Non-Oversampling TDA's

I've just spent the best part of 3 weeks playing around with TDA1543 and TDA1541 Non-oversampling Dacs and would like to make some observations and welcome any dissention or agreement. This thread may be useful for those contemplating getting into a DIY DAC.

I've been happy with a heavily modified Rotel 965 for a long time- to my ears it is more organic and less brittle than most contempory machines, but acquiring a Metaxas MASDAC with the UltraAnalog 23400 chip showed me that the Rotel really is very compromised.
Unfortunately the Masdac is beset with a noise problem and will only play when it is in the mood, and getting it repaired is impossible as Metaxas wont release circuits or provide any back-up.

There is something about the UltraAnalog chip that is just "right" and I can hear why it was so highly regarded for so long. I have heard that there were some reliability problems with it, and maybe that was the reason for it's demise, or maybe Wadia just wanted the design crew as they bought UltraAnalog out and then let the brand die.
If anyone has a circuit for a simple UltraAnalog Dac I would love to see it.

The TDA chips attracted me as the basis of a DIY DAC, mainly because of the good reputations for sound and the availability of information about them.

To get a feel for the different chips i bought one of Scott Nixon's DACKIT with a single TDA1543 , I purchased an assembled board from Scott and just had to enclose the board and provide sockets and a wall PSU.

The LITE DAC-AH from www.diyclub.biz is incredibly inexpensive for a completed ready to go dac - it uses 4 paralled TDA1543 with an OPA602 op-amp output stage.

Getting a TDA1541 DAC was a bit more difficult- there are published PCB's available , but I couldn't find any actually for sale. I settled on the www.diyaudiocraft.com Double TDA1541 kit.

I received the Lite DAC-AH first and was quite frankly dissapointed with it at first- for the price it is very good, but in very critical terms it sounded congested and dead- it displayed all the hallmarks of a conventional opamp output stage so I quickly bypassed the output stage and ran it unbuffered direct and cap coupled . Much more alive and less vieled but still a bit rough around the edges- so to be fair to it I connected it up to a spare transport and left it running for 3 days.
When I came back to it I received a very pleasant surprise for it was much better- almost smooth and quite listenable.
The BG coupling caps I used may even take longer to settle in, so I've still an open mind on this one.

Then i received the DACKIT from Scott Nixon in the US (Can I just take a short break to observe that on day 3 of the Olympics Australia with 20 million people has the same medal tally as the US with how many million? Perhaps that can be put down to the McDonalds culture and media propaganda that keeps TV screens on "reality" TV)
The assembled DACKIT PCB is a bit expensive for what you get, but a bargain in terms of sound quality. And bear in mind that you can get it going in under an hour with a minimum of skills.
It is a bare CS8414-TDA1543 with simple cap coupling to the output. It doesn't have a real low output Z so is best used with tube amps- which suits me just fine.
It's very transperant, lively and very revealing. A touch too revealing maybe- mine had Rubycon ZA caps, BG's might be better, but that will be taste, system dependent- most people would be very happy with The DACKIT.

The Double TDA1541A kit from www.diyaudiocraft.com arrived and looks very impressive - a very professional double sided board with gold plated through-plating mounting holes (a good DC soldering station and an equally good de-soldering gun is an absolute necessity)and good quality audiophile parts.
The manual is not the best , but it's not beyond most- the surface mount CX8414 freaked me out so I went for a DIL CS8412.
The output is quite low though and you cant use this direct, and I have some very high gain tube preamps here, which soon proved this point.

I tried 3 different types of output gain/buffers. First up was a parafeed step up trans- I was after simplicity, with the ability to tailor the frequency response both up top (for filtering) and in the bass. This was possible as the trans I used had very wide bandwidth, and parafeeding enables quite a range of variables.

I then tried a tube buffer, and finally Pedja Rogics 844/j-fet buffer stage.

This really got the Dac cooking! I'm an unashamed tube/transformer freak but this little SS gain stage was simply cleaner and more dynamic. It seems the output of the DAC is very load dependent- I've been told a high impedance load is not ideal for the TDA1541 and in my opinion, the tube gain stage, which was a very high quality unit, just didn't sound as competent as the AD844/2SK369.

Unfortunately, as good as it sounds, there is still a bit of high frequency noise present- I dont know if this is typical of the non-oversampling TDA1541and although I've been building tube gear for a long time I'm not much of an expert on digital.
At the moment the only filtering is a 1000p cap on the Dac output- Pedja's design has an optional further filtering cap on the output which I have yet to try.
Anyone with any suggestions on removing the HF noise very welcome!

So what's the verdict?

The TDA1541 has a cult following and I can hear why- it has a very natural sound and has also a sense of dynamics that is also best described as natural, rather than forced or electronic.

The TDA1543 is a detail machine, very transperant , very revealing, yet quite listenable and enjoyable.

Pedja seems to hear the same as per his quote from his web site:

".....it was interesting to note that though the TDA1541A is smoother (some claim it is the most analog sounding digital part ever made) and the TDA1543 is obviously a rougher performer that additionally could get into the saturation by the sonic content where TDA1541A stays soft, the TDA1543 has more liveliness and transparency. Then, with TDA1541A necessity for sin(x)/x compensation becomes more obvious. Hence I can, without any problem, understand both those claiming the TDA1541A is a higher class chip and those who opt for the TDA1543; I also understand those who describe the situation as a matter of taste."

I have to say though that I think the UltraAnalog 23400 kills both of them- and it probably should as it was just about impossible to buy it in a machine that didn't cost many thousands of dollars in it's heyday, plus the fact that I believe it was really a hybrid and virtually handmade- it certainly sounds remarkable to my ears.

But at the end of the day there is still that damn digital lack of true realism- it sounds very good, but there just isn't that freedom from constraint you get with a great vinyl system. The ease and flow just isn't there.

I think the problem isn't so much the DAC, but the whole gain/filtering process that follows. My experience with 30 years of building tube preamps and power amps is that any filter is bad news, whatever it's origins or components.

The CD system really is flawed . Like politics, it is really quite depressing. Thankfully I still have all my vinyl for when I want to listen to a more MUSICALLY dynamic , more organic and natural reproduction of music.

HF Noise

BTW I solved the HF Noise problem- it was a human error in assembling the PCB. Hard to find but easy to fix.

It's quite easy- I just cut the link to pin 3 of the opamp. There's some spare holes to connect a pair of 220uF Bi Polar Blackgates in Super E style from the output of the DAc (after the 270 ohm I/V resistor) direct to the output socket. Disconnect the wires from the opamp to the output socket, put a 100K resistor across output socket.
That's it. Email me if you need further help.

A tube buffer after the TDA1541 double dac is probably fine , but if your using a tube pre/power amp after that again ,I think you'll find the 844 buffer is much more neutral and ultimately natural.
Tubes are great! but too many is as bad as too much SS.