L272 (power op amp) for headphone amp

[composition4]

Hi all,

Looking at tacking on a headphone amp to a cab sim I've designed.  I don't need any (or very little) voltage gain as I'm already close to the 9V rails, just something capable of current gain.

Has anyone looked into the L272 power op amp? https://www.st.com/resource/en/datasheet/l272.pdf

Looks good to me - current production, cheap, more than enough power for headphones, 500k input impedance, good enough voltage swing, a fair bit of input offset voltage (15mV typical - 60mV max) but I'd really only be using it at unity gain.  Does anyone foresee any disadvantages to using this in this application?

Thanks in advance
Jon

[PRR]

It appears to run with zero bias in the output stage. This suggests high crossover distortion. Small signals are "hoarse" though loud sounds may seem fine. (Headphones may never get much out of the crossover zone.)

They market it as "motor controller", where a little crossover distortion does no harm. Very similar topologies with output stage bias are sold as "audio amplifiers". May be a reason for the distinction.

But hey, if they are in your hand, try them.

[composition4]

Thanks Paul,

I don't have any with me, didn't realise the potential for high crossover distortion.

Truth be told I'm looking for a simple headphone amp design with little or no voltage gain with currently in-production parts, but I'm struggling for options.  I don't trust that the op amp based designs i.e. NE5532 would drive low impedance headphones, and LM386 and other small IC amplifiers etc don't seem to run stable at low gains. All of the new amplifier ICs seem to be 5.5V max.

My priority is to keep parts count relatively low... Any good suggestions?

[J0K3RX]

https://www.mouser.com/datasheet/2/256/MAX9722-MAX9722B-1292355.pdf

[composition4]

Thanks for the suggestion, however that IC is 5.5V max supply, and I don't want to mess around with power rails etc. And I'm interested in keeping cost down so I can make a few, the design has way too many parts already.

After further searching, think maybe something similar to the MXR headphone amp here http://www.generalguitargadgets.com/pdf/ggg_mxr_hpamp_sc.pdf. Appears to be just a buffered opamp output, has anyone used this and is performance good?

[noisette]

Hi!
I have used this onehttps://tinyurl.com/yyva7nav, which is practically identical to the one you posted, only with fixed gain, and it leaves nothing to desire!

[composition4]

Oh, apparently the TDA2822 is still in production, not sure why I thought it was obsolete... Seems like a much simpler answer for me

Edit: oh never mind, apparently it's fixed gain at a pretty high level, won't go anywhere close to unity...

[bool]

Try OPA551 or 552.

[composition4]

A bit pricy unfortunately.. I think I've found my answer.  NJM4556 seems to be a decently high output current, cheap op amp; I think it should be good enough to drive pretty much any headphone with a 100R resistor in line with the output.

Thanks
Jonathan

[dschwartz]

I scratched my head a lot with the headphone amp for the omnicabsim.
I second Jokerx about the max9722..just add a 5v regulator and a capacitor..
This chip does not need output caps, so at the end, the extra cost of the regulator is not important. It will sound much better and louder than any opamp solution..

[Rob Strand]

If you want a minimum parts solution this type of booster might work,



The resistor between the base and emitters can be chosen so there's no zero crossing issues.  Perhaps around 220 ohm mark depending on the speed of the opamp and the output load specifics.    Just plop that 3-part output circuit into the MXR circuit posted above.   You don't have to use an inverting config like the pic and a 9V single rail is fine.  As with all these circuit you might (should) add a small cap ceramic from the opamp output back the the opamp -ve input.  (For the MXR ckt that means you probably should reduce C6.)

[PRR]

> currently in-production parts,

You planning a million units over the next 20 years?

Or just a few hundred for short-term?

Mouser has over 10,000 of LM380 in stock, 'bout a buck a pop in hundreds. That will keep you busy a while.
https://www.mouser.com/Texas-Instruments/Semiconductors/Audio-ICs/Audio-Amplifiers/LM380-Series/_/N-4gxtf?P=1yxyx85Z1z0zls6

Yeah, it's mono, so you may want two.

NJM4556  is indeed the go-to for many headphone jacks.

Rob's buffer has been used but as-drawn it risks burnout on a shorted phone (oops); also gain could be lower for most purposes. (Also 14 Ohms looks small for many opamps: 14r*0.030A is only 0.42V, not really gonna open the transistor up, >33r seems more likely.)

[Rob Strand]

Rob's buffer has been used but as-drawn it risks burnout on a shorted phone (oops); also gain could be lower for most purposes.

But if you "Just plop that 3-part output circuit into the MXR circuit posted above."  it will work fine (since it has the o/p series resistors and series caps).

Like-a-dis,

(Also 14 Ohms looks small for many opamps: 14r*0.030A is only 0.42V, not really gonna open the transistor up, >33r seems more likely.)

FWIW, the opamp only supplies o/p current for +/-0.7V o/p swing.  After that the transistors take over, the opamp then supplies constant current through the resistor and some base current.

[Rob Strand]

Rob's buffer has been used but as-drawn it risks burnout on a shorted phone (oops); also gain could be lower for most purposes.

But if you "Just plop that 3-part output circuit into the MXR circuit posted above."  it will work fine (since it has the o/p series resistors and series caps).

Like-a-dis,

(Also 14 Ohms looks small for many opamps: 14r*0.030A is only 0.42V, not really gonna open the transistor up, >33r seems more likely.)

FWIW, the opamp only supplies o/p current for +/-0.7V o/p swing.  After that the transistors take over, the opamp then supplies constant current through the resistor and some base current.

Having said that I'm not against the original MXR or LM380's 

[composition4]

Great thanks guys, I've gone from not having enough options to having quite a few - I'll explore them all!

[composition4]

It appears to run with zero bias in the output stage. This suggests high crossover distortion. Small signals are "hoarse" though loud sounds may seem fine. (Headphones may never get much out of the crossover zone.)

They market it as "motor controller", where a little crossover distortion does no harm. Very similar topologies with output stage bias are sold as "audio amplifiers". May be a reason for the distinction.

But hey, if they are in your hand, try them.

Randomly just thought of this reply from a month back... But if the op amp is biased to 1/2VCC, how would crossover distortion be a problem in non-inverting mode?

[PRR]

"Bias" means many things. Here we bias the center-point to 1/2V, *and* we bias the idle current to several mA so the transistors are not dead at idle.

[todd.dukes]

I scratched my head a lot with the headphone amp for the omnicabsim.
I second Jokerx about the max9722..just add a 5v regulator and a capacitor..
This chip does not need output caps, so at the end, the extra cost of the regulator is not important. It will sound much better and louder than any opamp solution..

Can you use this with a 78L05 or do you need more current capability?

[dschwartz]

I scratched my head a lot with the headphone amp for the omnicabsim.
I second Jokerx about the max9722..just add a 5v regulator and a capacitor..
This chip does not need output caps, so at the end, the extra cost of the regulator is not important. It will sound much better and louder than any opamp solution..

Can you use this with a 78L05 or do you need more current capability?

The chip is 130mW per channel, at 5V
That's 260mW/5V= 52mA..
Assuming the efficiency of the amp is over 80%, 52mA/80% = 65mA

It should work fine with a 78l05 (100mA max)

[composition4]

"Bias" means many things. Here we bias the center-point to 1/2V, *and* we bias the idle current to several mA so the transistors are not dead at idle.

Oh yeah current bias.. sorry, head was stuck in voltage land