hifi headphone amp?

Why omit the output transformer?

Isn't the 2503 transformer 4, 75ohm windings?
Could the three secondary windings be used to drive three separate sets of headphones simultaneously?

[quote author="Mike Keith"]Using a power amp for headphones only will not load the amp like it needs, it wants to see a 8 ohm load[/quote]
Most amps do not mind running without a load - when you connect headphones the speakers are switched off, remember?

Tube amps do NOT like running without a load though...

Best regards,

Mikkel C. SImonsen

[quote author="Ethan"]Correct me if I'm way off base in thinking, but couldn't just about any power amp be adapted for headphones?[/quote]

Generally speaking, I suppose so, but there's a couple of caveats that immediately spring to mind:

Crossover distortion in the handoff between PNP and NPN halves of the output stage is usually much less noticeable on speakers than on headphones. If you're averaging about 10 watts through speakers, as opposed to 10 milliwatts through headphones, there might be some problem. It's a voltage issue rather than a power issue, so high-impedance headphones would be much better off than low-impedance headphones. I'd only really consider this for high-impedance devices.

I'd consider PRR to be the forum oracle on such matters, having invested a decent chunk of several years to the matter. Take his ponderings over my drivel any day. :thumb:

Keith

I recommend the TPA6120 by TI. Follow the layout instructions carefully. It sounds good, and can be easily configured for balanced input. It's a surface mounted IC, though, so it would like to be mounted on a PC board. I built a quad headphone amp for my studio using four of them.

[quote author="Ethan"]Why omit the output transformer?

Isn't the 2503 transformer 4, 75ohm windings?
Could the three secondary windings be used to drive three separate sets of headphones simultaneously?[/quote]
Sure...

Or you might just omit the tx and parallel the headphones directly. Makes it easier, cheaper, and I really doubt that you'll need the galvanic isolation into headphones...

Using 3 individual outputs would just reflect the effectively paralleled impedance back in the same way... but with a pricy transformer in there.

Keith

OK after reading through the stuff PRR has posted here and at headwize, would a driving amp similar to this concept work?
http://sound.westhost.com/project56.htm
Its a mixed current and voltage feedback trying to have constant power at the load. Has anyone tried this with a headphone amp?

Perhaps this is a complete noob question, but that fits me to a tee.

How does one discover the impedance of a pair of headphones? My understanding is that, at least on loudspeakers, the impedance varies depending on frequency, so it can't be as simple as connecting an ohmmeter... eh, right?

some heresy, perhaps...

the popular 'ganged 5532' headphone amp works well and can be knocked together in one evening using parts you probably have already.

what Keith said. high wattage poweramps rarely sound good enough. i would treat headphones as small loudspeakers (which they are, duh) and make a small transistor PP Class A power amp, no transformers, no SE, no fancy current drive or negative output impedance

+/- 24V and capable of a few watts in 8 ohm

also using high power amps can create potentially dangerous situations and BLOW peoples eardrums. no good, deaf people don't buy studio time

it seems to be a majot headache for many studios and designers of gear. either the amps are topnotch but the distribution system blows or vice versa

> How does one discover the impedance of a pair of headphones? My understanding is that, at least on loudspeakers, the impedance varies depending on frequency, so it can't be as simple as connecting an ohmmeter... eh, right?

Eh, right. Speakers or headphones, measure the DC resistance, round up, call it the impedance.

Yes, it varies with frequency. But we mostly care how low it goes, not how high it goes. And for broadband transducers, the motional impedance is very low except in narrow bands like bass resonance. That motional impedance (outside resonant zones) runs 2Ω-0.1Ω. The coil has an inductance that will raise impedance at the top of the band, but rarely a whole lot and not at frequencies we make big power. So what you see looking into speakers or headphones is some very low impedance in series with the Copper Resistance, which you measure with an ohm meter.

"8Ω" speakers will show 6Ω or 7Ω DCR: speakers need to be somewhat efficient. Between the bass resonance and the treble inductance, impedance won't dip lower than 8Ω or 7Ω or so, and is higher over most of the band.

Headphones measuring 28Ω DCR will be around 300Ω over most of the audio band, with a 600Ω peak at resonance and rising to 450Ω at 20KHz. 300Ω near-enough.

You can get in trouble with VERY efficient speakers. 1930s theater horns approach 50% power efficiency, which means their motional impedance is similar to their electrical impedance, and the best will do this over much of their working bandwidth. JBL 2440 measures 12Ω DCR, is rated 16Ω, but in a BIG well designed horn (not a little 800Hz sectoral) it will be above 20Ω from ~200Hz to over 1KHz.

You can get in trouble with crossovers. A designer trying to fix a crossover suck-out may turn to high-Q crossover filters. Power can not be created, but the high-Q filters will transform 8Ω drivers to much lower impedance which will draw more power out of the amplifier. There are cases of "8Ω" speakers showing 3Ω at crossover. OTOH, I have some E-Vs with very efficient tweeters and to tame them, EV used a no-Q crossover that rises to 50Ω at 2KHz. Not a problem for typical stage amps, but could be piercing with some low damping audiophile amps.

Yeah, use the ohm meter. It is rarely far off.

Some digital ohm meters do not like odd loads. If in doubt, wire a 9V battery and a 900Ω resistor in series with the speaker, measure the DC on the speaker, do the math. 0.1V is 10Ω, near enough.

> would a driving amp similar to this concept work?

"Work"? Sure. In fact such schemes were used for speakers in days before Thiele and Small showed us how to figure our bass resonance.

The difference is that speakers have a known nominal impedance and need big power. Headphones come in 8Ω to 2,000Ω, popularly 32Ω to 300Ω, but use so little power that we don't have to care so much about efficiency (in home/studio use; pocket-amps are a special problem). Because of the wide range of impedances, you need like 10V and 300mA peak to cover both extremes of impedance at "ample" volume, but you never need or want both 10V and 300mA at the same time on the same phones.

To some degree, there are two types. Phones made to connect to a broadcast line-amp, 150Ω and 100mW (to some degree, these come from telephone practice with 1mW and 100-1,000Ω impedance); and WalkMan phones made to get loud on two 1.5V batteries. Sine wave on a 3V supply makes about 30mW in 30Ω which seems to make many people happy.

You could build several amps, one lo-Z one hi-Z. Tubes with transformers can use a tapped winding and get best power transfer over a wider range of loads. But for simple sand-state it is often best to run +/-12V to +/-20V rails (rarely higher for Senn 600) with 300mA peaks. Use a series resistor to balance power delivery in low and high Z phones. You will ALWAYS need a gain control because phones of the same impedance can vary more than 10dB in efficiency. The series resistor reduces accidents, and often makes amp protection a non-issue. If you really think phones need damping, put the resistor inside the feedback loop, even in series with the output devices' collectors.

TPA6120 will do the current (a little shy on voltage). Parallel 5532 do indeed drive hi-Z well and low-Z well enough for all but the worst uses. True 990s and several of its family (probably including Dan's GR amps and Joe's unreleased product) will drop +/-20V +/-200mA without any strain. LM1875 is killer for any load, and reportedly sounds better than you would expect from a chip-house. There are many ways to skin this cat.

My studies are based on RANE's studies which seem to cover most of the designs in current use (though model numbers and tailfins change, phone guts have not changed much in 20 years). We must exclude AKG K1000 which has shockingly low efficiency, and pretty much needs a Speaker Amp or a dedicated design. And old Brush crystal headphones, hi-Z magnetic cans used in classic radio and telephone industries....

A close enough method for checking impedance is to: know the ball park of the impendance, example 600 ohm, put a 1K pot in series with the headphone inject a 1Khz tone across the load say at 2volts adjust the pot until you have 1/2 or 1 volt across the pot and disconnect AC and check resistance of pot and that is what your impendance is...

[quote author="PRR"]> ...snip...

You can get in trouble with crossovers. A designer trying to fix a crossover suck-out may turn to high-Q crossover filters. Power can not be created, but the high-Q filters will transform 8Ω drivers to much lower impedance which will draw more power out of the amplifier. There are cases of "8Ω" speakers showing 3Ω at crossover. OTOH, I have some E-Vs with very efficient tweeters and to tame them, EV used a no-Q crossover that rises to 50Ω at 2KHz. Not a problem for typical stage amps, but could be piercing with some low damping audiophile amps.

[/quote]

I think the worst I heard of was some audiophile speaker, nominally 4 ohms, that dipped to 0.3 in a relatively narrow band. OUCH!

most speakers go from almost zero to hundred ohms in real life, porting the lf unit just makes it worse. my speakers dip below 2 ohms, but they're very easy to drive anyway because of the closed enclosure, 6dB/octave filters and very high sensitivity. much easier to drive than say a KEF or B&W

a loudspeaker load is very complex. an Apogee (not the converter mfg) or ESL63 can make even the biggest amp cry

> most speakers go from almost zero to hundred ohms in real life

"Almost zero" isn't clear. Less than DCR?

"8Ω" speakers rarely rise higher than 50Ω. There is no physical reason they could not be higher, but suspecnsion damping and box leaks usually won't let resonant Q exceed 10, and often more like 5.

> porting the lf unit just makes it worse.

Resonances are all on the far side of the DCR. I don't think you CAN get the driver working impedance lower than the DC resistance, only higher.

Show me an example where porting the box reduces impedance magnitude below DC resistance.

soundguy, did you mention any price limit?

I and some other have pcb's and you can can build a Spitzenklasse amp for rather little money but you have dozens of brand availabe. and a few of the with balanced inputs.

www.harmonydesign.se has two amps with balanced inputs (website not available this minute).

CIAudio and Meier Audio seems to have good products.

CIAudio has one amp made of TPA6120 which I also have used in my QRV-07 amp. Quite exciting IC!

http://www.ciaudio.com/
http://www.meier-audio.homepage.t-online.de/