DIY DAC....is this any good

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
The TDA1543 has both it's admirers and detractors. I believe the 1543 is utilized in the 11,500 British pound Vertex aq Aletheia DAC. Can't say I'm aware of too many folks who prefer the DIR9001 though. Looks like it would only cost you about 20 pounds to find out whether or not this ebay DAC any good to listen to.
 
I swapped it for a stack of TDA1387s which are SMT, so no not a direct replacement at all - a lot of rather fiddly soldering. The pinout though is basically the same. This was the end of a long line of mods - before that I experimented with an active stage in place of the passive I/V.

I found the best sound with the 1543 was to be had by removing the two green rectangular caps btw.
 
I've never used a 'brown dog' but yeah I think that could be used. I piled up 8 chips one on top of the other and ran the top 4 (paralleled) with inverted data, the bottom 4 with normal data. This gave a 4mA peak current output as each DAC is 1mA. I then built a subtracting and EQ stage using AD830 - overall this sounded really good. The pinnacle of simple NOS designs for me. :)
 
I'd hesitate to say it'd be a 'lot' better, but probably better. Not as good though as going over to a better performing DAC chip to start with (TDA1387).

I have a Muse 4 * 1543 which I modded the hell out of too, this also ended up with 1387s. 1543 is really nice as a beginner's chip but eventually I grew out of it. Building circuits that work well with CMOS DACs (like 1387) needs more expertise, I doubt I'd have figured all that out without starting on 1543.
 
Check out the website of Doede Douma.

Yep I agree his website is a very valuable resource. He so far is the only one I think who has shown how distortion gets better when more 1543 chips are stacked.

Cooling is the big issue when stacking up 1543s, especially so when running at higher supplies. I wanted to build DACs with extreme numbers of chips (my highest so far is 40) so I abandoned 1543 because it consumes 300mW compared to TDA1545A which is a super-cool dude at a mere 15mW.

<edit> DIR9001 is fine - NOS DACs are the least jitter-sensitive DACs and this chip's jitter is OK IME.
 
When stacking 1543 its pin-pin but keep pin7s isolated. In fact that rule applies to all the DACs in this 8pin family in my experience. Caps on the supply legs I haven't found too critical on 1543, much more so on 1387 because its CMOS and gulps current down on clock transitions.

The CMOS DACs take next to no current - one TDA1543 is about equivalent in current draw to 20 TDA1545s or 10 TDA1387s. A DIR9001 is 10-20mA I think so total is comfortably below 50mA - the other chip on that board is a CMOS logic (74HC00 I think) which takes almost nothing. Oh there's an optical Toslink receiver - I don't know how much that takes because I removed mine in a bid to keep power supply noise as low as possible.
 
I just read the datasheets and I can't see a whole lot of difference between the 1543 and 1545, but a couple of other threads here on diyaudio.com say that they are not a direct replacement! Is it something to do with the voltage supply to the chip. I notice on the schematic for the eBay DAC, that the supply to the 1543 is 3.3v via the 317 chip.
 
Wow - so many questions! :)

Firstly about pin7 on the 1543 - its a voltage output. I don't connect voltage sources in parallel (the tolerance is 2.1 - 2.3V) - rather connect a resistor to each one individually, or leave them all floating and use an external current source. I make no assurances about the guys over on TNT Audio :)

Secondly, only 1545A can be found nowadays. The original 1545 was I2S input but seems not to be around anywhere. So no, 1545A isn't a direct plug in (1387 is I2S though - one advantage for using this). 1545A also has different output current and different (read lower) output compliance limits so does require some redesign for a 1543 socket. Also note it won't work above 6V.

A supplly of 3.3V to a 1543? Can't think of the advantage of going so low - most people seem to prefer to run above the 5V recommended to get more output compliance range.

Finally I suspect the 1545A can be driven by one of the formats available from the DIR9001 though I've not tried it myself. I coaxed a CS841X into the right (EIAJ) outputs without much difficulty.

Have I covered everything? If not, get back to me...
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.