Headphone amp without any special components?

[Hides-His-Eyes]

So I'm at my parents for christmas, have nothing approaching a headphone amp, they go to bed very early and I'd like to play. I have brought my parts box with me but the postal service has Shut Down Everything because it's christmas/snowing (I know, I know...) I was thinking about that FET-champ emulation as a pre-amp, but still need the other half. I have:

2n3904s
2n3906s
5 2n5457s
~5 BC109Cs
TL072s
Ins/outs
A breadboard!

Can it be done?

Thanks for any advice!

[PRR]

So you want a custom design, with specified parts, NOW, but don't say what it should do (iPod or guitar) or what power you have available?

Or even how many TL072s?

Or what impedance your headphones are?

Or how much you will pay to avoid treking 27 miles on foot through the snow to be there when Banjo World re-opens to sell you a practice box?

You don't need gain. You don't need big voltage. You will need current.

I'm off to bed while you inventory your wallwarts, batteries, TL072s. zzzzzzz

[Taylor]

http://generalguitargadgets.com/diagrams/mxr_headphone_amp_sc.gif?phpMyAdmin=78482479fd7e7fc3768044a841b3e85a

I built one, pretty sure I used a 3904 and a 3906 instead of the spec'ed transistors.

[Hides-His-Eyes]

So you want a custom design, with specified parts, NOW, but don't say what it should do (iPod or guitar) or what power you have available?

Or even how many TL072s?

Or what impedance your headphones are?

Or how much you will pay to avoid treking 27 miles on foot through the snow to be there when Banjo World re-opens to sell you a practice box?

You don't need gain. You don't need big voltage. You will need current.

I'm off to bed while you inventory your wallwarts, batteries, TL072s. zzzzzzz

Custom design? No, I'm just looking for someone like taylor who has tried one of the many designs online with what I think of as "standard components" (Although I guess everyone has a different opinion of that)

My biggest power supply is 12VDC at 500mA, I have 5 TL072s, plenty of 9V batteries...

My standards are really pretty low, like, "makes the guitar audible without sounding like a fuzz face" low... It's christmas eve and I couldn't get to Banjo World even if it was open

Taylor: thanks. Will give that one a go this afternoon.

[geertjacobs]

Lazy solution:
plug guitar straight into PC mic input?

Then use breadboard to make a simple distortion/booster/cab sim, but be sure to add a level control so you don't make the mic input clip.

[merlinb]

My biggest power supply is 12VDC at 500mA, I have 5 TL072s, plenty of 9V batteries...

A TL07x running from 12V can drive headphones directly (with 470R protection resistor in series with the output). Volume is not earth-shattering, but it works. Even at 9V you can get a useful volume, although headroom will be nonexistent.

[PRR]

> It's christmas eve and I couldn't get to Banjo World even if it was open

They're sure to have an after-Xmas Sale. And the sales staff must be hungry to get ANY sale before the end of the month/quarter/year.... make an offer.

> http://generalguitargadgets.com/diagrams/mxr_headphone_amp_sc.gif
> I built one, pretty sure I used a 3904 and a 3906

That will work. Diode and Transistor selection is not too critical. I shudder to see NO emitter resistors and low-I diodes against high-I transistors, but bias will probably acceptable and the 100 ohm resistors protect against after-jack shorts (up to 15V supply). No protection for on-breadboard shorts.

I realized that you don't want "generous" power because bleed from the cans will wake the parents. You want very few milliWatts and are going to clip more/less for emphasis.

As Merlin says, TL07x will make some power, and is self-protected from shorts. His 470 ohm series resistor is safe and conservative, and does not strain the TL07x sonic quality much, but may not be loud enough. You can go much lower, and you can add more TL07x sections working together. Around 40 sections will make a LOUD CLEAN headphone amp. 2 to 6 sections is probably ample for your purpose. 2 to 3 sections is fewer connections than the discrete power stage.



I have stolen JD Sleep's "MXR" artwork and messed it up. I moved the Volume control so that an Audio Taper pot will work; MXR had large supply contracts, you don't. I changed the discrete output stage to massive pile-of-TL07x.

I've suggested ranges of values assuming that you do not have a full stock of every part.

[Hides-His-Eyes]

I think I'd be worried about my ears before waking the parents with headphone leakage, but thanks PRR, that's fantastic.

Just so that next time I'm designing it for myself and there's a lesson in every schematic...

-Large pulldown, input cap, to a non inverting amplifier; V_ref delivered with a large resistor to keep input resistance up

Voltage gain is ~100 at high frequencies, depending on complex inductance of the 4u7; bass cut to reduce the amount of farty bass

the serial 22k-100k with that cap forms a low pass filter to tame the high end (especially important when there's no speaker/cabinet to do that for us...?)

Each TL07* can provide limited current, so need multiple to drive the headphones, right? Call it a gain of 50x voltage, 50*0.05V RMS (single coils); V^2/R gives the power per op-amp; which is 125mW. Check the TL072 data sheet and the max is 680mW per op-amp, smashing.

Two chips gives four of those, about half a watt. stick all five pairs in there and I might get 1.25W, enough to drive a small speaker.

Just checked my resistor box and the choice is 10R, 47R or 100R for low value resistors. 10R is no good; that would give 5x the power per op-amp, which is dangerously close to the chip limit and easily enough to burn out my 1/4W resistor, so 47R it is.

Sound like I understand it?

[Hides-His-Eyes]

Should the Bass Cut cap not go to V_ref?

Built but not working, anyway. Tomorrow: debugging fun! After dinner, maybe.

[PRR]

> Each TL07* can provide limited current, so need multiple to drive the headphones, right? Call it a gain of 50x voltage, 50*0.05V RMS (single coils); V^2/R gives the power per op-amp; which is 125mW. Check the TL072 data sheet and the max is 680mW per op-amp, smashing.

Some gaps in that analysis.

What if we put in 200mV? 50*200mV= 10V output, which won't happen on a 9V supply.

If it did (and neglecting sine/square peak/RMS quibbles), at 32 ohms it would imply 10V/32 or 0.3 Amps or 300mA. The TL07x makes at most something like 20mA.

OK, take that 20mA guess and work out what can really be put into 32 ohms. 0.020A*32 is 0.64V. and 0.64V*0.020A is 0.0128 Watts or 13 milliWatts. That's peak or square, in audio we tend to compute as-if Sine, so 6mW.

That seems small. But as a rule-o-thumb, Watts in speakers are like MilliWatts in headphones. A 50 Watt speaker amplifier can get LOUD. A 50mW headphone amp can get LOUD. For clean unclipped music, in laptops and iPods, 30-50mW is a fair compromise between too-weak and too-greedy. Solo guitar is usually played UP TO clipping level and a bit more. You may need 50 Watts in a noisy club with 100 rowdy patrons, but in a quiet space a 5 Watt Champ is really pretty ample. Many folks use LM386 at under a half-Watt. So 6mW is not a silly-weak headphone amp for guitar.

However the TL07x will NOT really going to deliver 20mA on low voltages.... indeed the spec makes no claim on output current.

There IS a graph in the data (see figure), but it shows output falling to zero at 100 ohms which is not reasonable, probably hasty drafting.

Look at what's sorta-really inside.



Couple emitter followers, so we know we can't get the output within 0.6V of the supply. More transistors driving the bases, more drop, round-up to 1V loss each side.

With 9V supply we can only swing 9V-1V-1V= 7V peak to peak.

We only have one-way power. Audio is both-ways. To fake this out, we DC-idle the amp at about half-supply, 4.5V, and cap-couple that away at in and out.

So taking the idle point as "zero", we can swing up to 3.5V and down to -3.5V.

The TL07x has *large* resistors toward the output pin. TI's sheet says 64+128= 192 ohms, ST's sheet shows 100+200= 300 ohms. Experience, pessimism, and easy arithmetic says to use the 300 ohm number.

The MAXimum output current is into a dead-short. We can swing 3.5V either way, across 300 ohms, is 11.6mA.

Any load reduces max output current (but raises volage and power in the external load). Say we add 50 ohms external load. 300+50= 350 ohm total resistance. 3.5V/350 is 10mA maximum current. Likewise 100 ohms allows 7.8mA. As long as the external load is much less than the 300 ohms inside, the maximum load current will be near 10mA.

Let's consider the internal dissipation. If load is small, nearly all dissipation is inside the chip, between the 1V loss and 300 ohm resistance. It adds up to 4.5V at about 10mA. This is 45mW internal loss. Add the roughly 9V 2mA or 18mW idle loss, 63mW heat per amplifier. Two amplifiers per 8-DIP, 126mW internal heat. This is very comfortably below the 680mW limit.

Now what is our actual load? You did not answer my "what impedance your headphones are?" 32 ohm phones are popular. For Mono, this is really 16 ohms.

10mA in 16 ohms is 160 milliVolts. You can also write 0.010 Amps and 0.160 Volts.

0.010 Amps and 0.160 Volts is 0.0016 Watts, 1.6mW.

This is Peak or Square-Wave power. Sine RMS is half, 0.8mW.

If LM386 at 0.5W is ample in speaker, 0.8mW in headphones may be ample.

Doubling the current also doubles the voltage. Double the amplifiers makes FOUR times the Power. A whopping 3 milliWatts! Four amplifiers (two TL072) nearly quadruples power again, say 11mW.

So power output is a LOT less than your over-idealized computations. But it may be enough.

[Brymus]

Such a freakin awesome explanation.
I wanna try this now.