General DAC Questions/Tech Thread

[luvdunhill]

For the datasheet curious, here's the JT-11-EMCF transformer I use.

I tried a few different transformers, and yes some of them didn't do what they were supposed to in my opinion. Both to my ears and on the oscilloscope.

NekoAudio: just curious, are you of the crowd that believe that square wave response through a transformer equals all is well camp? I know it's a pretty heated discussion. I did use square wave response to tune my input transformer and it turned out well, but which I don't have a huge opinion on it (yet). The reason I ask is you mentioned both ears *and* oscilloscope

Dreadhead: if it measures transparent, it must be transparent, huh? We're not talking subjective transparency here, I'm saying it measured transparent. That should be an adequate defense in your book, right?

Edited August 17, 2009 by luvdunhill

[Dreadhead]

Dreadhead: if it measures transparent, it must be transparent, huh? We're not talking subjective transparency here, I'm saying it measured transparent. That should be an adequate defense in your book, right?

Absolutely. If I still looked like I was disagreeing I wasn't.

[luvdunhill]

Oops, okay, carry on then.

[dsavitsk]

Some of the knocks on transformers are deserved, but some are hyperbole. For instance, low Z transformers, such as the ones Neko uses, are relatively easy to make and are extremely linear. Even a cheap one with a steel core can perform quite well. Use a 50% or 80% nickel core and they are staggeringly good. Distortion is very low, and it is all low even harmonics and barely noticeable.

Input trandsformers are a lot harder as they require a lot more inductance and care in design. As such, they tend to impart their character a lot more. Steel cores tend to sound very warm and lack a bit of detail. Nickel is a lot better w/r/t/ detail, and they are less syrupy. However, something like the Lundahl amorphous core input transformers are utterly transparent. I've used them quite a bit and often can't tell whether they are in a circuit or not -- as close to wire with gain (sort of literally) as I've experienced.

Note that input transformers also impart the benefit of eliminating common mode noise which is often (usually) a big enough advantage that it overshadows any possible drawbacks.

Finally, it might be noted that studio work is often more about tuning the sound than it is about recreating it perfectly. As such, studio people will use transformers expressly for this purpose. For instance, if you want your band to sound like it was recorded in 1972, use an Edcor steel core. If you want your violin to sound like a violin, use the Lundahl amorphous. The fact that transformer manufacturers are not always trying to make a perfectly accurate device means that choosing what to use in neutral audio equipment must be done with some care. For a good example, Cinemag has a transformer comparison on their site -- you can download the files and listen to the transformer tuning -- Cinemag Audio Transformers

[luvdunhill]

I'm assuming you're referring to the LL1544A? I've shied away from it, as I never could find any data for it (graphs, etc.)

[dsavitsk]

Actually the LL1674. Similar transformer (Kevin at K&K says they are virtually sonically indistinguishable) but on a larger core than can handle higher dBu.

I think if you look around that Lynn Olson did some measurements on it, or maybe the LL1676.

[Azazel]

I used LL1674 on some DACs, I like it.

I love my S&B-based preamp.

MOAR IRON!

[aerius]

Actually the LL1674. Similar transformer (Kevin at K&K says they are virtually sonically indistinguishable) but on a larger core than can handle higher dBu.

I think if you look around that Lynn Olson did some measurements on it, or maybe the LL1676.

It was the LL1676. K&K Audio / Lundahl Transformers: Superb results with LL1676 in 1:1 mode by Lynn Olson

[NekoAudio]

NekoAudio: just curious, are you of the crowd that believe that square wave response through a transformer equals all is well camp? I know it's a pretty heated discussion. I did use square wave response to tune my input transformer and it turned out well, but which I don't have a huge opinion on it (yet). The reason I ask is you mentioned both ears *and* oscilloscope

No, I don't like to use steady-state as the only measurement. Checking a square wave is still good to do, but I played music to my oscilloscope. This way I was able to compare two transformers directly against each other.

The Jensen wasn't the most expensive I experimented with but it was the one I decided performed best. (I asked the various manufacturers for their cost-no-object-best-for-this-application transformers.)

[Azazel]

Wes, just curious, did you try some Sowter from UK? They have a fine range with fair prices...

Maybe it is time for the foundation of "Moar Iron Team"

[cetoole]

No, I don't like to use steady-state as the only measurement. Checking a square wave is still good to do, but I played music to my oscilloscope. This way I was able to compare two transformers directly against each other.

The Jensen wasn't the most expensive I experimented with but it was the one I decided performed best. (I asked the various manufacturers for their cost-no-object-best-for-this-application transformers.)

Did you use something to sum the input and output, to see differences in the two? I dunno, but at least what I listen to looks really, really squiggly on my scope, and I just cant see how that is useful.

[dsavitsk]

Did you use something to sum the input and output, to see differences in the two?

My wife has written/worked on software to do this. She uses it to compare sound waves against brainwaves generated from listening to said sounds. I'll find out what format she needs the waves to be in to do the comparison.

http://www.soc.northwestern.edu/brainvolts/documents/Lee_et_al_JofNeuro2009.pdf

[NekoAudio]

Wes, just curious, did you try some Sowter from UK? They have a fine range with fair prices...

Maybe it is time for the foundation of "Moar Iron Team"

Azazel, sent you a PM. I don't wish to discuss in public which transformers I decided against.

Did you use something to sum the input and output, to see differences in the two? I dunno, but at least what I listen to looks really, really squiggly on my scope, and I just cant see how that is useful.

Well, primarily I looked for additional distortion, which is more squiggles than less. I used a really crappy DAC as a guideline for what bad squiggles look like, because the distortion was audible with the crappy DAC. Certain songs worked better at creating the distortion. I also used headphones at the same time, in parallel, so I would know what I am hearing/seeing.

You can also see roll off in broad strokes to some degree. Since I was playing with my output stage, I could create roll off on purpose. That gave me a baseline for how the waveform would change as a result, and then I could look for and hear the difference when not doing it on purpose. Obviously bass roll off is easier because the wavelengths are really long.

Performing a difference wouldn't necessarily tell you if one was better than the other, just that they're different. But you can do it by summing with one of the outputs +/- swapped. Easy enough with a transformer-based output.

Instead, I liked to look at two signals simultaneously but with one swapped. I found this "mirrored" display easier for noticing differences, and how they were different, without hiding which signal was which.

I'm not sure anyone would get very far just looking at the scope. I needed to be listening to the music at the same time and identifying specific things to compare against. Takes some time to get in-the-zone, so to speak. I was spending hours straight, day after day, performing A/B testing with very specific music.

You might be able to get by with just the scope if you had a storage scope and a digital reconstruction of the waveform to compare against?

[NekoAudio]

http://www.soc.northwestern.edu/brainvolts/documents/Lee_et_al_JofNeuro2009.pdf

That's a really interesting read. Not really surprising to me, since it matches my experience, but it's cool to see people doing research into the subject.

I wonder if further differences would be found based on the type of musician? The musicians listed in the paper were primarily pianists and violinists, which allows you to mix notes, and I think feels very different to play compared to brass or wind instruments.

[Dusty Chalk]

My wife has written/worked on software to do this. She uses it to compare sound waves against brainwaves generated from listening to said sounds. I'll find out what format she needs the waves to be in to do the comparison.

http://www.soc.northwestern.edu/brainvolts/documents/Lee_et_al_JofNeuro2009.pdf

That's fascinating. I'll read the article completely later, but just from skimming it, it looks like she was looking for correlation? I'd be interested if some sort of function could be gleaned (e.g. "human ( input ) == output" ). I suspect that might also be a function of individuals, as well. Less evolved individuals might only be derivative functions, whereas us musicians would be complex, and hippies might have a mix of real and imaginary.

[cclragnarok]

That's fascinating. I'll read the article completely later, but just from skimming it, it looks like she was looking for correlation? I'd be interested if some sort of function could be gleaned (e.g. "human ( input ) == output" ). I suspect that might also be a function of individuals, as well. Less evolved individuals might only be derivative functions, whereas us musicians would be complex, and hippies might have a mix of real and imaginary.

Complex is basically a mix of real and imaginary....

What does that say about musicians and hippies?

[n_maher]

Did you use something to sum the input and output, to see differences in the two? I dunno, but at least what I listen to looks really, really squiggly on my scope, and I just cant see how that is useful.

Yeah, but we all know what you listen to!

And Wes, you can feel free to discuss the trafos that you tested. I think as long as you don't rip someone to shreds because you didn't prefer their trafo no one is going get up in arms about it.

[luvdunhill]

Wes, just curious, did you try some Sowter from UK?

yes, yes I have

[NekoAudio]

And Wes, you can feel free to discuss the trafos that you tested. I think as long as you don't rip someone to shreds because you didn't prefer their trafo no one is going get up in arms about it.

I don't think any Head-Casers will get offended, but there are companies that sell or make products incorporating transformers from these companies and I don't wish my comments to negatively affect them. It'd be one thing if I was talking as Wes, but it's something else when I have my Neko Audio hat on.

Anyway, I tried a few "audiophile" transformers. Oscilloscope + listening tests, single-blind A/B testing with a different transformer into each channel of headphones, and speaker listening tests. IMO the Jensen was noticeably better, and the single-blind A/B headphone testing supported that. (I wasn't super rigorous with the single-blind A/B testing, so take that as you will.)

[Dusty Chalk]

Complex is basically a mix of real and imaginary....

What does that say about musicians and hippies?

Exactly.

[The Monkey]

As most of you know, I have been (obsessively) comparing DACs lately. I've been finding that my preference leans toward a DAC with a "sweeter" midrange. No, what the fuck exactly does that mean? I'm, not quite sure, to be honest. So far, I have really liked the Dodson and the ECD-1 because the midrange (esp. voices) seems to stand out just a bit more from the rest of the presentation. It's subtle, but it's there. Like very shallow bas relief where the midrange is raised ever so slightly higher than the rest (and maybe some midbass too).

My perception when hearing the ECD-1 and the Dodson is that they are somehow clearer or revealing more information in those mids than comparators such as the Bryston, North Star, or HR UDAC, which seem to have a more recessed or "muted" presentation.

I have read certain criticisms (and I wish I remembered where) that postulate that the mids of some DACs, such as the ECD-1, are emphasized because that's basically a crowd-pleaser. But that in the end, it's really hiding the DAC's shortcomings. I'm not sure I agree with this, but I have insufficient knowledge to make a conclusion one way or the other.

What I do know is that I prefer those DACs where the mids give me a better sense of "presence." Is this a preference for artifice? Are the newer, more technically "advanced" DACs, with their seemingly more "neutral" presentation closer to what others would describe as analog sounding? Or is this all just a bunch of hooey?

[luvdunhill]

Are the newer, more technically "advanced" DACs, with their seemingly more "neutral" presentation closer to what others would describe as analog sounding?

I'd say definitely not, but that's just me.

[Azazel]

Monkey, maybe you are now in a good position to spend (under) $300 on something different: an eq (yes!) like the Behringer Ultracurve.

You can go digital in-digital out and it is quite transparent this way. Then, you can do almost anything your imagination may want, in particular you can eq a little bit to see, with knowledge, what do you hear and what happens if you have a little lees highs, or a litte more mids, or whatever...

Believe me, it is much fun, and very 'educative': afterwards you will be able to identify in auditions what is doing the hardware to the signal. For the price of, say, one cable.

Josep

[Torpedo]

I'd say definitely not, but that's just me.

x2 I think a DAC should sound natural and believable, specially in the midrange where most of the information resides. If the sound you seem to like isn't perfectly transparent to what was recorded, and has some tricks to sound nice, who cares? It's you who must be happy listening to his rig.

[spritzer]

I'd say definitely not, but that's just me.

Ditto.