How splitting a passive preamp's output into independent tube-buffered Class D channels solves the home audiophile's fundamental problem
I bought a Douk Audio T4+ because it looked good on a desk. Vacuum tube glowing through a metal housing, a VU meter with a bouncing needle, a solid aluminum volume knob — all for under $100. The intended use was simple: DAC into the AUX input, volume knob for level control, headphone output. A compact tube-flavored headphone amp with some visual charm.
It sounded warmer than a direct connection. The VU meter danced. I was satisfied. Then I made the mistake of looking at the PCB.
Three distinct circuit blocks on one board. A TL071+JRC2068 phono preamp for the MM/MC turntable input — irrelevant for my use case. A dual NE5532 op-amp providing gain and output drive. And the JAN5725 tube stage sitting on a daughter board above the main PCB, almost certainly configured as a unity-gain cathode follower whose only job is to impose its nonlinear transfer function on the signal passing through it. The tube colors. The op-amp drives. The phono circuit sleeps.
That opened a rabbit hole. If the tube is adding consistent harmonic coloration at line level, and the op-amp is handling gain and impedance transformation... this isn't just a preamp. It's a tube buffer feeding a solid-state output stage. What happens if you put a Class D power amp after it? What if you use two of them — one per channel — as independent monoblock front-ends? What if you move volume control upstream with a passive preamp, so the tube always operates on the post-attenuation signal?
The answer turned out to be an architecture. Not designed top-down from theory, but discovered bottom-up by asking what each stage is actually doing and whether it could do it better in a different position.
To understand why this architecture works, you have to understand what it's solving.
Audio equipment follows a logarithmic value curve:
where
Tube amplifiers achieve their sound through power-domain mechanisms: output transformer saturation, power tube compression, Class A bias excursion. These need current. At typical indoor SPL — 65 to 80 dB at the listening position — a speaker with 85 dB sensitivity draws maybe 1 watt. A 40W tube amp is loafing at 2–5% capacity, deep in its linear region, well below the operating point where any of those mechanisms engage.
You paid for tube sound. You're hearing solid-state behavior with tube maintenance costs.
The conventional answer is a more expensive amplifier. The engineering answer is to recognize that tube coloration and power delivery are independent functions that can be physically separated.
The abstract architecture:
Source → DAC (fixed output) →
Passive Volume Control → per-channel split:
├─ L → Tube-Hybrid Buffer → Class D Power Amp → Speaker L
└─ R → Tube-Hybrid Buffer → Class D Power Amp → Speaker R
Three things define this topology.
Post-volume placement. The tube buffer sits after volume attenuation, not before. The tube's bias point is set by its internal circuit, not by the signal amplitude, so it generates its harmonic signature regardless of how much signal you feed it. The key advantage of post-volume placement is that the tube buffer absorbs the passive preamp's variable output impedance — which is the passive topology's main weakness — and re-presents the signal at low, stable impedance for the power amp downstream.
Per-channel isolation. Each channel gets its own tube buffer and power amp. No shared chassis, no crosstalk path. This is a monoblock architecture assembled from separate boxes.
Functional decomposition. The DAC converts. The passive preamp attenuates. The tube buffer colors and transforms impedance. The Class D amp delivers current. Each stage does one job.
So what actually happens inside the "tube buffer" box? The answer matters, because commodity tube buffers aren't pure tube circuits — and understanding what's in there sharpens the architecture rather than undermining it.
The T4+'s AUX specs tell the story: 0.775V input, 1.5V max output — roughly 6 dB of voltage gain. A pure tube cathode follower produces ≤ unity gain. Something is providing gain on the AUX → Line Out path, and the NE5532 is the obvious candidate.
If you're a chi-fi manufacturer building one PCB to serve multiple functions, you make the tube a unity-gain cathode follower and let the op-amp handle gain. The tube's job is harmonic coloration — it doesn't need gain to do that. A cathode follower is cheaper to implement (fewer passive components), more stable across tube-to-tube variation (gain is always ≤ 1), and lets the op-amp provide precise, repeatable gain with low output impedance. The likely signal path:
AUX IN → JAN5725 cathode follower (unity gain, harmonic coloration) →
NE5532 (gain + low-Z output drive) → Volume Pot → RCA Line Out
The tube provides harmonic coloration through its nonlinear transfer function:
where
The op-amp that follows provides gain, low output impedance, and linear current delivery. The tube does what tubes do best (nonlinear harmonic enrichment at the voltage domain). The op-amp does what op-amps do best (precise gain, clean drive). This is a three-stage pipeline: tube buffer → solid-state op-amp → Class D power amp. Each stage has one job.
Every interface in the chain needs to satisfy the bridging condition: load impedance much greater than source impedance. The voltage transfer at any interface is:
When
The passive preamp's vulnerability is its variable output impedance — worst case at moderate attenuation, where the pot's wiper divides the resistance roughly equally. The tube buffer's high input impedance absorbs this variation. The op-amp output stage then re-presents the signal at low impedance for the power amp.
Every passive interface in a signal chain can only preserve or degrade SNR — never improve it. The source's SNR is the ceiling. Two things erode it: signal loss from impedance mismatch, and noise picked up from the environment. This gives us a bounded optimization:
The generic impedance chain:
DAC output: Low Z → easily drives passive preamp
Passive preamp: Variable Z (rises with attenuation — the vulnerability)
Tube buffer in: High Z → absorbs the variable impedance
Tube buffer out: Low Z → drives power amp cleanly
Power amp in: High Z → bridging condition met
Power amp out: Very low Z → high damping factor into speaker
Both tube buffers share a single linear power supply. The reason is mathematical.
The tube's transfer function
Without clean power (switching supply):
Because
With clean power (linear supply):
The tube processes only music. All nonlinear products are harmonics of the program material — musically consonant, perceptually warm. The LPSU ensures
The Class D power amp should be transparent — a wire with gain. At 75 dB SPL with 85 dB sensitivity speakers, required power is about 1 watt. A Class D monoblock delivering 100W+ provides headroom exceeding 100:1, deep in its linear region, below 0.01% THD+N. All tonal character comes from the tube stage. The Class D preserves it.
Running as monoblocks gives you zero crosstalk, independent power supply loading, and thermal isolation per channel. You can confirm what the tube buffer contributes by bypassing it — connect the passive preamp directly to the power amp and listen. The difference is most audible at low SPL, where the buffered path maintains harmonic richness that the direct path loses. At higher volumes, the paths converge as the program material provides its own harmonic density. That convergence at high SPL and divergence at low SPL is the signature of the topology working as designed.
This topology moves tube coloration from the power domain to the voltage domain. That's a deliberate trade-off.
It won't replicate power tube compression — the harmonic signature of a 6L6 or EL34 under load is different from a small-signal tube at line level. It won't add output transformer coloration — saturation and frequency-dependent phase shift from iron are real parts of traditional tube sound. It won't match the transient authority of a high-current Class AB design into difficult loads, though at normal listening levels this is irrelevant.
If you specifically want the sound of an output stage working hard, you need an output stage working hard. This isn't that. This is consistent tube coloration at any SPL, including the levels where people actually listen most of the time.
| Role | Component | Function |
|---|---|---|
| Source | Any DAC at fixed output | Clean D/A conversion |
| Passive Preamp | Tenealay FV2 (ALPS RK27) | Volume attenuation, L/R split |
| Tube-Hybrid Buffer × 2 | Douk T4+ (JAN5725 + NE5532) | Harmonic injection, gain, impedance transform |
| Power Amp × 2 | Fosi Audio ZA3 (TPA3255) | Current delivery, monoblock isolation |
| Power Supply | LPSU (50VA, 12V) | Clean DC for both tube buffers |
DAC (fixed 2V RMS) →
FV2 Passive Preamp → RCA L/R split:
├─ L → Douk T4+ #1 → Fosi ZA3 #1 → Speaker L
└─ R → Douk T4+ #2 → Fosi ZA3 #2 → Speaker R
(shared LPSU for both T4+ units)
T4+ AUX gain at roughly 3 o'clock for optimal tube drive level. ZA3 gain sets the system's maximum output ceiling. The passive preamp handles primary volume. Total cost is $300–500 depending on source and supply choices.
After building this system, I discovered that VTV Amplifier — a small US-based builder — sells monoblocks at around $2,000 per chassis that pair a vacuum tube input buffer with Purifi's Eigentakt Class D module. Their tube buffer board has a tube stage, a swappable premium op-amp (Sparkos, Sonic Imagery, Weiss — options ranging from $13 to $219), feeding a Purifi Eigentakt Class D power stage. The three-stage pipeline is: tube buffer → solid-state op-amp → Class D.
That's the same architecture. JAN5725 cathode follower → NE5532 → TPA3255 is the same functional decomposition at a different price point. I didn't reverse-engineer VTV's product — I reasoned from the indoor listening problem, impedance bridging constraints, and functional decomposition, and landed in the same place.
The convergence is the validation. Two independent paths — one from product engineering (how do we build a premium tube-Class D monoblock?), one from first principles (how do we get tube coloration at indoor SPL?) — arrive at the same three-stage topology. The component tier differs: a $0.50 NE5532 where VTV offers a $50 Sparkos 2590, a TPA3255 where they use a Purifi Eigentakt. But the architecture is identical. The topology is the insight. The components are the budget knob.
This project demonstrates something that applies well beyond audio: the limiting factors are usually not what industry marketing suggests. Room acoustics matter more than cable differences. Source material quality caps system performance. The conventional upgrade path — better DAC, better amp, better speakers — ignores the topological question of where and how coloration enters the chain.
The distributed monoblock is an engineering response to a real constraint. Listeners who want tube character at indoor listening levels cannot get it from conventional tube amplification at their actual listening levels. By moving harmonic generation from the power domain to the voltage domain, and by splitting the signal into independent mono channels after volume attenuation, this architecture delivers consistent tube coloration at any SPL.
Reproducible with commodity components. A fraction of the cost of the commercial equivalent. The validation is straightforward. It's not a compromise — it's a topology optimized for how people actually listen.