By late March, I checked on status. "Thanks for your email. It finally gave me an important 'aha!' moment about the Gradient Box so today I found a solution that'll help a lot. I now understand why we have this lack of intelligent remote-controlled analogue crossovers. There are no 'audiophile'-grade digital potentiometers. There are a very few good-quality digital volume-controller chips but they can't replace switched precision resistors due to their serious drawbacks: switch resistance that's too high, noise, measurable distortion etc. I think that's why current analogue chip tech can't yet efficiently integrate analogue switches and precision resistors into one IC. There are discrete op amps of course and soon there'll be 'discrete digital potentiometers' which will become the main building blocks of my xovers, later perhaps even analogue filters like smart phono preamps.

"I just laid out first parts placements. The Gradient Box needs 20 digital potentiometers (5 of the DDP above left). Without first creating a standardized DDP building block, designing different crossovers would become too much work and waste time. Soon though I'll have that new lego brick for my product library. Here's another brick, a 128-step R2R digital attenuator (above right). I made the design and PCB last May but assembled and tested it only a few weeks ago. It's very good indeed: 0.5dB steps, no noise, no distortion. The DDP module with firmware should be ready in April, the first prototype of the Gradient Box hopefully in May."

By now I'd made firm on adding a sub to the main system; and improving the upstairs sub's time-domain performance. To that end I bought a Dynaudio 18S dual 9.5" sub from a local dealer. Being compact, it could go right between the monitors so equidistant to the chair. Unless one digitally delays the mains because a sub sits in a corner well behind them to add meters of distance, physical time alignment is the only way. If you prefer your speakers as out in the room as I do, that does want a more compact sub. Otherwise a monster kills your optics then doubles as a big reflector for the mains. Dynaudio's Sub6/18S do include digital delay. That must obviously route speaker signal through their A/D, DSP, then D/A. In this price range, I'd not expect this to be the best solution¹.

The Dansk sub freed up the 135cm tall Zu Submission for the main system until Martin Gateley's RiPol Vox Sub D15 arrived. Then the Zu would move into the video system.

The moral of all this is basic but irrefutable. To undertake then ace certain jobs mandates proper tools or we botch up. A chef needs a sharp knife. Perfect subwoofer integration needs its own. Hello Gradient Box. To bridge my gap until its arrival, I ordered from Pál a carbon copy of the fixed filter built into my icOn 4Pro SE. That would follow the Vinnie Rossi tube preamp. It's all walking my talk that properly integrated subwoofers add things which I'd not want to be without. Could this be true for you? I couldn't possibly know. I'll simply describe the effects I'm getting. Then you decide.

¹ Purely theoretically, I asked Pál whether analog delay could be a feasible option and if so, how could it be made variable to allow one to dial in the distance by which the mains need to be delayed? For example, if a sub sits 1m, 1.5m, 2m or more behind them relative to the seat, the speakers should be delayed accordingly to insure coincident time arrival at the ear. I had a vague recollection of delay lines in original Quad electrostats to create a center-out wave pattern on their panels but no notion how that might apply to our application. "A 1m difference means a ~3ms delay. In the analog domain that's impossible to correct without extraordinary difficulty whilst creating far bigger quality losses than digitizing. In active crossovers, all-pass filters correct phase errors between drivers within very short distances. To provide some numbers, I've built a reasonably flat analogue delay line for 1ms with 5kHz bandwidth using 25 inductors and capacitors. To compensate for one meter so 3ms across the full audio bandwidth, a rough estimate would use 300 inductors. It's not something you'd want to do!"