[matt_garman]

Considering the recent excitement over the TI tpa311x amps, I was wondering if there isn't a DAC equivalent.  That is, a "diamond in the rough" so to speak among DACs.  I figure, we've got the TI chip amps for amplification, the Pioneer SP-BS22-LR for speakers... now how about the DAC for the "ultimate reference" (budget) system?

Basically, I'm looking for a DAC.  Currently I'm using a Cambridge Audio DACMagic (the original), and I like it.  But I want to re-purpose it.  My theory was, given the quality of the tpa3110/3116 amps---particularly at their price---shouldn't we have reached a similar level with DACs?  That is, a "giant killer" board, maybe requires a little soldering or case work, but something that sets a new bar for price-performance?

Anyway, I read DAC reviews until I was nauseous, and finally settled on the "Muse Mini TDA1543x4 DAC".  You can get these for $50--60 on ebay (multiple sellers, direct-shipped from China); it looks like Amazon even sells them.  This is nothing new; in fact, my board is dated 2010.  The technology itself is rather old, being of the NOS (non-oversampling) variety, using these TDA1543 chips from yester-year.  There is a super long thread on Head-Fi concerning this DAC and many modifications.

(The more common flavor of DAC today is the "sigma-delta" variety (as opposed to NOS).  There are some interesting articles out there that describe the differences between the two.  I have a very vague notion about the differences, but won't even try to explain them here.  As with "all things Internet", there seems to be an on-going debate over which is better.  I won't pretend to have any opinion on this, but there does seem to be agreement that the two kinds of DACs sound different.)

In contrast to my opening statements, I wasn't expecting this to be a giant killer, but I was hoping it might at least be as good as my DACMagic.  I was also thinking (hoping) it might sound different, but in a pleasant way.  I thought, at this price level, maybe I'll find that the rest of my system isn't resolving enough to hear differences between a respectable DAC and a cheapie.

In short, I'm disappointed.  I received it Friday night, played with it a bit, let it run all day Saturday and Sunday, and checked in from time to time, switching back and forth between it and the DACMagic to compare the sound.  A couple notes: I didn't even bother with the included power supply, but instead used this Regulated 3-12 VDC 2A 6-Way PSU from Parts-Express.  Saturday night I removed the output DC coupling capacitors and replaced them with a wire.

At first I thought it sounded pretty good, and I was patting myself on the back, thinking I achieved my goal: the performance I wanted at a much lower price.  However, the more I listened, the more I started to doubt myself.  It came to a head when I started listening to Gov't Mule's new album Shout!.  What jumped out at me was the cymbals: they sounded like somebody saying "Tsss tsss tsss" rather than an actual cymbal.  I switched back and forth several times, and indeed, the definition just isn't there.  Two other things became obvious: the soundstage seemed "collapsed" with the Muse.  By that, I mean the music sounded like it was coming from two speakers, rather than the "holographic" effect where the speakers seem to disappear and the music is "just there".  Likewise, instrument separation was diminished with the Muse; there's simply a lot more nuance to the music with the DACMagic.

As per the Head-Fi thread I linked, there are a lot more mods that could be done to this DAC, but I'm not sure if they are worth it.  Bypassing the output caps and a better power supply seemed to be the two "must do" mods; the rest are increasingly complex, along with debatable performance improvements.  So some questions I have after this little experiment: is what I'm hearing a limitation of these tda1543 chips?  Or is Muse's implementation simply poor?  Or, do I just have a bad sample (curiously, my PCB is red, and it looks like everyone else's is blue or white)?  Is it worth letting this thing go for another 100+ hours, just to see if "burn-in" makes a difference?  Where do I go from here?

I'm curious, what's everyone else in the C&C community using for a DAC?  Have you found something <$100 that bests $500+ equipment?

[srb]

So some questions I have after this little experiment: is what I'm hearing a limitation of these tda1543 chips?

As you said, the TDA15xx chips are really liked by some and fairly quickly dismissed by others.  I've heard quite a few and I tend to lump them into a lower dynamics / lower details category.  Smooth?  Yes.  Non-offensive?  Yes.  But they all sound to me like a thin pillowcase was thrown over the speaker frequency wise, and also lacked punch.

I owned an MHDT Paradisea which was regarded to be an exemplary example of the genre at the time, but it just didn't do it for me, despite it's capable power supply and fairly high quality parts.

Steve

[Mudjock]

My experience with TDA1543 is that, yes they are a little rolled off at the extremes, particularly high frequency, but sound pleasing and unoffensive overall.  I had to take it out of my main rig because it was causing me to voice my speakers with a little too much treble.  As far as imaging, I have had that trouble with a couple different DACs, including the 1543, and the cure has been jitter reduction.  I used a Monarchy DIP 24/96 in front of the 1543 DAC back in the day, but that isn't quite as cheap and cheerful...

[JohnR]

is what I'm hearing a limitation of these tda1543 chips?  Or is Muse's implementation simply poor?

Um, well, it's cheap... having said that, I also prefer more modern DACs. The ones from hifimediy might be worthwhile, haven't gotten around to trying one. You could probably search for comments and there are threads in C&C on others.

[Folsom]

Try battery power with some large caps for plate noise.

I've been very impressed for the money with these little guys, so long as they aren't running unconditioned, regular (regulated/linear/smps) power supplies. They need a few changes like better caps.

Couldn't describe the sound the way you do.

[wushuliu]

I really liked it. Actually I loaned mine out to another member and I don't think he ever gave it back!

It's not worthwhile looking for a 'TPA' equivalent in DACs. The TPA and similar amps are great because they are relatively simple designs. DACs are just plain complicated to do well and the direct from china boards just aren't worth it now that you can get a Dragonfly for $150 designed by a well respected expert in the field or an AMR, etc. There is a really sweet looking ES9023 DAC that seems very well made on ebay but even that clocks in at around $150-200 shipped.

[matt_garman]

It's not worthwhile looking for a 'TPA' equivalent in DACs. The TPA and similar amps are great because they are relatively simple designs. DACs are just plain complicated to do well...

But don't you think that's just a limitation of the current state of technology at this moment?  Surely it's only a matter of time before there is a TPA DAC equivalent... unless the problem is completely intractable, which I don't think it is.  Seems to me, at some point, some clever person will figure out a relatively simple way to do a great job of digital to analog conversion, bake that knowledge into an IC, manufacture it in bulk, and now we've got the TPA DAC equivalent.  And/or, manufacturing technology will enable once cost-prohibitive designs to be realized at a reasonable cost.

There is a really sweet looking ES9023 DAC that seems very well made on ebay but even that clocks in at around $150-200 shipped.

Which one would that be?

Another question: I have a belief about consumer products in general (not just audio equipment): that value exists on a non-linear scale.  Since I'm a geeky guy, I think about this in terms of a graph: imagine cost on the vertical scale, and quality/performance on the horizontal scale.  Products will fall all over that graph, but I think the general trend is a roughly linear trend until a certain point in quality, at which cost increases dramatically for minimal performance gain.  At some point the graph is almost a vertical line: you're paying exponentially more, but getting virtually nothing for it.

So for any product, I'm interested in the point on the graph just before it stops being linear.  What is is that price-point with DACs?

To put it another way: let's simplify and say there are only four DACs available in the whole market: dacA at $50, dacB at $100, dacC at $200, and dacD at $500.

Clearly, the trick is quantifying "quality and performance" for something like a DAC that has a huge subjective component.  But for the sake of argument, let's say we can assign a number to the quality/performance of a DAC.  So, the performance numbers for our hypothetical DACs are: dacA=5, dacB=10, dacC=20, dacD=30.

Here, you can see, with A through C, for each increase in cost, I get a proportional increase in performance.  But, with dacD, compared to dacC, you pay 2.5x as much but get only 1.5x the performance.  If this were reality, I'd want dacC, because to me, it represents the optimal value point.  Yes, dacD is better in terms of absolute performance, but it's clearly worse in terms of performance/price.

Now, for kicks, let's say some new/disruptive DAC technology comes along, and now we have dacE for $75 with a performance number of 15.  If I hadn't already bought dacC, this is the one I want to own: yes, it's now only third best in terms of absolute performance, but it's off the charts in terms of performance/cost.  Unless you absolutely need top-tier performance, this represents the best value; it's the TPA of DACs.

Just to make it easier to follow:

Code: [Select]

Model  Price  Perf  Value (Perf/Price)
dacA   50     5     0.1
dacB   100    10    0.1
dacC   200    20    0.1
dacD   500    30    0.06
dacE   75     15    0.2

As I said, this whole theoretical framework kind of falls apart when performance has a subjective component... but despite that, I think you can still get a reasonable approximation of a performance metric of sorts.

Note that this also gives DIY a clear edge, since (in theory) there's less cost going to non-performance-adding things like branding, marketing, distributor profits, retailer profits, etc.  This of course assumes you enjoy DIY, so the time cost is zero (or arguably negative if you really enjoy it).

Just kind of rambling on about how I think about these things. 

[Folsom]

There's a lot of reasons it's hard to have a DAC that's as killer as the TPA amps. Power is a big one. TPA is still good with decent power, cheap DAC's have one bad regulator and need 3 or more good ones on board. They also need design between multiple chips, TPA are all in one. The list continues.

Just bread board copy Peter Daniels's if you can. It'll cost you hundreds in parts but the product will be literally world class and the best NOS DAC ever made (his tweaked version).

[Folsom]

But again, the cheap 1543's just need better power to become full and musical. Don't expect big refinement.

[JohnR]

Seems to me, at some point, some clever person will figure out a relatively simple way to do a great job of digital to analog conversion, bake that knowledge into an IC, manufacture it in bulk, and now we've got the TPA DAC equivalent.

Maybe, but it won't be found in a decades-old design like the TDA1543... The ES9023 that wushuliu mentioned is probably the chip you are looking for (as of now).