More pictures at the
end of the document.
Remarks
This design was for an individual project in an
undergraduate lab session. With the limitations of using components from lab
and keeping cost down many things were compromised. The overall sound quality
and repeatability of performance will suffer from low quality parts including
resistors, capacitors, and type of printed circuit board used. The battery
performance will suffer because of the low efficiency op amps; the power rails
are ran off of an LM324, which are power hungry op amps in a comparator
configuration. This configuration was chosen to minimize the number of
components and overall size of the end circuit and also to keep refrain from
ordering a compatible high efficiency quad op amp. With all of these
restrictions, this project has become a prototype design project.
The goals in mind:
Fit in an Altoids box:
The design started with the
challenge of fitting the circuit in an Altoids box. I have seen many projects
online about different ‘Altoids Box Projects’ but wanted to make my own. The
altoids box proves to be difficult because of the rounded corners and Tin
bottom. A printed circuit had to be made to accomplish this goal.
Solid design/schematic:
The signal handling is based
largely on the NWAVGUY’s schematic for his ‘Objective 2’ headphone amp design.
NWAVGUY used two 9-volt batteries, a 200mA LED, a DC supply option, and an
added gain switch; all features I have left out of this design to save space.
Handle a large variety of headphones (in ear/over ear/on
ear):
There are HIFI headphones that
are extremely current hungry for example the 38 ohm planar HiFiMan HE-4/HE-5LE
and headphones like the Beyer DT880-600 that need large amounts of voltage for
being dynamic/planar headphones (Design
Process NWAVGUY).
Ability to add a charging port (future).
The following steps were taken to design the headphone
amplifier.
The first step was to design the power handling stage. The
Objective 2, also referred to as the O2, uses a dual supply off of two 9-volt
batteries and has a DC supply that will charge the rechargeable 9-volt
batteries when not in use. To fit the circuit in an Altoids box the DC supply
option was taken out and one of the two batteries were removed.
To accomplish a clean output signal dual rails were created
off of a single battery. Using a voltage divider and a two-stage voltage
follower from an LM324 to source enough current and create a virtual ground.
To detect a low battery a comparator circuit was used and
implemented with an op amp. The operational amplifier chosen has four channels
and handles plenty of current for this application. The circuit from the
Objective2 amplifier was modified to set the cutoff voltage to 7.0 Volts (see
calculations section). So when the
circuit is running off of a 9-volt battery, and the battery runs low the rails
will be cut off to protect the sensitive circuitry to follow.
MOSFETs were used to switch the output rails from on to off.
The source of the MOSFET was connected to the input rail, drain was connected
to the output rail, and the gate was connected to the respective outputs from
the op amps.
When the battery supplies sufficient voltage (above 7.0 volts)
the P-MOS (on the top rail) gets a low voltage make VSG greater than
its threshold and the N-MOS (on the bottom rail) gets a high voltage, making VGS
greater than the threshold. When the output rails are supposed to be cutoff the
gate voltage on each MOSFET are equal to the source voltage, so VSG
=VGS = 0.
Electrolytic and ceramic capacitors along with a power
switch were added to the design to stabilize the key voltages and complete the
circuit.
The capacitors from the output of the op amp to the rail
stabilize the gate voltage so the MOSFETs do not turn off more than they need.
The capacitors from the rail to ground (input or output) are used to supply
momentary voltage to the rails when an unbalanced output is pulled from the
rails
The signal handling was straightforward. To supply the added
voltage gain for the voltage hungry headphones a gain stage was the first stage
the signal went through. For current hungry headphones a current stage was
added before the output.
A non-inverting amplifier configuration was used to give a
voltage gain and save the phase angle. The capacitor to ground was used for
coupling to provide better response to lower frequencies. The capacitor in the
feedback loop is to kill any DC signals and very low notes (less than 10 Hz).
Note: only one channel shown.
Two stages of current gain were added per channel using a
high current op amps in a voltage follower configuration. Pull down resistors
were added at the non-inverting input to kill any amplification on no signal
inputs (ie noise) and to reduce the no signal or ideal output.
The volume control was achieved by putting an audio
potentiometer between the voltage gain stage and the current gain stage. The
input was fed to one sided of the POT, the output was taken from the wiper of
the POT, and the other end was connected to ground. As the potentiometer is
turned from lowest audio output to loudest output the resistance is decreased
allowing more voltage to go to the current gain stage. This is shown in the
schematic section.
Full Schematic: Note the Battery on the left is represented
as a capacitor so the leads of the 9-volt connecter can be properly soldered.
Close up of the power handling:
Close up of the signal handling:
View of the routed PCB board:
Pseudo ground planes were added to the signal side to help
clean any capacitive noise in the output signal.
Power Circuit
5 * 330 KΩ resistor
1 * 120 KΩ resistor
1 * 560 Ω resistor
4 * 100 μF electrolytic capacitor
5 * 0.1 μF ceramic capacitor
1 * Red LED
1 * DPDT Latching Switch
1 * 9-Volt battery clip
1 * LM324 operational amplifier
1 * BS170 N channel MOSFET
1 * ZVP3306 P channel MOSFET
Signal Circuit
2 * 120 Ω resistor
2 * 10 KΩ resistor
2 * 1 KΩ resistor
2 * 330 KΩ resistor
1 * 10 KΩ Dual Audio Potentiometer
4 * 220 pF ceramic capacitor
2 * 1 μF ceramic capacitor
2 * 3.5mm Headphone jack
2 * NJM4556AD High current Operational Amplifier
1 * NJM2068D Low Noise Operational Amplifier
Low power ratios:
Low battery voltage = 7.0
Negative terminal to the first op amp:
This voltage is slightly greater than the Vd of
the RED LED causing the output rails to be zero.
Frequency Response:
The following MultiSim test was performed to visualize the
pass band of the amplifier.
The amp has great pass band region. The top graph shows the
output at the potentiometer at zero percent, and the bottom graph shows when
the POT is at 100%.
The output current was measured using a moderate load from a
300mV signal.
The amplifier was built and tested as follows. Pictures of
finished product are at the bottom of the page.
The output response was measured at maximum volume at
different frequencies. The following O-Scope shots detail the results.
At 20 Hz:
At 100Hz:
At 20KHz:
At 10Hz:
G=Vo / Vi =2.485 V/V
At 100Hz:
G = 2.507
At 1KHz:
G=2.507
At 10KHz:
G = 2.507
At 20KHz:
G = 2.507
It is easy to see that the amplifier has only amplifies the
signal and does not attenuate it.
The output of the amplifier was measured when ground was
wired into the input. This test shows the output of the Amp when no input is
connected or when the music is paused. Note the scaling factors no the probes
(Bottom left):
The input signal is right around ground with added noise
from the environment. The output signal in red shows the ideal condition
response. The ideal response ~= 35mV. The input signal is often around 350mV so
the added distortion to the signal can be generalized and approximated to be ~
10%.
The amplifier works and sounds good. The testing results
from the frequency response is promising but the idle condition test yielded a
very poor result. Many things will be changed in the next version:
PCB Routing:
The PCB Routing could have been
much more efficient. The PCB can also be much smaller when a professional
service is used to create the circuit. The battery had to be stood upright to
fit in the enclosure. If the board was shortened the battery could be flat to
fit under the lid.
Cost:
Smaller resistors and more
efficient op amps can be used to save space and power.
References:
NWAVGUY. "NwAvGuy." : O2 Documentation. Web. 18
Apr. 2015.
The first sub-test was a load test
to see how stable the section is at near maximum output. The section was
discharged at 300 amps, then the sections voltages were checked to make sure
they were within a given specification. To do this test the section first had
to be at a standardized initial state. A subvi was created to ‘Precondition’
the section for the load test to verify each cell of the battery section were
within a given requirement and there is no more than twenty millivolts
difference between any two cells. Once the Precondition subvi was passed the
battery would then go through the load test. This test discharged the battery
at 300 amps for 30 seconds. To do this, initialization of the remote channels
of the ABC150 were established and -300 amps was requested at a specific time.
A subvi was created to communicate with the ABC150 so that when current was
requested it would poll the output and not continue until the requested current
was received. Safety checks were added to the test in addition to an emergency
stop button, so that the battery would safely make it through all tests.
Internal thermistor voltages were constantly polled so that if the temperature
of the battery ever got out of hand it would kill the test. After the battery
section successfully made it through the 30 seconds at 300 amps the test would
then poll the cells to discern if the cells were within a required voltage and
had an acceptable voltage difference between the cells. A bad voltage
difference, whether before or after the test, indicates a bad cell. There are
two types of cell failure, an imbalance between the cells and a faulty or bad
cell. When viewing the graph of the cells it is easy to discern whether the bad
voltage difference is due to a bad cell or an imbalance: a bad cell will start
higher (or even) with the other cells, but drop much more rapidly than the
other cells, this shows that the cell has a lower capacity than the other. One
would see the bad cell cross the voltage level of the other cell. In the
situation of an imbalance, an unbalanced cell will stay parallel with the other
cells, but be higher or lower than the rest of the cells during the whole test.
If the battery section makes it through the 300 amp test with an acceptable
voltage level and difference then the program will continue to the next test.
A server power supply was used to
create a constant current output to charge the new ‘Tennant’ battery. The
supply has constant voltage with short circuit protection, because the battery
is such a low resistance the charger’s voltage rail would crash out when
connected. So a circuit was added to its output to make the supply a
freewheeling constant current supply. This was done by connecting a current
sensor to a comparator circuit that controlled the gate of a FET. Power is
supplied to the battery, when the current sensor detects zero current it would
drive the power FET’s Gate high and charge the battery, when the current sensor
detected this power it would turn off the supply. This made PWM-like output to
the battery that would continually charge tenant with constant current.
Multiple large heat sinks were added to the components including the voltage
regulators, the FET, and the diode. The diode was added to the design for the
inductive kick that will come off of the battery when the voltage supply goes
off. The diode allows that kick to be recirculated through the battery so that
it does not damage the power source or electronics. A copper wire was placed
under the solder traces on the proto-board and so the traces could carry more
current.
