CD
Converter.
Objective:
After the end of the box speaker, after the adjustment of the
amplifiers, it
remains a last objective to be realized: improvement of the CD output,
i.e. to reduce the imperfections due to a moderate
purchase price.
Then my project consists to use the DF1706 and PCM1704 of Burr-Brown.
After going on the site of Burr-Brown, which beame a part of Texas
Instrument now, I discovered that there were demonstration
board
which integrated the main part of the components to use to
make my idea.
I hesitate a few times, then I decide to buy
this evaluation module.
Initially it is very well to realize, ground plan for the digital,
a second ground plan for DACs and the output amplifiers, 3 different
supplies, a good diagram!
It was difficult for me to do better, and considering that there is a
digital circuit, it was perhaps better to directly use
this evaluation module.
DEM_DAI1706:
Document of Burr-Brown is here: dem_dai1706.pdf.
This module of evaluation perhaps bought at the cost of 399$ directly
near TI at the following address: http://focus.ti.com/docs/toolsw/folders/print/dem-dai1706.html
And here the photograph of this module of evaluation such as I received
it:
Photograph
It contains all the integrated necessary circuits,
the amplifier are assembled on 8 pins support in order to be able
to change them, the condensator are ELNA Cerafine, this is a
sμFficient quality
I think for the application.
So I envisages a few updates:
- Change of the OPA2134 counters modules with double AD825.
- Addition of an external quartz in the immediate future.
- Maybe replacement of this quartz by very high stability
clocks such as
for example: Superclock
III in a few times when all goes well.
This circuit has another advantage: all is configurable easily with
wires or switches. So an update in future, due for example
to a purchase of a new apparatus like SACD, will be
very easy to do!
Integrated circuits
Here is the DF1706, stereo 24bits, 192kHz, digital filter interpolation
with 8 times oversampling:
Photograph
The PCM1704, digital to analog converter, BiCmos
with signed amplitude:
Photograph
And then another circuit placed discreetly by Burr-Brown (because it is
one of
these competitors), the CS8414 of Hook Corporation, an audio digital
receiver, 32... 96kHz - 16,18,20 or
24 bits:
Photograph
Supply Choice.
Of course, I have to add the supply.
It is necessary to realize:
- 5V supply for the digital. A integrated regulator
on this circuit will decrease this voltage at 3.3V for the
DF1706.
- ±5V
symmetrical supply for both DACs. These DACs have a good supply
rejection, like -120dB at 10Hz, therefore it is useless to envisage
complex references voltage. It is enough if is
cleanest as possible. For that I would use 4 condensators of 10000μF,
just to clean completely the supply.
- ±15V symmetrical supply for operational amplifiers: even
notices
that the preceding one, 4 condensators of 4700μF will fully filter the
supply after the
regulator.
- For regulator, I will use standard 7805, 7815, 7905, 7915. It should
be enough, considering the cargo of
wrapped 4700μF which will follow after.
- On transformers side, I will use three
R-core transformer, considering which they are not too
expensive, even a little less expensive than the toric in around 30VA
power.
Schematics of the Supply.
There is 5 independents supplies. 4 of the 5 supplies
have a PI-filter after the diodes rectifier, based on large values of
10000μF. I use that to strongly value to remove
all possible defects of the sectors 220V. The PI-filter uses
also
inductances of 180μH. I could have taken resistances. The condensators
before the regulators are standard models of Philips,
so traditional models. But the condensators after the
regulator come from the japanese Rubycon, these are blackgate
condensator (standard series) of 4700μF,
that are famous for their high
quality in audio. Their value is enormous in order to remove all
the transitory defects of the regulators.
A ceramic condensator of 100nF is also added in
parallel on each
chemical condensator.
All the ground are separate in the supply. They will be connected between
them only on the demo-board circuit.
Supply of DACs:
A transformer 2*9V is connected on the diodes rectifier. Once rectified, the continuous voltage is strongly filtered
thanks to C1, L1, C2 for the positive voltage, and C3, L2,
C4 for the negative voltage. At this point I should obtain close
to ±15V. Then I use a traditional regulator 7805
and 7905 to decrease them at ±5V.
Photograph
Power supplies of the operational amplifiers:
A second transformer 2*18V is connected on the diodes rectifier. Once rectified, the continuous voltage is strongly
filtered thanks to C1, L1, C2 for the positive one, and
C3, L2, C4 for the negative one. at this point I should obtain close to ±-30V. Then I use a traditional
regulator 7815 and 7915 to decrease it at ±15V.
Photograph
Supply of the digital circuits:
A third transformer of 2*9V is
connected on the rectifier diodes, the two secondaries in parallel.
Here, I filter more simply the rectified voltage thanks to C15. At this
point I should obtain close to +15V. Then I use a traditional regulator
7805 to decrease is at +5V. But I add 2200μF on regulator output. I could have put
less capacity here, but I had these condensators in stoc!
Photograph
Final Results:
Then it is necessary to put all that in a metalic box, while being arrange
so that the catches of optics and digital entries of the evaluation module are directly accessible since the exterior.
It gives this:
Photograph
The 3 transformers are on the left, followed immediately on the right
by the
5 diodes bridges screwed on the frame of the box. Then
the rectified voltage are sent on a test board of 10*20cm
comprising all the
supply circuits. Then I just have to connect the régulated
voltage on the evaluation module of
Burr-Brown. The supply circuit is screwed on the bottom of the box
with screws and traditional braces. Same fixing system for
the evaluation module on the right.
To fix this one, it is necessary to pay attention to bore holes
for digital entries, because they are placed against
the backside of the
box. It allow to have a easily access. Remain the audios
output. I use copper wire covered with PTFE, to
connect the CINCH adittionnally screwed on the back side of
box. CINCH are insulated versions, which
makes it possible to limit possibility of ground loops.
The first listening...
The first evening, the first 10 minutes of listenings give a
surprising impression. It is the first time that I listen electronic source equal or lower than my NAD512.
Initially I am surprised by the acute ones, they
seems more present, and the sound is very
clear. But it is midnight, and it is necessary to go to bed.
The following day I listen to the 9th symphony of
Dvorak (CD collection Deutsch Grammophon, Gold series). The impressions
is better. The acute are clear and forceful. It is even surprised,
it is the first
time that I hear that. Coppers are impressive! At such degree that
I
have the impression that my NAD (and not that him!) attenuate the
acute. The meiums are clear also, without
burs, and the low ones are there too and strike well when it is needed.
CD starts with a soft melody. I thus waits a moment where it would
have a strong orchestral push, because it is where it become bad
in general.
And when it occur, all the instruments stay at
their place, they remain all light, and to tell the truth the
distortion which I had 1 year ago with my preceding installation
becomes very frankly difficult to detect. Finally, I forget everything
and I
listen the music...
In short, I have took a risk, and I do not regret it. Ah, that cleans
the ears these Burr-Brown circuits (DF1706, PCM1704)!
Nevertheless, they are not badly...
Measurements of the digital signals...
Using an 300MHz oscilloscope, and a 500MHz probe,
I can show you the most useful signals
Usual SPDIF digital entry:
We sees the data bits intermingled with the clock.
Photograph
Here a batch of 4 images showing the edge of this
input signal.
The TTL gate of the NAD512 is capable to generate a decreasing edge of 5ns
and an increasging edge of 10ns.
Photograph
Photograph
But the NAD512 uses an output transformer on the digital output.
The edge are spread out over 30ns... and they arrive like that on
the entry digital of our convertor.
Photograph
Photograph
Input signal of the DF1704: BCK.
It is the clock signal which goes with DATA. The frequency is 64 times
huigher than the input signal. It is made by the CS8414.
Photograph
Input signal of the DF1704: CLKO.
It is the frequency of operation of the DF1706. This signal results
from the data flow, and it is unfortunately impossible to make
differently with this board.
Photograph
Input signal of the DF1704: DATA.
It is the data bits converted into voltage usable by the CS8414.
The printed frequency does not correspond to anything.
Photograph
Input signal of the DF1704: LRCK
It indicates in alternation the data which correspond to the left ways and the right ways. I check also that my NAD512
work well at 44.1kHz.
Photograph
Input signal of the DF1704: SCLK.
It is the output clock of the DF1706 and is the entry of PCM1704.
It is at 11.2896MHz (= 256*44.1) and is identical to CLKO. It makes
it possible to drive the digital bits DOL and DOR.
Photograph
Output signal of the DF1706: DOL.
Left data left output of DF1706 and entry of PCM1704. They work
with WCK and BCK.
Photograph
Output signal of the DF1706: DOR.
Right data output of DF1706 and entry of PCM1704.
They work with WCK and BCK.
Photograph
Output signal of the DF1706: WCK.
This is data indicating the beginning and the end of each package of
data. It is has 8 times 44.1kHz, because the DF1706
oversample 8 times.
Photograph
Output signal of the DF1706: BCK.
It is the clock drive DOL and DOR for the DAC.
Photograph
Measurements of the analogical signals...
Output signal of the current voltage converter.
The action of the DF1706 combined with the DAC appears on this 20kHz signal.
The 2 following photographs show a 20kHz signal then a zoom...
Photograph
Photograph
Audio output with a musical signal:
Photograph
Audio output with a sinusoide of 20kHz:
Photograph
Measurements of the Supply...
Here are what gives all the Supply.
Here are two photographs, and I have always this type of signal.
No low frequency signals, and the high frequency signals can
partly be due to the setup of measurement (HF
ground mass loop with the scope).
Photograph
Photograph
realization of a CD converter - assembly of a CD converter -
use of DEM_DAI1706 - diagram of a CD converter - measurement with the
oscilloscope of a CD converter - oscilloscope converter CD
Create : 01 june 2004.
Last modification it : 11 november 2007.