The evolution of a non oversampling DAC based on the TDA1543.

The evolution of a non oversampling DAC based on the TDA1543.


This boards were made for comparing the following issues:

5V vs. 8V supply voltage for TDA1543.

0ne vs. several chips in parallel.

Different output stages:first board from the left:

Active I/V according to Philips datasheet (with opa627).

second board from the left: Passive I/V with active buffer (AD847) according to this webpage:

third board from the left: Passive I/V just with resistors.

alt As I use a passive preamp, I preferred the 8V supply because of the higher signal level. More than one TDA1543 in parallel offered more micro dynamics, authority and a bigger soundstage. The two active OP output stages couldn't compete with the completely passive version. The ambience was gone, treble wasn't that airy anymore. After removing the filter caps on the OP circuit, the sound improved a bit, but still I decided to go passive for further development.

This is about testing the benefits of a clean power supply (comparable to the supply that AudioNote uses in their DACs - scheme posted by Elso and described here: against a standard voltage regulator. Well, the benefits are obvious: less nervous, better bass, better focus. It's really worth it. And again I compared different numbers of  parallel TDA1543. The more the better.... (i know, some people disagree)


This is a version with 4 TDA1543 in parallel built into a Philips DVD 963SA. The converters are fed with a I2S signal. The good news is that it works, the bad news is that the I2S is an oversampled signal. The sound is less bright than the original output, so it might be a good alternative.



Philips 963 CN1400 TDA1543
pin 14 DATA pin 3
pin 15 Digital Ground pin 4
pin 16 Word Clock pin 2
pin 18 Bit Clock pin 1

DCF 1.0

A complete DAC with optical and coaxial input. The DAC chips run at 8.5V. They get really hot. I measured 68°C between the fourth and the fifth chip when placed inside the housing. The sound is powerful, fast, warm, non fatiguing, sweet, soft, and completely non-digital.

DCF 1.0

Detail view: Mounting the chip tower horizontally allows some airflow for cooling. The supply caps aren't any special audio types. A big 2200uF extra low ESR from Rubycon, five Roederstein 47uF, and two polypropylene film capacitors (Epcos and Wima FKP2).


The big brother, now with Elso's asynchronous re-clocker. In this setup it is possible to switch  between normal and re-clocking mode within two minutes. But that's not necessary to hear the difference. Both versions sound basically similar. They have the same tonal balance, the same unstressed and involving sound without irritation. With the re-clocker, everything is more exact like looking at a sharp picture. Sharper means more detail, more over all and micro dynamics, better separation between instruments, more speed and rhythm. It's a must.

DCF 1.0

Same as above. But with 16 Chips in parallel. It definitely improves again. Without the fan for cooling, the chips would get too hot. I took a brushless extra silent fan, but unfortunately you can still hear it. Probably the tower wouldn't overheat, if the space between the chips is increased by one or two millimetre. I'll go for a better cooling solution.

I also listened to different coupling caps. After the DAC there's a 10klog D.A.C.T. potentiometer and a 1uF coupling cap (Mundorf MCap Zn) at the amps input in the signal path. So I always had two caps in line. Nevertheless it was possible to clearly hear that each capacitor has a certain sound.

The Elkos (100uF Nichicon Fine Gold / 47uF Elna Cerafine) sounded somehow grainy, harsh and less dynamic. I've not yet tried Black Gate's.

I compared MKT's from Wima and MKP's Mundorf and SCR. They all had different values from 1.5uF to 15uF. It's hard to say what's best. The smaller values tend to sound thinner and maybe more precise. The larger ones seem to have more bottom,  maybe at the expense of precision. It's probably trade off between low end extension and precision. Unfortunately I don't have identical values from SCR, WIMA and Mundorf caps, so it's impossible to say whether the sound depends more on size or brand. Anyway, the caps are a fine tuning instrument.

Right now I'm happy with a 4.7uF polypropylene from SCR.

I haven't yet A/B compared different sizes and brands of supply caps. That's one thing I'll do in near future.


Looks better from behind...

DCF 1.0

Testing the DIY DAC's with the following equipment:

Teac VRDS 25X
Wadia Digimaster X32
Audio Research Preamp
Pass Aleph 0
Fischer "Netzleiste"
Infinity Kappa 9a

Home Equipment

DCF 1.0

Everything is DIY except for the CD transport.
Speakers: 2-Way D'Appolito. 12 dB filter, crossover frequency at 2.6kHz. Tweeter: Eton ER4, MT: Audax HM 170 Z18 (selected by German distributor)
Amp: Thel IQ-Amplifier with integrated passive D.A.C.T volume attenuator.

Listening Test:

The Wadia converters are well known for their musicality. The only one i know is my friend's Digimaster X32 which is my personal reverence. It never looses control, is very colourful, and has impressive bass slam.

My first built DAC (CS8420/DF1704/PCM1704/OPA627) couldn't reach the natural performance of the Wadia. I expected that a design with today's IC's would easily beat the old Digimaster. Well, I was wrong and so I tried this widely discussed TDA1543 converter. It's clearly a winner. I'm very happy with it.

Like the Wadia, the TDA1543 based DAC plays music different from most other digital gear. It remains solid as a rock, even in very complex music. Everything else is outstanding. The Wadia still offers more dynamics and a more solid soundstage, but the higher frequencies are not as natural as with the TDA1543. My friend - the owner of the test equipment - calls this DAC the "wooden Wadia."  ;-)

I call it      TDAC1543

I'm looking forward to compare the DACs on my system. My friends room acoustics are suboptimal and his speakers may not show all the positive virtues of a non oversampling DAC.