- Paralleled TDA1541 NOS/OS DAC -


I bought this TDA1541 DAC off eBay from seller in Hong Kong. The arrival of DAC was kind of late compared to my other purchase from Hong Kong sellers. Worse, the DAC is DOA. I found the +5v regulator was completely malfunctioning and output +12v to all digital power supplies. There was a PCB bug in the first LM317 stage, which left the ground floated. And somehow the second LM317 stage was not doing its job either. After I fixed both LM317 regulator, the DAC was still not working. I pulled out my oscilloscope and decided the input buffer chip was dead. I emailed the seller, and fortunately he is very cooperative in getting me a replacement DAC. After a long nearly four week wait, the replacement DAC arrived. Still, I need to fixed the PCB bug in the first stage of +5v power supply by soldering a short wire. But after all, it produced some kind of sound.

Power Supply

The power supply transformer is a 110VA toroidal with four outputs - 2 x 15V, 18V and 9V. It is quite sufficient for any DAC.

There is one thing that need to be carefully avoided. The bolt to fasten toroidal transformer should be in contact with both bottom a upper metal case. Otherwise an AC loop will form and cause noise and overheat of the transformer.

The -5V and +/-15V power supplies are built from discrete components. I cannot tell how much better it is than the conventional linear IC regulators, but it works fine. The only issue I had is that the way transistor were mounted to heat sinks caused connection problem in two cases. I had to loosen the mounting screws to reduce the excessive stress.

DAC

The DAC can be easily switched from OS to NOS by four switches. My preferred mode is NOS. So the SAA7720 chip is taken out.

Initially, each of the current outputs of the TDA1541 chips had one 100 ohm serial resistor - for protection purpose I assume. But it is not needed for current output pin. It caused about 20mv voltage drop at the TDA1541 output pins, which is much higher than the max 25mv specified by manufacture. I removed the four 100 ohm resistors replace then with 4 short wires.

The receiver section includes three chips plus one digital transformer. It is the most complicated receiver I ever have. However, even though the DIR9001 spec looks pretty good, it have not received a raved review in the web.

I/V and Output Stage

The I/V stage design failed to ensure the TDA1541 output pin be keep within 25mv range. It actually measured 200mv. That is why the DAC sound so distorted when it was first powered on.

After some simulation in ORCAD, I got a pretty good understanding of the I/V circuit. To fine adjust the voltage at TDA1541 output pins, I added a 10K ohm multi-turn trimmer in parallel with R2. By making sure Q2 have the similar quiescent current as Q4, I can set TDA1541 output pin voltage within 5mv.

I somehow don't like the cathode follower after R4. So I bypassed the stage completely.

I replace all capacitors in the low pass filter with WIMA MPK capacitors.

Paralleled Chip Noise Issue

The DAC is working all fine with a single tda1541 in either of the socket. But when both tda1541 are plugged in, I can heard apparent noise/distortion - especially in the quieter music passages. The noise is not present when it is not playing music or it is between the gap of tracks. I switched different tda1541 chips in and out, some particular matches of tda1541 produce less noise, however it can not eliminate the noise completely.

This has led me to wonder if the I/V circuit is sensitive to matches of the tda1541a chips. Or two tda1541a may be in contention for bias current?

The work to figure out the noise issue is still going on...

- 2010 -

- Addendum April 04 2010 -

To debug the noise issue for paralleled TDA1541, I first started off with ORCAD simulation. The R5 sets the quiescent current for the whole IV stage. The simulation shows that Q7 could be driven to saturation if R5 sets the quiescent current too high. But if the quiescent current too low, Q7 could be cut off when the DAC is operating in high current area. To test to see if the noise results from quiescent current, I added a 150 ohm resistor to parallel with R5 to bring the resistance value to around 45 ohm. But no luck, the noise persisted after the change.

Again, to decide if the IV stage is source of the noise, I built a simple opamp based IV stage to replace the onboard discrete IV. The IC used is a AD826 - one of opamp with the fastest settling time. The build was relative straight forward. Once it was done, I sat down for listening test. As expected, the opamp IV produces a sound with much tighter low end and faster attacks. But the relaxation and spaciousness was gone. Apart from the sonic difference, the background noise is still present with the two original TDA11541A chips with match batch number. Replacing one of the chip with a different one reduces the noise to almost inaudible except for one or two occasions - actually I am not sure it was not caused by something else. So the current conclusion is that the noise in parallel TDA1541A DAC is the result of the mismatches of chip parameters. To resolve this issue by circuit design is not an easy one - at least for now.

It is time to wrap up. I've decided to switch back the discrete IV with only one TDA1541A in the socket. I am happy with the sound, and DAC should do it for me for a while from now on.