect impedance.  Some headphones are designed for an impedance of 120Ω, and for these I suggest that a 120Ω resistor be installed in series with the output.  Many modern phones will prefer zero ohms impedance.

The opamp power supply pins were not shown in the diagram for clarity.  These pins are bypassed by 10uF and 100nF decoupling caps.  100nF caps should be placed between the supply pins (as close to the IC as possible), and the 10µF caps between each supply and ground (typically at the point where power enters the board).

Only one channel is shown, so two units are needed for stereo.


For the prototype, I used Veroboard and have found the amp to be very tolerant of layout.  I've made boards based on the prototype and ESP may make them available to others when there is enough demand to offset costs.  Needless to say, these boards would make construction a breeze.  I used 1/4W carbon resistors throughout.  Considering the level of ambient noise that a portable system has to put up with, the volume level would usually be high enough to make it impossible to discern noise from signal - however I still recommend metal film resistors for best results.

fig 2
Figure 2 - Work In Progress

I chose the NE5532 for this project.  Since the source is a PDA's internal DAC, I didn't see the need to use premium opamps.  Of course, if it makes you feel better, you can always use higher quality (expensive) opamps.  Just make sure the opamp is capable of driving low impedances.  LM6171, OPA2134, OPA2132, OPA134 and OPA4134 (dual) are some possible substitutes.  It's likely that there are others.  IC sockets are therefore a good idea if you have plans to upgrade the opamps.

The volume pot should be a linear type and would give, with the 15k resistor in parallel, the benefits outlined in ESP's A Better Volume Control.

The crossfeed is on a separate board in the prototype.  I mounted it vertically on the main board using hot-melt glue.  All the switches, jacks and volume control were also mounted on the enclosure using a hot-melt glue gun.  I used generous amounts of hot-melt glue around the bases of all the capacitors as they are more susceptible to lead and track breaking due to vibrations.

For the enclosure, I chose what used to be a part of a plastic school lunch box.  I measured and marked the spots for the cuts I had to make.  A sharp hobby-knife, a drill bit, a tabletop vise and a steady hand were all that I needed for the job.  When working with plastic, it's a very good practice to measure twice and cut once (he spake from bitter experience).

fig 3
Figure 3 - Testing ... testing ... 1, 2, 3 ...

The belt clip was made for Nokia and came into my possession when I bought a 7110 aeons ago.  If you use a belt clip, level it to make sure that the amp, with all the jacks sticking out, doesn't get in the way of your belly when you sit down.  Trust me, it can be very painful!

By far the most expensive part of this project is the paint job.  I went through a can of flat black and a can of clear lacquer to get the finish.

Power Supply

My prototype uses two 9V alkaline batteries to give 9-0-9V supply.  I get around 20 hours of operation at normal portable listening levels.  The effect of the demise of a few pairs of alkaline batteries on my wallet has decided me to switch over to rechargeable batteries.  A battery charger is now under construction.

This amp can also be powered by Project 05 using a 15-0-15V transformer.  A 5VA transformer should have oomph enough for the job.

A star ground was not necessary for my battery powered version but is recommended for a mains powered one.  Use the common point of the power filter caps as the ground return and employ a ground loop breaker if you use a metal enclosure.

The Fruits of Labour?

The first thing that you notice about the sound is the authority of the low frequency.  The sensation of the kick drum's 'kick' (pun intended) on the earlobes greatly enhances the listening experience.  Only now do I realise the full potential of the Sennheiser PMX 60.

With the crossfeed on, the vocal that used to seem to be right on top of one's nose is pulled forward.  The perceived depth in the sound stage and the bass is very much dependent on the source material.  For some materials, loss in bass is experienced with the crossfeed on.  This is due to the cancellations of the unrealistic, out-of-phase signals.

I find the crossfeed to be satisfactory for listening to classical and pop, rock gets mixed results and death metal is less confusing because of the cleared up sound stage.

At the time of writing, I had just finished a 15-0-15V power adaptor based on the Project 05.  The improvement in the sound brought forward by the increased voltage is just amazing! As my ears were recovering, I was on my way to hunt down an enclosure that would house the project along with four 9V batteries.  That's ±18V - I must be crazy!

ESP Comments

The project shown here has the potential to be extremely good, depending on the construction.  The output opamps (U2/3) need to be able to supply enough current to get acceptable sound levels, but with most phones this isn't likely to be an issue.  While Meraj doesn't consider the NE5532 to be 'premium', it's still an awesome opamp, and there aren't many that can equal it even at much higher cost.  There are many opamps that cost a great deal more than the NE5532, but don't even come close to its performance !

The power supply is obviously important, and the usefulness of a battery powered headphone amp is (IMO) somewhat dubious.  Using a mains supply (based on P05 or similar) will generally give better results.  If you do use 9V batteries, I suggest that you use a minimum of 1,000µF for supply bypass.  The internal impedance of 9V batteries isn't low enough to provide good performance with widely varying current drain.

The case used for the unit shown here is a suggestion only - for a fixed (mains powered) unit, a more traditional aluminium enclosure would be better.  Make sure that it's big enough t allow good separation between the electronics and the power transformer.  It's surprising how little noise becomes very audible with headphones!


HomeMain Index ProjectsProjects Index
Copyright Notice.This article, including but not limited to all text and diagrams, is the intellectual property of Meraj Salek and Rod Elliott, and is © 2004.  Reproduction or re-publication by any means whatsoever, whether electronic, mechanical or electro-mechanical, is strictly prohibited under International Copyright laws.  The publisher (Rod Elliott) grants the reader the right to use this information for personal use only, and further allows that one (1) copy may be made for reference while constructing the project.  Commercial use is prohibited without express written authorisation from Rod Elliott.
Page Created and Copyright © Rod Elliott - 27 Oct 2004