Apologies in advance for the long winded reply here but there’s a lot to cover. Those not interested skip over now.
Well I have to declare, “mission accomplished” (thank G-d). It did lift the relative blanket off the sound. The sound is much more live(ly) now. Just what we were after here. Bravo! Now we can hear more clearly how well the rendering is going, and its rendering bee-yoo-tee-ful as well. No other way to put it. Trumpets sound particularly amazing. Vocals as well.
Think we’ll call it the “blanket lift” mod. Based on my experiments, think can conclude:
a) The MFB topology hampers the rendering output of the DS. My guess is, the step response is involved. Whatever it is, it got much better sonically when changed to plain old topology.
b) the OA as closed loop LPF does give a higher noise floor that impacts clarity and liveliness. The “blanket” source if you will.
By all means- YMMV on the above conclusions 
So for this we left the OA factory front end R’s, but the filter C between them is now much smaller. We removed the feedback C and have a larger feedback R connected to remove the MFB). Now the OA deliverers buffered digital bitstream basically to the passive LPF on the output.
On the output we have a simple RC RC cascade using tantalum SMD R and PPS SMD C. Since it’s just a first order cascade it has perfect step response. Simple, doable. The output impedance is also ~ similar (enough IMO) to what was there before. Not saying this is the ultimate output filter, just what we have for now and working with what we have. But based on the sound I feel no need to mess with it so far. We then wire that direct to the transformer.
So we’ll take the other pair of OA and convert to this topology as well. Then we’ll have 2 OA per channel as Ted suggested we need due to the fw algorithm.
But think I’ll then let each OA +/- drive a separate primary coil (there are 2 in parallel now in this transformer). That should be interesting as well.
We will also see about trying output caps again as well. Need to compare. Ultimately what we could do is no caps, no transformer iff we set it up for bi-polar PSU to the OA. Maybe we’ll create new modules and do just that. We’ll see how caps sound vs transformer first and go from there.
BTW the tinfoil hat slipped off for a moment and I bought 2 pr of Bybee Cu bullets. I’m thinking we’ll try them soon as well to connect the transformer outputs to the XLR. We’ll see I guess; either it works audibly or it doesn’t. Re: the slipped hat, as I said the other DAC drove a foundational level shift in thinking about all this. Ultimately that’s a good thing I guess and its already providing some large benefits…
I feel the experimental DAC is about ready to go on local high end tour. I’m excited and the gang will be impressed, no doubt. One also has a massively bi-amped, very transparent, best cables, etc, same speakers I have that is going to absolutely blow us away I think… He also has heard the other DAC (in fact he brought it over here) so knows what it’s up against.
Turbo
On another OA pair, also cut loose from the output nets and wired direct to transformer.
Now to see how it sounds…
