Product: TNT 1541
Company: not for sale, TNT-Audio free DIY design
Approx. cost: 200$/Euro (just components)
Author: Giorgio Pozzoli - TNT Italy
Published: January, 2003
The TNT Convertus has been one of my favorite designs. It was probably one of the first DAC DIY projects published in Italy, and attracted a lot of attention (well, given the size of the market, obviously...).
A few changes have been made to the original Convertus design, but I never had the time to test another DAC: it would have required a complete overhaul of the project, including a new PSU.
Anyway, rumours about the outstanding quality of another DAC chip have loudly continued - the DAC in question, the TDA1541A, has a slight problem: it is no longer in production.
As I wanted to present an easy to build DAC, this was a condition I was not willing to accept, but after building it, I realized that the design includes a few interesting elements that can be applied to any other dac type. Even though the DAC is now rather difficult to find, especially in the selected versions, it is still available, and in the end the situation is no worse than that of a few well known NOS tubes....
So here is the DAC.
One more note. I will not go back in detail over all the circuitry I have already described in the Convertus, Convertus decima and Convertus Decima Digital articles. Please read them through before reading the present one.
For the receiver I used a CS8412. At the time I’m writing this article it is still present in the RS catalogue in a DIL package. The CS8414, the current generation, seems to work a little better, but is not available in DIL format: if you dare working with a surface mount chip, it can be a valid pin to pin substitute.
In order to be able to select any output configuration, the mode pins are controlled by a quadruple microswitch... this lets me select any possible output configuration with ease. Mode 3 is to be used (M0=1, M1=1, M2=0, M3=0), unless you do not wish to implement the digital decimation circuit. In that case, you must select mode 1 (I2S compatible, all signals output).
More interesting are the digital inputs. In fact, two inputs are shown on the schematic, but you can add as many as you want, provided that you check the power supply regulation is up to the job.
Each input connection is completely de-coupled using a high frequency transformer. I used a ferrite toroid from RS Components as a core, part number 212-0831, and just wound up both primary and secondary coils with five turns of insulated copper solid core wire (the one used for normal power transformers secondary coils). Unfortunately the original ferrite toroid is no longer available; try finding something similar of small dimensions.
The digital input socket can be either RCA or BNC. I always use BNC, which can specified for a precise 75ohm impedance, but this sometimes creates problems (or requires an adapter to be used...). In either case they must be female sockets, insulated from the panel.
The digital input signal comes out of the secondary of the transformer and is taken to a digital receiver, a DS34C86. This chip contains two differential line receivers and two drivers. The drivers have no use for us, at least for the moment. The receivers work according to the RS-422 electrical standard, which is very similar to S/PDIF electrical standard.
A very interesting aspect of the chip is that the receiver has a three-state output: this means that its output can be disabled, and in this situation presents a very high impedance. The technique is obviously used in buses, where more than one device can get control of the bus depending on the situation.
In our case it allows us to have several different receivers with output in parallel on the same digital line, provided that only one receiver output is enabled at any time. This is achieved by using a simple switch (SW101). If you plan to add more inputs, a rotary switch is recommended.
It is now clear why it is so easy to add other digital inputs: it is just necessary to add one more receiver chip and connect its output to the common output line and its selection input to the input selector.
Note how the receiver outputs are coupled to the 8412 input via a capacitor.
The output signal EN_SPDIF_IN is for future development (guess what for?).
I am not going to describe this module in detail, as it has been already described in the article about the Convertus Decima Digital.
If you want to avoid implementing it, you must simply select CS8412 mode 1 through the multiple microwsitch SW101/A-D and just connect FSYNC to LRCK, SCLK to BCK, SDATA to DATA.
As there is switch SW201 that allows to enable or disable decimation, I strongly suggest to test it. If you don't like the results, you can simply bypass the decimation module.
I must admit that while decimation improves the sound of the Convertus in a significant way, the TDA1541A treats high frequencies with such care and detail that decimation can be considered optional.
C204, C205, C206, C207 are bypass capacitors placed as near as possible to U201, U202, U203, U204 respectively, and connected between the power supply and ground pins of each IC.
Even the DAC circuit is not anything special. The DAC, configured as in the schematic, directly supports the I2S interface, so connection to the receiver is really straightforward, apart from the decimation circuit.
The DAC chip requires three voltages: +5V, -5V, -15V. As you see there are three separate voltage regulators on board. They should be placed as near as possible to the corresponding DAC chip pin.
In this area all components should be of the highest possible quality: polypropylene capacitors must be used, especially for the 14 capacitors to be connected to the DAC chip and the I/V conversion resistors (R310, R311).
Some explanation is perhaps required for DAC chip quality levels. From old Philips documentation ("Semiconductors for Digital Audio", Designer's Guide, August 1997) it appears that there are four versions of TDA1541A; from the datasheet of February 1991, it appears that the selection was based on the highest linearity attainable, but only 3 selections are defined. I have also included TDA1543 info in the table just as a reference (in referring to Convertus take into account that a quad DAC increases SNR by 6dB).
Type | Description | Data Format | Typ. THD+N at 0dB * | Typ. THD+N at -60dB * |
Typ. SNR | Typ. Output Current | Channel Separation | Max diff. linearity error |
dB(%) | dB(%) | dB | mA | dB | ||||
TDA1541A | high-performance 16- bit DAC | I 2 S, up to 8f s | -95( 0.0018) | -42( 0.79) | 112 | 4.0 | 98 | bit 1-16 EdL < 1 LSB |
TDA1541A / R1 | high-performance 16- bit DAC | I 2 S, up to 8f s | -95( 0.0018) | -43( 0.7) | 112 | 4.0 | 98 | bit 1-16 EdL < 2 LSB |
TDA1541A / S1 | single crown 16- bit DAC | I 2 S, up to 8f s | -95( 0.001) | -47( 0.4) | 112 | 4.0 | 98 | bit 1-7 EdL < 0.5 LSB bit 8-15 EdL < 1 LSB bit 16 EdL < 0.75 LSB |
TDA1541A / S2 | double crown 16- bit DAC | I 2 S, up to 8f s | -97( 0.002) | -47( 0.4) | 112 | 4.0 | 98 | N/A |
TDA1543(T) | economy 16- bit DAC | I 2 S, up to 4f | -75( 0.018) | -33( 2.2) | 96 | 2.3 | 90 | N/A |
* A-weighting
The best version of TDA1541A is obviously the double crown, which was used only on really high-end converters. And this happened not centuries ago: in 1998 it was still the best Philips converter available and in fact when the Marantz CD-7 cd-player was produced, it was chosen for that specific reason, according to Marantz documentation.
I have not been able to find any double crown, and I have used a single crown instead. I also used during first tests a normal version (no crown) just to spare the better ones. There is more refinement and detail in the single crown, but the difference is not so huge as when comparing the quad TDA1543 of Convertus with a single TDA1541A.
Given the huge costs the single crown has reached these days, I do not dare imagine the cost of a double crown, assuming it is possible to find one. Anyway, should you by chance have any spare double crown, please e-mail me...
The C310-323 capacitors must be very high quality polypropylene capacitors (in choosing them, take into account that normally film and metal sheet capacitors are better than metallized film ones) and should be placed as near to the corresponding pin as possible. A package as small (thin) as possible is therefore required.
Also C304,C306,C308 are very high quality polypropylene capacitors, and must be placed as near as possible to the corresponding power supply pin. C301, C302, C303 are OS-CON capacitors placed as near as possible to the previous ones.
For R310-R311, the I/V conversion resistors, I used Holcos. Given the fact there are only two, and their importance in defining the resulting sound is expectedly high, you could experiment with more expensive varieties.
TDA1543, according to Kusunoki, is the only DAC able to drive directly a load up to a normal line output level, and hence the only one not requiring any output stage with gain.
In the case of TDA1541A such a stage is mandatory. I have used a very simple solid state differential stage followed by an emitter follower, in order to achieve the required gain (around 22dB) with a very low output impedance.
There is however a peculiarity: note the position of R421/423. These introduce a small of amount of feedback. In general I agree that the less input/output feedback the better, but here the case is different: this is positive feedback, reducing the stability margin of the stage but increasing its speed. The difference in sound with these resistors in place is really notable: far higher push, impact and apparent speed - a really lively sound.
A few other designs use twin TDA1541As to increase dynamics: I do not really think it is necessary, unless you are a real lover of high dynamics. That said, in our solution, it requires only four more resistors...
The circuit is perfectly stable and has a good deal of local feedback, which helps in giving rather low distortion in spite of the simplicity of the circuit (0.05% at 20kHz at full output level from simulations: direct measurements are not so easy due to characteristics of the zero-oversampled signal...).
Here the highest quality components are again mandatory. In the prototype all resistors are Holcos, all capacitors are polypropylene or OS-CON electrolytic ones. Matching transistors for the same gain and possibly lowest noise cannot do any harm, but I didn't bother and no trouble has arisen.
There is also a low pass filter, composed simply by C407/408 and the collector resistors of the differential couple. This causes only a tiny drop at 20kHz, but also a reduced drop at higher frequencies: in one word, the output is (very) dirty. In compensation, the phase shift due to the output filter is very, very low.
The output is AC coupled via a very high quality polypropylene capacitor. Use the best film capacitor you can think of, and afford...
I definitely suggest avoiding paper in oil capacitors here, as the sound (for my taste) already has enough passion, sparkle and glitter without paper in oil caps!
The circuit has been built on the same board as the DAC, as near as possible to the I/V conversion resistors to reduce the risk of interference.
Given the high number of local regulators, I decided to limit the number of power supplies to two - one for digital components (receiver and decimation circuits), and one for the analogue side. If you wish, you can mount a larger number, dividing their usage between the different modules.
All power supplies are regulated using a simple capacitance multiplier circuit, using either an N- or a P-channel MOSFET depending on the polarity.
Do not underestimate the importance of the pre-regulator: without it, significant mains ripple filtered through to the output, so it is definitely mandatory.
I used a large number of power supply capacitors in parallel in both PSUs. To an extent, the more the better, but don't be excessive.
The diodes are ultrafast rectifiers, with a 25ns recovery time; this reduces spikes and switching disturbances. You could even use the SBYV27 series, which has a recovery time of 15nsec.
Given the distance from the audio circuits, I do not think it is the case of using special capacitors here: standard industrial quality should be enough.
Important note: all the MOSFETs must be mounted on small heat sinks.
All the circuits are mounted on single sided printed circuit boards. As usual, the copper side is used as the ground plane and all the components are mounted on this side. Have a look at the assembly instructions for more information.
The internal connections, between the boards and from these to the RCA connectors) are in rather thin, solid-core, silver plated copper wire. When the connections are longer, I used insulated solid-core wire from some UTP Category 5 network cable.
Follow the the order of the schematics exactly in wiring the input pins. As already said the input jacks are isolated from the back panel, and also the output sockets are insulated. All ground leads meet only in one point in each board. Follow the instructions and schematics carefully.
I am not going to review the unit at all. I am not so arrogant as to pretend to be able to be objective (yes I know, I have some limitations...).
Even so, I understand it is important to give the reader an idea of the sound it produces. Anyway I warn you in advance to consider whatever follows a very biased and personal view...
The sound is lucid and passionate, with highs going up to heaven and further without any hint of hardness. The bass is deep, solid when necessary, controlled and at the same time warm and rounded - you are sometimes surprised to hear a bass line where normally one would only know a bass was playing.
From the imaging point of view, it is precise; the soundstage is naturally deep.
A comparison with Convertus? I used as a reference the latest Convertus version with decimation, and source follower output stage. Tests were made with decimation both active and inactive. All that I can say is the 1541A is in another league. The sound is far, far more lucid, smooth, to such an extent that this seems to be a tonal abberation compared to Convertus, with the highs taking priority. But then the bass comes in - neat, deep, and you feel it is correct this way.
When you come back to Convertus you really miss the smoothness and detail, the softness of 1541A. There is only one area in which Convertus seems to be slightly better, and it is punch: but this is probably down to the laid back and rough high frequencies in Convertus, which as said seems to highlight the bass.
This is not a simple project to build. As said, there are also problems in finding the TDA1541A. You can find the normal version for around 25 EUR, but the cost for the selected version is an issue.
Anyway, the result is so good that I really think you should at least give it a try. If you have a Convertus, you can just mount a board with the DAC and output stage, and give it a test. I am pretty sure that if you don't like the result, you'll be able to re-sell the chip for at least at the same price you bought it...
Kind of investing in chips? Well, let's say... cannot be much worse than stock, these days, no? And remember, guys: stock does not sound at all...
Receiver
Code
Value
Rating
Notes
C101
220uF
25V
High Quality Electrolytic
C102
0.1uF
35V
Polypropylene
C104
220uF
25V
High Quality Electrolytic
C105
0.1uF
35V
Polypropylene
C107
0.01uF
35V
Polypropylene
C108
0.01uF
35V
Polypropylene
C111
0.1uF
35V
Polypropylene
C112
220uF
25V
High Quality Electrolytic
C115
0.047uF
35V
Polypropylene
L101
1000uF
RF Choke
L102
1000uF
RF Choke
L103
1000uF
RF Choke
R101
75
1/4W 1%
Holco
R102
75
1/4W 1%
Holco
R103
100
3.5W
Resista
R104
100
3.5W
Resista
R105
1k
1/2W 1%
Holco
R106
1k
1/2W 1%
Holco
R107
1k
1/2W 1%
Holco
R108
1k
1/2W 1%
Holco
R109
10k
1/4W 1%
Holco
R110
10k
1/4W 1%
Holco
R111
680
2W
Resista
R112
10k
1/4W 1%
Holco
R113
10k
1/4W 1%
Holco
R114
10k
1/4W 1%
Holco
R115
10k
1/4W 1%
Holco
R116
1k
1/2W 1%
Holco
R117
1k
1/2W 1%
Holco
R118
1k
1/2W 1%
Holco
SW101
1 way 3 positions
U101
DS34C86C
RS422 Line Driver/Receiver
U102
CS8412
Digital Receiver
U103
TL431
Precision Voltage Reference
U104
TL431
Precision Voltage Reference
U105
TL431
Precision Voltage Reference
BNC female
75ohm
for panel, insulated, qty:2
Decimator
Code
Value
Rating
Notes
C201
220uF
25V
High Quality Electrolytic
C202
0.1uF
35V
Polypropylene
C204
0.1uF
35V
Polypropylene
C205
0.1uF
35V
Polypropylene
C206
0.1uF
35V
Polypropylene
C207
0.1uF
35V
Polypropylene
L201
1000uF
RF Choke
R201
330
1W
Resista
R202
1k
1/2W 1%
Holco
R203
1k
1/2W 1%
Holco
R204
1k
1/2W 1%
Holco
R205
1k
1/2W 1%
Holco
SW201
1 way 2 positions
U201
74HC161
Binary Counter
U202
74HC161
Binary Counter
U203
74HC74
Dual D-type Flip Flop
U204
74HC00
Quad 2-input NAND gates
U205
TL431
Precision Voltage Reference
DAC
Code
Value
Rating
Notes
C301
220uF
25V
OS-CON Electrolytic
C302
220uF
25V
OS-CON Electrolytic
C303
220uF
25V
OS-CON Electrolytic
C304
0.1uF
35V
Polypropylene
C305
0.1uF
35V
Polypropylene
C306
0.1uF
35V
Polypropylene
C310
0.1uF
35V
Polypropylene
C311
0.1uF
35V
Polypropylene
C312
0.1uF
35V
Polypropylene
C313
0.1uF
35V
Polypropylene
C314
0.1uF
35V
Polypropylene
C315
0.1uF
35V
Polypropylene
C316
0.1uF
35V
Polypropylene
C317
0.1uF
35V
Polypropylene
C318
0.1uF
35V
Polypropylene
C319
0.1uF
35V
Polypropylene
C320
0.1uF
35V
Polypropylene
C321
0.1uF
35V
Polypropylene
C322
0.1uF
35V
Polypropylene
C323
0.1uF
35V
Polypropylene
C324
470pF
35V
Polypropylene
L301
1000uF
100mA
RF Choke
L302
1000uF
100mA
RF Choke
L303
1000uF
100mA
RF Choke
R301
100
3.5W
Resista
R302
180
3.5W
Resista
R303
180
3.5W
Resista
R304
1k
1/2W 1%
Holco
R305
5k
1/2W 1%
Holco
R306
1k
1/2W 1%
Holco
R307
1k
1/2W 1%
Holco
R308
1k
1/2W 1%
Holco
R309
1k
1/2W 1%
Holco
R310
33
1/2W 1%
Holco
R311
33
1/2W 1%
Holco
U301
TL431
Precision Voltage Reference
U302
TL431
Precision Voltage Reference
U303
TL431
Precision Voltage Reference
U304
TDA1541A
High Performance DAC
Gain Stage
Code
Value
Rating
Notes
C401
220uF
25V
Electrolytic OS-CON
C402
0.1uF
35V
Polypropylene
C403
220uF
25V
Electrolytic OS-CON
C404
0.1uF
35V
Polypropylene
C405
10uF
100V
Polypropylene very high quality
C406
10uF
100V
Polypropylene very high quality
C407
100pF
35V
Polypropylene very high quality
C406
100pF
35V
Polypropylene very high quality
L401
1000uF
RF Choke
Q401
BC109B
NPN transistor
Q402
BC109B
NPN transistor
Q403
BC109B
NPN transistor
Q404
BC109B
NPN transistor
Q405
BC109B
NPN transistor
Q406
BC109B
NPN transistor
R401
330
2W
Resista
R402
330
2W
Resista
R403
5k
1/2W 1%
Holco
R404
1k
1/2W 1%
Holco
R405
5k
1/2W 1%
Holco
R406
1k
1/2W 1%
Holco
R407
6.8k
1/2W 1%
Holco
R408
6.8k
1/2W 1%
Holco
R409
270
1/2W 1%
Holco
R410
270
1/2W 1%
Holco
R411
4.3k
1/2W 1%
Holco
R412
6.8k
1/2W 1%
Holco
R413
6.8k
1/2W 1%
Holco
R414
270
1/2W 1%
Holco
R415
270
1/2W 1%
Holco
R416
4.3k
1/2W 1%
Holco
R417
10k
1/2W 1%
Holco
R418
10k
1/2W 1%
Holco
R419
220k
1/2W 1%
Holco
R420
220k
1/2W 1%
Holco
R421
10k
1/2W 1%
Holco
R422
150
1/2W 1%
Holco
R423
10k
1/2W 1%
Holco
R424
150
1/2W 1%
Holco
U401
TL431
Precision Voltage Reference
U402
TL431
Precision Voltage Reference
RCA Pins
gold plated
for panel, insulated, qty:2
PSU A
Code
Value
Rating
Notes
C501
1nF
35V
Ceramic
C502
0.47uF
35V
Polypropylene
C503
4700uF
35V
Electrolytic
C504
4700uF
35V
Electrolytic
C505
4700uF
35V
Electrolytic
C506
4700uF
35V
Electrolytic
C507
4700uF
25V
Electrolytic
C508
4700uF
25V
Electrolytic
C509
4700uF
35V
Electrolytic
C510
0.47uF
35V
Polypropylene
C511
0.47uF
35V
Polypropylene
C512
1nF
35V
Ceramic
C521
1nF
35V
Ceramic
C522
0.47uF
35V
Polypropylene
C523
4700uF
35V
Electrolytic
C524
4700uF
35V
Electrolytic
C525
4700uF
35V
Electrolytic
C526
4700uF
35V
Electrolytic
C527
4700uF
25V
Electrolytic
C528
4700uF
25V
Electrolytic
C529
4700uF
35V
Electrolytic
C530
0.47uF
35V
Polypropylene
C531
0.47uF
35V
Polypropylene
C532
1nF
35V
Ceramic
D501
BYV27-100
Superfast Diode
D502
BYV27-100
Superfast Diode
D503
BYV27-100
Superfast Diode
D504
BYV27-100
Superfast Diode
M501
IRF630
N-channel MOSFET
M521
IRF9630
P-channel MOSFET
R501
33k
1/2W 1%
Holco
R502
4.7k
1/2W 1%
Holco
R521
33k
1/2W 1%
Holco
R522
4.7k
1/2W 1%
Holco
TR501
PSU Transformer
30VA
Prim:220V, Sec:25+25V
PSU D
Code
Value
Rating
Notes
C601
1nF
35V
Ceramic
C602
0.47uF
35V
Polypropylene
C603
4700uF
25V
Electrolytic
C604
4700uF
25V
Electrolytic
C605
4700uF
25V
Electrolytic
C606
4700uF
25V
Electrolytic
C607
4700uF
25V
Electrolytic
C608
4700uF
25V
Electrolytic
C609
4700uF
25V
Electrolytic
C610
0.47uF
35V
Polypropylene
C611
0.47uF
35V
Polypropylene
C612
1nF
35V
Ceramic
D601
BYV27-100
Superfast Diode
D602
BYV27-100
Superfast Diode
D603
BYV27-100
Superfast Diode
D604
BYV27-100
Superfast Diode
M601
IRF630
N-channel MOSFET
R601
33k
1/2W 1%
Holco
R602
1k
1/2W 1%
Holco
R603
10k
1/4W 1%
Holco
VR601
2.2k
1/2W
Trimmer potentiometer
TR601
PSU Transformer
30VA
Prim:220V, Sec:15V
Mains Filter
with IEC Socket
with power switch and fuse
© Copyright 2003 Giorgio Pozzoli - http://www.tnt-audio.com
Supervisor: Tom Browne