I don't have any meaningful difference in experience between listening to these DACs via Stax headphones vs. Zu speakers. For anyone who has history with Stax, the SR5 was a mid-1970s issues and it was/is a very natural, relaxed, holistic-toned headphone. The later Stax, beginning with the SRX, were in my opinion substantially "hotter" on the top end, which some people might say were more extended, but they sounded unnaturally bright to me, compared to the earlier large diaphragm SR3 and SR5 phones. Compared to other phones of their era, the SRX was still a comparatively excellent headphone, but having tried them, I reverted to the SR5, of which I have 3 pair dating back to 1975/6. They have been in continuous use for 40 years, so I think it's fair to say that that college-era purchase I stretched for was worth the expense!
The Zu speakers are built around Zu's 10.4" full range driver, supplemented by the lovely Radian compression tweeter on a simple high pass filter above 12.5kHz. The Zu speakers are crossoverless. They have an unusual coherence, tonally, spatially, dynamically and wrt phase, that eludes crossover-intensive speakers with multiple, disparate drivers. They are also quite efficient at 101db/w/m and hence bursty and vivid, just like my Stax SR5s. With the Stax, it is a little easier to easily hear and comprehend the finer differences in DACs, which takes a little more concentration and quietude to equally discern via speakers, but I see no reason for a MHDT R2R DAC to be more enjoyable and accurate via speakers than via headphones.
A lot of contemporary hifi gear, and the audiophiles who listen and judge it, is designed for what I'll say is excessive resolution. That is to say, there is a striving for extreme leading edge detail that one doesn't actually hear in live music. We've been trained to want the ear candy of xray resolution, by close-mic'ing, manufactured soundstages, multi-tracking, hot microphones and engineers mastering for cheesy playback. If you are focused on truly realistic timbres, tonal veracity and a naturally-occurring level of transient detail, much of what passes for "high end" in hifi today will sound over-etched and odd if you deliberately consider what you're hearing.
The common thread of disappointment in delta-sigma DACs ala the ESS 9018, is the flawed ratio of definition relative to tone. Yes, HOW the DAC is implemented is hugely important, and some of the FPGA solutions like Chord uses are much more acceptable. But for the most part, delta sigma has a trait that undermines what's impressive about it by giving tonal fidelity short shrift.
I happened upon mhdt around 2009 when I saw outboard DACs becoming viable again, and bought several to experiment with. I liked the design simplicity of MHDT's Havana, the adherence to R2R, and the use of the tube buffer. Over the course of several months I tried a few ESS implementations, a few Burr Brown delta-sigma, the Audio Note and the mhdt R2R in the form of Havana Balanced. At the time I only cared about 16/44 and hadn't decided how I would migrate to a server. As precursor to ripping my large CD collection, I just wanted to get better CD sound.
When I first listened to a Havana Balanced, I wasn't fully sold on it. It was organic and lovely in some respects, but not as dynamic as I expected or preferred. The Zu speakers are vividly dynamic and they are "fast." The closest a dynamic speaker gets to a Quad ESL in quickness. Stax SR5s are similar. I traced the Havana's "laziness" to three factors: 1/ the stock 5670 tube; 2/ need for extended break-in; 3/ the stock feet. Digital electronics are very sensitive to vibration and what they are sitting on. I found an outstanding improvement by simply setting the Havana Balanced on audio bearings. I then found that the Bendix 2c51 or the 6385 tube sharply improved its dynamics and clarity of natural detail. And I found that four months of active break-in helped every aspect of sound along.
The Havana Balanced was energetic, tonally coherent and harmonically complete. I later tweaked it for a little more resolution in practice by first upgrading the PCM56J ships to K chips, and later got another notch up in natural resolution by replacing the PCM56K dac chips with AD1856. Altogether excellent. Sean Casey, founder and designer/engineer/chief-of-everything at Zu said I had the best digital sound he's ever heard. I also replaced the output caps in the Havana Balanced with Hovland Supercaps, which not only easily fit, but are affordable and sound convincing, but that isn't really necessary. By and large, the mhdt dacs sound quite fine stock, though tube substitution is worth experimenting with until you find *your* tube.
I used my Stax SR5s to identify the very fine points of difference between DACs at that time, and between chips, tubes and caps changed in the mhdts themselves. I use Aurios Media Bearings under my MHDTs. They aren't made anymore, but any bearing product, including the inexpensive balls/cups from Herbie's will work.
The R2R chips have a reputation for sound quality, which is borne out by the MHDT implementations. I regularly hear current delta-sigma-based DACs in Zu-based systems, and others, at high expense compared to MHDT. I can afford them. I don't elect to buy them.
Lately, I have been interested in having 24/96 and 24/192 capability as I at last make a move to put my CD content on a server, and mix that library with higher-rez downloads. I considered "upgrading" to a high-rez + DSD dac, but haven't been convinced by their essential tonal fidelity and imaging. I hadn't followed what MHDT is doing for a couple of years, so when I circled back to them and found they have PCM1704 and AD1862 DACs for 24/192 with 24 and 20 bit output respectively, with discrete transistor I/V rather than using integrated opamps, I found it easy to ditch DSD from my immediate consideration.
I will write comparative notes after I get the SE Pagoda, but as soon as I replaced the Havana Balanced in my Zu Definitions system with a Pagoda Balanced, I could hear the gain in useful resolution, spatial realism and lower noise without *any* sacrifice in tone. And that is before break-in or any other interventions. I did replace the stock 5670 tubes with 2c51, for further snap and vividness. The 20 bit Atlantis arrived last week. I moved the Pagoda Balanced to the Druids system where the preamp has balanced inputs, and have been listening to the SE Atlantis, immediately equipped with a NOS Hytron 2c51 for the last few days on my Definitions system. Just as folks have written here, it is startlingly lovely and instrumental tones are fully realized in a way I've never heard from any delta-sigma dac. What I'm interested in is actively comparing SE Pagoda with SE Atlantis. Both have legendary late-era R2R DAC chips. Both have the same discrete analog section. I will be posting my Havana Balanced DACs for sale soon, and will either sell the SE dac that "loses" the comparison, or I will keep both and use the slightly lesser on for my office system.
More to come, for anyone interested. I'm in Los Angeles, in case anyone is interested.
Phil
Please do tell more, the comparisons and your way of explaning sound is very easy to follow. The knowledge is very much appreciated!
