headphone amplifier. Which IC? Differential amplifier?

[Chaos_Klaus]

Hey EEVBlog-Forum,

I am building a device that has a headphone output. It is upposed to be used in a (semi) professional music studio for monitoring purposes during live recording. However it is powered with only +5V.

Typical studio headphones have impedances between 32Ohms and as high as 250Ohms or higher. Now I am looking for an IC that can do the final amplification for the headphone output.

I browsed the ti.com and found these two and can not quite make out which one would suit me, or if they are possibly both too week to drive enough power into a 250 Ohm load.

http://www.ti.com/lit/ds/symlink/tpa152.pdf
http://www.ti.com/lit/ds/symlink/lm4880.pdf

Someone on another forum suggested I should consider a differential amplifier. I work with a virtual ground reference anyway. I found this one:

http://www.ti.com/lit/ds/symlink/ths4522.pdf

It's quite expensive, but it the specs look extremely good from an audio point of view (at least to my eyes). It also has rail-to-rail output which is handy in the 5V application. However I am confused by "nagative rail input". Does that mean the inputs need to be negative?  Because that would disqualify this chip.

What is the advantage of a differential amplifier? Less noise?

Thanks for your time!

Cheers!

Edit: I am a bit concerned, because it would be a shame if the device would not give enough output. I need some volume if a drummer is upposed to hear something while playing even with closed headphones. I hooked up my oscilloscope to the headphone output of my USB-audio-interface and 4V peak-to-peak on my is already extremely loud on my 62Ohm headphones.

[tszaboo]

Hi. Check out the O2 (Objective 2) headphone amp. That was made by an engineer.
Not sure why you would consider using a 100Mhz+ opamp to amplify 20Khz signals.

[Chaos_Klaus]

Not sure why you would consider using a 100Mhz+ opamp to amplify 20Khz signals.

Well, It did not cross my mind until I read the datasheet where it is specificly mentioned audio applications ...

As for the O2 ... I obviously thought about using a seperate device, before. It's not a viable option. Too many cables. An additional wall wart.

Maybe I should rephrase my question:

Can someone recommend an IC that can act as a headphone amplifier with loads between 32Ohm and 250Ohm? It should work with 5V single supply and have high output swing.

[Rick60]

If your budget can stretch http://www.ti.com/product/tpa6120a2 

edit just finished layout with this , but it needs -5v so not what you require 

[ciccio]

With a single +5 V supply, even if you find an amplifier that can swing  rail to rail, you have 5 V p-p, that means less than 1.8 V rms.
That results in 98 mW on 32 ohm load, 1 mW on 2500 ohm load.
That power level is absolutely inadequate in a noisy studio. even with efficient headphones.
With a bridge amplifier (if you can correctly wire the headphones with a 4 pin connector) you'll get 4 times that power, still inadequate (it's the same power you could get with a 10 V supply).
In standard intercom systems, used in live performance, the supply ranges from 24 to 30 V, and the headphones are very efficient, but not high fidelity. Sometime the operators say the level is too low...
I suggest a DC-DC converter, from 5 V to 24 V, with suitable filtering. In this case you can use standard op-amps (NE5532), paralleling  2 or 3 of them for an increased current output.

Best regards

[Zero999]

That results in 98 mW on 32 ohm load, 1 mW on 2500 ohm load.
That power level is absolutely inadequate in a noisy studio. even with efficient headphones.

Your hearing must be bad then.

32Ohm headphones powered by a 3V amplifier is unbearably loud for me at full volume. I've never tried 250Ohm headphones so you may be right about those but 32Ohm should be more than enough for a 5V amplifier

[ciccio]

It's not my hearing (obviously it was better when I was younger..).
The cameramen at a live music concert, or the musicians in a recording studio,  are often in a very high sound environment.
Headphones must be loud, to overcome the external noise.
High power is required for good dynamic, or you'll get distortion.

Best regards

[Zero999]

It's not my hearing (obviously it was better when I was younger..).
The cameramen at a live music concert, or the musicians in a recording studio,  are often in a very high sound environment.
Headphones must be loud, to overcome the external noise.
High power is required for good dynamic, or you'll get distortion.

Best regards

No wonder tinnitus and deafness is so much for a problem for musicians when they get older.

[dom0]

TPA6138A2

Well. Actually it's for 3.3 V systems, but that shouldn't be a problem.

[retrolefty]

That results in 98 mW on 32 ohm load, 1 mW on 2500 ohm load.
That power level is absolutely inadequate in a noisy studio. even with efficient headphones.

Your hearing must be bad then.

32Ohm headphones powered by a 3V amplifier is unbearably loud for me at full volume. I've never tried 250Ohm headphones so you may be right about those but 32Ohm should be more than enough for a 5V amplifier

 Without knowing the sensitivity ratings of a specific headphone it's hard to say how much voltage swing is too much/loud. 32 ohm rating is an impedance rating of the headset, not it's SPL sensitivity (efficiency) rating.

[dom0]

The average headphone delivers about 90 dB SPL/1 mW. Some a few dB more, some a few less. Ballpark stays the same...

[Richard Crowley]

With a single +5 V supply, even if you find an amplifier that can swing  rail to rail, you have 5 V p-p, that means less than 1.8 V rms.
That results in 98 mW on 32 ohm load, 1 mW on 2500 ohm load.
That power level is absolutely inadequate in a noisy studio. even with efficient headphones.
With a bridge amplifier (if you can correctly wire the headphones with a 4 pin connector) you'll get 4 times that power, still inadequate (it's the same power you could get with a 10 V supply).
In standard intercom systems, used in live performance, the supply ranges from 24 to 30 V, and the headphones are very efficient, but not high fidelity. Sometime the operators say the level is too low...
I suggest a DC-DC converter, from 5 V to 24 V, with suitable filtering. In this case you can use standard op-amps (NE5532), paralleling  2 or 3 of them for an increased current output.

Best regards

Yes I absolutely agree with @ciccio.  Your 5V supply makes this proposal a complete non-starter.
Furthermore, any kind of bridged/floating solution is completely inappropriate for conventional headphone applications. However powered.

If someone came to me with those specs I would reply: No Bid.  Come back when you have a better grasp on reality.

[dom0]

The state-of-the-art are amplifiers with integrated charge pumps (like the one I mentioned above), so no, they're not bridged. About 5 Vpp or so is enough output swing for any headphone.

[Richard Crowley]

Most certainly there are marvelous modern circuits that can operate well down at even 3.3V
But those are designed for portable use with ear-buds for iGadgets, et.al.

NOT SUITABLE for a studio situation.  You will not find any such wimpy circuits in studio headphone systems.

[Chaos_Klaus]

Thank you guys for all the replys.

With a single +5 V supply, even if you find an amplifier that can swing  rail to rail, you have 5 V p-p, that means less than 1.8 V rms.
That results in 98 mW on 32 ohm load, 1 mW on 2500 ohm load.
That power level is absolutely inadequate in a noisy studio. even with efficient headphones.
With a bridge amplifier (if you can correctly wire the headphones with a 4 pin connector) you'll get 4 times that power, still inadequate (it's the same power you could get with a 10 V supply).

There is lots of truth to that, I guess.

There are several problems in a studio environment.

During recording the instruments are really loud. I'm not talking about roaring guitar amps or anything. I record mostly Jazz bands. Even acoustic instruments are really loud. This is especially a problem when I am tracking multiple musicians in the same room. I'm a musician myself (not actually a full time audio engineer) so I know the problem first hand.

Also, a piece of music that was recorded, mixed and mastered has had its dynamic range severely reduced by the various compressors involved. That makes it possible to get high RMS levels without clipping the peaks. In a monitoring scenario, the signals are mostly unprocessed. Therefore there is potentially high peaks and low RMS levels. So lots of headroom is needed, or some form of limiter to attenuate the peaks.

4Vpp is plenty to listen to music really loud. But I guess there is not enough headroom for producing good RMS levels when dynamic range is so high.

8Vpp only gives me 6dB more headroom. To get 12dB I already need 16Vpp. With the NE5532 I get an output swing of about Vcc-2x2V which means I need a Supply that can give me 20V, which means using a 24V wall wart.

That is kind of the classical approach.

My DAC runs on only 5V and draws at least 100mA ... that means I will lose lots of power here.

About headphone sensitivity:

I don't need to worry about completely high impedance and inefficient heaphones. If someone really comes to me and wants to use it, I just use a seperate powerful headphone amp.

But I might sometimes run into the situation where someone brings his beloved Beyerdynamic DT770 Pro and he has bought the 250 Ohm version which is actually meant for mixing, while there is an 80 Ohm version for recording.

So this is basically my "typical worst case" that I want to have covered. The DT770 Pro 250 Ohm has a sensitivity of 96dB(SPL)/mW@500Hz.

http://europe.beyerdynamic.com/shop/dt-770-pro.html?SID=13d123e7ed1c60159b44a60d3f9a62b8&___store=en&___from_store=de

Maybe I can find a solution that uses an only slightly higher supply voltage (maybe 10V) but uses an amplifier with an integrated charge pump?

Maybe someone can point me to one of those ICs, becaus on ti.com I can not find one. I am in Germany, by the way.


EDIT: I just took some readings of a piano track in my audio software. rms level is about 20dB below the peak level. So 4Vpp output means about 0.4Vrms which is about 84dB(SPL) on a DT770Pro ... I guess it's back to the drawing board for me.

[dom0]

I mentioned one above, but those are usually only for 3.3 V systems because that's where you really need it (mobile devices running of a single lipo cell).

If you want the large output swing, you're likely not getting around a boost converter. ~1:3 ratio is rather moderate, so I don't expect issues there.

[Chaos_Klaus]

I mentioned one above, but those are usually only for 3.3 V systems because that's where you really need it

Well, the TPA6138A2 only works with 3.3V power supply and can give 40mW into a 32Ohm load. The average 5V headphone amp gives around 70mW into 32Ohm. So that is kinda out of the question.

Step-Up boost converters ... wow there is a lot of them:

http://www.ti.com/lsds/ti/power-management/step-up-boost-converter-products.page#p32=1&o4=ACTIVE&p238max=5;75&p634max=15;75


EDIT:

I am super confused now.

I just tested all the headphones I own:

55Ohm, 91dB SPL / mW
62Ohm, 91dB SPL / mW
32Ohm, 96dB SPL / mW

I put a typical unprocessed monitoring signal through and turned up the volume.

At high levels where listening gets really uncompfortable (but not jet painful), the output swing of my headphone output stays well below 4Vpp.

Also that 4Vpp seems to be all my Audio Interface is giving me anyway. When I force it into digitally clipping (max output) it still stays inside that range while visually clipping on the oscilloscope.

Honestly ... that never was too low for me, or anyone.

In the datasheets, i can not find any specification of the maximum output swing. They only ever give the maximum sustained power.

[tszaboo]

I mentioned one above, but those are usually only for 3.3 V systems because that's where you really need it

Well, the TPA6138A2 only works with 3.3V power supply and can give 40mW into a 32Ohm load. The average 5V headphone amp gives around 70mW into 32Ohm. So that is kinda out of the question.

Step-Up boost converters ... wow there is a lot of them:

http://www.ti.com/lsds/ti/power-management/step-up-boost-converter-products.page#p32=1&o4=ACTIVE&p238max=5;75&p634max=15;75


EDIT:

I am super confused now.

I just tested all the headphones I own:

55Ohm, 91dB SPL / mW
62Ohm, 91dB SPL / mW
32Ohm, 96dB SPL / mW

I put a typical unprocessed monitoring signal through and turned up the volume.

At high levels where listening gets really uncompfortable (but not jet painful), the output swing of my headphone output stays well below 4Vpp.

Also that 4Vpp seems to be all my Audio Interface is giving me anyway. When I force it into digitally clipping (max output) it still stays inside that range while visually clipping on the oscilloscope.

Honestly ... that never was too low for me, or anyone.

In the datasheets, i can not find any specification of the maximum output swing. They only ever give the maximum sustained power.

There are headphones with 600 Ohm coils, and music, where parts are +30dB above the average (Overture). For these kind of applications you need a good +/- 15V or at least +/-9V powered amplifier. Or Hifiman headphones requiring literally 1000 times the power.

[Chaos_Klaus]

There are headphones with 600 Ohm coils, and music, where parts are +30dB above the average (Overture). For these kind of applications you need a good +/- 15V or at least +/-9V powered amplifier. Or Hifiman headphones requiring literally 1000 times the power.

Well, I actually know that there are very demanding headphones and that there is very demanding material with high dynamic range. That however is not important for my application.

250Ohms with 90dB SPL/mW is my worst case scenario. And If I were to record classical music, I would certainly not use headphones.