FIRST, look at your player configurations. Where do you adjust volume, buffer sizes, and host based processing? If you have any over-sampling DAC, be sure to test reducing the player gain or if it has any soft clipping features. ( read that "peak compressor) Some have found software like host based oversampling to be better than the DAC's, though that may be with older DACS. It gets complicated and can make bigger differences than the hardware.
I find clear differences in DACs for a specific problem I am sensitive to. So far, the very best DAC I have found for midrange smoothness and not producing "digital' sounds that for me show up as sibilance in vocals and horns is the JDS Atom +. It is $99. The Schiit MODI is also about $99 but for my ears, the JDS is smoother. ( I have the ESS card in my Asgard). I do not hear, nor necessary believe , all the "instrument separation, stage width, bass weight, air", and other hyperbole that reviews are full of. Those are normally attributes of first the signal, then the room and speaker. What I do hear, but does not show up in any A/B testing is fatigue. I can listen to the JDS all day. The older AKM card in my Asgard was not as smooth long term.
DACs have improved greatly in the last few years. What used to be pretty well revived like the Cambridge is below mid-pack today. Yea, even the CI-FI have gotten pretty good. DAC chips spec better, but it is the USB input and reconstruction side that have the biggest effect on the sound. Newer USB inputs have just about eliminated all the jitter and noise issues. Some have even improved the PCM inputs.
I am very interested in the newest low end ( relative) R2R such as Bifrost, Aries, Soekris, Musician, Gustard R26, etc. I may like the added distortion from the R2R just as many like the distortion from tubes. Please do not read "distortion" always as bad. It is a parameter that can be used to make the music more enjoyable.
Also looking at the SMSL DO300, RME, Gustard X16, and for what I consider very very expensive, the Cord Qutest. The differences between my Topping DX3pro+ and the JDS are very very small and in all truth, may be more my frustration with the poor range of the Topping remote and some sub-conscious bias against cheap Chinese products. One I have not heard is the Geselli. Specs not great for a chip DAC, but it may sound better than the spec as the R2R DACs may.
Do consider the JDS. Too cheap not to try. FWIW, I found bigger improvements by switching to JRiver and making several player configuration changes. The -3dB issue is as real on a Mac as on a PC. Far more real that I realized before making the adjustments. The RME feature set is fantastic. I do wonder if it holds up to newer implementations.
Many of these have direct sales and return policies. Many are available on AMAZON with return policy. Only YOUR ears will tell you the truth. You may find an Apple dongle to be as good as any. It is not a bad DAC.
Consider the features and if a decent headphone amp, balanced outputs, remote control volume or in the case of the RME, a parametric eq, are important to you. Do you need preamp function? Headphones etc. That can greatly shift the end price point. An Aris gets a lot more expensive if I have to add a Topping P90 or Schiit Frea on top of it where the RME looks like a lot better deal, and my DX3 may actually be all I can hear. I use the Atom Amp+ on my desk and it is as good as I can hear as preamp or headphone amp.
FWIW, I do hear differences with burn in. It is a valid issue explained by the laws of physics on how dielectrics "form" with some use. Not just caps, but possibly even circuit boards and wire insulation. It has not been my experience it takes longer than a few hours. They can also reverse if left without charge over long periods.
I also know that out brain will change our interpretation over time and we process out issues as "environment" . This may explain why some things get better over longer times. The devoice may not have changed, but you get used to it. A good thing actually. Mechanical devices, speakers and cans, can take much longer and change over time. Plasticizers leach out, bonds break etc. They can keep changing forever.
Finally, ( coffee is ready!) R2R is very difficult to do well. The overall cost to quality puts a $1000 R2R on about the same level as a $100 D-S. Same basic level in build, packaging, power supply, analog stage etc. That does NOT mean they sound the same and the traditional measurements of the D-S will be far better than the R2R. We do know that the traditional measures are not the entire story. Helpful, but we have no idea what is the "musical" parameter is to measure. The question is, at any given price point, which do you prefer, and for any step in price, is is worth it to you? I may be willing to buy a Qutest if that is what it takes to really tame Joni Mitchel, Judy Collins and my big band CDs. Or, my new tweeters and crossover may have fixed it by better control of my woofer breakup.
