TDA1541A single vs parallel?

Joined
Feb 14, 2011
Messages
403
Reaction score
3
I've seen top end CD players and standalone DAC units with single TDA1541A chips (e.g. Naim) two chips (e.g. Marantz) and even four chips (e.g. Cambridge Audio and Satch DAC).

I'm considering building TDA1541A based DAC and I've seen kits on offer which use both single and parallel TDA1541A chip configurations.

Other than the cost of the extra components what are the benefits/disadvantages of having parallel TDA1541A chips? Is two really better than one?
 
It's attempted because , in theory, you gain about 3dB in improved signal:noise for each extra dac - if the output noise is truly random. (SNR gains proportional to SQRT[no. of devices in parallel] : 4devices = 6dB)

You then also have to sort out the I/V conversion for the higher total current; that's not exactly trivial, actually.

I have several problems with the parallel-chip idea. First, the TDA1541A has SNR of -110dB so that's not a limitation - it is, still, a very quiet dac. That number is small enough as it is that the average analogue output stage will rather spoil it...

Secondly, there's no guarantee that the 'noise', the unwanted signal components are random. I'd bet dollars to doughnuts that almost all of it of it will be correlated to signal (whether analogue output or digital feedthrough) so the mooted benefit is very much smaller than the marginal theoretical increase - which already is no limitation...

Finally, there are so many dac boards out there that don't extract what the one dac chip (this or other chipsets!) offer in the first place - by reason of errors in design choice, layout or (often) silly parts choice, like 'audiophile' film caps with long leads meant to decouple HF signals - one of my bugbears.

Adding a whole extra dac or three in parallel - at a sizable cost - simply can't make up for that kind of thing, and only presents extra sources of potential problems, too: more places to c**k-up, such as keeping the digital side tidy when you've one I2S source trying to feed two /four dac digital inputs in parallel. That's a nice fan-out problem just for starters.

Personally I'd stay with using one dac, and doing it as well as you can ;)
 
Last edited by a moderator:
Hi Martin, thanks for the quick response.

Sorry I've not been very active on here recently but I've been a bit preoccupied with other hobbies, more specifically learning to build bicycle frames :)

I've also been more than happy with my system lately so haven't felt the need to go chasing upgrades.

I'm still delighted with the sound of modded Arcam Alpha 5 and I'm slowly but surely becoming a devotee of the TDA1541A sound.

Most people would be happy to have such an excellent CD player but I'm just not wired that way and I can't help but wonder what the TDA1541A is capable in a 'no compromise' implementation.

I've been keeping an eye out for a truly top end TDA1541 based CD player or DAC (AMR CD-77, Marantz CD 7 etc) but they don't exactly come up often and when they do come up for sale they fetch very high prices.

I've therefore decided that perhaps my best route to a really special TDA1541A sound might be to go down the route of building a DAC kit. I've looked at various options and I like the look of the PCB sold by Analog Metric.

http://www.analogmetric.com/goods.php?id=428

This design seems to have a lot going for it:
  • Good quality board and thick plating.
  • Separate regulated power supplies for each IC and each voltage input for the TDA1541.
  • Plenty of room around the TDA1541A for decent bit decoupling caps.
  • Plenty of room for decently sized film DC blocking output caps.
  • It also retains a SAA7220P/B filter chip giving me the option of retaining oversampling/filtering (I'm not convinced about N.O.S).
  • And finally it's good value!

I was all set to order one today when I noticed that they also sell a more expensive alternative with parallel DACs hence my post.

http://www.analogmetric.com/goods.php?id=1766

Martin, what do you think of the schematic for the single DAC version? Can you see any obvious problems with it?
 
My opinion is -

(oh hang it, in for a penny..)

I have to say it looks all a bit bit perfunctory, the proverbial Mug's Eyeful. There's an awful lots of parts thrown at datasheet-based application notes with no real thought, for example a lot of electrolytics pointlessly-paralleled with small film types for the sake of it (regardless of whether or not it helps or makes a resonant mess*); there's no detail at all on the actual physical layout - the other 50%, the 'hidden' schematic that matters so very much when when working with HF and digital signals; and whoever plonked AD797s into that circuit did so because they wanted it to appear good on paper, without actually reading the 797 datasheet or ever playing with one to realise those extra parts and layout hints in the datasheet are there for really good reasons. Opportunities utterly unexamined, not merely tossed away here I think.

As a pile of parts to play with as a diy usb-input dac it actually may be pretty good value as a thing to tinker with until insane, perhaps worth a go if you want to indulge the time. But I bet your fettled Arcam Alpha even in its simplest incarnation would give it a solid kicking, out of the box or after you've been 'fixing' the dac for months ;) And objectively, spending a comparable amounts on the current Musical Fidelity V-dac )or whatever it's called) will buy you a working dac that measures superbly well meanwhile, is flexible, compact and future proof and comes with a warranty. You might even like it.

M.

*Alright, a simple example - look at the regulator schematic. Adjustable 3-pin regs with both 100uF, and a 0.1uF film cap in parallel on the output (no!!); then all followed with a 47uH inductor (why? to contain the mess you just made? It separates the HF film bypass from the circuit served if nothing else. So neither part is doing any goddam good, and wouldn't even if they were physically placed to do any good. But given the contradictions I've littel hope of that.

- So then you probably have both an unhappy reg, and a thing uselessly raising it's output Z in the audioband, but there because it looks like 'low noise'. Smacks of pandering to the kinds of people who review with their eyes and no actual measurement. Frankly I'd rather see a cheap 7805 with cheapo CRC in front of it. Cheaper in parts and guaranteed to always work, and probably well enough until proven otherwise. And even then - possible to get wrong if you dont route the reg's 0v reference directly to the 0v reference of the thingummy it powers.

And this kit has USB-ii2s conversion, digital filter and a dac requiring three good regulators to itself, all on the same board. Oh and a TCXO, a classic sign that the designer knows very little about what might matter. But it is shiney.
 
That's definitely one of my favourite posts on this forum, I had a good chuckle reading that little rant from you Martin. ;)

OK, so I admit to being duped by the appearance of the kit. It certainly looks the part to the untrained eye, very shiny. I think you hit the nail on the head with your comment about it pandering to peoples perception of what they think is good.

So since the Analog Metric kit has been comprehensively shot down in flames I guess that brings me back to my 'old faithful' project Arcam.

I never attempted the DEM reclocking on my Arcam, I'd love to give it a go if I can find a simple circuit diagram to follow.
 
I've not played with DEM clocking. There's a lot of discussion over on DIYaudio - member 'ecdesigns' has posted a lot on it - his current favourite option after trying many other is to use the standard DEM cap and tie both DEM pins to -15v with 6800ohm resistors. Quite what that is supposed to do I'm not sure.
 
Back
Top