Different headphone amps

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The schematic of this headphone amp is attached.

It is different because:

1. It uses Class aP amplification. It switches analog [music] signals via precision rectification to generate analog Pulses [aP].

2. It is a current source amp.

3. The normally-shared ground lead of stereo headphones is separated so to cater to each headphone. Needed with any current source amp!

4. A common [+/- 15 V] is used for both channels, and an additional +/- 9 V [or higher] for each channel. All 3 supplies are very low power; preferably regulated.

The above may sound complicated; but the prototype sounds fine despite its switching operation.

The schematic is an image of a simple LTSpice file. Examine performance at the circuit points which are enclosed in rectangles.

Please find the thread "Class aP amplification" in the Pass Labs Forum for additional info.

Best
Anton
 

Attachments

  • ClassaP current source headphone amp.png
    ClassaP current source headphone amp.png
    37.5 KB · Views: 1,003
How does this headphone amp work?

It works as follows:

1. The BJTs idle at ~50-100 uA. They are barely turned on. A trivial Class A!

2. The OpAmp [OA] inverts the phase of the input signal. When the voltage of the input signal goes positive [relative to ground], OA's output voltage goes negative. This negative-going signal of OA turns off NPN fully and simultaneously turns on PNP fully.

3. The inverting port of OA is at a virtual ground; because it tracks the non-inverting port which is set at actual ground. Virtual ground is a s good as actual ground. Its value is it locks the emitters of both BJTs [via their degeneration resitors] to a fixed reference which is ground.

4. As OA's voltage output goes negative, it removes forward bias to NPN, and increases it for PNP. This is analog switching,

5. A positive-going input signal from the generator pushes current [I+] toward virtual ground. [I+] cannot enter the inverting input of OA; because it is high impedance. It maybe the gate of the input FET or the base of a BJT diff amps! So..

6. The negative-going signal at OA's output fully turns on PNP which then routes this [I+] to its collector and to the load [32 Ohms]. OA is a trafic controller for injected input current.

7. When the input signal goes negative, OA's output goes positive which fully turns on NPN and fully turns off PNP.

8. Input signal generator cannot pull [I-] from virtual ground because it is high impedance. Instead, it pulls [I-] from the emitter of NPN because it is fully turned on. Again, OA is the traffic controller of input current. The resultant current [I-] finds its way to the load.

9. The above jives with the objective results for the LTSpice model in post# 1; namely switching at the emitters of BJTs, and a fully satisfactory music Vout signal across the headphone load.

To be continued..

Best
Anton
 
Harmonic distortion

The LTSpice model in post #1 gave H2 and H3 at 70dB and 100 dB below fundamental. This objective performance can be readily improved by increasing the forward bias for BJTs.

H2 has a "likeable" subjective sonic value per the teaching of Mr. Pass in his thread which is entitled H2, and his two related articles in the FirstWatt website.

Best
Anton
 
That's actually just a variant of class B - each output device is on for 180 degrees of the cycle, which is the definition of class B.

I'd worry about the linearity of those 9V batteries given their relatively high internal resistance compared to the load resistance.


100 ohm input impedance is a non-starter for many applications though, if you have a source capable of driving 100 ohms
without distorting, you already have a headphone or power amp!
 
Yes the "likeable " sound of a tube amplifier of second order harmonics,--that's good for third order .
The human ear is forgiving of even harmonics much more than the "harshness" of odd order.

I take it that's the value across the audio spectrum ?

Thanks duncan2 for your accurate post.

The objective performance at 20 KHz c/o the LTSpice model gave similar results like that at 1 KHz.

I increased the idle current of the BJT's from 50 uA to 5 mA each. I was surprised that the objective performance was unchanged. The OpAmp has done its work at either idle levels to clean up objective performance.

I now have a Class A instead of Class [B or aP] current source amp. This Class A amp may present a different and more pleasing subjective performance.

Best
Anton
 
Input impedance at V5? (Looks like 101 Ohms.)

Current gain? (Looks like unity.)

The part shown it totally symmetric so should not have second order unless that opamp has a bad lean.

Hello PRR. Your observations are accurate. The input impedance is low and the current gain is =1.

The presence of H2 across the load in this symmetrical circuit may be a bad lean by the OpAmp. However, the switching therein upsets this apparent symmetry because one feedback branch is fully on and the other is off. The OpAmp output [Vout +error] has a minimal H2 and only shows H3. A low or absent H2 suggests cancellation by adding H2+ phase from one branch with an equal magnitude H2- phase from the other branch.

Best
Anton
 
That's actually just a variant of class B - each output device is on for 180 degrees of the cycle, which is the definition of class B.

I'd worry about the linearity of those 9V batteries given their relatively high internal resistance compared to the load resistance.


100 ohm input impedance is a non-starter for many applications though, if you have a source capable of driving 100 ohms
without distorting, you already have a headphone or power amp!

Hello Mark Tillotson. Thanks for you post and its accuracy.

You and PRR showed concern about the amp's low input impedance. The bolded statement in your post describes exactly the following amp's objective which is really needed for its operation:

Drive the low impedance input of this prototype amp by a power or a headphone voltage-source amps so as to transform this power input to a power output for a current source amp at a current a gain =1.

Why? I wish not preach to the choir. A headphone is a full-range speaker. It has an Impedance versus Frequency performance graph like that of a stand-alone woofer driver. And it thus behaves similarly. Mr. Pass has excellent articles on this subject in his FirstWatt website.

The reality is that the subjective performance of any headphone driven by a voltage source amp is totally different from it driven by a current source amp.

The diyer has now an additional and different subjective format to drive any headphone.

Best
Anton
 
Simplify, simplify and improve....

The attached image is a simple LTSpice circuit which transforms the input voltage [Vin] at the non-inverting input of an OpAmp buffer to a power output for a current source amp [CSA]

Its objective performance is highly accurate. View the signals at Vout and at the output of the OpAmp. Almost distortionless. View the currents flowing through the PSUs. It showed switching.

One may wish to experiment with a favorite OpAmp. Two [[+/- V] PSUs are stille needed with this like any other [CSA].

One may wish to direct the output voltage of the OpAmp [voltage source] via headphone to one ear and the Vout via its twin phone to the other ear [current source]. No need to separate the leads of the a stereo headphone because each phone shares a common ground.

Best
Anton
 

Attachments

  • uA741 CSA.png
    uA741 CSA.png
    43.2 KB · Views: 400

PRR

Member
Joined 2003
Paid Member
> a current source amp [CSA]

OK, I did not know that was your intent. And your #11 plan is neat and tidy.

But a classic way to force a current output is the Howland.
howland current source - Google Search
See:
The Howland Current Pump Circuit
Jump down to "Let's reset our examination..."
Howland%20Current%20Pump%201.png


This forces the input voltage across a, say, 100 Ohm resistor to force a current. This form allows the load to be grounded, as we want for usual TRS phone plugs.

There is a vast literature on Howland design and optimization. In the simple form taught on TubeCAD the NFB loop diverts 'metered' current from the load, but this is easily buffered or designed-out.

The Howland also allows using dual-amp or funny-power chips if desired.

howland current source - Google Search
https://www.ti.com/lit/an/snoa474a/snoa474a.pdf
 

Attachments

  • Howland%20Current%20Pump%201.png
    Howland%20Current%20Pump%201.png
    13.8 KB · Views: 363
Hi Guerilla.

Thanks for your note. Found octaver and harmonizer [schematics too] in a Google search. Interesting! Octaver can divide frequency; which may make my voice "deeper-manlier" or double it to make me a great sopralto. Disguising voice?

Fortunately, these devices are commercial to experiment with.

Best
Anton
 
A Mono Headphone Amp

The attached image is a simple LTSpice circuit which transforms the input voltage [Vin] at the non-inverting input of an OpAmp buffer to a power output for a current source amp [CSA]

Its objective performance is highly accurate. View the signals at Vout and at the output of the OpAmp. Almost distortionless. View the currents flowing through the PSUs. It showed switching.

One may wish to experiment with a favorite OpAmp. Two [[+/- V] PSUs are stille needed with this like any other [CSA].

One may wish to direct the output voltage of the OpAmp [voltage source] via headphone to one ear and the Vout via its twin phone to the other ear [current source]. No need to separate the leads of the a stereo headphone because each phone shares a common ground.

Best
Anton

The attached schematic shows this mono HA amp driving stereo headphones. Mine are model SR80 by Grado Labs. Note that I had cut their plug and separated each's common lead [Black for phase identification]. Why?

Vout [VSA] and Vout [CSA] are out of phase and of equal amplitude. They are highly correlated [1:1] and thus highly inter-dependent. The
headphone leads were reversed so as to establish the correct phase for both ears; especially for base.

The system sounded great which led to this experiment which turned out not begnin:

1. The phones were still used as the load to the amp.

2. Vout [VSA] and Vout [CSA] were fed to a stereo power amp which drives loudspeakers in the room.

3. One speaker leads were reversed so as to get equal phase.

4. The system howled and buzzed [osillated]; which was triggered by the 1:1 inter-dependence of the two signals. In-phase and correlated sound signals in any room cause positive acoutic feedback due to regeneration.

But; this led to this hypothesis. Did the stand-alone headphones/amp sound great because of a well-tempered "Intra-Cranial Positive Acoustic Feedback"? .
Best
Anton​
 

Attachments

  • uA741 Mono Headphone Amp.png
    uA741 Mono Headphone Amp.png
    41.9 KB · Views: 323
Your last paragraph Antoinel --"intra-cranial positive acoustic feedback ".


Thats why for headphone professional testing "Neuman Heads" were made to model the physical acoustical properties of resonance in the human skull .


Stax Electrostatics ( of which I own a pair ) made use of this method to design one type of electrostatic headphone that "angled " the sound to try to reproduce a more exact scientific method of using an electrical device ( the headphone ) to model to the more "natural " properties of a living human being rather than just using the standard means.

They were called the Stax Sigma .
 
A Mono Micro Headphones Amp

Thanks duncan2 for your post.

I had a STAX Sigma until my teenage daughter cranked up its volume and blew it.

I have a broad collection of Op Amps or micro amps. Mono's, dual, and quads.

The attached schematic is for a prototype head amp. It is a "tester" I suggest using to discern the subjective performance of micro amps. Which one of the following dual OpAmps has the best sound? It is simple, efficient and has the OpAmp as sole variable understudy.

LF353, TL082, NE5532, OPA2134, LM4562 .

I was not surprised that each dual has a characterisic sound which overall for this group is clean and satisfying as mono.

This tester may help clarify the old performance debate of discrete versus integrated microamps.

The tester's schematic for a dual OpAmp shows the following:

1. Each channel accepts an audio input [L or R] and is connected as a buffer.

2. Both channels share the same power supply rails. Thus, Vout [current source] is automatically the sum of both channels.

3. A Zobel is needed at Vout [current source] to prevent/remove instability; which I observed testing OPA134PA but not uA741.

4. The Vout [voltage source] of each channel is joined by using an [0.2 Ohm] in series with each output as shown encased in rectangle. I've seen/read this method to join voltage source outputs in one the forums.

Best
Anton
 

Attachments

  • Mono micro HA.png
    Mono micro HA.png
    38.9 KB · Views: 224
JLH came to the same conclusion that quality OpAmps correctly built into a small signal audio circuit were no less capable of high quality reproduction .


I can understand why a 741 wouldn't oscillate , I have used all the ones you post except the LM4562 , D.Self always recommended the NE5532 for a lot of his designs
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.