As is this writer's MO, narrative unfurls in real time, not backwards from a conclusion once listening has wrapped. My approach captures process to confront personal expectations, even misconceptions. That's why I pen previews well before the products ever arrive. I get to assume the position—present what I think the product will be about—and dialogue with the designer. Once the loaner shows up, I may have to change my position; or challenge the designer's claims. Now the reader gets to accompany me on that journey rather than suffer a neatly linear exposition which knew how it all ended before the first word was ever put to screen. Hindsight tends to rewrite history.

After learning of Vinnie's reluctance to use output transformers, I wrote him that "my question/concern was more about running LIO DHT as preamp into amps other than your own. Your >20KΩ input impedance recommendation now makes perfect sense." To which he responded that "I'm still looking into lowering that a bit so when using the higher tubes already mentioned (except for the 101D/205D), we could come in below 1.5KΩ; and for the 2A3, 300B, PX4 and PX25 below 800Ω." I then confided to Vinnie that one reason why I got out of tube amps was that I felt output transformers killed bandwidth, speed and introduced phase shift in the frequency extremes. This was rubbed in by quality transistor amps from Nelson Pass which emphasized that difference. Vinnie concurred. "They seem to take away from what I am after by running my DHT with the least amount of influence. I lose that crisp leading edge and the microdynamics when their plates are loaded by transformers. Thus far the lone nude Vishay Z-foil resistor is as pure as I've heard. It lets through all the textural finely nuanced differences not only between various tube types but different brands of the same type, i.e. a 300B from Emission Labs, Genalex or ElectroHarmonix. With an OPT, they sound far too similar.

"The interesting thing is that in my volume control, using a transformer/autoformer had the opposite effect as you already know from comparing LIO's resistor volume to the AVC. Of course that's assuming a source output impedance low enough to drive the AVC properly. Luckily most are. In this DHT circuit, OPT as plate loads are killing off some musical soul. Why use 'em just because the measured output impedance can be even lower? If it doesn't sound as good, why should I care? But I still look forward to trying a few more transformers before I settle down." I told Vinnie that my preference had become to pursue tube flavour in the line-level domain even though that usually means 12AX7/6922 types which won't do what direct-heated triodes can. Hence my personally charged interest in his DHT project as a sneaky way of having the DHT cake and eating it.

Vinnie had an opinion on that, too. "I don't think I can go back to 12AX7/6922 since I've been playing with different DHT. Before this project, I was working on a 6SN7 preamp stage. It was actually up and running and sounding good yet it didn't touch the big triodes. It sounded rather lifeless by comparison even with some really great Siemens CCa E188CC and such. I feel very spoiled now. The other thing to reiterate is that running 2A3, 300B etc. in a linestage with minimalist topology really shows off what they sound like, more so than any SET amp I've tried them in. And I'm not talking about their typical power limitations. Even with the 'help' of an OPT, I just don't think they sound as good having to drive low-impedance loads like speakers. Meanwhile they really sing in the linestage but again, they're ultra-hyper-super sensitive to noise, hum and physical vibrations.

"That's exactly where I'm trying to come in with LIO. The ultimate would be battery power into the custom Belleson filament regulators but I think that'll be a later option. Customers would have to source their own battery and I'd do the proper charging circuit in the external supply. On something like this, a 2A3 would go for about 10 hours, a 300B more like 15. SV811/SV572 of course use 4A filaments. With an 8A demand, you'd 'only' get 6 hours from a 50Ah battery. They do make >100Ah batteries but I don't want to be the one shipping them. And not everyone wants to be thinking about batteries, charging and eventual replacement even though those Optima batteries last for years." As you can see, certain items from the prior page weren't yet set in stone when I started writing. I thought it more interesting to look over the designer's shoulder whilst final decisions were being made. Now I wanted to know how he'd pursue lowering the output impedance below the values given on page 1 without falling back on transformers or NFB. What other options were there? Relative to microphonics, I sent Vinnie links to photos on how Sasa Cokic of Trafomatic Audio and Thomas Mayer of Vinyl Savor attack DHT microphonics; with spring and viscoelastic tube-socket suspensions respectively. Allnic Audio even designed their own gel-filled tube sockets for the very same reasons. DHT can be real prima donnas. Most designers give up on using them for preamps. Thankfully for us, some are bewitched enough by their sonic promises to fight the good fight. They go to 'heroic' extremes to combat their annoying mechanical aspects. Call it a labour of love, not calm commercial reason and greed.

On further lowering output impedance, "I'm playing with a lower plate load resistor. That cuts down on gain and max output voltage a little but most people are fine with a 6dB-ish linestage and most power amps don't need much more than 2Vrms to achieve full output. Another idea is something like a John Broskie 'super triode' which is not the usual Mosfet buffer/voltage follower. Instead it's a crafty way of giving the triode full control over the Mosfet's current conduction. It puts the Mosfet in parallel with the tube, generates more output current, hence lower output impedance. Per the left image, using a single class A-biased FET in parallel and replacing the transformer with a load resistor whilst dealing only with low milliamps of current, it's a pure class A circuit with the triode boss. That'd be another way of doing it without feedback, transformer or buffer because it really isn't a buffer. Hopefully I won't go down that route and end up with sufficiently low output impedance the minimalist way instead." Relative to softly decoupled tube sockets, "they will be raised off the board via two standoffs per socket. Standoffs go to module, module goes to motherboard, motherboard goes to lower standoffs which go to the bottom panel. If you don't tap the tube or knock on the enclosure, it's fine. I've heard of cases where a DHT was so microphonic that if you clap your hands a few feet away, you'd hear it in the speaker. That is definitely not the case. But I am still playing with small O-rings in certain places to improve it. Plus, I haven't yet tried tube dampers and different isolation feet under the LIO. I'm curious also about using shorter standoffs and attaching them to those StillPoints interfaces right] to arrive at equal height."

Trafomatic Audio's Sava hybrid combines a 10Y DHT gain stage with Mosfet current buffer and suspends its tube sockets with these horizontal springs to combat microphony.

I next pointed Vinnie at this photo of an Art Audio Carissa amplifier with optional Pyrex chimneys surrounding the exposed 845. Isolating microphonic tubes from air turbulence—which happens the moment you play music and gets more intense as volumes increase and the distance between speakers and tubes shrinks—can be another wrinkle in the diva treatment. "I'll be testing this with a speaker placed/aimed right at the tubes from a different source/amp whilst running LIO's headphone output via a long extension cable to another room (to isolate my ears from the speaker) to a pair of isolating headphones. That will show me how much noise if any gets picked up from air vibrations; and what I might do to minimize it. Right now I'm getting good results from using small soft O-rings between standoffs and socket. I will still try putting a thin larger-diameter O-ring between tube base and socket as well and more between PCB and standoffs. The bigger issue is tube rush which happens even with theoretically perfect power supplies. It's the noise created inside the tube by the electron flow from cathode to plate and sometimes a little to the grid. The EH gold-grid 2A3 have very little of this. The Emission Labs mesh plates have more although when music plays, you don't notice it and the EML 2A3 Mesh sound incredibly good. All the 300B I've tried so far have this noise. You could tie their grid directly to ground and try every other external trick in the book without eliminating this noise. It's internal to the tube especially with direct-heated triodes. That's the price to pay for DHT sound. As long as you can't hear it in your listening chair when no music is playing, you certainly will not hear it when music is playing. If you do, you have a really noisy tube." 

This ties directly to the old debate between AC or DC heating a DHT; and in general, on negative feedback or not. One may measure better, one may sound better. It had my old boss Randy Smith of Mesa Engineering declare that he ought to call the high-feedback position on his adjustable knob 'measure' and the no-feedback position 'listen'. And it applies to all listeners who follow their ears over measurements. That includes tube fanciers whose amps rarely measure as good as transistor equivalents. Yet if they sound better, tech arguments don't make the listening seat proof.

Given the very real challenges of using direct-heated triodes in a modern LIO application, detractors could insist that the 'D' in DHT is really short for diva, that the following HT is not for hi but hollow or horrible tech. The continued relevance and popularity of tubes including such triodes simply shows that many disagree; or couldn't care less. If you were still foggy on what makes a direct-heated triode, recall that triodes consist of three elements: the grid, the cathode and the plate/anode. To emit electrons, the cathode requires energy. This comes from heat provided by a separate filament aka the heater - in most tubes that is. But in direct-heated tubes, the cathode is the filament. That double tasking is at the root of their potential self noise; and why whatever powers the filament/cathode becomes so critical. It's also why for the DHT module, the stock laptop-style switching PSU powering the ultracaps is replaced by a linear supply with split voltages - one for the ultracaps using the existing power inlet; and one dedicated to just the DHT filaments using a new second power inlet.