NE5532 Headphones Preamplifier/Amplifier with EQ

[afternath]

Hi guys this is my first DIY audio project, it's a Headphones Preamplifier/Amplifier based on a NE5532 opamp, BD139/140 transistors and ELNA SILMIC II electrolytic capacitors.
The power source comes from a 12V AC to DC wall plug, this is why i split supply.
The project is currently in progress, all was tested successfully except the EQ stage.
I test all my audio setup with wave generator, scope and when it's ok i launch a RMAA audio test.
I simulate all the circuit with ISIS, the PDF file is attached.
RMAA results are available if you want.
I will really appreciate your comments

Regards

[Art M]

D8, D9 configuration looks wrong ?

[kroto]

33dB (+ EQ boost) of gain is... wooow...
I think it's too high for a headphone amp.

use less than 20dB is good value.

[afternath]

Originally Posted by Art M

D8, D9 configuration looks wrong ?

Why does it look wrong? This is the basic push-pull setup with crossover diodes.

Originally Posted by kroto

33dB (+ EQ boost) of gain is... wooow...
I think it's too high for a headphone amp.

use less than 20dB is good value.

It's depend of the input level, i've tested it on several headphones with ipod input at 50% and it's the best IMO.

[!]

PSU has a regulated positive, then depends on a theoretical perfectly matched set of resistors for the voltage divider to create a virtual ground. Scrap that, use a normal center tapped transformer and positive and negative rail regulators. If you absolutely insist on using the R3/R4 voltage divider instead, use lower value resistors and of course match them... dropping to 470 ohm for R3/R4 for example you're still only at 12mA loss, 9mA difference (negligible unless it's going to be battery powered but if it is, you don't need the regulation stage).

Why NE5532? It's ok, but I wouldn't use it just because someone else put it in a schematic if that's why it is the default choice. If you need a dual opamp DIP8 package, consider TI's OPA2107. Granted, it will make a little less difference because you have a buffer (push pull subcircuit) but I still feel it's a worthwhile change.

What is the C10, 10uF capacitor going to be? Giant expensive film cap or some little electrolytic? IMO, you can use a smaller value high quality film cap with a redesign.

Don't rely too much on your signal readings, that's not telling you how it's driving a load. Where is the push pull subcircuit? Put output from that in the feedback loop for the opamp "volume stage".

Also, 12V input and THEN split in half? That's close to the minimum spec of +-5V for the opamps. OK for low-z cans, but NE5532 can operate up to +-22V, better to use a 24V center tapped transformer and end up with roughly +-15VDC after positive and negative rail linear regulators and other losses... unless you just want to use parts you already have.

What's going on with the PSU R2 and R6 resistors? A decision that it's important to dial in an exact rounded off voltage (number)? That's a human fault, thinking it matters to have a round number, the circuit does not "care" so R6 is wasted and should be removed. Also, the voltage divider is a bit screwy, who specs use of an uncommon value of 82 ohms for R1? Ok it's not as rare as hen's teeth but ideally a circuit would use the most common components when possible. Not to nitpick, if you have an 82 ohm resistor lying around ok, otherwise pick more common values.

As for the choice of BD139/140, they are a fine choice, just be aware that to gain higher current (relative to typical TO92 package transistors commonly used in audio) you will lose a little resolution, smoother but less detailed audio. To some people this is a good thing, to others it is not. Using NE5532 will offset this a little, it's a pretty *honest*/uncolored sounding opamp while my suggestion above for OPA2107 with BD139/140 in the buffer could sound a little sedated to some people (again a personal/subjective decision...)

[!]

Also regarding BD139/140, and I am NOT necessarily suggesting you change them, but keep in mind that with typical >=32 ohm cans, at +-6V power rail voltage your output is limited to a little over 250mA so while you may or may not like these transistors the best, 1.5A transistors aren't "necessary"... I'd probably use them anyway, I like power margin in parts.

[Mooly]

Originally Posted by Art M

D8, D9 configuration looks wrong ?

I agree...

If you take the output stage in isolation and leave L_EQ "floating" or in other words "not connected" then Q3 and Q4 will conduct heavily being biased on via R7 and R8.

If L_EQ is now taken to ground potential (signal ground) which is the same DC potential as the opamp output driving it, then D8 and D9 do not conduct... they don't do anything...

In practice it may actually work (in that you get audio out) because the gain of the BD139/140 is low, R7 and R8 are "high", and the supplies are "low" so there may not be enough bias to fully saturate the transistors.

Have you measured the DC voltage at the output as it appears not to be under any kind of global feedback. Is it at 0.00volts with respect to signal ground.

Edit... The PSU is OK but an OP07 is overkill. A 741 type is all you need. You don't need any kind of precision at all. Signal ground is your reference point. R5 seems a touch low not leaving much current for the BD139/140. It will certainly all work though.

[afternath]

Originally Posted by !

PSU has a regulated positive, then depends on a theoretical perfectly matched set of resistors for the voltage divider to create a virtual ground. Scrap that, use a normal center tapped transformer and positive and negative rail regulators. If you absolutely insist on using the R3/R4 voltage divider instead, use lower value resistors and of course match them... dropping to 470 ohm for R3/R4 for example you're still only at 12mA loss, 9mA difference (negligible unless it's going to be battery powered but if it is, you don't need the regulation stage).

First, i don't want using main power directly in my amp (i'm a beginner in this).
Using lower values for R3/R4 is useless because of the huge opamp input impedance (120G !!), about matching i use 1% metal resistors in all the amp.

Originally Posted by !

Why NE5532? It's ok, but I wouldn't use it just because someone else put it in a schematic if that's why it is the default choice. If you need a dual opamp DIP8 package, consider TI's OPA2107. Granted, it will make a little less difference because you have a buffer (push pull subcircuit) but I still feel it's a worthwhile change.

About using NE5532, i was very sceptic about its poor stability without the feedback cap and its offset (~7mV measured in the schematic setup). This is why i started with a TL072, much more stable and with low offset. But after measuring output noise i decided to take the NE5532 which was so quiet.

Originally Posted by !

What is the C10, 10uF capacitor going to be? Giant expensive film cap or some little electrolytic? IMO, you can use a smaller value high quality film cap with a redesign.

C10 is a 10µF ELNA SILMIC II Electrolytic cap (is it a good choice?). I would like you to explain me why lot of people use film cap?

Originally Posted by !

Don't rely too much on your signal readings, that's not telling you how it's driving a load. Where is the push pull subcircuit? Put output from that in the feedback loop for the opamp "volume stage".

What is a push pull subcircuit?

Originally Posted by !

What's going on with the PSU R2 and R6 resistors? A decision that it's important to dial in an exact rounded off voltage (number)? That's a human fault, thinking it matters to have a round number, the circuit does not "care" so R6 is wasted and should be removed. Also, the voltage divider is a bit screwy, who specs use of an uncommon value of 82 ohms for R1? Ok it's not as rare as hen's teeth but ideally a circuit would use the most common components when possible. Not to nitpick, if you have an 82 ohm resistor lying around ok, otherwise pick more common values.

I agree with you about R6 but i'm a little perfectionist
I have a 10ohm to 1Mohm 1% metal resistor E12 kit so i don't care common values...

Originally Posted by !

As for the choice of BD139/140, they are a fine choice, just be aware that to gain higher current (relative to typical TO92 package transistors commonly used in audio) you will lose a little resolution, smoother but less detailed audio.

I don't understand, can you explain more?

Originally Posted by !

To some people this is a good thing, to others it is not. Using NE5532 will offset this a little, it's a pretty *honest*/uncolored sounding opamp while my suggestion above for OPA2107 with BD139/140 in the buffer could sound a little sedated to some people (again a personal/subjective decision...)

I LOVE uncolored sound !!
IMO an amplifier must be transparent

[Anteaus]

D8, D9 are the wrong way. They need to be in the conducting direction. The way you have it, smoke instead of sound as the output pair will conduct hard.

Also, You may need to select the diodes to give the required quiescent current, as there is a fair tolerance on diode characteristics. Too little and you will have distortion at low sound levels, too much and the output stage will run hot.

As for the supply. I'd be inclined to either use a single-rail design, or if you have AC available, a proper dual rail design. Having the signal lines at non-ground potential could lead to all sorts of issues. Not to say it won't work, but he who ventures onto untried ground... finds out whether it works or not, and maybe saves others the trouble!

Carbon resistors are quite OK unless you are working at millivolt levels such as on a mic input. Metal film are preferable for audio circuits, though.

Finally, the standard and most important advice in amp design, watch how you return your grounds to common. If the load returns its current to the PSU via a wire shared with a signal ground, you will have instability. Keep them separated and you are OK. This is one of the things that can't be determined from a schematic, but where actual physical wiring matters.

HTH.

[!]

Originally Posted by afternath

First, i don't want using main power directly in my amp (i'm a beginner in this).

Ok, "want" is fair, I was only suggesting a better way with center tapped tranfo.

Using lower values for R3/R4 is useless because of the huge opamp input impedance (120G !!), about matching i use 1% metal resistors in all the amp.

"G"? Can we talk in recognized terms? I'm not suggesting it won't work, only that relative to the time and expense, there isn't a good reason to do it this way.

About using NE5532, i was very sceptic about its poor stability without the feedback cap and its offset (~7mV measured in the schematic setup). This is why i started with a TL072, much more stable and with low offset. But after measuring output noise i decided to take the NE5532 which was so quiet.

I don't know where you are getting opamp tips from, but TL072 is garbage, out of the question to use. NE5532 is merely cost effective when pennies matter to a mass manufacturer.

C10 is a 10µF ELNA SILMIC II Electrolytic cap (is it a good choice?). I would like you to explain me why lot of people use film cap?

Sound quality. It's always interesting that people challenge decades of research and wisdom.

What is a push pull subcircuit?

Transistor arrangement following the opamp to buffer the current.

I agree with you about R6 but i'm a little perfectionist
I have a 10ohm to 1Mohm 1% metal resistor E12 kit so i don't care common values...

There is nothing at all "perfectionist" about it, it's not only not perfect it's worse. The arbitrary (meaning illogical and self-defeating, and worse) human idea that a rounded off number matters, is basically an insanity with nothing at all good about it.

I LOVE uncolored sound !!
IMO an amplifier must be transparent

It never is. Through experimentation you can reach what you feel sounds best and that is all.