Philosophical DAC question

I am working toward a new DIY speaker/amp system. This isn't my primary system, so I'm trying to keep cost somewhat in lline. The speaker will be a 3 way fed by digital distributed audio (RPi, SPDIF HAT, data over wi-fi) and I am planning on using a DSP for crossover and room correction, all staying in digital. This means I'm going to have 3 DACs per channel and that could get pricey. I have made a compromise and will use some nice but inexpensive DS dac boards from diyinhk for two of the three dacs per channel. I will have the option of something a bit better for one of the dacs, probably a Modi Multibit, Airist RDAC, or Metrum Flint.

The question is where should I use the better R2R dac? The woofer should be fine with a DS dac. But I'm having a hard time deciding between the mid range or the tweeter for the better dac. Many people say that the image mainly comes from the mid range and I think the musicality is often from the mid range as well. But detail can be more influenced by the tweeter, can't it?

As for the drivers, they are all high quality leftovers from other projects. Some nice 10" Vifa Denmark ally woofers. Some old but fine Dynaudio mid range drivers. The famous Dayton RS28F non-ferro fluid silky tweeter.

So, let me know what you think. At this moment, I am leaning toward using the better dac with the mid range for the sake of musicality, but I'm just not sure.

 

Where are the crossover cutoffs between drivers? Might influence your decision.

Id lean towards using the best dac with the mid band.

Depends on your taste in music as well - most modern pop/rock releases (last 10 years or so) compress the hell out of the high frequencies and have highly shaped dithering, so resolve here might not be as important...

 

Good question. I'm going to try the mid range open baffle. My first attempt. The woofer mid crossover will be dictated by the travel/distortion of the bottom of the mid range in open baffle. I expect the woofer mid crossover to be in the 300 to 500 Hz range, but I have some distortion testing to do before I can finalize it. The mid is a 17 cm, so I want to keep the mid tweeter crossover fairly low. The silky tweeter is clean down to 1500 Hz, but a little safer at about 1800 Hz.

As for music, I tend toward small group acoustic and cleanly recorded electronic music. I wouldn't worry about the compress pop music. When it plays, I don't have high expectations for quality.

 

For the price of a new Airist or Flint, you can get a used Modi Multibit x2... then at that point you might as well get a third Modi for the woofer and spend a couple bucks on the Schiit floor-wart for some cleaner cable management.

 

What XO frequency points are you thinking, how steep slopes?

Imaging, soundstage contours and depth information is mostly in treble frequencies (1280 Hz and up).
And yes, at soundstage ability R-2R dacs are superior, up until crazy priced dacs with atomic clocks.
Modi Multibit might not be best bet for highs based on numerous reports here on tonality and soundstage capabilities.

Mids are where I'd use a R-2R anyways, the tonality and dynamic performance are most important there.

Also, don't underestimate dac impact for lows. A wimpy dac will result in wimpy bass.
There are many SD dacs with capable low range. Ime, bass is more impacted by dac filters and psu than architecture.

Lastly, I'd be cautious with mixing different sounding dacs through out the drivers. It might lead to timber issues.

 

I'll say: use the same DAC chips for everything, and if possible, the same or very similar sounding amps for everything, to get the most coherent sound.
But I'm curious about this experiment, whatever you do, please report the results

 

Be aware that different types of receiver/filter/etc. might lead to different end-to-end latency if you're mixing DAC types, it will require fudging of a delay to get even approximately right.

There is a bunch of unintended consequences lurking here, from over-complicating things. There are reasons why people often go through a phase of massive piles of gizmos and then simplify down to fewer, better-quality components eventually.

 

Thanks guys. Very interesting comments. It gives me a lot to think about.

The reason for this project is that I built a Linkwitz LXmini a couple of years ago. It's pretty successful but I worry about the minidsp changing analog back to digital for the sake of dsp, than back to analog, all with cheap chips. I wanted to try dsp but keep it digital through the dsp stage. Also a learning process on dsp. My primary system is all analog using the keep it simple approach.

As for xo slopes, I have typically used 4th order acoustic in analog systems and would probably start there.

The more I think of this, the more I think I should try an R2R DAC in different xo locations to understand those unintended consequences. The beauty of dsp is it should be easy to swap.

I will also have to come up with a latency test. Delays are theoretically easy in dsp, but I have yet to discover the reality.

This whole thing may drive me to matching dacs. I have a Flint on pre-order, so it's easy to try one R2R DAC, but I get the risk of mismatching dacs. Thanks for the input.