Want to build a headphone amp.

[jdraughn]

Hi all.

I am looking to build a little headphone amp and am looking at building this one http://sound.westhost.com/project113.htm

Could I get some advice on whether that is a decent design or not? Also, I would like to make it as small as possible and I like to work with 0402 and 0603 size resistors and caps if possible.  Can anyone tell me what parts should be rated for some wattage? I am guessing that the 4.7 ohm resistors should be of higher wattage. How low can I go? I was thinking of running it with 3 or 4 Li-Ion batteries for the top rail and 3 or 4 Li-Ion batteries for the bottom rail to make it portable.

Any gotchas I should be aware of?  I have a pair of Sennheiser 555 headphones that I want to build this for. They sound great just running out the headphone jack on my phone or laptop, but from what I have read they should sound a bit better with some real power behind them.

I am not set on that schematic and am open to any other circuits.  Thanks

[Richard Crowley]

So if you plug your Sennheiser 555 headphones into a high-power source like a traditional stereo receiver, can you really hear a significant difference?

The Elliot design is certainly "decent" but it is not very "portable" in any practical sense of the word. When you have so many (relatively) giagantic components like all the electrolytic capacitors and output transistors, it is not at all clear why you would be concerned about the size of the smallest components (the resistors).  Getting them down to 0402 or even 0603 seems pointless when they are already such a tiny part of the size of the whole circuit.

[homebrew]

I don't see the 'superb' characteristics either. It might be a working amplifier - I mean why not. Yet another Class-B design. But I would be almost certain that standard integrated one-chip solutions might have even better audio performance ...

[DmitryL]

Why not to build a more decent thing from the same author ?
http://sound.westhost.com/highspeed.htm
I'm pretty happy with that device.

Not for the beginner though..

[Fungus]

I don't see the 'superb' characteristics either. It might be a working amplifier - I mean why not. Yet another Class-B design. But I would be almost certain that standard integrated one-chip solutions might have even better audio performance ...

Agree.

a) In whose opinion is it 'superb'? Compared to what?
b) Taking the exact same signal you're listening to now and passing it through an amplifier can only make it louder, it can't possibly improve it.

If you have USB output then you'll have a really hard time beating something like this: http://www.ebay.com/itm/141767703384

(seller chosen at random, yes I own one of those and have listened to it a lot)

And... that one comes in a nice box. You can get a PCM2704 DAC devices without a box for far less than that.

[dom0]

prefer to use a diamond buffer, instead of an emitter follower.

What if I told you the diamond buffer is just an emitter follower, too?


e:
I don't really like highspeed.htm for their fixed ~10 ? output impedance. There are better ways around that.

[tszaboo]

Could I get some advice on whether that is a decent design or not? Also, I would like to make it as small as possible and I like to work with 0402 and 0603 size resistors and caps if possible.

I made the same mental mistake years ago. Think of a resistor, where sound goes through it. With higher volumes/amplitudes, it will self heat. Say, temperature of the resistor changes 1 degrees, and you used a 100ppm/degrees resistor. 100ppm is -80dB. It will depend on the circuit whether or not this is audible, or at least it will worsen your THD and other results. just 600uA over a 1K resitor is almost a miliwatt, it will heat up a 63 miliwatt reistor. So I'm afraid, those 0402,0603 resistors are too susceptible to these secondary effects.
 

[DmitryL]

I don't really like highspeed.htm for their fixed ~10 ? output impedance. There are better ways around that.
[/quote]

Hmm.. remove 10 ohm resistor and you'll have much lower output impedance

[suicidaleggroll]

b) Taking the exact same signal you're listening to now and passing it through an amplifier can only make it louder, it can't possibly improve it.

:rolleyes:

I know this forum is crazy anti-audiophile, but come on...you can't possibly think that any device capable of generating a 20-20kHz tone is also capable of preserving that tone, without distortion or attenuation when driving any load between 10-1000 ohms, can you?  Why even bother having an amplifier or a buffer in the first place?  Why not just drive the load straight off the DAC?

Why is it that many people here will spend thousands of dollars on multiple 8.5 digit multimeters to measure standard circuits that couldn't possibly need more than 3.5 digits of accuracy, but as soon as any audio-related topic is brought up, anything more than 8 bits of resolution and a single op-amp buffer is "audiophoolery"?

The headphone outputs in many devices are shit.  Not all, many decent stereo receivers have perfectly adequate ones, but phones and laptops are notoriously crap.  Plug a decent set of headphones into most of them with and without an amp and you'd be hard-pressed NOT to hear a significant improvement with the amp.  Of course the DAC in most of them is crap too, so adding an amp is only half the battle...

OP - I know nothing about that amp, but I do have experience with a lot of AMB's designs.  The Mini3 is very portable and works great, and with the optional li-ion battery it lasts for a very long time between charges.

[Marco]

Still, I would prefer to use a diamond buffer

Assuming a very low impedance source does a diamond buffer actually have better linearity than an A/B stage?

[dom0]

Not really. If you use transistors with very good complimentary-ness approximately identical collector currents in the first and second stages you'll get some distortion cancellation. It's not really relevant in a closed-loop system.

Could I get some advice on whether that is a decent design or not? Also, I would like to make it as small as possible and I like to work with 0402 and 0603 size resistors and caps if possible.

I made the same mental mistake years ago. Think of a resistor, where sound goes through it. With higher volumes/amplitudes, it will self heat. Say, temperature of the resistor changes 1 degrees, and you used a 100ppm/degrees resistor. 100ppm is -80dB. It will depend on the circuit whether or not this is audible, or at least it will worsen your THD and other results. just 600uA over a 1K resitor is almost a miliwatt, it will heat up a 63 miliwatt reistor. So I'm afraid, those 0402,0603 resistors are too susceptible to these secondary effects.

This increases low frequency distortion (the thermal mass plus the thermal resistance to ambient acts pretty much as a 1st order low pass), incidentally low frequency distortion of speakers and headphones is already pretty bad. I would worry more about risk of thermal failure if someone inputs too high input levels and such.

[DanielS]

you can't possibly think that any device capable of generating a 20-20kHz tone is also capable of preserving that tone, without distortion or attenuation when driving any load between 10-1000 ohms, can you?  Why even bother having an amplifier or a buffer in the first place?  Why not just drive the load straight off the DAC?

I would expect any decent signal generator to do a fairly good job from 50 ohms to infinity at the very least - driving 50 ohms is what they are designed for. But that signal is not straight out of a DAC, it is already internally buffered by op-amps, just like most audio sources do too, but not all sources have low impedance output drivers and even when they do, they are only designed for standard 1-2Vpp line-level output which is nearly useless for driving high impedance headphones like the 120 ohms 555.

[Fungus]

b) Taking the exact same signal you're listening to now and passing it through an amplifier can only make it louder, it can't possibly improve it.

:rolleyes:

I know this forum is crazy anti-audiophile, but come on...you can't possibly think that any device capable of generating a 20-20kHz tone is also capable of preserving that tone, without distortion or attenuation when driving any load between 10-1000 ohms, can you?

Assuming current capability isn't the problem; the only difference between a 10 Ohm and a 1000 Ohm load is the voltage needed to power it to the required level. He's not complaining about the volume level so I don't think increasing the voltage will help. Current drive capability? That's what capacitors are for. Almost any old capacitor will do at 20kHz.

Why even bother having an amplifier or a buffer in the first place?  Why not just drive the load straight off the DAC?

Because a typical DAC isn't designed to output any possible voltage range, it's assumed you'll connect it to an amplifier. They don't give DACs much current drive ability either, and for the same reason.


His laptop/phone already has an op-amp in it and he says his headphones "sound great" when connected. The problems he has (if any) aren't voltage/current related, they're about noise floor, dynamic range, etc. Getting an external DAC is therefore a good start. OK, it doesn't solve the problem of connecting them to his phone, but it's a start. Probably a much better start than building that amplifier. For $8 I'd get one for the laptop. If the phone suddenly sounds a lot worse than the laptop then start working on that problem.

PS: If you want a PCM2704 with an amplifier they do those too, eg.: http://www.ebay.com/itm/121292071586  Yes, I own one. No I can't hear any difference between that and the non-amplified one (not with my headphones anyway, YMMV).

Don't worry, there's a whole subculture dedicated to swapping the capacitors in those boxes if that's your thing (or you can just pay a couple of bucks extra for ELNA brand capacitors - the audiophile's choice)

[krivx]

b) Taking the exact same signal you're listening to now and passing it through an amplifier can only make it louder, it can't possibly improve it.

:rolleyes:

I know this forum is crazy anti-audiophile, but come on...you can't possibly think that any device capable of generating a 20-20kHz tone is also capable of preserving that tone, without distortion or attenuation when driving any load between 10-1000 ohms, can you?  Why even bother having an amplifier or a buffer in the first place?  Why not just drive the load straight off the DAC?

Why is it that many people here will spend thousands of dollars on multiple 8.5 digit multimeters to measure standard circuits that couldn't possibly need more than 3.5 digits of accuracy, but as soon as any audio-related topic is brought up, anything more than 8 bits of resolution and a single op-amp buffer is "audiophoolery"?

The headphone outputs in many devices are shit.  Not all, many decent stereo receivers have perfectly adequate ones, but phones and laptops are notoriously crap.  Plug a decent set of headphones into most of them with and without an amp and you'd be hard-pressed NOT to hear a significant improvement with the amp.  Of course the DAC in most of them is crap too, so adding an amp is only half the battle...

OP - I know nothing about that amp, but I do have experience with a lot of AMB's designs.  The Mini3 is very portable and works great, and with the optional li-ion battery it lasts for a very long time between charges.

Hear, hear! Pun intended.

[suicidaleggroll]

Assuming current capability isn't the problem; the only difference between a 10 Ohm and a 1000 Ohm load is the voltage needed to power it to the required level. He's not complaining about the volume level so I don't think increasing the voltage will help. Current drive capability? That's what capacitors are for. Almost any old capacitor will do at 20kHz.

Of course current capability is the problem!  What does rail capacitance have to do with the source's output impedance?  Or are you actually suggesting putting a capacitor on the output of the source, as if that would help drive a low impedance load?

Because a typical DAC isn't designed to output any possible voltage range, it's assumed you'll connect it to an amplifier.

A 5V DAC has plenty of voltage swing for most headphones.

They don't give DACs much current drive ability either

Same with the crappy output stage in most phones/laptops...

[Alex Nikitin]

Hi all.

I am looking to build a little headphone amp and am looking at building this one http://sound.westhost.com/project113.htm

Could I get some advice on whether that is a decent design or not? Also, I would like to make it as small as possible and I like to work with 0402 and 0603 size resistors and caps if possible.  Can anyone tell me what parts should be rated for some wattage? I am guessing that the 4.7 ohm resistors should be of higher wattage. How low can I go? I was thinking of running it with 3 or 4 Li-Ion batteries for the top rail and 3 or 4 Li-Ion batteries for the bottom rail to make it portable.

Any gotchas I should be aware of?  I have a pair of Sennheiser 555 headphones that I want to build this for. They sound great just running out the headphone jack on my phone or laptop, but from what I have read they should sound a bit better with some real power behind them.

I am not set on that schematic and am open to any other circuits.  Thanks

I've designed a number of production headphone amps (the first one was in 1994, IIRC and there are some still manufactured). To choose a suitable circuit you need to look at the requirements.

Well, let's list main points:

1) Portable with a battery supply

2) 120 Ohm nominal headphone impedance

3) More power and better quality than a usual portable device output.

4) Reasonably small size

5) Not overcomplicated.

#1 and #4 more or less exclude the circuit you've linked.

#2 and #3 means a fairly healthy output voltage (and hence the supply voltage too) - at least 3-4V RMS, ~12V p-p, so at the very least 12V supply is required, more is better.

If you can leave with two 6V (or even two 12V) batteries, a good choice would be just a quality dual opamp with healthy current output capabilities, for example the LM6172, set for an inverting gain about 5, with 10K volume control on the input and 10 ohm series resistors on the output.

Cheers

Alex

P.S. - here is the circuit, you would need a dual switch for the batteries and perhaps an LED to show the power is on.

[Lightages]

It is fun to experiment. It is fun to design circuits and build them. There is great satisfaction in having a nice project that is complete and working like you want at the end. If these are your goals in building the headphone amp then be prepared to spend much more than you can buy one for, such as recommended by Fungus. Designing, building, and redoing it all over it again many times until you are happy is not the cheap route.

@Fungus:

The first USB DAC you pointed to looks like a steal. I am going to get one to play.

[calexanian]

The op amp driving a B class complementary output assist transistor circuits work reasonably well, but they are a far way off then it comes down to THD from purpose built headphone drivers. If you are just playing around and like the way elliott presents his projects i would recommend just building two low voltage discrete full complimentary power amps at say +/- 10 volts. You will learn about long tail pair stages, biasing BJT's and the basic complementary power amplifier circuit that is at the heart of all of the amp on a chip systems such as the 3886 and such.

If you just simply want to build a project there are many many kits out there with everything ready to go.

If you want to save money, just buy a cheap rolls headphone amp from guitar center or amazon for $20

[retrolefty]

Why is it that many people here will spend thousands of dollars on multiple 8.5 digit multimeters to measure standard circuits that couldn't possibly need more than 3.5 digits of accuracy, but as soon as any audio-related topic is brought up, anything more than 8 bits of resolution and a single op-amp buffer is "audiophoolery"?

 Pot calling kettle black your saying? Difference is the volt-nuts don't obtain their 8.5 meters to measure 'standard circuits' but rather to measure and compare with their other meters. Sort of measurement masturbation?

 There is some subjective excesses in most all of us when you are talking a favorite endeavor.

[Fungus]

Difference is the volt-nuts don't obtain their 8.5 meters to measure 'standard circuits' but rather to measure and compare with their other meters. Sort of measurement masturbation?

8.5 digit meters are also objective. You can tell if they're working, you can measure exactly how much better instrument A is than instrument B.

With audio there's a fair amount of subjectivity (and personal taste!) involved. Some people actually want distorted (aka "warm") sound.

It also varies depending on what you had for lunch.

[jdraughn]

Everyone - thank you for your input. I am going through it all and studying your answers now. 

[krivx]

Why is it that many people here will spend thousands of dollars on multiple 8.5 digit multimeters to measure standard circuits that couldn't possibly need more than 3.5 digits of accuracy, but as soon as any audio-related topic is brought up, anything more than 8 bits of resolution and a single op-amp buffer is "audiophoolery"?

 Pot calling kettle black your saying? Difference is the volt-nuts don't obtain their 8.5 meters to measure 'standard circuits' but rather to measure and compare with their other meters. Sort of measurement masturbation?

 There is some subjective excesses in most all of us when you are talking a favorite endeavor.

The response to the 8.5 digit buyers is not nearly as dismissive or even hostile as that for people doing anything to do with audio. An extremely objective design case backed with rigorous data will still get derailed with jokes about magic crystals or claims that amplifiers can only make signals louder. I think it's a shame - low-noise and low-distortion design can be very interesting but it's not received well on the forum.

[Lightages]

I think it's a shame - low-noise and low-distortion design can be very interesting but it's not received well on the forum.

Yeah, I think that there are many on the forum here who jump on audio design as being woowoo even if it is practical and science based design. I think that they are just parroting the behaviour of the audiphool bashers without understanding the reasons and therefore bash anything as soon as it refers to audio design.

To me a good headphone design needs to be low noise, low output impedance (<2 ohm), and of course low noise/distortion on the order of less than 0.05%.

The original circuit referenced by the OP is a good start to play with. Some people just run headphones directly from NE5532 op amps. If this is your first design and project, then going for the smallest parts is not going to make it easy. Start with through hole parts and make it easy to make changes.

[dom0]

btw. here's a lot of data on headphones: http://www.innerfidelity.com/images/AllGraphs.pdf

[retrolefty]

Difference is the volt-nuts don't obtain their 8.5 meters to measure 'standard circuits' but rather to measure and compare with their other meters. Sort of measurement masturbation?

8.5 digit meters are also objective. You can tell if they're working, you can measure exactly how much better instrument A is than instrument B.

With audio there's a fair amount of subjectivity (and personal taste!) involved. Some people actually want distorted (aka "warm") sound.

It also varies depending on what you had for lunch.

 Both groups, volt-nuts and audio-nuts, are married to their passions.

 The audio group judges their equipment and improvements by listening to music not sine waves, so measurements will never be the end all of attempting improvements in their journey to audio nirvana.

 The volt-nuts (along with time-nuts) are seeking the nirvana of exact repeatable measurements at their finger tips to go forth and make judgments and decisions on circuit and other equipment improvements they dwell in.

 Both are just examples of human passions pushing towards the edges of extreme behaviors at times. Outside of a commercial endeavor there is probably little justification for anyone to obtain a 8 1/2 digit DMM. Conversely a high end audio reproduction system does not require a budget of tens of thousands of dollars chasing a subjective immeasurable goal.