Class-A Headphone Amplifier
Intro: Class-A Headphone Amplifier
This is an amplifier designed to provide plenty of high-quality power for discerning headphone listeners. It can be used to boost the output level from smartphones or digital music players suitable for studio-quality heaphones, or provide a headphone-level output from a line level signal.
CAUTION: high sound levels damage hearing! This is intended for use with lower-sensitivity (or higher impedance) on-ear headphones. It's not recommended for use with standard smartphone earbuds. As the constructor and user you are responsible for keeping your ears protected from dangerous sound exposure levels.
STEP 1: About the 'JLH' Class-A Amplifier
In 1969 John Linsley-Hood published an article in Wireless World magazine describing an audio power amplifier based around a four-transistor circuit. It is an engineering masterpiece - it uses a small number of readily-available components, gives dependable and repeatable performance, but it's not so simple as to need obvious improvements. With modern components the performance is at least as good as the original.
This is a 'Class-A' amplifier - that's a technical term, meaning that its transistors are always turned on (passing current) even when there is no signal. This has the drawback that the amplifier uses significantly more power than the alternative 'Class B' (or Class-AB) types, but the advantage that distortion vanishes away as the output level decreases.
The original JLH amplifier was designed to give an output power of 10 Watts into an 8 ohm loudspeaker. For a headphone application the power output can be much less (1 Watt is more than enough), so I've scaled the supply voltage and operating currents appropriately. This means the amplifier produces a fraction of the heat of the 10 W version, and can use an inexpensive, ready-made power supply. The schematic, with adjusted component values, is shown above.
There is plenty of information on this circuit on this website: https://sound-au.com/tcaas/index-1.htm
STEP 2: Matching Power Transistors (Optional)
The JLH circuit achieves best performance (lowest distortion) when the two output transistors Q1 and Q2 have the same current gain (hFE). In this step we measure the gain for each transistor in a batch, and pick pairs (one pair for each channel) which have closest gain.
This step is optional - building the prototype amp I bought a set of 10 TIP3055 transistors, and none of them were more than 5% different from another. Modern manufacturing is much more consistent than in 1969, and it's very likely any pair from the same batch will give excellent performance.
You will need:
- A 5V power supply
- A meter for measuring current up to about 300mA (a bench power supply with an output current meter is ideal)
- A 2k2 resistor
- Test leads (e.g. crocodile clips)
- A heatsink or piece of flat metal
The procedure is to connect each transistor up as shown in the diagram, apply power and measure the current drawn by the transistor. (The gain is equal to the collector current divided by the base current - for matching all we really need is the total current). For a typical TIP3055, the gain is 100 and the total current in the circuit above will be about 200mA.
Make a note of the current reading for each transistor, and then you can pick pairs with similar readings. If you have a choice, higher gain (higher current) is better.
The transistors will heat up when power is applied, which will affect the gain! If you're finding it difficult to measure the current within a couple of seconds of applying power, clamp the transistor to handy piece of metal to keep the temperature stable.
STEP 3: Start PCB Assembly
As always, it easiest to begin assembling with the lowest-height components first, in this case the resistors.
I've shown the component positions for the eBay PCB (see 'Supplies' for the link). Note that some of these will be different from the values printed on the board itself.
Note regarding R2A and R2B
R2 is used to set the operating current of the circuit, which affects the maximum output power, and the heat produced. The eBay PCB has a trimmer in the R2 position. You can follow this approach (a 5K trimmer would be appropriate) but to keep 'moving parts' to a minimum I opted to use a fixed 2k7 resistor (R2A) in parallel with a second resistor (R2B) chosen during testing. See step 5 for more details.
STEP 4: Completing the Board Assembly
Next step is to fit the capacitors and transistors, and complete the board assembly.
Be careful with the two small (TO92 case) transistors: not all TO92 transistors have the same arrangement of Emitter, Base and Collector leads. The eBay circuit board uses a 2SA970 part for Q4, which does not match the BC558 transistor which I had available. This is not a big problem, as with care you can bend the leads (see photo) to fit. Alternatively you can look for a 2SA970 or similar low-noise, small-signal PNP transistor with the right pinout.
For the output capacitor (C2) I used two 470uF capacitors in parallel, equivalent to a single roughly 1000uF component. This was mostly to make ordering the components easier (a pack of 10 is enough for both channels, with spares) but also helps to keep the ESR low.
Before you solder the TIP3055 power transistors in, figure out how you will mount them in whichever case you're using. The leads will need bending accurately so both PCBs have the same dimensions.
STEP 5: Testing and Setup
In this step, we will check each circuit board for correct operation, and adjust the R2 resistor to set the operating current.
When the boards are assembled it is important to ensure the power transistors are mounted on a heatsink of some sort before applying power. The picture shows the two boards mounted on the aluminium channels that form the case described in the next step. The TIP3055 transistors need insulating pads between the cases and the heatsink - the case is connected to the collector terminal (see picture for details). Don't over-tighten the nut, or it will damage the transistor case.
If you have a bench power supply (with variable output voltage, and current limiting), it is highly recommended to perform a 'smoke test' to check the PCB for faults.
If you have used a trimmer for R2, set it to half-way. Set the power supply to 12V, with a current limit of 250mA, and use a multimeter to monitor the voltage between the positive end of C2A/C2B and ground. When you apply power this should slowly (over 10 seconds or so) increase from zero to roughly half the supply voltage (i.e. 6V). If this works you're good to go to the next stage.
If this doesn't happen, check for the following common faults:
- Component leads which you forgot to solder
- Short circuits caused by solder blobs, or from Q1 or Q2's cases to the heatsink
- Transistors incorrectly inserted
- Incorrect resistor values
In this step we choose a value for R2B to set the operating current for each amplifier channel. The optimum current is around 180mA, which allows full power to be delivered to a 32 ohm load.
You'll need to apply 24V power to the amplifier, while measuring the supply current (again - a bench power supply is ideal). When you apply power there will be a high current draw (up to 400mA) until the DC voltages stabilise, and then it should eventually settle to some stable value (perhaps 100-150mA). When this is stable, record the value.
You can then pick a value for R2B in order to reach 180mA (or just over) operating current, according to the following guide:
- 10K - increases current by 25%
- 8K2 - increase by 33%
- 6k8 - increase by 40%
- 4k7 - increase by 60%
- 3k3 - increase by 80%
You can quickly tack the resistor on to the PCB, then apply power and measure the current when stable. When you're happy, trim & bend the leads and attach it properly (see picture).
If you're using a trimmer for R2, start with it in mid-position, and slowly adjust it to reach 180mA. A lower resistance will increase the current.
Setting the current precisely is optional - it mostly affects performance at maximum output power, which hopefully you won't need very often. If you are unable to perform this step, use a fixed 2K2 resistor for R2, or 2K7 in parallel with 6K8 (=1.93K).
STEP 6: Building a Case
Obviously there's no reason you couldn't use a ready-made case here - the only important thing is to ensure all the power transistors have adequate cooling: each transistor dissipates approximately 2 Watts, so use a 10 °C/W heatsink per transistor, or 5°C/W per pair of transistors.
My case was built around two 200mm lengths of 50 x 25mm aluminium U channel, 3.2mm thick, which form the sides. The PCBs are mounted between them by bolting the power transistors to the channels. This is somewhat flimsy, so the next step is to mount them to a base plate.
The base plate here is 200mm long by 135mm wide, and is bolted to the sides with four M4 Allen bolts, one at each corner. It's important not to stress the transistor leads when the bolts are tightened, so ensure there is enough clearance in the bolt holes to allow adjustment.
One the base is attached to the sides, the rest of the case can be built up. The front and rear panels shown were made from a thick brushed aluminium plate I salvaged from an old 19 inch rack case, each 135mm wide x 63mm high. They are attached to the case sides using short lengths of 20 x 20mm aluminium L-section (see photo), one at each corner. With this arrangement it's quite easy to remove and replace the panels for drilling and wiring up.
The top of the case is the same size as the base plate, and fits between the front and rear panels. I made the top from a thin piece of aliuminium sheet with a piece of 6mm plywood cut to size on top. I applied a couple of coats of Danish oil, to complete the overall "1970's hi-fi" vibe of the finished case.
STEP 7: Wiring Up
Wiring up the amplifier is relatively straightforward: we have the signal input and volume control, the power supply input, and the headphone output socket.
The signal input wiring uses thin (e.g. 3mm diameter) screened cable from the input sockets to the volume potentiometer, and from the potentiometer to the amplifier input. See the photo for the wiring. I used isolated sockets here (i.e. the 'ground' side of the socket is not connected to the case).
The power supply input jack has a 220nF capacitor directly attached directly to help reduce any high-frequency noise from the power supply - although I didn't find this to be an issue with the PSU I used.
The headphone output socket is wired as shown in the photos - the left ground and right ground are connected together. The socket I used was a non-isolated type, so the case is connected to ground through this point.
When all is done, you're ready for a first listen. Start with the volume right down, and increase slowly!
One final note: this amplifier uses roughly 10W of power when operating. Please don't leave it permanently switched on! Five minutes' warmup time is plenty for it to reach stable operating conditions.
STEP 8: Measured Performance
Here is the measured performance of the ampifier as built.The frequency analyser plots were made with a Focusrite Clarett audio interface, and NAK T-100 analyser software.
Maximum output power
Measured at 1Khz, 0.5% THD
1.8W rms (21.6V pk-pk) into 32 Ω
1.0W rms (23.2V pk-pk) into 64 Ω
Gain/frequency response
x13 (+22dB) at 1KHz
-0.7dB at 20Hz, 32 Ω load
< -0.1dB at 50KHz
Distortion
32 Ω load
0.015%, 1Vrms out (31mW)
0.025%, 100mW out
64 Ω load
0.005%, 1Vrms out (15mW)
0.011%, 100mW out
28 Comments
rfrisbee 1 year ago
If simplicity is what you're after then paralleling up a few NE5532's with 10-ohm output resistors would provide enough drive for a headphone amplifier with lower idle power dissipation, lower noise + distortion and no need to match components.
For those who balk at the idea of the "op-amp sound", how many NE5532's do you think the recorded music you are listening to has passed through already?
Didactech 5 months ago
If you are in any doubt about this then I suggest you read his book:
- The Art of Linear Electronics (Oxford, Butterworth-Heinemann, 1993
All your questions, including the NE5532, and many more will be answered.rfrisbee 5 months ago
Doug Self has measured the performance of the NE5532 with a loaded output. I stand by my original recommendation of using several of these devices paralleled as a headphone driver would be superior in virtually all respects, including distortion, ease of construction, thermal stability and quiescent power consumption than the design presented in this instructable.
TheSpodShed 1 year ago
Yes, I did look at audio op-amps. The NE5532 datasheet generally gives specs when driving a 600R load; you'd need to parallel ~20 of them to get comparable performance into 32R. The maximum output current is given as 38mA typical; you'd need 8 of them to get 300mA peak output current (as this amp does).
You can use fancier devices - Ti's OPA1688 (https://www.ti.com/product/OPA1688) can drive 75mA out and (under different conditions) has great distortion numbers. But it's hard to buy and the tiny package is hard to solder.
Neither of these seemed particularly appealing as construction projects, whereas having ready-made PCBs widely available was a big factor in choosing the JLH circuit.
(As step 2 makes clear, transistor matching is entirely optional, and probably makes no audible difference. But neither is it difficult or time-consuming).
rfrisbee 1 year ago
a) Have the least sensitive headphones ever made
b) Are deaf
c) Want to be deaf
I would suggest 6 parallel NE5532 outputs with 10-ohm load sharing resistors per channel (3 8-pin ICs per channel) would be more than adequate for a 32 to 64-ohm headphone driver and significantly superior in every respect compared to the JLH circuit used for this purpose. I suspect the distortion of the JLH circuit rises steadily above 1kHz.
A PCB would not be required: It would be very simple to construct on stripboard as each paralleled stage is just a unity gain buffer and 10-ohm output resistor. If 6 parallel stages did not give enough output drive then leave some space on the end of the board and add a few more ;-)
The parallel NE5532's could be driven by a NE5532 configured as a non-inverting amplifier if the source did not provide enough voltage to drive the headphones at sufficient volume.
In total seven NE5532 IC's would be needed. Quiescent power < 4W. Distortion x10 better than the JLH amplifier, more at higher frequencies. No need to heatsink any parts.
Having said all that, if you enjoyed making and listening to it then it was a worthwhile project for you. I just don't think the design has got anything to recommend it in the 21st century compared to similarly simple to construct alternatives.
ilhemedu31 6 months ago
a bad amplifier which cuts the very low and very high frequencies can be advantageous..
To sum up, you have to listen and not see the oscilloscope.
these amps can be very damaging to the ears..
a bad amp has the advantage of not lasting while listening. musical comfort can make listening last longer and cause wear and tear on the ear in the long term.
like a car, the more comfort there is, the greater the risk of accident.
kooth 1 year ago
AntonX2 1 year ago
KenC90 1 year ago
The "not fully on or off" in another reply is not the definition of Class-A.
Here's the definition (from wiki) of Class-AB:
In a class-AB amplifier, ... each one of the two active elements conducts more than
half of the time.
So "not fully on or off " would apply to Class AB, they are biased so that each is always conducting (that's the bias current), neither one ever turns fully OFF. Class A OTH, is always fully ON. In Class A, there is no modulation of the current from one device to another.
I have seen some Class A where it appears to be two transistors being modulated, but one is a constant current source. That can improve efficiency in certain cases.
rfrisbee 1 year ago
Er, no. If the above were true with a load connected then that would violate kirchhoff's current law.
In Class-AB and Class-B operation one of the push-pull transistors does turn off. A simulation on LTSpice can verify this for you. See attached image:
Class-AB amplifier Bias ~18mA; each transistor's current goes down to <60uA (effectively off) each half-cycle for 5V into 8-ohm load. For small signals (~100mV into 8-ohm) it operates in Class-A.
In Class-A operation neither transistor turns off because the bias current has been set high enough that subtracting the load current still leaves some left to flow through the other transistor. I.e. both transistors are conducting 100% of the time, though with different currents flowing through them (modulated by the input signal) except when there is no current flowing into a load.
A push-pull amplifier's Class of operation simply depends on the bias current, load impedance and the size of the input signal. I.e. It will operate in Class-A so long as the load current remains below the level of the bias current. Above that it will transition into Class-B operation, with one of the devices turning off.
KenC90 1 year ago
https://www.electronics-tutorials.ws/amplifier/amp...
You will see the load is in series with the transistor, their currents are the same, no violations of any laws.
In Class AB, one device carries more current than the other, on opposing half-cycles. That leads to non-linearity, as the devices won't be perfectly matched, and possible cross-over distortion (depending upon bias).
There is no requirement for a Class AB circuit to turn "fully off" (or even 'nearly so'). The Class AB (as opposed to B) is biased to reduce crossover distortion. But that doesn't make it Class A.
rfrisbee 1 year ago
If the bias is high enough that neither transistor turns off over a full cycle then it is Class-A.
If for a limited output range neither transistor turns off over a full cycle, but at larger expected output levels one transistor does turn off then it is Class-AB.
If one transistor is normally off and the other conducting then it is Class-B.
These are the standard definitions of push-pull amplifier operation. But don't take my word for it.
From Doug Self's book "Audio Power Amplifier Design Handbook":
Class-A
In a Class-A amplifier current flows continuously in all the output devices,
which enables the non-linearities of turning them on and off to be avoided.
They come in two rather different kinds, although this is rarely explicitly
stated, which work in very different ways. The first kind is simply a Class-B
stage (i.e., two emitter-followers working back-to-back) with the bias voltage increased so that sufficient current flows for neither device to cut off
under normal loading. The great advantage of this approach is that it cannot abruptly run out of output current; if the load impedance becomes
lower than specified then the amplifier simply takes brief excursions into
Class AB, hopefully with a modest increase in distortion and no seriously
audible distress.
The other kind could be called the controlled-current-source (VCIS) type,
which is in essence a single emitter-follower with an active emitter load
for adequate current-sinking. If this latter element runs out of current capability it makes the output stage clip much as if it had run out of output
voltage. This kind of output stage demands a very clear idea of how low
an impedance it will be asked to drive before design begins.
Class-AB
This is not really a separate class of its own, but a combination of A and B.
If an amplifier is biased into Class-B, and then the bias further increased,
it will enter AB. For outputs below a certain level both output devices
conduct, and operation is Class-A. At higher levels, one device will be turned completely off as the other provides more current, and the distortion
jumps upward at this point as AB action begins. Each device will conduct
between 50% and 100% of the time, depending on the degree of excess
bias and the output level.
Class AB is less linear than either A or B, and in my view its only legitimate
use is as a fallback mode to allow Class-A amplifiers to continue working
reasonably when faced with a low-load impedance.
Who is Doug Self?
https://en.wikipedia.org/wiki/Douglas_Self
KenC90 1 year ago
As you set the bias current higher, you move from Class B to Class AB - agreed. And yes, you are approaching Class A operation as you increase the bias. But you never get there. One transistor is going to be conducting less than the other to drive the load, and that's what Class A avoids.
Let's say we drive 1A peak into an 8 Ohm load, 8 W peak. If the bias is greater than 1A, each of the transistors will conduct 100% of the time. But that doesn't make it Class A. You still have the current increasing in one transistor while it decreases in the other, leading to a non-linearity that is different from what a Class A would experience. Class A will have non-linearity too, but it is a different type, the non-linearity of that single device, not a non-linearity of two devices at different currents.
Also, this statement you (or Doug Self) made is wrong (perhaps it was a typo?):
Class AB is less linear than either A or B,
That's just wrong - Class AB is more linear than Class B, less linear than Class A. The whole purpose of increasing bias current in the Class B topology is to make it more linear, at the expense of increased power dissipation. If Class AB used more power and was less linear than Class B, there would absolutely no reason for it (it probably wouldn't even have a name!), and it would not be as common as it is in high-fidelity amps.
rfrisbee 1 year ago
"Let's say we drive 1A peak into an 8 Ohm load, 8 W peak. If the bias is greater than 1A, each of the transistors will conduct 100% of the time. But that doesn't make it Class A."
Yes, it does! The Wikipedia page you quoted in an earlier reply:
https://en.wikipedia.org/wiki/Power_amplifier_clas...
"In a class-A amplifier, 100% of the input signal is used (conduction angle Θ = 360°). The active element remains conducting all of the time.
Amplifying devices operating in class A conduct over the entire range of the input cycle. A class-A amplifier is distinguished by the output stage devices being biased for class A operation."
Per the above definition: If the bias of a push-pull amplifier is set such that neither device turns off for the entire cycle then the output stage is operating in Class-A.
Class-AB is not more linear that Class B and Doug Self has done the THD measurements to prove it. If I recall correctly it has something to do with the change of transconductance between the Class-A and Class-B modes of operation of the output transistors, but I'd have to look at the details again to be sure.
The main reason to bias an amplifier for Class-AB would be if for most listening levels the amplifier operates in the more linear Class-A mode, only going into Class-B for very loud listening levels. However, most Class-AB amplifiers are biased such that at normal listening levels the amplifier is transitioning between Class-A and Class-B operation, increasing the level of distortion over optimally biased Class-B.
HiFi is full of nonsense I don't think using an argument based on popularity in the HiFi industry as to why Class-AB is better than Class-B has any merit whatsoever. Biasing a crappy amplifier circuit into Class-AB might improve its performance, but not a Class-B one where thoughtful engineering and measurement has been used to maximise its linearity.
At this point I have provided you with repeated explanations, a simulation and expert information on what Class-A, Class-AB and Class-B operation mean. I suggest rather than firing off another response to me you go and read a few books by Doug Self (he has written several) or similar experts in audio engineering to educate yourself.
And just for the record I am an electronics engineer with over 25 years experience in mixed-signal design.
KenC90 1 year ago
Class-AB is not more linear that Class B...
and
However, most Class-AB amplifiers are biased such that at normal
listening levels the amplifier is transitioning between Class-A and
Class-B operation, increasing the level of distortion over optimally
biased Class-B.
OK, I found the source for that:
http://www.douglas-self.com/ampins/dipa/dipa.htm#1
In short, Class-AB offers lower distortion than Class-B below the AB threshold but more above it.
It all seems rather moot to me - no one is going to use a straight Class B as a "Hi-Fi" amp - the crossover distortion at low/moderate levels would be awful. And he even says:
This vital fact is little known, presumably because gm-doubling distortion is at a relatively low level and is obscured in most amplifiers by other distortions.
A Class AB amplifier is an "optimally biased Class B". That's the only difference between Class AB and Class B - the bias (higher bias on Class AB to reduce cross-over distortion). We seem to have the disconnect over biasing vs topology, so I'll leave it at that.
Hey, it was a 'fun' discussion - it's always good (IMO!) to have beliefs challenged, I did some digging and learned/relearned a few things. Have a good day!
FatlyCatly 1 year ago
rfrisbee 1 year ago
The current passing through the output transistors with no signal present is set high enough so that even with a signal present neither transistor fully turns off.
JohnW51 1 year ago
TheSpodShed 1 year ago
JohnW51 1 year ago