Standard power output for a headphone amp?

[user 37518]

Im designing a battery powered headphone amp using a 9V battery, the design is nothing special, im using an opamp with an output push-pull stage in a feedback loop. Im not looking for anything fancy or ultra low distortion,  however I dont know about the typical recommended power output for a headphone amp, from my simulations Im getting around  6.5 mW with a 600 ohm load at 0.1% THD, 17 mW with a 220 ohm load at  0.1% THD, and 50 mW with a 16 ohm load also at 0.1% THD, So my question is: Is it too much power? too little?

Edit: I've been reading about headphone specs, apparently the sensitivity of the headphones plays a huge role, they are rated in dB SPL/mW. The thoughest load from my simulation measurements is a 600 ohm load, it will depend on the sensitivity of the headphones if the power is too much or too litle, so I guess my question should be: Whats the typical sensitivity of 16 ohm headphones? and that of higher impedance headphones such as 600 ohms?

 

[PRR]

Sensitivity in SLP/mW does not differ THAT much.

For a voltage-limited amp, mW in X Ohms varies a LOT over the range of all headphones ever made, from 3 Ohms to 2K.

9V supply means at-most 3Vrms. This is 3,000mW in 3 Ohms, 300mW in 30 Ohms, 30mW in 300 Ohms, 4mW in 2K Ohms.

As maybe most headphones in use today are 20-60 Ohms, and most phones get loud with 50mW, this may be OK (450mW-150mW). If you have cherished 300+ Ohm phones, and need LOUD, 9V batt is limiting.

 

[Rochey]

there are some laws regarding this matter, but most codecs I see drive around ~50mW or so, some of the bigger single chip amps drive up to 120mW+
There's also the ol' TPA6120A2 amplifier that could drive 1.5W, but that one was reserved for drummers only.

 

[volker]

https://docs.google.com/spreadsheets/d/1qBPZrG73Gbzy8RsU7rs63LoG4YIP5rXBLDAsn9m_s1M/edit?type=view&gid=0&f=true&sortcolid=0&sortasc=true&rowsperpage=500#gid=0

Rochey, new kid on the block is the OPA1622, no?  The IC package might not be to everyone's taste though. What a shame TI dropped so many of National's LME parts...

 

[Matador]

I'll quote from PRR from another thread as his advice worked perfectly for me:

[quote author=PRR]
This over-high level can be done over most headphone impedances with 7V behind 29 ohms. 300r phones get 6.4V (136mW), 32r gets 3.5V (380mW), the ancient 4r phones would get 0.8V (180mW). With thrifty output stage, this can run on +/-12V power.
[/quote]

 

[user 37518]

https://docs.google.com/spreadsheets/d/1qBPZrG73Gbzy8RsU7rs63LoG4YIP5rXBLDAsn9m_s1M/edit?type=view&gid=0&f=true&sortcolid=0&sortasc=true&rowsperpage=500#gid=0

Rochey, new kid on the block is the OPA1622, no?  The IC package might not be to everyone's taste though. What a shame TI dropped so many of National's LME parts...

That spreadsheet is pure gold! thanks!

 

[user 37518]

Sensitivity in SLP/mW does not differ THAT much.

For a voltage-limited amp, mW in X Ohms varies a LOT over the range of all headphones ever made, from 3 Ohms to 2K.

9V supply means at-most 3Vrms. This is 3,000mW in 3 Ohms, 300mW in 30 Ohms, 30mW in 300 Ohms, 4mW in 2K Ohms.

As maybe most headphones in use today are 20-60 Ohms, and most phones get loud with 50mW, this may be OK (450mW-150mW). If you have cherished 300+ Ohm phones, and need LOUD, 9V batt is limiting.

Makes sense, I guess I should stick to the +-15V rails

 

[PRR]

Volker, thanks for the extensive data. It does not disagree with the old RANE data (headphones have to fit the existing systems), but adds hundreds more and newer (and more extreme) examples.

Taking a VERY high (deafening) listening level:

for 115dB SPL
Voltage: 0.2V..22Vrms  (0.3V..30V peak) (>0.6V..60V total supply)
90% below 3.8Vrms  (5.3V peak) (>10.6V total supply)
80% below 2.4Vrms  (3.4V peak) (>7V total supply)
Current: 5mA..384mArms  (7mA..540mA peak)
90% below 78mArms (110mA peak)
80% below 60mArms (84mA peak)

I once designed for 115dB SPL for a live recording rig, so-that "any" headphone in my gig-bag would NOT distort in the headphone amp (so I knew any distortion was in the recording chain not the monitoring). I would NEVER listen at 115dB SPL more than some Seconds.

It is not clear that all of these headphones are safe for 115dB SPL.

105dB SPL is a much safer listening level. All the above V and mA become 1/3rd of the values for 115dB SPL.

Unless you have special low-sensitivity hi-Z phones, and need LOUD, I do not think +/-15V (30V) supplies are necessary or even safe for most headphones. Note that 90% of the phones in that table will get very loud (or burn) with 11V peak-to-peak, so +/-6V should be ample.

 

[benb]

https://docs.google.com/spreadsheets/d/1qBPZrG73Gbzy8RsU7rs63LoG4YIP5rXBLDAsn9m_s1M/edit?type=view&gid=0&f=true&sortcolid=0&sortasc=true&rowsperpage=500#gid=0

Rochey, new kid on the block is the OPA1622, no?  The IC package might not be to everyone's taste though. What a shame TI dropped so many of National's LME parts...

There's a nice thread on that chip, mostly about driving headphones (!) on diyaudio:
http://www.diyaudio.com/forums/vendor-s-bazaar/283672-audio-op-amp-opa1622.html

 

[Rochey]

https://docs.google.com/spreadsheets/d/1qBPZrG73Gbzy8RsU7rs63LoG4YIP5rXBLDAsn9m_s1M/edit?type=view&gid=0&f=true&sortcolid=0&sortasc=true&rowsperpage=500#gid=0

Rochey, new kid on the block is the OPA1622, no?  The IC package might not be to everyone's taste though. What a shame TI dropped so many of National's LME parts...

Agreed on all points. Only difference is I don’t keep up with what TI is releasing any more!

 

[Bo Deadly]

Unless you have special low-sensitivity hi-Z phones, and need LOUD, I do not think +/-15V (30V) supplies are necessary or even safe for most headphones. Note that 90% of the phones in that table will get very loud (or burn) with 11V peak-to-peak, so +/-6V should be ample.

Agreed. Even +/-6V would be super loud into even 100 ohms.  The problem is how to make +/-6V. If you just make a simple regulator circuit from +/-15V for OPA1622 which limits current to +/-150mA. That 9V 150mA dropped across your regulator is 1.35W peak which is ugly. But is that really going to happen? Probably not. 32 ohm phones would burn out.

 

[Andy Peters]

https://docs.google.com/spreadsheets/d/1qBPZrG73Gbzy8RsU7rs63LoG4YIP5rXBLDAsn9m_s1M/edit?type=view&gid=0&f=true&sortcolid=0&sortasc=true&rowsperpage=500#gid=0

Rochey, new kid on the block is the OPA1622, no?  The IC package might not be to everyone's taste though. What a shame TI dropped so many of National's LME parts...

OPA551PA will probably be just as good, and in a package you don't need special tools to work with.

 

[ruairioflaherty]

Agreed on all points. Only difference is I don’t keep up with what TI is releasing any more!

New gig?  Do tell.

 

[PRR]

> The problem is how to make +/-6V.

12V CT and a FWB.

> 32 ohm phones would burn out.

The plan cited in Reply #4 tends to adjust power versus impedance, to a not-too-insane level for any likely headphone.

Sample values, 7.4 volts through 29 ohms into various impedances:
_ Ohms _______ Watts
_ 8 __________ 0.30
_ 32 _________ 0.46
_ 60 _________ 0.41
_ 120 ________ 0.29
_ 250 ________ 0.17
_ 600 ________ 0.08
_ 2000 _______ 0.03

The drop of power from 32r to 250r is in-line with an observation that the classic old hi-Z phones tend to be more sensitive, the new-age 32r phones tend to be less sensitive. These levels tend to approach 120dB SPL (not 115 as I remembered).

 

[emrr]

Rock club remote recording observation: a MOTU Monitor 8 interface stating max 80mW (into?) headphone output driving 63 ohm Sony's did not cut it for monitoring inputs back at FOH.  FWIW. 

 

[Bo Deadly]

> The problem is how to make +/-6V.

12V CT and a FWB.

Or something like Mean Well DLC03A-05 might be good for a headphone amp supply. It would be cheaper, easier, smaller, ...