If I put my notes here, I might be able to find them again later!
B-board Boxer Project : A Low Voltage Headphone Amplifier for 16 ohm Loads
I always seem to end up optimizing my headphone amplifier circuits for higher impedance headphones, this mostly happens because I own a pair of 300 ohm HD-600s and it is tedious to design for both the voltage requirements of high impedance headphones and the current requirements of low impedance headphones.
Not impossible, just, for the class-A designs I seem to be building recently, increasingly large, heavy, and impractical.
Complimentary transistor circuits, however, offer the opportunity to swap voltage for current at something close to the same design cost. They are therefore a practical topology for efficient class-A power delivery into low impedance headphones. As a design experiment, my aim is to discover how far I can leverage an ultra-low-voltage, unity gain circuit for compactness without sacrificing sound quality.
Ok. Back-of-the-envelope calculations:
A typical 16 ohm in-ear-headphone has a sensitivity of 100-105 dB/mW. At typical listening the total voltage attenuation is about -30 dB from 2 Vrms input, so the working output signal is about 100 mV and 5 mA (typ. max.). Applying an arbitrary 5x overbuild sets the clipping target at 500 mV and 25 mA. These are the ballpark estimates for the voltage rails and output bias current, respectively.
While 16 ohms headphones don't need much voltage the current requirement are considerable.
What I will look at, next, is whether the B-board diamond buffer can be run at ultra-low voltage while maintaining low distortion, and whether the whole layout can be shrunk down to "trouser pocket" dimensions.
**
I hadn't originally planned to make this portable, but: Typical NiMH AA cell : 1.4 V, 2000 mAh. That's 40 h operation at 50 mA. Plenty. Very doable. We may be fitting this into trouser pockets yet!
**
Uploaded screen cap of LTSPICE window, B-board Boxer AA. The output bias current is about 20 mA, and the voltage clipping onset is about 700 mV peak. As demonstrated, the circuit can run on 2xAA cells workably well, putting out 75 mV / 5 mA peak into 16 ohms with very low distortion, and going to to 500 mV / 30 mA peak at about 1% THD. At just below voltage clipping the amp manages 700 mV / 43 mA at 3% THD. Class A output power is 5 mW.
Static current draw is 120 mA stereo. So 15 h run time on two AA cells. 10 h real world battery life seems like a fully realizable goal.
**
Preliminary schematic uploaded. The amp should end up "Chu Moy" sized, so inputs and outputs are 3.5mm stereo jacks. A volume control if you want, a 10k input resistor otherwise, or a crossfeed filter.
I figure I can build a "mule" from a couple of b-board rev 11 boards, leaving the regulators unpopulated. This is a test bench build, obviously the final version would have a dedicated, smaller board with the jacks and volume included.
If the circuit works out I will try to design a board that fits in standard cases people tend to use for these sorts of projects.
**
Note: you can scale to higher supply voltage by increasing the emitter resistors: drivers 33 ohms x V+, output 0.1 ohms x V+.
Note: At V+ = 1.4 V there's no room in terms of voltage for any kind of protection, but neither is there enough voltage to damage 16 ohm headphones. If you add electrolytics to the power supply, however, you should be careful to get the batteries connected in the correct polarity before switching on.
**
Uploaded schematic and board of "alpha build" rev. 0.4. The stereo board is 10 cm x 4 cm. Is that small enough? I'm not sure...
**
Details and results of the prototype build, here.
Not impossible, just, for the class-A designs I seem to be building recently, increasingly large, heavy, and impractical.
Complimentary transistor circuits, however, offer the opportunity to swap voltage for current at something close to the same design cost. They are therefore a practical topology for efficient class-A power delivery into low impedance headphones. As a design experiment, my aim is to discover how far I can leverage an ultra-low-voltage, unity gain circuit for compactness without sacrificing sound quality.
Ok. Back-of-the-envelope calculations:
A typical 16 ohm in-ear-headphone has a sensitivity of 100-105 dB/mW. At typical listening the total voltage attenuation is about -30 dB from 2 Vrms input, so the working output signal is about 100 mV and 5 mA (typ. max.). Applying an arbitrary 5x overbuild sets the clipping target at 500 mV and 25 mA. These are the ballpark estimates for the voltage rails and output bias current, respectively.
While 16 ohms headphones don't need much voltage the current requirement are considerable.
What I will look at, next, is whether the B-board diamond buffer can be run at ultra-low voltage while maintaining low distortion, and whether the whole layout can be shrunk down to "trouser pocket" dimensions.
**
I hadn't originally planned to make this portable, but: Typical NiMH AA cell : 1.4 V, 2000 mAh. That's 40 h operation at 50 mA. Plenty. Very doable. We may be fitting this into trouser pockets yet!
**
Uploaded screen cap of LTSPICE window, B-board Boxer AA. The output bias current is about 20 mA, and the voltage clipping onset is about 700 mV peak. As demonstrated, the circuit can run on 2xAA cells workably well, putting out 75 mV / 5 mA peak into 16 ohms with very low distortion, and going to to 500 mV / 30 mA peak at about 1% THD. At just below voltage clipping the amp manages 700 mV / 43 mA at 3% THD. Class A output power is 5 mW.
Static current draw is 120 mA stereo. So 15 h run time on two AA cells. 10 h real world battery life seems like a fully realizable goal.
**
Preliminary schematic uploaded. The amp should end up "Chu Moy" sized, so inputs and outputs are 3.5mm stereo jacks. A volume control if you want, a 10k input resistor otherwise, or a crossfeed filter.
I figure I can build a "mule" from a couple of b-board rev 11 boards, leaving the regulators unpopulated. This is a test bench build, obviously the final version would have a dedicated, smaller board with the jacks and volume included.
If the circuit works out I will try to design a board that fits in standard cases people tend to use for these sorts of projects.
**
Note: you can scale to higher supply voltage by increasing the emitter resistors: drivers 33 ohms x V+, output 0.1 ohms x V+.
Note: At V+ = 1.4 V there's no room in terms of voltage for any kind of protection, but neither is there enough voltage to damage 16 ohm headphones. If you add electrolytics to the power supply, however, you should be careful to get the batteries connected in the correct polarity before switching on.
**
Uploaded schematic and board of "alpha build" rev. 0.4. The stereo board is 10 cm x 4 cm. Is that small enough? I'm not sure...
**
Details and results of the prototype build, here.
Total Comments 7
Comments
-
Hello rjm
Please see the older thread entitled " A headphone transconductance amp for a change". It has a similar topology as you describe here except that the system [ground] is taken at the junction of the emitter and load resistors. The load now is at the opposed collectors of the complemetary pair of high output impedance relative to the one you describe. Will the LTSpice modelling be any different between a voltage source and current source topologies of the same circuit elements?. Your insight is valuable.Posted 13th July 2012 at 02:22 PM by Antoinel
-
You mean your headphone003 schema I suppose? To answer your question, "current source" and "voltage source" are names people give to circuit elements, the devices themselves are oblivious, and behave/model consistently.Posted 14th July 2012 at 12:16 AM by rjm
-
Hello rjm
Yes to your question above. Thank you for replying. The headphones must [and did] sound different by using the different [current versus voltage source] approaches. Best regards.Posted 14th July 2012 at 12:56 AM by Antoinel
-
Hello rjm,
The simulation above shows a 16 Ohm load resistor. Should it be a resistor in series with an inductance? If [or since] simulation is a predictor of actual acoustical performance, why is it predict equal performance [voltage versus current source] when the two load connections do not sound the same?.Posted 15th July 2012 at 07:51 PM by Antoinel
-
Hello rjm
Have you considered using Germanium complemetary alloy transistors; e.g. NTE132 and NTE155? They are medium power in a TO-66 package.Posted 15th July 2012 at 08:25 PM by Antoinel
-
I don't expect the simulation to accurately tell me the performance into a real headphone load. I only rely on it to give me an idea how well it drives 16 ohms resistive, and I extrapolate from there, based on the circuit topology, listening tests, and experience.
No, I had not considered the obsolete and hard to find NTE132/155 JFETS. I doubt Vgs is sufficient temperature stable or in close enough tolerance to be usable in a diamond circuit, but if you know better please explain -Posted 17th July 2012 at 12:26 AM by rjm
-
Hello rjm,
Thank you for the clarification on the subject of simulation. I was not referring to "fet'" transistors; but rather to "bjt" where the amplifying substrate is made of the older technology Germanium rather than the Silicon-based Epitaxial ones you use. I searched Bing for "complementary germanium transistors". It showed high and medium power ones; albeit Japanese. I did not search further for lower power devices to sit on your PCB. The Vbe of Ge transistors is ~0.2 V. Thus their use will probably give your existing circuit more amplifying headroom to better manage distortion.Posted 17th July 2012 at 04:12 PM by Antoinel
