This is only a preview of the February 2012 issue of Silicon Chip. You can view 24 of the 104 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "Crystal DAC: A High-Performance Upgrade":
Items relevant to "SemTest: A Discrete Semiconductor Test Set; Pt.1":
Items relevant to "Simple 1.2-20V 1.5A Switching Regulator":
Purchase a printed copy of this issue for $10.00. |
Crystal DAC
For the very best performance from 24-bit/96kHz
recordings – uses the Crystal CS4398 DAC and a
discrete transistor output stage
This new DAC board can be substituted for the original board
used in our Hifi Stereo DAC project (Sept-Nov 09) without any
major changes, effectively replacing the Burr-Brown DSD1796
DAC IC with the high-end Cirrus Crystal CS4398. Its harmonic
and intermodulation distortion figures are significantly lower
than before although some people will have difficulty discerning
the differences. Try it and find out for yourself.
T
HE INSPIRATION for this project came from our review of the
Marantz CD6003 CD player, which appeared in the June 2011 issue. At the
time, we made some measurements
using our Audio Precision System One
and discovered that it not only had a
very low harmonic distortion figure
24 Silicon Chip
for a CD player but it was practically
flat across the audible frequency band
(20Hz-20kHz).
We figured that this was partly due to
its Crystal (Cirrus Logic) CS4398 DAC
(digital-to-analog converter) IC. This
is mounted on a large PCB, amongst a
forest of discrete and passive compo-
nents. So we thought, hmmm . . . could
we do something similar for our DAC
design? We suspected they were also
doing some fancy digital processing
using a DSP (digital signal processor)
to get that level of performance but that
the CS4398 DAC must also be pretty
good for such an excellent result.
siliconchip.com.au
By NICHOLAS VINEN
It turns out we were right on both
counts. The CS4398 is very good but
Marantz seem to be doing some digital
interpolation (possibly increasing the
sampling rate to 96kHz or 192kHz)
to keep the distortion so low. While
our new DAC board does not have
the benefit of digital interpolation,
it is clearly superior to the previous
design, especially when processing
24-bit/96kHz program material.
If you have already built a Stereo
DAC kit and would like to try out
this new board, it’s pretty easy. You
just build the new PCB and swap it
for the old one; it’s the same size and
the critical parts are in the same locations. You then reprogram or swap the
microcontroller on the input board and
Bob’s your uncle.
Like the Marantz, we designed the
filtering hardware using all discrete
components (ie, bipolar transistors
and passives).
There was some controversy on the
internet (unheard of!) over our choice
of op amps in the original DAC design
(SILICON CHIP, September, October &
November 2009). This time we have
avoided using those “evil” little black
boxes, which should make the extreme
audiophile cognoscenti happy (impossible!).
The resulting circuit has a lot more
components than it would if we had
used op amps but they are all cheap
siliconchip.com.au
and commonly available. The resulting wide bandwidth compared to an
op amp means that the output filtering
works very well.
Performance
We tested both the original and new
DAC designs extensively, using both
our Audio Precision System One and
the newer Audio Precision APx525
with digital processing. We also
performed numerous listening tests,
including blind A/B tests.
The first result that became clear
from all this testing is that the original
design really is very good. Its distortion and noise are low (including intermodulation distortion), its linearity
is very good and it generally sounds
excellent. However, the new DAC design measures even better, with lower
distortion (especially at high frequencies), even lower intermodulation
distortion and astounding linearity
down to -100dB.
Fig.1 shows a comparison of the
harmonic distortion between both
channels of the original and the new
DAC design. These tests were performed on the same unit with just the
DAC boards swapped, so they give an
apples-to-apples comparison.
Note that noise has been digitally
filtered out of this measurement completely, for a couple of reasons. First,
both DACs have quite a bit of high-
frequency switching noise in their
output (but a lot less than some DVD
and Blu-ray players we’ve tested!)
and this can mask the distortion if
we set the bandwidth wide enough
to capture harmonics of high audio
frequencies. Second, the 20Hz-20kHz
residual noise of both the original
and new boards is similar and this
too means that a THD+N comparison
would tend to understate the reduction
in harmonic distortion obtained with
the newer design.
As you can see, harmonic distortion
with the CS4398 is substantially lower
than the original design, both at high
frequencies (above 3kHz) and low
frequencies (below 100Hz). The differences between channels are due to
asymmetries in the PCB layout as well
as mismatches between the two channels within the DAC ICs themselves
(eg, due to resistor ladder tolerances).
Fig.2 shows the channel separation
for both units. The lines labelled “left”
show how much signal from the right
channel couples into the left and the
lines labelled “right” show the opposite. In both cases, channel separation is very good and is generally
better than -100dB across the audio
spectrum. The older design is slightly
better in this respect, although the difference is largely academic.
Fig.3 compares the linearity of both
DACs. This plot shows the deviation
February 2012 25
Performance Graphs
Harmonic Distortion vs Frequency, 90kHz BW
05/12/11 12:31:22
Crosstalk vs Frequency, 90kHz BW
05/12/11 14:45:09
0
0.01
Left Right
Left Right
CS4398
DSD1796
0.005
CS4398
DSD1796
-20
Crosstalk (dB)
Harmonic Distortion (%)
-40
0.002
0.001
-60
-80
0.0005
-100
0.0002
0.0001
20
-120
50
100
200
500
1k
Frequency (Hertz)
2k
5k
10k
20k
Fig.1: harmonic distortion (ignoring noise) versus freq
uency for the original (DSD1796-based) and new (Crystal
CS4398-based) DACs. The newer design has lower
distortion overall but especially above 2kHz. The channels
differ slightly due to layout asymmetries and differences
in the ICs themselves. The spikes at 1.2kHz and 9kHz are
due to aliasing between the test and sampling frequencies.
between the expected and actual
output level for a sinewave at a range
of levels between -60dB and -100dB.
Both DACs perform extremely well in
this test but the CS4398 is especially
good, with a maximum deviation of
no more than 0.25dB at -100dB! Its
deviation is essentially zero above
-84dB while the DSD1796 still shows
some deviation up to -70dB.
Note that all of the above test results
were obtained with the Audio Precision APx525 (which can test in the
analog or digital domain) using 24-bit
96kHz signals fed into a TOSLINK
input of the Stereo DAC project.
Fig.4 shows the FFT frequency
spectra for the updated DAC with
one channel in magenta and the other
in khaki. This was computed with a
one million sample window, an equiripple algorithm and 8x averaging.
The test signal is at 1kHz and the
bandwidth is 90kHz. The harmonics
of the test signal are clearly visible at
2kHz, 3kHz, etc. Also visible is some
50Hz and 100Hz mains hum at around
-120dB, as well as various intermodulation products of this hum with the
fundamental and its harmonics.
As we said earlier, both DACs are
very good but the updated design generally has better figures. We also ran
the SMTPE intermodulation distortion
test on both. This involves sending a
26 Silicon Chip
-140
20
50
100
200
500
1k
Frequency (Hertz)
2k
5k
10k
20k
Fig.2: a comparison of channel separation (ie, crosstalk)
for the original and new DAC boards. The original is
slightly superior but both are very good, with less than
-93dB crosstalk at any frequency and separation of at least
100dB up to 1kHz. As is typical, there’s more coupling
in one direction (for the new design, left channel to right
channel) than the other, again mainly due to asymmetry.
4:1 mix of 7kHz/400Hz sinewaves to
the test device. These frequencies are
then filtered from its output (400Hz
with a high-pass filter and 7kHz
with a notch filter) and the remaining harmonics measured. These will
generally be the sum and difference
frequencies of 6.6kHz and 7.4kHz but
possibly other harmonics too.
The old design gives an intermodulation distortion level of around
0.0018% (-95dB) while the new design
gives 0.0006% (-105dB); a significant
improvement.
Listening tests
The results of our listening tests
were somewhat controversial. We
used our 20W Stereo Class A Amplifier (May-September 2007) and the M6
Bass Reflex Loudspeakers (November
2006), while the 3-Input Selector presented last month was used to switch
between the original and updated
Stereo DAC prototypes. The original
prototype was set to a volume of
-0.5dB and the levels matched almost
perfectly, giving seamless switching
between the two.
The two Stereo DACs themselves
were fed with digital audio from a
Blu-ray player with separate TOSLINK
and S/PDIF outputs.
Some staff members could not
tell the difference in sound quality
between the two DACs while others
claimed to be able to hear a distinct
difference between the two on certain
passages, although the difference was
not obvious on other passages. With
complex choral music, two of the
“guinea pigs” were able to pick the
updated DAC as sounding “brighter”.
On other types of music, a difference
could be discerned but we could not
reliably pick which DAC we were
listening to.
You’ll have to make your own mind
up about whether the new design gives
an audible improvement. However, we
can be certain that this upgraded DAC
design gives far superior performance
compared to virtually any CD, SACD,
DVD or Blu-ray player on the market.
And for those people who think that
Blu-ray players are generally superior
in terms of sound quality, our limited
tests demonstrated that this is not
necessarily true. Cheap Blu-ray players are just that – cheap!
Circuit description
Fig.5 shows the circuit diagram
for the new board. IC1 is the CS4398
DAC chip and this is wired to 16-pin
IDC socket CON1. Its configuration is
identical to that of the original DAC
board, carrying the 3.3V supply from
the control board as well as audio data
(pins 4, 6, 8 & 10) and serial control
siliconchip.com.au
Linearity
05/12/11 14:01:58
Frequency Domain Plot
+1.0
+40
Left Right
+0.8
CS4398
DSD1796
+20
0
+0.4
-20
+0.2
-40
Level (dBr)
Output Deviation (dB)
+0.6
0
-0.2
-60
-80
-0.4
-100
-0.6
-120
-0.8
-140
-1.0
-100
-160
-90
-80
Nominal Output Level (dBr)
-70
-60
Fig.3: a comparison of the linearity of the original and
updated DAC boards. Delta-Sigma DACs typically have
good linearity and in fact both are excellent. However,
the updated board (with the CS4398) is the best of the two
with an astounding deviation of less than one quarter of a
decibel at levels down to -100dB! (The dynamic range of
CD-quality audio is just 96dB).
data (pins 7, 9, 11 & 13). There are
also two mute feedback lines (pins
15 & 16), allowing the micro to sense
output silence.
IC1 has a dual 3.3V and 5V power
supply with multiple supply pins for
each internal section. Both rails have
100µF bulk bypass capacitors. Each
supply pin also has a 100nF bypass
capacitor for lower supply impedance
at higher frequencies (>100kHz).
VLS (pin 27) is supplied 3.3V to suit
the audio serial data levels while VLC
(pin 14) is at 5V to match the microcontroller’s I/O levels. To avoid switching
noise feeding back into the 5V rail,
which also powers analog circuitry,
a 100Ω stopper resistor is included.
VD (pin 7) is the supply pin for the
DAC’s digital core (digital filtering and
so on). This runs off 3.3V while the
internal analog circuitry (op amps, etc)
runs off a 5V rail connected to VA (pin
22). This 5V rail is also fed separately
to VREF (pin 17) for the DAC reference
voltage. Capacitors at FILT+ (pin 15)
and VQ (pin 26) smooth IC1’s internal
reference voltages.
VQ is the quiescent output voltage
and generally sits at half supply (ie,
2.5V). We aren’t using the DSD (Direct
Stream Digital) input pins on the IC so
they are tied to ground.
The microcontroller’s serial I/O pins
connect to header CON1 via LK1-LK4.
siliconchip.com.au
.03
.05
.04
.1
.2
.3 .4 .5
1
Frequency (kHz)
2
3 4 5
10
Fig.4: a frequency domain plot (ie, spectrum analysis) of
the output of the updated DAC for a 1kHz sinewave. Eight
FFTs were averaged to reduce noise. The harmonics are
clearly visible at multiples of the fundamental (2kHz, 3kHz,
etc) as well as mains hum at 100Hz. You can also see the
various intermodulation products of the fundamental and
its harmonics with 100Hz.
These are closely-spaced pads on
the bottom of the PCB which can be
bridged with solder. The CS4398 can
operate without a microcontroller and
to do so, pins 9-12 are connected to
either ground or VLC (+5V).
This arrangement allows those pins
to be connected to configure the DAC
correctly, even in the absence of a
microcontroller. However, if this is
done, many features of this design do
not operate properly, such as volume
control, automatic input scanning and
muting. As a result, we suggest that
constructors simply bridge LK1-LK4
and reprogram the micro with the new
software. All the features of the original design will then work normally.
Analog filtering
The DAC IC we used previously
(Burr Brown DSD1796) has differential current outputs while the CS4398
has differential voltage outputs. That
means we no longer need current-tovoltage converters; they are internal
to IC1. However, we still need to filter
the outputs to remove the DAC switching noise and convert the differential
(balanced) signals to unbalanced, to
suit the inputs of a typical amplifier.
We have used the recommended
filter, a 2-pole Butterworth low-pass
arrangement, consisting of six resistors
and five capacitors for each channel.
These are shown just to the right of IC1.
The operation of this filter is quite
complicated since the two RC filters for
each channel interact with each other.
Let’s look at the left channel; the right
channel circuit is identical. The noninverted output from IC1 comes from
pin 23 (AOUTA+) and the inverted
signal from pin 24 (AOUTA-).
The waveforms from each pin are
(theoretically) identical but opposite
in polarity, ie, one swings up when
the other swings down and vice versa.
Both signals are attenuated, with a gain
of around 0.45, by a pair of resistive
dividers. While the division ratios
are very similar, the actual resistor
values differ: 620Ω/510Ω for the noninverted signal and 1.6kΩ/1.3kΩ for
the inverted signal.
These resistors also form singlepole, low-pass filters in combination
with the 18nF (non-inverted signal)
and 6.8nF (inverted signal) capacitors. The attenuating resistors are effectively in parallel with each other,
for a -3dB point of around 32kHz in
both cases.
These are then followed by another
set of RC low-pass filters – 270Ω/4.7nF
for the non-inverted signal and
680Ω/1.8nF for the inverted signal. In
isolation, these have corner frequencies of around 130kHz.
Note that the bottom ends of the
February 2012 27
20
DIGITAL
INPUT/OUTPUT
+3.3V
1
+5V
3
100 F
100nF
100nF
7
22
VD
27
100 F
620
VA
VLS
VLC
100
14
510
100nF
100nF
18nF
100 F
4
6
6
4
8
3
10
5
5
13
Vref
MCLK
SCLK
LK1
9
9
LK2
10
7
LK3
11
13
LK4
12
15
25
16
18
2
1
12
2
14
28
1.6k
100nF
SDIN
680
100 F
LRCLK
6.8nF
1.3k 1.8nF
RST
IC1
CS4398
100k
11
17
CDIN
AOUTA+
AOUTA–
23
+2.5V
24
+2.5V
20
+2.5V
19
+2.5V
CCLK
CDOUT
AOUTB+
AD0/CS
AOUTB–
AMUTEC
BMUTEC
FILT+
DSD_B
DSD_SCLK
VQ
DSD_A
REF GND
15
26
16
100nF
CON1
IDC-16
DGND
AGND
21
8
10 F
100 F
10k
620
510
18nF
100 F
+15V
D5 1N4004
K
POWER IN
CON2
1
220
+15V
100 F
2
3
SC
2012
100 F
680
6.8nF
REG1 78L05
IN
1.6k
A
+5V
OUT
GND
1.3k 1.8nF
100 F
0V
10k
–15V
–15V
STEREO CRYSTAL DIGITAL-TO-ANALOG CONVERTER
Fig.5: the circuit is based on a Cirrus Logic (Crystal) CS4398 stereo DAC chip (IC1). This has differential outputs (pins
23 & 24 and 20 & 19) and these drive discrete audio output stages based on transistors Q1-Q12 in the left channel and
Q15-Q26 in the right channel. Q14, Q28 & dual N-channel Mosfets Q29a-b & Q30a-b mute the outputs when there is no
signal from the DAC. Power comes from an external ±15V supply, with REG1 providing a +5V rail for IC1.
28 Silicon Chip
siliconchip.com.au
100
K
D1
1N4004
220
A
Q5
BC559
270
E
47 F
2.2k
B
2.2k
B
B
100
47 F
E
C
C
E
47 F
Q7
BC559
–15V
100
C
B
E
2.2k
47 F
10k
E
220
C
10k
Q2
Q1
BC559 BC559
B
E
C
C
100
4.7nF
Q6
BC559
+15V
VR1
5k
B
B
C
E
Q10
BC549
Q11
BC549
TP1
10
TP2
47 F
+2.5V
100pF
1nF
10
C
Q3
BC549
B
B
E
D2
1N4004
K
C
E
68
100
Q8
BC549
B
C
10nF
Q12
BC559
D
+5V
B
C
2.2k
Q14
BC559 E
Q9
BC549
E
G
100pF
B
100k
100
ZD1 18V
D3
1N4004
220
A
Q19
BC559
270
E
47 F
2.2k
B
2.2k
B
B
47 F
10k
E
C
C
Q21
BC559
–15V
100
C
B
E
VR2
5k
B
B
C
E
Q24
BC549
Q25
BC549
TP3
10
TP4
47 F
+2.5V
10
C
K
B
E
B
E
68
100
Q22
BC549
B
C
10nF
Q26
BC559
D
+5V
2.2k
2.2k
C
Q28
BC559 E
Q23
BC549
E
G
100pF
B
S
S
G
Q30b
IRF7905
C
100k
100
A
100k
D
100
ZD3 18V
K
–15V
BC549, BC559
D1–D5: 1N4004
A
siliconchip.com.au
Q30a
IRF7905
C
B
68
100k
E
B
E
Q18
BC549
RIGHT OUT
CON4
100
100pF
1nF
C
K
ZD2 18V
47 F
2.2k
100 47 F
E
E
K
+15V
220
C
10k
Q16
Q15
BC559 BC559
B
E
C
C
100
4.7nF
Q20
BC559
A
A
K
100
D4
1N4004
Q29a
IRF7905
D
100k
–15V
Q17
BC549
S
S
G
C
100
A
K
Q29b
IRF7905
C
2.2k
68
100k
E
B
E
Q4
BC549
LEFT OUT
CON3
100
ZD1–ZD4
A
K
A
K
ZD4 18V
78L05
B
E
A
COM
C
IN
OUT
February 2012 29
Silicon Chip
Binders
REAL
VALUE
AT
$14.95
PLUS P
&
P
Features & Specifications
Output Level .................................................................................. 1.9V RMS
Signal-To-Noise Ratio ........................................................................-112dB
Idle Channel Noise ...........................................................................<-124dB
Channel Separation ........................................~100dB <at> 10kHz (see Fig.2)
Harmonic Distortion (see Fig.1) ... <0.001% <at> 1kHz, <0.002% 20Hz-20kHz
THD+N .............................................................................. 0.0014% <at> 1kHz
Intermodulation Distortion .................................. <0.001% (400Hz/7kHz 4:1)
Frequency Response .........................................-0.25,+0.05dB 20Hz-20kHz
Supported Sampling Rates .......... 32kHz, 44.1kHz, 48kHz, 88.2kHz, 96kHz
Signature ________________________
1.3kΩ resistor and 1.8nF capacitor
are connected to the output of the
following differential amplifier, rather
than ground. Because the output is
out of phase with the inverted signal
from pin 24 of IC1, this acts like a
virtual ground. So there is twice the
voltage across these compared to the
non-inverted signal filter, hence the
higher resistance values (keeping the
current from each output approximately equal).
The overall filter response (determined by simulation) is -3dB at 45kHz,
which is above the 30kHz or so you
would expect if the filters operated
in isolation. This is partly due to
their interaction and also partly due
to the connection from the differential
amplifier’s output to the inverting
signal filter. As we said earlier, it’s
complicated!
The resulting response is -0.1dB at
20kHz. Including the DAC’s internal
filtering and the additional filtering at
the output, the overall response for the
circuit is -0.25dB at 20kHz, which is
quite acceptable.
The active filter gives around 13dB
of attenuation at 100kHz, increasing
at around 12dB/decade. This is ultimately limited by the bandwidth of
the differential amplifier circuit and
so the filter is ineffective at very high
frequencies (many MHz).
This means that the 1.8nF capacitor in the filter network can couple
very high frequencies through to the
output but their level is too low to
cause problems.
Name ____________________________
Discrete op amps
These binders will protect your
copies of S ILICON CHIP. They
feature heavy-board covers & are
made from a dis
tinctive 2-tone
green vinyl. They hold 12 issues &
will look great on your bookshelf.
H 80mm internal width
H SILICON CHIP logo printed in
gold-coloured lettering on spine
& cover
H Buy five and get them postage
free!
Price: $A14.95 plus $A10.00 p&p
per order. Available only in Aust.
Silicon Chip Publications
PO Box 139
Collaroy Beach 2097
Or call (02) 9939 3295; or fax (02)
9939 2648 & quote your credit
card number.
Use this handy form
Enclosed is my cheque/money order for
$________ or please debit my
Visa Mastercard
Card No:
_________________________________
Card Expiry Date ____/____
Address__________________________
__________________ P/code_______
30 Silicon Chip
As noted above, we have used
discrete transistors in this circuit
instead of op amp ICs. The design is
very similar to that used in the Hifi
Stereo Headphone Amplifier (OctoberNovember 2011).
Again referring to the left channel
only, the base of NPN transistor Q1 is
the non-inverting input of the differential amplifier while the base of Q2
is the inverting input. Both transistors
have 100Ω emitter degeneration resistors to improve linearity.
PNP transistor Q5 acts as a constant
current source for the long-tailed pair
and this is set to around 3mA by a 220Ω
resistor. NPN transistors Q3 and Q4
form a current mirror collector load,
with 68Ω emitter resistors to improve
current sharing.
The current into the base of NPN
transistor Q8 is proportional to the
difference in voltage between the two
inputs (ie, between the bases of Q1 &
Q2). Q8 and NPN transistor Q9 act
as a beta-enhanced transistor (like a
Darlington) and operate as a commonemitter amplifier. PNP transistor Q7
acts as a constant-current collector
load at around 3mA.
Together, Q8 & Q9 form a trans
impedance amplifier, converting the
current delivered to the base of Q8 into
a voltage at Q9’s collector. This voltage controls the output stage which
consists of NPN transistor Q11 and
PNP transistor Q12 in a push-pull,
emitter-follower configuration.
NPN transistor Q10 forms a VBE
multiplier. This generates an adjustable bias (set by trimpot VR1), so that
both Q11 & Q12 are conducting full
time, giving Class A operation.
The 100pF and 1nF capacitors
between Q9’s collector and Q8’s base
provide frequency compensation. The
two constant current sources (Q5 & Q7)
limit their charge and discharge currents and so set an upper limit on slew
rate and frequency, reducing gain at
siliconchip.com.au
very high frequencies below the level
required for sustained oscillation.
With this 2-pole compensation
scheme, the 2.2kΩ resistor to the -15V
rail increases the open loop gain
available at higher audio frequencies
(see “A Look At Amplifier Stability &
Compensation”, July 2011). At low frequencies, this resistor shunts much of
the current passing through the 100pF
capacitor so that it never reaches Q8’s
base but at much higher frequencies,
the capacitor’s impedance so low that
it has no effect.
PNP transistor Q6 provides the
bias and negative feedback for current sources Q5 and Q7, keeping the
voltage across their emitter resistors
constant. Its own collector load is a
bootstrapped constant-current sink
formed from two 10kΩ resistors and
a 47µF capacitor. This prevents variations in the supply rail from affecting
the current regulation, as this would
increase inter-channel crosstalk and
reduce supply hum rejection.
The signal output appears at the
junction of the 10Ω emitter resistors
for Q11 & Q12. The output voltage has
a 2.5V DC offset which is removed by
a 47µF DC-blocking capacitor with a
100kΩ bias resistor. The audio signal
then passes through an additional RC
low-pass filter (100Ω/10nF) before
passing to the output RCA connector
CON3 (CON4 in the right channel).
Since the output signal swing is
about ±2.7V (1.9V RMS), the 100Ω
resistor limits the short-circuit output
current to 27mA. Otherwise, Q11 or
Q12 would quickly burn out with a
shorted output.
Muting
As suggested in the CS4398 data
sheet, we have added muting circuitry
to the outputs. This consists of a dual
Mosfet for each channel, the Mosfets
operating as analog switches. These
short the output to ground when there
is no signal from the DAC.
This suppresses any clicks or pops
that may occur when the sample rate
changes or the DAC selects a different input and so on. It also makes
the signal-to-noise ratio appear to be
better, by reducing the idle channel
noise. But it doesn’t affect the actual
signal-to-noise ratio during playback
since the muting Mosfets are then
switched off.
These components are not strictly
necessary but don’t add much cost or
siliconchip.com.au
complexity to the circuit.
The example circuit in the CS4398
data sheet uses 2SC2878 NPN transistors rather than Mosfets. These are a
special type of bipolar transistor with
an unusually high reverse hFE of 150,
compared to around 1-2 for a normal
NPN transistor. So they can operate
normally even with their collector and
emitter reversed; in this case, when
the collector voltage (ie, signal) swings
below ground.
2SC2878 transistors are available
but not widely so. By contrast, the dual
Mosfets we have used instead can be
bought from many different sources.
The CS4398 DAC automatically determines the polarity of its AMUTEC
and BMUTEC outputs (for the left and
right channels, respectively) based on
the external biasing arrangement. In
this case, they have a resistive path
to ground and so the chip drives them
low to mute and high otherwise.
When the mute output is low, current is sunk from the base of PNP
transistor Q14 via the 100kΩ resistor,
turning it on. Q14 then pulls the gates
of Q29a & Q29b high to 5V via a 100Ω
resistor. The 100Ω resistor creates a
low-pass filter with the Mosfet gate
capacitance, preventing voltage spikes
due to stray inductance.
The two Mosfets in each pair are
connected source-to-source, with one
drain connected to the output and the
other to ground. As a result, the two
parasitic body diodes are connected
anode-to-anode so that regardless of
the output signal voltage polarity, at
least one is reverse-biased. If we had
used a single Mosfet instead, the signal
would be clipped to within one diode
drop to ground when the body diode
was forward-biased.
These diodes also clamp the sources
of both Mosfets to no more than 1V
above ground. So when the gates
are at +5V, both Mosfets have a gatesource voltage of at least +4V. The
on-threshold for the IRF7905 is no
more than 2.25V so they are turned
on hard in this situation, shorting the
output to ground.
When the AMUTEC mute output
goes high, Q14 turns off and so the
gates of Q29a & Q29b are pulled to
-15V via a 100kΩ resistor. This is well
below the lowest output signal voltage
of -2.7V and so both Mosfets switch off
and the signal is unaffected.
When off, the Mosfets do have
some capacitance, due mainly to the
Parts List
1 PCB, code 01102121, 94 x
110mm
1 16-pin PCB-mount vertical IDC
connector (CON1)
1 3-way mini PCB-mount
terminal block, 5.08mm pitch
(CON2)
1 white PCB-mount switched
RCA socket (CON3)
1 red PCB-mount switched RCA
socket (CON4)
2 5kΩ mini sealed horizontal
trimpots
M3 nuts and flat washers (may
be required to adjust new
PCB height to suit holes in
existing case)
Semiconductors
1 CS4398 Stereo DAC IC (IC1)
(Element14 1023397)
1 ATMega48 programmed with
0110212A.hex (or reprogram
existing micro)
2 IRF7905 dual N-channel
SMD Mosfets (Q29,Q30)
(Element14 1791580)
1 78L05 5V linear regulator
(REG1)
14 BC559 PNP transistors
(Q1-Q2, Q5-Q7, Q12, Q14-Q16,
Q19-Q21, Q26, Q28)
12 BC549 NPN transistors
(Q3-Q4, Q8-Q11, Q17-Q18,
Q22-Q25)
5 1N4004 1A diodes (D1-D5)
4 18V zener diodes, 0.4W or 1W
(ZD1-ZD4)
Capacitors
9 100µF 16V electrolytic
10 47µF 35V/50V electrolytic
1 10µF 16V electrolytic
6 100nF MKT
2 18nF MKT
2 10nF MKT
2 6.8nF MKT
2 4.7nF MKT
2 1.8nF MKT
2 1nF MKT
4 100pF NP0/C0G
Resistors (0.25W, 1%)
7 100kΩ
2 510Ω
6 10kΩ
2 270Ω
10 2.2kΩ
5 220Ω
2 1.6kΩ
17 100Ω
2 1.3kΩ
4 68Ω
2 680Ω
4 10Ω
2 620Ω
February 2012 31
+
10k
10k
2.2k
Q1
Q2
680
1.3k
Q5
3 x 100F
CON1
16
2
1
220
+15V 0V -15V
15
DIGITAL I/O
REG1
4004
100nF
100F
2.2k
220
2.2k
2.2k
100F
47F
18nF
D5
100k
100nF
+
+
100
+
(UNDER)
Q6
1.8nF
CAD latsyrC
CS4398
100nF 100nF
+
100nF
4004
100
100
510
620
270
1.6k
6.8nF
10F
Q7
1nF Q8
+
100nF
D2
D1
4004
+
620
1.6k
6.8nF
100F
100pF
+
510
Q14
47F
Q9
47F
' 2012
+
01102121
18nF
68
68
100pF
12120110
+
Crystal DAC
1.8nF
100k
4.7nF
2.2k
2.2k
VR2: 5k
100
100
2.2k
D4
D3
4004
4004
100k
100F
Q15
100
100
680
270
1.3k
4.7nF
+
100F
100k
10k
Q28
Q19
100F
Q10
VR1: 5k
Q3
1nF
Q18 Q17
Q16
100k
10k
100pF
47F
47F
18V
100
100pF
Q22
220
68
68
10k
2.2k
10k
18V
100
+
Q24
Q12
+
220
2.2k
100k
10nF 2 x IRF7905 10nF Q4
18V
18V
(UNDER)
ZD3,4
ZD1,2
+
Q20
47F Q11
TP2 TP1
100k
100
47F
Q23
Q21
100
+
+
47F
100
100
100
2.2k
220
+
100
CON4
100
Q26
CON3
+
TP3
10
10
+
Q25
47F
R
OUT
100
L
+
47F
10
10
TP4
RIGHT
(RED)
LEFT
(WHITE)
TOP SIDE OF BOARD
CON2
Fig.6: follow this layout diagram to install the through-hole parts on the
PCB. Take particular care with the transistors. There are two different
types (BC549 & BC559) – don’t get them mixed up.
Left: this is the
fully-assembled
PCB. Note the
orientation of
the IDC socket.
32 Silicon Chip
drain-source capacitance which is at
a maximum of about 350pF when the
drain-source voltage is zero. However,
most of the time, the two capacitances
are in series and so there is effectively
no more than 200pF additional capacitance at each output. This is swamped
by the parallel 10nF capacitors and so
has no effect on distortion.
A pair of back-to-back 18V zener
diodes between the gates and sources
of each Mosfet protect them from
damage in the case of a voltage spike
or static discharge. Due to the low
currents normally involved, the zeners
will conduct below 18V, clamping the
gate-source voltages below the 20V
maximum rating.
The 100pF capacitor between the
emitter and collector of Q12 helps
keep it on when power is first applied,
preventing start-up clicks or pops.
Q12 is then held on by the resistors
between its base and ground until the
DAC IC begins actively driving the
mute outputs.
Power supply
The ±15V supply for the amplifier
circuitry is provided by an external
power supply board (as used in the
original Stereo DAC), wired to CON2.
This powers the output stages directly,
while the rails feeding the input stages
are applied via RC filters. These filters
each comprise a 100Ω resistor in series
with each rail plus a 47µF capacitor
between the two rails.
This improves the channel separation by preventing supply voltage variations to the input stages due to current demands from the output stages.
Diodes D1 & D2 in the left channel and
D3 & D4 in the right channel prevent
the 47µF capacitors from pulling either
supply rail to the wrong side of ground
during power-up or power-down.
The +5V supply is derived from the
+15V rail using REG1. D5 prevents
REG1 from being damaged if the +15V
rail collapses faster than the +5V rail.
The associated input and output capacitors ensure regulator stability and
reduce output noise, while the 220Ω
resistor reduces dissipation in REG1
and helps filter any ripple from its
input supply.
Building it
All the parts are mounted on a
double-sided PCB coded 01102121
and measuring 94 x 110mm. Fig.6
shows the parts layout.
siliconchip.com.au
siliconchip.com.au
UNDERSIDE OF BOARD
R
OUT
IRF7905
L
IRF7905
01102121
2012
IC1
CS4398
LK1
LK2
Crystal DAC
LK3
LK4
The DAC IC (IC1) should be fitted
first. This device is in a 28-pin TSSOP
(thin shrink small outline package)
with a 0.65mm lead pitch and is installed on the underside of the PCB
– see Fig.7.
That’s done by first placing the PCB
copper-side up, with IC1’s pads to the
left and right (ie, with the board rotated
90°). That done, apply a very small
amount of solder to the upper-right
pad with a clean soldering iron (use a
medium to small conical tip).
Next, pick up the IC with tweezers
and position it near the pads with the
correct orientation (ie, with its pin 1
dot positioned as shown on Fig.7).
That done, heat the tinned pad, slide
the IC into place and remove the heat.
Now check its alignment carefully,
using a magnifying glass if necessary.
It should be straight, with all the pins
over their respective pads and an equal
amount of exposed pad on either side.
If not, reheat the solder joint and gently
nudge the chip in the right direction
until its position is perfect.
The diagonally opposite pin should
now be soldered, after which you
can solder the remaining leads. Don’t
worry about solder bridges; they are
virtually inevitable and can easily be
fixed. The most important job right
now is to ensure that the solder flows
onto all leads and pads.
Once the soldering is complete, apply a thin smear of no-clean flux paste
along the leads, then remove the excess
solder using solder wick. Once the flux
is heated to boiling point, this should
happen quickly. Be sure to trim the
end off the wick if it gets solder-logged.
You should now make a final inspection to ensure that there are no
remaining solder bridges and that the
solder has not “balled” onto a lead
without flowing onto its pad. If there
are still bridges, clean them up with
more flux and solder wick.
For further information on soldering SMD packages, refer to these two
articles: (1) “Soldering SMDs – It’s
Becoming Unavoidable”, December
2010; and (2) “How To Hand-Solder
Very Small SMD ICs”, October 2009.
Mosfets Q29 & Q30 go in next. These
are also SMDs but come in SOIC-8
(small outline integrated circuit)
packages with much wider leads and
greater pin spacing than the DAC chip.
The leads can be soldered individually
although it’s a good idea to add a small
amount of flux paste and use solder
–15V INPUT
DIGITAL I/O
Fig.7: this diagram shows how the SMD parts are installed on the bottom
of the PCB. Note that you also have to install solder bridges for links
LK1-LK4 but temporarily leave these out if you want to test the completed
board without reprogramming the microcontroller – see text & panel.
wick to remove excess solder when
you have finished. This also helps to
reflow the solder, ensuring good joints.
Again, be careful with the orientation. The Mosfets may not have a dot to
indicate pin 1. Instead, SOIC packages
normally have one bevelled edge and
pin 1 is located on that side.
Links LK1-LK4
The next step is to bridge the solder
pads for LK1-LK4 (see Fig.7). This
connects pins 9-12 of IC1 to CON1 and
it’s simply a matter of soldering across
the four pairs of closely spaced pads.
However, be careful not to bridge adjacent links or to bridge to the 0V and
5V pads on either side of the four links.
Note: if you want to test the board
without reprogramming the microcontroller, leave these links open and
connect pins 9-12 to either 0V or +5V,
as detailed in the accompanying panel.
Through-hole parts
The larger through-hole parts can
now be installed, starting with the
resistors, diodes D1-D5 and zener
diodes ZD1-ZD4. Table 1 shows the
resistor colour codes but you should
also check each resistor with a DMM
before installing it, as some colours
can be difficult to read. It’s also a bit
of a hassle to remove an incorrectlyplaced part from a PCB with plated
through-holes.
If you do need to remove a resistor
or diode, first cut the lead off one side,
near the body. That done, heat the pad
on the opposite side and gently pull
the body until it comes away. Finally,
grab the remaining lead with pliers,
heat its pad and again pull it out.
Once the part is out, you can then
clear the holes with a solder sucker.
Other parts can be removed in similar
fashion, ie, by cutting away the body
and then removing the leads one at
a time.
Check that each diode (and zener
diode) is orientated correctly before
soldering its leads. The 78L05 regulator (REG1) can then go in. Orientate
it as shown and bend its leads with
February 2012 33
Table 1: Resistor Colour Codes
o
o
o
o
o
o
o
o
o
o
o
o
o
o
No.
7
6
10
2
2
2
2
2
2
5
17
4
4
Value
100kΩ
10kΩ
2.2kΩ
1.6kΩ
1.3kΩ
680Ω
620Ω
510Ω
270Ω
220Ω
100Ω
68Ω
10Ω
pliers to match the holes on the PCB.
Now for the transistors. There are
two different types, BC549 (NPN) and
BC559 (PNP), so don’t get them mixed
up. Crank their leads so that they mate
with the pads, then push them down
onto the PCB as far they will comfortably go before soldering their leads.
Follow with the two horizontal
trimpots, then mount the ceramic and
MKT capacitors. That done, solder the
electrolytic capacitors in place. These
are all polarised so be sure to orientate
them correctly.
That just leaves the four connectors
(CON1-CON4). Make sure that the IDC
socket is installed with its notch towards the edge of the PCB and that it is
pushed down fully before soldering its
pins. It’s best to solder two diagonally
opposite pins first and check that it’s
sitting flat before soldering the rest.
Similarly, terminal block CON2 must
go with its wire entry holes towards
the edge of the PCB and must be flush
against the board.
Be sure also to push the RCA sockets down as far as they will go before
soldering their pins. The red socket
is mounted on the righthand side as
shown on Fig.6, while the white (or
black) socket goes to the left.
Chassis mounting
Once the assembly is complete, the
PCB can be mounted in the chassis.
Assuming you built you Stereo DAC
from an Altronics kit, it’s just a matter
of removing the old DAC board and
mounting the new board in its place
(the mounting holes are in the same
locations).
Note, however, that you may need to
34 Silicon Chip
4-Band Code (1%)
brown black yellow brown
brown black orange brown
red red red brown
brown blue red brown
brown orange red brown
blue grey brown brown
blue red brown brown
green brown brown brown
red violet brown brown
red red brown brown
brown black brown brown
blue grey black brown
brown black black brown
install some washers under the spacers
to get the RCA sockets at the correct
height. If so, install these between the
spacers and the bottom of the case. If
you put the washers under the PCB,
they could short some of the component leads to earth.
The connectors are also in essentially the same locations, so the new
PCB should slot straight in to any case
that’s already in use for the original
Stereo DAC.
Reprogramming the micro
You will now need to either reprogram the Atmel microcontroller
on the Input PCB or replace it with a
micro that has the new software. The
hex file (0110212A.hex) is available
for download form the SILICON CHIP
website. If you don’t have an Atmel
programmer, you can either purchase
a programmed micro from SILICON
CHIP or send yours in to have it reprogrammed for a fee (contact SILICON
CHIP for details).
Input board modifications
There are other changes we suggest
you make to the Input Board. First,
the original design had 33pF capacitors between each TOSLINK receiver’s
output and ground. These were recommended in the data sheet for the Jaycar
ZL3003 16Mbps TOSLINK receivers
we used originally. However, we subsequently found that these capacitors
caused some TOSLINK receivers to
oscillate under no-signal conditions
and published an errata in June 2010
which recommended increasing the
capacitor values to 100pF.
The problem with this is that with
5-Band Code (1%)
brown black black orange brown
brown black black red brown
red red black brown brown
brown blue black brown brown
brown orange black brown brown
blue grey black black brown
blue red black black brown
green brown black black brown
red violet black black brown
red red black black brown
brown black black black brown
blue grey black gold brown
brown black black gold brown
Table 2: Capacitor Codes
Value
100nF
18nF
10nF
6.8nF
4.7nF
1.8nF
1nF
100pF
µF Value
0.1µF
0.018µF
0.01µF
.0068µF
.0047µF
.0018µF
.001µF
NA
IEC Code EIA Code
100n
104
18n
183
10n
103
6n8
682
4n7
472
1n8
182
1n
102
100p
101
the 100pF capacitors, the TOSLINK
inputs can no longer reliably receive
data with a 96kHz sample rate. As a
result, we removed these capacitors
altogether from our unit (there were
no ill effects) and were then able to
test it at 96kHz.
So if you want to use the DAC
with 96kHz data, first check that you
have TOSLINK receivers capable of
16Mbps. The aforementioned Jaycar
ZL3003 are suitable and Altronics
now stock a similar part (Cat. Z1604).
If you do swap them over, be sure to
check that the link selecting 3.3V/5V
operation is in the correct location.
You must then remove the 33pF (or
100pF) capacitors at the outputs of the
TOSLINK receivers. While you are at
it, be sure to change the 300Ω resistor
across the S/PDIF input socket (CON1)
to 82Ω (see Notes & Errata, December
2011).
Setting up & testing
The new DAC Board can now be
tested but first a warning: never apply
power to the unit without both CON1
and CON2 (on the DAC board) wired
up. If you do, you could damage IC1.
siliconchip.com.au
The new DAC Board (top, right) is a drop-in replacement for the older board. Be sure to connect both the I/O cable and
the supply leads befor applying power, otherwise you could damage the DAC chip.
Check also that the power supply
polarity to CON2 is correct before applying power.
Before switching on, turn trimpots
VR1 and VR2 fully anti-clockwise,
then back clockwise about a quarter
of a turn. That done, apply power and
check the voltage between TP1 & TP2
using a DMM. You don’t need PC pins;
just push the probe tips into the test
point holes.
The reading should be below 10mV.
If it’s higher, switch off and check for
faults. Also, check the voltage between
TP3 & TP4; it should also be less than
10mV.
Assuming these readings are OK,
monitor the voltage between TP1 &
TP2 and slowly turn VR1 clockwise
until you get a reading of about 20mV.
That done, repeat this procedure by
monitoring TP3 & TP4 and adjusting
VR2.
This sets the quiescent current
through the output transistors in each
channel to around 2mA. That’s sufficient for them to operate in class A
mode for any load of 1.3kΩ or more.
For lower load impedances or highly
capacitive loads, the circuit will automatically switch into class B mode.
siliconchip.com.au
Testing The PCB Without Reprogramming
Communications between the DAC (IC1) and the microcontroller on the other
board (via CON1) go via LK1-LK4 which are closely spaced pairs of pads on the
underside of the PCB. These are normally shorted with solder.
We could have used permanent tracks instead but this way, it’s possible to test
the DAC board without having to reprogram the microcontroller. This is because
the CS4398 has multiple different configuration modes and the simplest involves
tying pins 9-12 either high to +5V (VLC) or tying them low. These are the same
pins used for serial communications and they are connected to LK1-LK4.
Most constructors should just short the four links as shown on the overlay
diagram, then reprogram the microcontroller. However, if you want to test the new
board out first, you can instead connect pins 9-11 of IC1 to the small, nearby 0V
pad and pin 12 to the adjacent 5V pad. In this mode, many DAC features do not
work properly (eg, the volume control, input scanning and muting) but you can
at least verify that the new board is functioning and use it in a limited manner.
If for some reason you want to drive a
600Ω load in class A mode, increase
the quiescent current to 6mA by adjusting VR1 & VR2 for 60mV between
the associated test points.
There’s no thermal feedback between the VBE multipliers and output
stages but at these current levels, transistor self-heating is low and thermal
runaway should not occur. Changes
in ambient temperature will be compensated for though, as it will affect
all transistors more or less equally.
Finally, connect a signal source and
check that the sound is undistorted. It’s
also a good idea to check that the volume control, scanning, muting and so
on are all working correctly. This will
confirm that the microcontroller can
communicate with the DAC IC (IC1).
Once it’s up and running, its operation is identical to the original Stereo
DAC – see the November & December
SC
2009 issues for further details.
February 2012 35
|