Headphone amp

;D

yes


BTW you don't need to use 1% resistors

JR

The typical problem with a 9V rail is limited Vp-p output.

But ignoring that no worries.

JR

Looks like you are running over 7mA through those 470 ohm resistors. does it need to be that high?

Cheers

Ian

Also the pot divider across the op amp plus input uses 10K resistors. This make a 5K nput impedance which unnecessarily load the 10K vloume pot. I would change the 10K to 100K.

Cheers

Ian

If you insist on running off 9V, I'd suggest that you bootstrap the output transistor driving.

That way you don't have to loose the two diode drops and the two B-E drops. Those 2.4V can make a big difference for the power available in this circuit..


Jakob E.

gyraf said:

If you insist on running off 9V, I'd suggest that you bootstrap the output transistor driving.

That way you don't have to loose the two diode drops and the two B-E drops. Those 2.4V can make a big difference for the power available in this circuit..


Jakob E.

Click to expand...

And that allows increasing seriously these base resistors, probably tenfold for a similar output swing. It would also reduce the idle current in the output pair.

gyraf said:

If you insist on running off 9V, I'd suggest that you bootstrap the output transistor driving.

That way you don't have to loose the two diode drops and the two B-E drops. Those 2.4V can make a big difference for the power available in this circuit..


Jakob E.

Click to expand...

Forgive my ignorance but do you mean tie the two bases together? Even then you must lose the Vbe surely or is there something smart going on?

Cheers

Ian

> 470 ohm resistors. does it need to be that high?

The large-signal output impedance is 470/hFE. Assuming low hFE like 100, this is 5 Ohms. Plus the 1 Ohms. Makes 6 Ohms. Assuming we want large swing in 16 Ohm load, this is about as high as it should be.

OTOH, all the 16r phones I know will drive loud with a Volt.

I would suspect "OK" with 1k5-1k8 resistors here.

If he is battery-operation, this power efficiency should be considered. If he has a guitar-pedal 9V wart, it is inconsequential.

That direct-split bias will stuff half the B+ crap into the amp. If the supply is not dead-clean, you want like the two 10K, with a 10uFd to ground at the junction. Then run 100K over to the opamp to bias it.

BTW: the old-old LM386 (get JRC) will do as well or better on 9V and 16r-up. You need two, but the total pincount is less than an 8DIP and two Qs. THD is not stellar but may be no worse than a 358 and basic buffer. Wire the '386 for low gain (20) so its hiss is not bad. Do not forget the RC load on the output to suppress MHz squirreliness.

PRR said:

BTW: the old-old LM386 (get JRC) will do as well or better on 9V and 16r-up. You need two, but the total pincount is less than an 8DIP and two Qs. THD is not stellar but may be no worse than a 358 and basic buffer. Wire the '386 for low gain (20) so its hiss is not bad. Do not forget the RC load on the output to suppress MHz squirreliness.

Click to expand...

I made a tinnitus box for my wife from a358 with the gain tuned right up. Makes an excellent white noise generator.

Cheers

Ian

I have advocated both sides of a TL072 as a hiss source. Cheaper and better hiss than the old pseudo-random chips. For calibration you need to measure and compensate the 1/f rise, which is very variable. For tinnitus or A-weighting, it is probably moot.

> the circuit simulator takes so long to charge the 100uf capacitor

?? The usual SPICE approach is to remove all caps, find all DC voltages, pre-charge the caps, then start time.

You can "skip initial transient solution" to watch power rise from zero. Important for some circuits. Not this one.

If at all concerned about battery life TI has more than 20 class D amp chips. 

In my first generation drum tuner running from a 9V supply I did something simple with an op amp section and a couple bipolar transistors (no diodes, just tie NPN and PNP base to opamp output and run a small R from bases to output.) 

For my current generation tuner to improve battery life I went class D and that sucker sips current while running from lower voltage. In fact the chip I use now can't even tolerate 9V rail voltage, but many do.

JR

PS: You may have to be willing to use SMD as TH may not even be available in these modern parts.

Using only 9vdc is not the way to go. You need to run it on +- 15vdc at least, in order to drive decent phones like Senn.  Beyer etc.
You will not be able to drive Orthodynamic phones with this circuit; they require plenty of power.  My new headphone amp; due to hit the market soon has output of 2.8 watts into 50 ohms & 1.4 watts into 100 ohms, thd .0007%.

If you have any questions email <[email protected]>

Bill Wilson
Wilson Audio Developments

user 37518 said:

How would you bootstrap this circuit?  I imagine something like splitting the 470 ohm resistors in half and adding a capacitor from the output to the the joint between the 2 resistors?

Click to expand...

Yes, that's it. Attached an example of this, .jpg

Jakob E.

Or use Sziklai output to maximize output swing and power. Although limited output swing is not necessarily bad for a headphone amp IMO. Even 2V into 16 ohms is a quarter watt peak which is very very loud.

ruffrecords said:

Forgive my ignorance but do you mean tie the two bases together? Even then you must lose the Vbe surely or is there something smart going on?

Cheers

Ian

Click to expand...

Actually this is a common topology for not very hifi applications using modern (fast) op amps. The common collector NPN and PNP output buffer devices, with the bases tied together still exhibit a normal Vbe drop because the other complementary base-emitter junction in parallel is reverse biased and not conducting half the time.  Adding a modest resistor from base to emitter, allows the op amp to supply some  current during the crossover region between -0.5V and +0.5V.  For outputs greater than +/-0.5V the buffer transistors conduct and supply current as expected.

As I may have already mentioned, I used this topology in my first generation drum tuner to drive two speakers (in series). Low enough distortion to wiggle drum heads, while not audiophile. No class A bias current in output stage makes it a little more battery friendly, but not as friendly as class D.

JR

JohnRoberts said:

Actually this is a common topology for not very hifi applications using modern (fast) op amps.

Click to expand...

The resulting performance was deemed adequate by the Studer engineers, who had a reputation of never sacrificing performance to petty savings.
They used it in conjunction with positive current FB in order to minimize transformer distortion.

abbey road d enfer said:

The resulting performance was deemed adequate by the Studer engineers, who had a reputation of never sacrificing performance to petty savings.
They used it in conjunction with positive current FB in order to minimize transformer distortion.

Click to expand...

The old 301, with the feedforward compensation, was relatively fast for back in its day...  (I wonder if the current feedback into the + input, bothers the FF compensation that only works with inverting topology? Probably not much voltage swing there from current into the 6.8r .)

Still not my first choice for driving speakers or low Z cans.  At least for bench specs on paper it won't measure that great at 20kHz, probably sounds just fine.

JR

user 37518 said:

Whats the purpose of the resistor (R34) between the base and emmiters in the Studer schematic?

Click to expand...

To deliver current to the output before the transistors are turned on. 

When the current through the 330 ohm R developes more than 0.5V the transistors get turned on and supply the rest of the current needed.

For low current the op amp is driving the output by itself through the 330 ohm. For higher currents the transistors do the heavy lifting.


JR