Nauscopio Scipiorum: Acronos RS-1, single-ended KT150 EN

Traffic light — 15-second read

In short: the Acronos RS-1 is a stereo single-ended valve amplifier, 2× KT150, pure class A, with 25 W maximum and 18 W @ 0.01 % THD declared. EC86 NOS → ECC803S NOS → KT150 chain. Ultralinear, with NFB, IEC C14 RF/EMI filter, two threaded fuse holders and custom OPTs.

Clearly established: three house traits confirmed by the manufacturer (NFB in all Acronos, UL in the valve models, RF/EMI filter in all) and custom OPTs different from those of the M1, confirmed in a public video. Probable but unconfirmed: the chassis is inherited from the M1 (five identical parameters); DF ≈ 50 is an OPT design goal, according to the manufacturer's email; the power transformer may hum at 100 Hz due to rectifier peaks. Unknown: NFB depth and type in dB, %UL, real B+, real sensitivity (100 vs 1000 mV), verified mechanical hum, RF/EMI filter brand, fuse values.

Practical decision: with speakers below 90 dB/W/m, the RS-1 is tight. At 3.5 m or in large rooms, 25 W in pure class A leave less than 1 dB of headroom at 100 dB peak. If you expect real measured THD of 0.01 % or DF above 20 with a real load, measure before buying.

GETTING STARTED — WITH NO PRIOR KNOWLEDGE (Basic guide G1–G6)

GETTING STARTED — WITH NO PRIOR KNOWLEDGE

Six ideas are enough to follow the report without having studied electronics. Only A-level physics is needed: logarithms, power and impedance.

Basic guide

G1. What a single-ended amplifier is

An amplifier converts the weak signal from the player (~1 V) into a signal capable of driving a speaker. In a valve amplifier, the preamp increases voltage and the power stage supplies current. Everything is powered by a high DC voltage called B+.

Most domestic amplifiers use push-pull: two power valves work in opposition, and their currents partially cancel. The RS-1 is single-ended (SE): a single power valve per channel. There is no pair, no cancellation.

PUSH-PULL (most):                   SINGLE-ENDED (RS-1):

Valve A ------+                     Single valve ---+
              +-- OPT primary                     +-- OPT primary
Valve B ------+  (balanced                         ¦   (biased
                  current)                         ¦    current)
                                                   ↓
                                        DC current flows
                                        permanently through
                                        the primary → core
                                        with gap.

Three valves per channel: EC86 NOS (preamp), ECC803S NOS (driver + cathode follower) and KT150 (power). Six valves in total.

Advantages of SE. No crossover distortion, linear small-signal transfer, simple structure. Drawbacks. The quiescent current magnetises the core (requires a gap), efficiency is low and Zout without correction is high: it interacts with the speaker and colours the sound.

Pure class A: the valve conducts throughout the entire cycle, without cutting off. The most linear and least efficient: 20–30 % in real use with low THD.

Three house traits of Acronos: all use feedback, the valve models are ultralinear and all incorporate an RF/EMI filter on IEC C14. Starting point of the analysis, not a hypothesis.

G2. Which data is reliable and which is not

A serious report does not treat a measured figure, one published by the manufacturer and a suspicion the same way. Each statement carries an evidence label. The key distinction: DECLARED ≠ CONFIRMED.

Label	Meaning
CONFIRMED	Official datasheet, schematic or identifiable independent measurement.
MANUFACTURER	Design fact confirmed by the manufacturer.
MANUFACTURER-VIDEO	Manufacturer testimony in a public video (the RS-1's custom OPTs).
DECLARED	Published without independent verification.
OBSERVED	Visible in photographs (8 trimmers, fuse holders).
CALCULATED	Formula applied to known data (class A ceiling).
PROBABLE	Well-supported inference (Schaffner brand of the filter).
ND	Not determinable (NFB depth in dB).
MEASURED-EXT	External laboratory measurement (Hi-Fi News).

No statement rises above the category of its source evidence. The rule applies to the report's own citations too.

Why it matters. We know that the RS-1 uses NFB, is ultralinear, has an RF/EMI filter and has custom OPTs. What we do not know is the NFB depth, the %UL, the filter brand, the fuse values, nor whether the power transformer hums mechanically.

G3. How many watts are needed: decibels, distance and sensitivity

The required power depends on the speaker, the distance and the type of music. The dB is logarithmic: +10 dB ×10 power, +3 dB ×2.

For the reference speaker (89 dB/W/m, 3 m):

1 W at 1 m: 89 dB.
Distance loss: 20·log10(3) ≈ 9.54 dB.
1 W at 3 m: 79.46 dB.
95 dB peak (one speaker): ≈ 35.8 W.
100 dB peak (one speaker): ≈ 113 W.
Decorrelated stereo: 95 dB = 17.9 W/channel; 100 dB = 56.6 W/channel.

Manufacturer's warning. JC Valvular recommends a minimum of 90 dB sensitivity. With 89 dB speakers you are below.

In-room (typical +4 dB): 95 dB ≈ 7.13 W/channel; 100 dB ≈ 22.5 W/channel.

In this particular device. Orchestra peaks at 100 dB with 89 dB/W/m speakers at 3 m: the 22.5 W/channel over 25 W leave only 0.46 dB of margin. Less than half a dB of headroom.

▸ See the interactive SPL and required-power calculator

G4. Power supply and ripple: why SE is more demanding

After rectifying 50 Hz, the ripple appears at 100 Hz. In PP the two valves receive the ripple in phase and the output transformer rejects it. In SE there is no possible cancellation: the ripple goes straight into the quiescent current, is amplified and appears at the speaker. That is why SE amplifiers use several filtering stages (LC or RC in series).

The B+ ripple with a single capacitor is:

ΔV ≈ I / (f · C)

For I = 130 mA and f = 100 Hz, a bank of 220 µF leaves 5.9 V of ripple (1.31 % over B+ 450 V). With 2200 µF it drops to 0.59 V (0.13 %), equivalent to −57.7 dB at the output. To leave it below −80 dB, additional LC or RC stages are needed.

The input RF/EMI filter is not a substitute for B+ filtering. They are two distinct functions: the IEC C14 filter works against conducted high-frequency interference (150 kHz–30 MHz); the HV bank works against low-frequency ripple (100 Hz). An SE needs both.

Consequence. DF ≈ 50 can only be honest with a well-filtered B+. The RF/EMI filter is an indication of construction care, not proof. The mathematical development of the ripple is in §2.3.

G5. Impedance, current and damping factor

A "6 Ω" speaker does not always measure 6 Ω: its impedance changes with frequency. Amplifier and speaker form a voltage divider: if Zout is very low, the response is flat; if it is high, it follows the speaker curve and colours the sound. The parameter that summarises it is the damping factor:

DF = Z_load / Z_out

The RS-1 paradox. Et Sona declares DF ≈ 50 (Zout ≈ 0.16 Ω over 8 Ω). Precedents with the same valve and topology stay an order of magnitude below. With NFB confirmed, the question is not whether there is a loop, but whether the available depth reaches.

DF is not a single figure. It depends on 12 factors (§4.5) and varies with frequency and load. A DF of 50 measured at 1 kHz with fixed resistors does not describe behaviour at 20 Hz over a real speaker.

G6. The four key ideas

The essentials

The RS-1 is single-ended (one KT150 per channel), ultralinear, with NFB, with an IEC C14 RF/EMI filter and with two threaded fuse holders. All by house policy. The output transformers are custom-made, confirmed by the manufacturer in a video.
It declares two THD points (18 W @ 0.01 % and 25 W @ 0.5 %) and a DF ≈ 50 published by Et Sona. Precedents with SE KT150 are far below.
With 89 dB speakers at 3 m, the 25 W are at the limit. Manufacturer recommends 90 dB minimum.
The class A ceiling constrains B+: 25 W with 130 mA bias require B+ = 385 V theoretical, ~478 V practical. The available NFB is not enough either for DF 50 or, much less, for THD 0.01 %. The power transformer may hum at 100 Hz.

PART I — ACRONOS RS-1 REPORT

From the declared figure to the verifiable hypothesis.

0. Executive summary

Acronos RS-1, general view of the amplifier with its two KT150 valves visible. First image in the post, used as og:image / twitter:image. — Acronos RS-1 — general view. Image: JC Valvular / Et Sona. *Click to enlarge.*

20-second read

What it is. Stereo single-ended, 2× KT150, pure class A, 25 W max., 18 W @ 0.01 % THD. EC86 NOS → ECC803S NOS → KT150 chain. Ultralinear, with NFB, IEC C14 RF/EMI filter, two threaded fuse holders and custom OPTs (MANUFACTURER-VIDEO).

Central finding. Under the §4.3 model, the 0.01 % THD at 18 W demands 49–54 dB of reduction; the declared DF of ~50 corresponds to 31.7 dB as an impedance ratio. These are conditional figures, not physical equivalences.

Loop calculation. Available: 21–29 dB with 100 mV sensitivity. Below the threshold: under the model, neither H1 (0.1–0.45 %) nor H2 with cathode feedback (0.025–0.089 %) reaches 0.01 %.

Class A ceiling. 25 W with 130 mA bias require B+ = 384.6 V theoretical; ~478 V practical.

General description

The Acronos RS-1 is a single-ended pure class A valve amplifier, with one KT150 per channel, 25 W maximum and 18 W @ 0.01 % THD declared. Chain: EC86 NOS → ECC803S NOS → KT150. Evolution of the Acronos M1 (KT120, 20 W). Manufacturer: JC Valvular (Juan Calatayud), Málaga.

Three house premises, confirmed by the manufacturer: all Acronos use feedback, the valve models are ultralinear and all incorporate an RF/EMI filter on IEC C14. Furthermore, in a public video, the manufacturer confirms that the RS-1 output transformers are custom-made, commissioned with specifications different from those of the M1 MANUFACTURER-VIDEO.

Where the declared figures come from. The commercial data have been verified against Et Sona. Full coincidence, except for three notation corrections: "100Khm" → "100 kΩ", "ECC86" → "EC86" and "ECC803 nos" → "ECC803S NOS". The DF ≈ 50 is the only critical data point that comes exclusively from Et Sona.

Most sensitive variable of the report: sensitivity. All the arithmetic of §4 depends on whether the declared 100 mV is for a pure power stage or for an integrated with the potentiometer at maximum. With 100 mV, the closed-loop gain is 43.01 dB and the available loop is 21–29 dB. With 1000 mV, the closed-loop gain drops to 23.01 dB and the loop rises to 41–49 dB: it is the only scenario where THD 0.01 % revives without cathode feedback. This data point is not confirmed: ND. See §4.3, §4.4 and Block 1 of Appendix A.

Provisional conclusion

The RS-1 sits between commercial KT150 SE amplifiers (DF ≈ 7–9) and declared figures of another category. We know there is NFB, that the OPTs are custom-made and that the chassis is probably inherited from the M1. But the loop gain calculation (§4.3) leaves a narrow or negative margin. The outcome depends on a data point the manufacturer can provide in one sentence: how much NFB, in dB, and of what type, does the RS-1 have? And on an acoustic verification: does the power transformer hum?

Ethical closing. This report does not measure an RS-1. It models what NFB would be needed for its declared figures to be coherent. All critical figures remain DECLARED until a specific unit is measured with a two-load protocol and a THD vs. power curve.

1. Scope, method and limitations

Key idea of chapter 1

This chapter does not assess how the RS-1 sounds. It explains what we know about it, where each data point comes from and how reliable it is. It is an exercise in reverse engineering without a schematic from the manufacturer. That is why every statement carries a label.

Starting limitation. We do not have our own measurement of the RS-1 or its schematic. Everything that follows is inference from documentation, manufacturer testimony and commercial precedents.

1.1 Sources of evidence

Direct contact with JC Valvular. It confirms three house policies that apply to all Acronos: 1) All use negative feedback, NFB. 2) All valve models are ultralinear, UL. 3) All carry an RF/EMI mains filter on the IEC C14 connector. Brand unspecified, probably Schaffner. MANUFACTURER
Public video of Juan Calatayud. He declares that the RS-1 transformers are different from those of the M1 and that he had to commission them custom-made with other specifications. This data point is key: it resolves the debate about whether the same transformer is reused. MANUFACTURER-VIDEO
RS-1 commercial spec sheet. 25 W maximum, 18 W @ 0.01 % THD, 130 mA bias, 195 W consumption, 20 Hz – 40 kHz response, 100 kΩ input impedance, 100 mV sensitivity, 14 kg weight, 45×26×11 cm dimensions. Valves: 1× KT150 + 1× ECC803S + 1× EC86 per channel. The 100 mV sensitivity is ND: we do not know whether it is for a pure power stage or for an integrated with the potentiometer at maximum.
Et Sona review. It confirms lineage: "On the website you will find the M1 with KT120, but this time we have the RS-1 which uses KT150s". It publishes DF ≈ 50 and describes a 1.5 mm carbon steel chassis, solid wood panels and 22 mm feet.
Official M1 spec sheet. From jcvalvular.com, January 2023. 20 W max., 18 W @ 0.01 % at 1 kHz, 100 kΩ, 100 mV, 14 kg, 45×26×11 cm, KT120 in class A ultralinear. Same 5 parameters as the RS-1. DECLARED
Measured precedents. Icon Audio ST30SE (Hi-Fi News 2019.11): 26 kg net / 28 kg packed, 28 W in UL, Zout 0.85–1.17 Ω. Dared Saturn Signature: 25 W SE KT150, 18 kg. Fiio EA13: push-pull class A 4× EL34, 17.7 kg, with documented mechanical hum at 100 Hz.
Datasheets. KT150 Tung-Sol: 70 W maximum plate dissipation, heater 1.75–2.0 A. Example point in pentode class A1: 400 V, 150–180 mA, 20 W with 14 % THD. EC86 Philips: VHF triode, µ = 68, gm = 14 mA/V, ra ≈ 4.86 kΩ.
Photographs. Interior (two boards, 8 trimmers, point-to-point) and rear panel (IEC, fuses, RCA, terminals). OBSERVED
Reference system. Crossover A v2.4 MR-A-01 (6 Ω nom / Zmin 4.9 Ω / 89 dB/W/m), listening at 3 m.

1.2 Catalogue and M1 → RS-1 lineage

Catalogue context — JC Valvular range

Model	Topology	Power	Category
Acronos M1	SE Class A with KT120, ultralinear	20 W	Genealogical predecessor of the RS-1
Acronos RS-1	SE Class A with KT150, ultralinear	25 W max. / 18 W @ 0.01 %	Subject of this report
Acronos P52	Push-Pull Ultralinear (4× 6550)	40 W nominal	Dedicated report
Acronos M2	Hybrid	60 W	High native DF without relying on the OPT

A single-ended (SE) uses a single valve per channel. It is pure class A, always conducting. A push-pull (PP) uses two valves that alternate. It does not need a gap and weighs less per watt.

House policy. Three common traits: feedback in all, UL in the valve models, IEC C14 RF/EMI filter in all. The M2 hybrid achieves high DF without relying on the OPT, but also uses NFB.

Key idea of the lineage

M1 and RS-1 share five exact parameters. That points to an inherited mechanical platform. The manufacturer's video clarifies the key point: the RS-1 OPTs are custom-made, different from those of the M1. The fact that total weight coincides suggests that the redesign has not been massive, but it does not prove it. The full physical analysis is in §2.2.

The five shared parameters

Parameter	M1 (KT120)	RS-1 (KT150)
Total weight	14 kg	14 kg
Dimensions	45 × 26 × 11 cm	45 × 26 × 11 cm
THD at 1 kHz	18 W @ 0.01 %	18 W @ 0.01 %
Input impedance	100 kΩ	100 kΩ
Sensitivity	100 mV	100 mV

Five parameters nailed. To the millimetre and to the decimal. The maximum power does change: 20 W in the M1, 25 W in the RS-1. The power valve also changes: KT120 vs. KT150. But the distortion point —the most extraordinary figure in the report— does not move.

The three lineage scenarios

Scenario	Chassis	OPT	Status
A — Full same platform	Inherited from M1	Inherited from M1	DISCARDED by the video: the OPTs are custom-made.
B — Inherited chassis, custom OPTs	Inherited from M1	Redesigned, different from M1	MANUFACTURER-VIDEO for the fact of the redesign.
C — Modified chassis	Slightly adapted for KT150	New	PROBABLE but marginal.

What does "custom-made" mean in an SE?

It does not automatically mean "bigger". It means that electrical parameters are changed to adapt the KT150:

B+: plate supply voltage.
Ibias: 130 mA that the primary must support without saturating.
Raa: impedance seen by the valve, typically 3 kΩ in KT150.
Gap: core air gap. It sets how much DC current it withstands.
Lp: primary inductance to reach 20 Hz.

If the total weight remains at 14 kg and the external volume does not change, what is probable is that the redesign is moderate in mass. There is no obvious room for much more copper or iron. What changes more easily is N (number of turns) and the gap.

Consequence for the 18 W @ 0.01 %. With custom OPTs, the logical thing would be a different THD point. The fact that it remains nailed at 18 W @ 0.01 % suggests that it is a family design goal, not a new measurement of the RS-1. PROBABLE.

Final lineage label. The video confirms that the OPTs are different and custom-made MANUFACTURER-VIDEO. The inherited chassis is PROBABLE, not explicitly confirmed in the video. The magnitude of the redesign is ND: the coincidence of weight and dimensions suggests a moderate change, but does not prove it.

Summary of chapter 1 in 3 points

We work without a schematic, with spec sheets, photos, precedents and video testimony from the manufacturer. Every statement carries an evidence label.
M1 and RS-1 share 5 identical parameters (weight, dimensions, THD at 1 kHz, 100 kΩ, 100 mV), which indicates the same mechanical platform. The inherited chassis is PROBABLE.
Calatayud's video confirms custom OPTs, different from the M1. The magnitude of the redesign is ND: the identical weight suggests a moderate change, but does not prove it.

2. Architecture and power supply

Key idea of chapter 2

Architecture is the set of functional blocks and how they are connected. In the RS-1 there are three: power supply, signal chain and output. Topology SE ultralinear with NFB and input RF/EMI filter, by house policy. The OPTs are custom-made. The diagram is a hypothetical reconstruction, not a manufacturer's schematic. This chapter examines the diagram, the chassis, the transformers, the power supply and the risk of mechanical hum from the transformer.

2.1 Block diagram

MAINS 230 V
 ¦
 +-- [IEC C14 · RF/EMI filter --- MANUFACTURER] --- brand/model ND · Schaffner?
 ¦      ¦
 ¦      +-- [2× threaded fuse holders rear panel --- OBSERVED]
 ¦             (live+neutral or mains+B+? ND · values ND)
 ¦
 +-- Power transformer --- OBSERVED · hum? ND
       ¦
       ↓
  SOLID-STATE RECTIFICATION (probable) → LC/RC filtering (topology ND)
       ¦
       ↓
 B+ (voltage ND) ------→ KT150 channel L (UL)
                     +--→ KT150 channel R (UL)
 SIGNAL CHAIN PER CHANNEL (DECLARED by Et Sona)
   [RCA in ×3 + Muting] → [EC86 NOS · preamp]
                        → [ECC803S NOS · driver + cathode follower]
                        → [KT150 · UL power]
                        → [Custom SE output transformer · gap] → [4 Ω / 8 Ω speaker]
                        +-- NFB loop from secondary to driver (MANUFACTURER)

  Ultralinear mode --------------- MANUFACTURER
  NFB (exists) -------------------- MANUFACTURER
  NFB depth in dB ----------------- ND (central question)
  %UL ----------------------------- ND
  Bias 130 mA --------------------- DECLARED
  Sensitivity 100 mV -------------- DECLARED · nature ND
  IEC C14 RF/EMI filter ----------- MANUFACTURER
  Threaded fuse holders ----------- OBSERVED · function and values ND
  Custom OPTs, different from M1 --- MANUFACTURER-VIDEO
  B+ = 385 V for 25 W class A ----- CALCULATED (§3.2)
  8 trimmers --- OBSERVED · individual function ND
  Power transformer hum ----------- ND · PROBABLE risk

Acronos RS-1 printed circuit boards: signal board (left) and power supply board (right). — Figure 1 — Internal boards. Left: signal board with 4 trimmers. Right: power board with another 4 trimmers and filter bank. Image: JC Valvular. OBSERVED

2.2 Chassis, weight and transformers

Key idea of §2.2

Total weight does not say whether an amplifier sounds better. But it says a lot about the size of the transformers. The comparison between the three commercial KT150 SEs is what matters: Icon 26 kg, Dared 18 kg, RS-1 14 kg. Same valve, same physics, 12 kg difference. This block also examines the physical reason why an SE weighs more per watt than a PP: the OPT air gap.

Why an SE weighs more per watt than a PP

In SE the quiescent current always flows through the primary of the output transformer. That DC magnetises the core and would saturate it. To avoid it, a cut is made: the air gap or gap.

The gap lowers the effective permeability from 10,000 to about 200. The formula governing inductance is:

L_p ∝ N² · μ_eff · A / l

If μ_eff drops 50 times, to maintain the same L_p, N must be multiplied by 7. That is 7 times more copper and a much larger window. In practice, an SE OPT is 3 to 5 times bulkier than a PP one.

The minimum inductance to avoid losing bass is:

L_p ≈ R_aa / (2·π·f_c)

With R_aa = 3 kΩ and f_c = 20 Hz → L_p ≈ 24 H minimum. For flat response at 20 Hz, 60–70 H are needed.

Comparative table with kg/W, volume and density

Model	Topology	Power	Weight	Volume	Density	kg/W	Evidence
Icon ST30SE	SE KT150, UL	28 W	26 kg net	ND	ND	0.93	MEASURED-EXT
Acronos M1	SE KT120, UL	20 W	14 kg	12.87 L	1.09 g/cm³	0.70	DECLARED
Dared Saturn	SE KT150, triode	25 W	18 kg	ND	ND	0.72	MEASURED-EXT
Acronos RS-1	SE KT150, UL	25 W	14 kg	12.87 L	1.09 g/cm³	0.56	DECLARED
Cayin Jazz 90	PP AB1 KT88	50 W	27.5 kg	34.21 L	0.80 g/cm³	0.55	CONFIRMED
Fiio EA13	PP A EL34	18 W	17.7 kg	21.10 L	0.84 g/cm³	0.98	MEASURED-EXT

Documentary asymmetry. The 14 kg of the RS-1 and of the M1 are DECLARED: they come from the manufacturer's spec sheet without independent verification. The 26/28 kg of the Icon, the 18 kg of the Dared and the 17.7 kg of the Fiio are MEASURED-EXT. The lightest data point in the table is also the least robust from a documentary standpoint. The density of the RS-1 (1.09 g/cm³) is the highest in the group: it is not a light amplifier, it is a small but compact one.

Acronos Commercial SE KT150s PP reference

What weight says and what it does NOT say

The coincidence of weight and dimensions between M1 and RS-1 is compatible with a moderate redesign in mass: same mechanical platform, different electrical specifications (gap, N, Raa). But it does not prove it: the same weight could hide a different OPT with a very different internal mass distribution. Without measurements, comparable photographs or core and winding data, the exact magnitude of the redesign is ND.

What is ruled out is a massive redesign. If the new OPT had much more iron or copper, the total weight would have risen from 14 to 16–18 kg. It has not. The 12 kg difference with the Icon ST30SE admits three explanations, all compatible with the data: (a) smaller RS-1 OPTs; (b) lighter chassis (Et Sona documents 1.5 mm steel + solid wood + 22 mm feet); (c) some component not included in the declared weight. It is resolved by measuring L_p and the 20 Hz response at full power.

Weight does not say which one sounds better. A heavier amplifier is not better for being heavier. Nor does it say whether it is undersized: the right question is not "why does the RS-1 weigh so little?" but "is the light one undersized or the heavy one oversized?". The Jazz 90 case (0.55 kg/W, almost identical to the RS-1) illustrates that the ratio is not comparable between devices with different integration: the Jazz carries DAC, Bluetooth, phono and protections; the RS-1 is a minimalist amplifier with three RCA inputs.

Terminological precision. The ratio 195 W / 14 kg = 13.9 W/kg is a declared power / total mass ratio. It is not thermal density (W per unit volume or area) and does not allow predicting hum. It is cited here as a descriptive indicator; its acoustic interpretation is developed in §2.3.

Summary of §2.2 in 3 points

An SE weighs 3–5× more per watt than a PP because of the OPT air gap. The RS-1 declares 14 kg, 12 kg less than the Icon ST30SE with the same valve.
The coincidence of weight with the M1 suggests a moderate redesign, but does not prove it. The exact magnitude is ND. The volumetric density (1.09 g/cm³) is the highest in the comparison group.
Power/mass ratio 13.9 W/kg: descriptive, not thermal density. The 12 kg difference with the Icon is resolved by measuring L_p and 20 Hz response at full power.

2.3 Power supply and hum

Key idea of §2.3

Everything behind the IEC C14 and everything that can vibrate there. The RF/EMI filter works against conducted high-frequency interference (150 kHz–30 MHz). The HV bank of the B+ works against low-frequency ripple (100 Hz). They are two distinct functions and an SE needs both. Added: the power transformer, with pure class A as the regime of maximum demand, can vibrate mechanically at 100 Hz from two motors —Lorentz and magnetostriction—. It is the third finding of the report.

B+ ripple translated to dB at the output

In an SE, the B+ ripple is not cancelled. With ΔV_ripple ≈ I / (f · C), for I = 130 mA and f = 100 Hz:

HV bank capacitance	ΔV_ripple	Ratio to B+ 450 V	Theoretical output attenuation
220 µF	5.9 V	1.31 %	−37.6 dB
470 µF	2.8 V	0.62 %	−44.2 dB
1000 µF	1.3 V	0.29 %	−50.8 dB
2200 µF	0.59 V	0.13 %	−57.7 dB

For the ripple to remain below −80 dB at the output with a typical PSRR of 30–40 dB in an SE, additional LC or RC stages adding 30–40 dB more filtering are needed. A single capacitor is not enough.

Consequence. Without a schematic of the HV bank, it cannot be verified whether the RS-1 reaches that −80 dB of ripple. It is an ND.

RF/EMI filter, fuse holders and safety

The RS-1 has two threaded fuse holders on the rear panel, old style: screw or bayonet caps, separate from the IEC C14.

Acronos RS-1 rear panel: IEC C14 with filter, two threaded fuse holders, RCA and speaker terminals. — Figure 2 — Rear panel. Image: JC Valvular. OBSERVED

Why two fuses?

Configuration	Implication	Probability
Live and neutral on mains	Double pole. Extra safety if the plug is reversed.	More probable
Mains + B+ HT	One slow (T1.6A–T2A), one fast (F315–500mA).	Probable if the values are different
One per channel	SE with two independent transformers.	Unlikely

Double fuse + RF/EMI filter on IEC C14 is typical of a product intended for sale: safety regulation compliance. An indication of a commercial product, not a home workshop.

Pending verification. Without a photo of the caps with the engraved value, it is ND whether they are T1.6A, T2A, F500mA or others. The brand and model of the RF/EMI filter are also ND (probably Schaffner).

Compatibility with the reference speaker

Scenario	Zout	Δ level Zmin ↔ Znom	DF @ 6 Ω
If DF 50 were real	0.16 Ω	≈ 0.28 dB	37.5
Icon ST30SE (UL)	1.0 Ω	≈ 1.6 dB	6.0
Icon ST30SE (triode)	0.5 Ω	≈ 0.85 dB	12.0
No NFB (typical SET)	2.5–4 Ω	≈ 3.5–5 dB	1.5–2.4

With 0.28 dB the response is essentially flat over this speaker. With 1.6 dB (Icon UL) there is already audible coloration in the woofer crossover region. With 3.5–5 dB (SET without correction), the response literally follows the speaker's impedance curve. You can explore the effect with the Zout calculator.

Mechanical hum from the power transformer

Pure class A subjects the power transformer to a constant and maximum load. The current peaks from the rectifier can make the windings vibrate at 100 Hz due to two motors: Lorentz force on the copper and magnetostriction of the core. The chassis can amplify it like a sound box. It is not DC on the mains: it is load stress. The Fiio EA13 documents the mechanism. The RS-1 uses the same 1.5 mm steel chassis with solid wood panels as its predecessor M1. That chassis was designed for a KT120 at 20 W. With the KT150 at 25 W, the transformer works harder on the same anti-vibration mounting.

Cause-effect table

Cause	Frequency	DC Blocker?	Mitigation
DC on the mains	50 Hz (hum)	Yes	DC blocker
Rectifier peaks	100 Hz (buzz)	No	More HV capacitance, NTC
Core magnetostriction	100 Hz	No	Core with glued joints
Lorentz force	100 Hz	No	Fixed coils, silentblocks
Chassis-transformer coupling	100 Hz amplified	No	Wood panels, silentblocks

Precedent: Fiio EA13

Push-pull class A, 17.7 kg, 21.1 L. Users report audible buzz at 3–4 m. Fiio tried to prevent it with OPTs encapsulated in resin and sand, a unibody aluminium chassis and anti-vibration feet. It still hums: what vibrates is the power transformer. The RS-1 has a power/mass ratio of the same order (13.9 W/kg vs 12.4–13.0 W/kg for the EA13) and less mass and volume. The verification is binary and without instrumentation: listen at 10 cm after 1 h hot with the Spectroid app.

Absence of reports is not proof of absence. Label: PROBABLE for the RS-1. It is the third finding of the report, together with the class A ceiling and the loop gain calculation.

Summary of §2.3 in 3 points

Supply: RF/EMI filter (150 kHz–30 MHz) + HV bank of the B+ (100 Hz) are two distinct functions. Ripple: a single capacitor gives −37 to −51 dB; for −80 dB, LC/RC stages are needed. ND.
Two threaded fuse holders + IEC C14 filter indicate a commercial product, not a home workshop. Values and model: ND.
Hum: two motors at 100 Hz (Lorentz + magnetostriction). EA13 precedent with less margin. Acoustic and binary verification. Risk PROBABLE.

3. Chain and power stage

Key idea of chapter 3

Three valves per channel: EC86 (preamp), ECC803S (driver + cathode follower), KT150 (UL power). This chapter covers the chain that feeds the KT150 and everything around the KT150: operating point, pure class A, bias, dissipation, class A ceiling and ultralinear operation. Two dense sub-sections.

3.1 EC86 → ECC803S → KT150 chain

Total: six valves (two KT150, two ECC803S, two EC86).

Stage	Valve	Type	Function
Preamp	EC86 NOS	UHF triode	Initial amplification. NOS for linearity and low noise. µ = 68.
Driver	ECC803S NOS	Double triode	One triode drives; the other, cathode follower.
Power	KT150	Beam tetrode	Pure class A ultralinear.

EC86 data

Single triode for UHF. Data confirmed from the Philips/NXP datasheet at 175 V and 12 mA: µ = 68, gm = 14 mA/V, ra ≈ 4.86 kΩ CONFIRMED. Max plate: 220 V. Equivalent: 6CM4.

ECC803S data

Long-plate premium version of the ECC83 / 12AX7 family. One triode amplifies voltage, the other acts as a cathode follower (gain ~1, very low Zout) to isolate from the KT150 input capacitance (Miller effect).

The cathode follower improves HF, but is not responsible for DF ≈ 50. DF is determined in the output stage (KT150 + OPT + NFB), not in the driver.

Gain per stage

With typical ideal loads, A = µ · R_L / (R_L + r_a):

Stage	µ	r_a	R_L	Av (dB)
EC86	68	4.86 kΩ	47 kΩ	35.8
ECC803S	100	62 kΩ	100 kΩ	35.8
KT150 (UL)	—	3.5 kΩ	1.5 kΩ	25.5

Total at the plate: 61.6 × 61.7 × 18.9 ≈ 71,899 ≈ 97.1 dB.

3.2 KT150: operating point, UL and class A ceiling

The KT150 is octal, successor to the KT120 and KT88, with 70 W maximum plate dissipation according to Tung-Sol CONFIRMED. Heater: 1.75–2.0 A.

Feature	KT150	KT120
Max plate dissipation	70 W	60 W
Max plate voltage	850 V pent. / 650 V triode	700 V
Heater	1.75–2.0 A	~1.7–1.95 A

The 130 mA bias is consistent with pure class A. At B+ = 400–450 V: 52–58.5 W per valve, within the 70 W maximum with a margin of 16–26 %. The declared consumption of 195 W breaks down into heaters (~32 W), plates (~104 W), drivers (~10 W) and losses (20–30 W). That gives 166–176 W: the declared figure is 10–15 % higher, plausible as a catalogue figure rounded upwards.

Pure class A: implications

Practical efficiency: 20–30 % with low THD and real music.
Theoretical efficiency limit: ~48 % (25 / (400 × 0.13)).
High linearity: no crossover distortion.
Distortion dominated by even harmonics.
Soft clipping: progressively compresses.

Dissipation per valve according to B+

B+	Estimated P_plate	% of maximum (70 W)
425 V	55.25 W	78.9 %
450 V	58.50 W	83.6 %
478 V	62.14 W	88.8 %
500 V	65.00 W	92.9 %

Above 83 % of maximum throughout the 450–500 V range. Indicative service life of 2000–4000 h before emission drops noticeably (ND specific to the RS-1). Expected temperature: OPT 45–55 °C, transformer 55–70 °C. It implies not enclosing it, leaving 5–10 cm free and not stacking it. Without published thermography, it is PROBABLE.

Bias: the 8 trimmers and thermal drift

The 130 mA bias sustains the entire class A ceiling. If it drifts, P_max changes and with it the 25 W figure. In the photos, 8 blue and black Bourns trimmers can be seen:

Trimmer	Board	Hypothesis
1–2 (blue Bourns)	Signal	Bias per KT150 (one per channel)
3–4 (blue Bourns)	Signal	EC86 operating point or ECC803S driver balance
5–8 (2 blue + 2 black)	Power	NFB loop compensation (pole, zero), per channel

The whole map is ND until the board is traced with a multimeter. Indirect measurement without a schematic: measure the drop across the cathode resistor (typically 10 Ω), I = V / R. For 130 mA and R_k = 10 Ω: 1.30 V expected.

Safety. B+ 450–500 V is lethal. Switch off, unplug, wait 10 min, discharge the capacitors with 100 kΩ / 5 W down to < 20 V before touching. With B+ ≈ 450 V, each KT150 dissipates ~58.5 W (84 %). At 500 V, 65 W (93 %). No manufacturer data on drift at 500 h: ND.

Class A ceiling

In a class A SE, the maximum theoretical power is limited by B+ and quiescent current:

P_{max,class A} = B+ · I_bias / 2

Table A — Ideal theoretical limit (without primary Rdc or Vknee)

B+	P_max	Margin over 25 W	Achievable in practice?
384.6 V	25.00 W	0 W — exact limit	No: losses lower it
425 V	27.62 W	+2.62 W (10.5 %)	No: Vknee and Rdc consume the margin
450 V	29.25 W	+4.25 W (17 %)	Tight: with low losses
500 V	32.50 W	+7.50 W (30 %)	Yes

Note. The formula B+·I_bias/2 is ideal: it does not include the primary winding resistance (Rdc) or the KT150 knee voltage (Vknee). That is why 384.6 V is not achievable in a real circuit. The practical minimum is calculated in table B.

Table B — Practical minimum with real losses

B+_{min practical} ≈ (I_bias · R_aa + V_knee) / η ≈ (0.13 · 3000 + 40) / 0.9 ≈ 478 V

B+	Interpretation	Compatible with 25 W class A?
385 V	Theoretical limit without losses	No
430 V	With V_knee, without ohmic losses	No
478 V	Practical minimum	Tight
500 V	Comfortable zone	Yes

450–500 V is not optimistic, it is the minimum practical range. If the real B+ were below ~430 V, the 25 W figure would include excursion outside class A.

Operating mode: ultralinear

Manufacturer confirms UL in all valve Acronos. In UL, the screen grid is connected to an intermediate tap of the OPT primary. The %UL decides how much it resembles a triode or a pentode. Typical values 20–50 %.

Mode	Expected power	Status
Pure pentode SE	~20 W @ THD = 14 %	DISCARDED
Pure triode SE	12–18 W	DISCARDED
Ultralinear SE + NFB	20–30 W with low THD	CONFIRMED

Double reading. UL reduces THD compared to pentode by a factor of 2–4×, not 4.7×. With NFB confirmed, the gap is closed. The question: how much NFB? See §4.3.

Summary of chapter 3 in 3 points

Chain: EC86 (µ = 68) → ECC803S (dual function) → KT150 (UL power). The cathode follower does not explain DF ≈ 50. Gross gain at the plate ~97 dB.
KT150: 70 W max. 130 mA bias, 8 trimmers (map ND), dissipation 79–93 % depending on B+. No drift data at 500 h.
Class A ceiling: B+ = 384.6 V theoretical (without losses), ~478 V practical. UL confirmed, %UL ND. Service life 2000–4000 h. Expected temperature OPT 45–55 °C, transformer 55–70 °C.

4. The central question: how much NFB?

Traceability note on DF ≈ 50. The DF ≈ 50 figure that circulates for the Acronos RS-1 comes from the Et Sona review, not from an official JC Valvular spec sheet. The manufacturer does not publicly claim that figure. Its measurement methodology is not documented: with the available documentation it is not possible to distinguish whether Et Sona measured a physical unit or reproduced figures provided by the manufacturer. In this report it is recorded as DECLARED and treated as pending independent verification. The developments in §4.3, §4.4 and §5 start from that figure, but describe what would be needed for it to be coherent, not what has been measured. The measurements that would close the matter are in §6.1 (priorities 1, 2, 3, 6 and 7).

Key idea of chapter 4

Manufacturer confirms NFB in all Acronos. What is not known is how much. NFB lowers Zout (raises DF), lowers THD and widens bandwidth. It is the candidate mechanism to explain the two extraordinary figures. The question: is the available NFB enough for DF ≈ 50 and THD 0.01 % at 18 W?

Methodological warning — unmeasured variables. This chapter works under explicit hypotheses. The real gains of each stage, the open-loop distortion, the reflected impedance of the OPT and the exact operating point are not measured in the RS-1. The model starts from: (a) open-loop THD in UL between 3 % and 5 %; (b) gross gain between 92 and 100 dB; (c) OPT transformation ratio n = 25. The results describe what happens if those inputs are accepted; they do not constitute a measurement or a proof of impossibility.

4.1 The two extraordinary figures: THD is the one that breaks the mould

Source	Power	THD	Configuration
Tung-Sol KT150	20 W	= 14 %	Pure pentode, class A1, no NFB
Acronos RS-1 (declared)	18 W	0.01 %	Ultralinear + NFB

Three orders of magnitude. It is not a contradiction: it is what UL + NFB contributes.

How much NFB the THD demands

3 % / 0.01 % = 300 → 20·log10(300) ≈ 49.5 dB

5 % / 0.01 % = 500 → 20·log10(500) ≈ 54.0 dB

49–54 dB of loop gain for THD alone. The declared DF of ~50 corresponds, in the §4.3 model, to 31.7 dB. These are two different quantities —one is loop gain, the other an impedance ratio— and should not be compared as if they were the same measurement.

IEC 60268-3. The standard requires specifying the measurement bandwidth and the filter (typically 20 Hz – 20 kHz, with or without an 80 kHz filter for high harmonics). A "0.01 %" without that information is not comparable to another "0.01 %". The RS-1's declared data point is DECLARED and lacks IEC context: it may be more lax or more strict than the figure of a manufacturer that does detail it.

Structural argument. A buffer would lower Zout but not THD. Only NFB lowers both at once. NFB exists. What remains to be known is its depth.

4.2 How to measure DF correctly

DF = R_load / Z_out

Standard procedure: two known, different resistive loads.

Never without load. The OPT is inductive; without load, overvoltages can destroy valves or the transformer itself.

Two de facto standardised methods for measuring Z_out. The technical reference is IEC 60268-3, which describes the general procedure for measuring electrical parameters of amplifiers by the voltage-drop method (On-Off): measure the voltage without load, measure it with a known resistive load, and solve Z_out from the voltage divider. However, the standard does not fix specific conditions —frequency, bandwidth, type of load— under which DF must be expressed as a catalogue figure. That is where the disparity of figures between manufacturers and laboratories comes from.

There is a preferable variant for valve amplifiers: the two-load method, which avoids measuring in open circuit. Since the output transformer is inductive, measuring without load can generate overvoltages that damage the transformer or the valves. That is why this report uses the two-load method, which is also the one JC Valvular applies in its design routine.

Two-load method

Load R₁ = 8 Ω, inject 1 kHz, measure V₁.
Load R₂ = 4 Ω, measure V₂.
Calculate:

Z_out = R₁ · R₂ · (V₁ − V₂) / (R₁ · V₂ − R₂ · V₁)

DF = 8 / Z_out.
Repeat at several frequencies (20 Hz – 20 kHz). The curve matters.

4.3 Loop gain calculation

Base model. KT150 in UL with effective ra ~3.5 kΩ, OPT with n = 25, 100 mV sensitivity.

Open Zout. 3500/625 + 0.7 ≈ 6.3 Ω → DF_ol ≈ 1.27.
Necessary reduction. From 1.27 to 50: factor 39.4.
Loop gain: β·A ≈ 38.4 → 31.7 dB.

Unified model sensitivity table

ra_UL	Raa	Loop needed for DF 50	Verdict with 21–29 dB available
3.5 kΩ	5 kΩ (n = 25)	31.7 dB	Does not reach
1.5–2 kΩ	5 kΩ (n = 25)	26–28 dB	H1 revives
3.5 kΩ	3 kΩ (n ≈ 19.4)	~36 dB	Does not reach
3.5 kΩ	3 kΩ, with CF ×5	~12–15 dB + CF	Yes

Is 31.7 dB achievable?

Scenario	Gross	After OPT	Closed	Loop	Exact DF
Optimistic	100 dB	72.04 dB	43.01 dB	29.03 dB	37.2
Typical	95 dB	67.04 dB	43.01 dB	24.03 dB	21.5
Realistic	92 dB	64.04 dB	43.01 dB	21.03 dB	15.6

In the best case, 29 dB, below the 31.7 dB needed. For THD: with 21–29 dB of loop and 3–5 % open-loop in UL, the final THD would be 0.1–0.45 %. 20–33 dB missing.

Central finding — traffic light table

Declared figure No. 1 · DF ≈ 50. Loop needed 31.7 dB · Loop available 21–29 dB. 3–10 dB short. Semantic note: the 31.7 dB are the Zout reduction factor (from 6.3 Ω open-loop to 0.16 Ω with loop), not a voltage loop gain. They coincide numerically because the model uses a simple loop with constant β; conceptually they are different magnitudes.

Declared figure No. 2 · THD 0.01 % @ 18 W. Loop needed 49–54 dB · Loop available 21–29 dB. 20–33 dB short.

Alternative scenario · If real sensitivity = 1000 mV. Loop needed 41–49 dB · Loop available 41–49 dB. Just reaches. It is the only scenario where H1 revives without cathode feedback.

Semantic precision. In §4.3 there appear two figures in dB that are not the same magnitude and should not be compared as if they were:

31.7 dB is the Zout reduction factor needed to go from 6.3 Ω (open loop) to 0.16 Ω (DF ≈ 50). It is an impedance ratio.
49–54 dB is the voltage loop gain needed to reduce THD in the declared proportion. It is a voltage ratio.

They coincide numerically in the model only because the simple loop with constant β makes them interchangeable. Conceptually they are different magnitudes and both must be achieved simultaneously if DF ≈ 50 and THD 0.01 % are real at the same time. Under the estimates of this report, the available loop gain (21–29 dB with 100 mV) is not enough for either one separately.

Catalogue figure vs. real behaviour. A DF figure obtained at 1 kHz with fixed resistive loads describes the amplifier's behaviour at a very specific point, not across the entire audible band or with the reactive impedance of a real speaker. It is a valid data point as a design indicator —because DF is decided in the OPT design, as the manufacturer confirms in §4.5—, but it does not allow predicting how the device will behave with a real load.

Analogy. If you measure a car's speed downhill on the best stretch of a motorway, you get 200 km/h. If you measure the average speed over a complete journey with curves, climbs and descents, you get 80 km/h. Both data points are true, but they describe different things. The same can happen with DF: the catalogue figure and the overall behaviour of the amplifier with real speakers do not necessarily coincide. That is exactly the point that forces measurement, not extrapolation.

Methodological note. The declared figures are DECLARED; the available loop is CALCULATED from typical gains at the plate and 100 mV sensitivity. The three results change completely if the real sensitivity is 1000 mV (§4.4).

100 mV sensitivity. It implies 43 dB closed, anomalously high. Possibilities: (a) integrated data point with the pot at maximum; (b) erratum (1000 mV?); (c) real.

4.4 The 1000 mV scenario

If the real sensitivity were 1000 mV instead of 100 mV, the closed-loop gain would drop to 23.01 dB and the loop would rise to 41–49 dB. It is the only scenario where H1 revives without cathode feedback.

Sensitivity	A_cl	Loop	THD without CF	THD with CF ×5
100 mV	43.01 dB	21–29 dB	0.1–0.45 %	0.02–0.09 %
1000 mV	23.01 dB	41–49 dB	0.011–0.045 %	0.002–0.009 %

Conclusion. If the real sensitivity were 1000 mV, the final THD would fall within 0.011–0.045 % without CF and 0.002–0.009 % with CF. It is the only scenario where H1 revives without cathode feedback.

THD vs. power chart — the four key curves

Pure pentode (Tung-Sol 14 % at 20 W) UL without NFB (3–5 %) UL + 21–29 dB (0.1–0.45 %) UL + 41–49 dB (0.011–0.045 %) Declared point

4.5 Manufacturer's response

Juan Calatayud has replied by email to the report's questions.

Key points

DF is complex in valves ("infinitely easier" in solid state).
It is calculated in the OPT design, not measured at the end.
The DF ≈ 50 figure is an "average DF", measured at 1 kHz with fixed resistive loads.
It is not constant: it varies with frequency and with the reflected speaker impedance.
Does not mention how much NFB in dB, nor %UL, nor cathode feedback.

The 12 DF factors according to the manufacturer

#	Factor
1	Internal impedance of the output valves
2	Stage configuration (pentode, tetrode, triode, UL)
3	OPT transformation ratio
4	Primary winding resistance
5	Secondary winding resistance
6	Global negative feedback
7	Local feedback
8	OPT leakage inductance
9	Primary magnetising inductance
10	OPT parasitic capacitances
11	Wiring and connections
12	Operating frequency and load

What the email clarifies. DF ≈ 50 is not a "side effect" of NFB: it is an OPT design goal combining the 12 factors. It reinforces H2/H4.

What it does not clarify. It does not say how much NFB in dB, nor whether there is cathode feedback, nor the %UL.

Methodological reading of the email. The fact that the manufacturer expresses DF as an average at 1 kHz with fixed resistive loads is coherent with widespread practice in the sector and describes exactly a catalogue figure. It is not a false figure: it is a design data point obtained at a very specific point. But it does not describe the amplifier's behaviour across the entire audible band or with the reactive impedance of a real speaker. The manufacturer's sentence confirms, without saying so, what this report has argued since §4.2: the declared figure and the real behaviour can describe different things without contradicting each other.

Methodological warning — the Fezz case

click to expand

The Fezz Titania line: PA MK2 (declares DF > 40) and PA Evolution (measured by HiFiNews with DF = 1). They are different models, but they illustrate that declared figures can differ radically from measured ones.

Lesson. The only way to resolve it is to measure the specific device with the correct protocol. That is exactly what the verification plan in §6.1 does.

Summary of chapter 4 in 3 points

Central finding: the model estimates that THD 0.01 % at 18 W requires 49–54 dB; the declared DF of ~50 corresponds to 31.7 dB. The three verdicts in the traffic light table of §4.3.
Loop needed: 31.7 dB for DF 50; available 21–29 dB with 100 mV. With 1000 mV: loop 41–49 dB, THD 0.011–0.045 % without CF. It is the only scenario where H1 revives.
Manufacturer's response (§4.5): DF is designed in the OPT, 12 factors, average at 1 kHz with resistive loads. It confirms the methodological suspicion, reinforces H2/H4, leaves NFB open.

5. Precedents and hypotheses

Key idea of chapter 5

Precedents are the empirical floor: what others have achieved with the same valve in an SE. Hypotheses are the theoretical ceiling: which architectures would explain the declared figures. Together they define the space of the possible. The Icon ST30SE is the most relevant precedent (same valve, same SE topology with gap). The Fiio EA13 provides the hum precedent. The Dared Saturn is inconsistent and does not serve as a ceiling.

Comparative table of precedents

Parameter	Icon ST30SE	Dared Saturn	Fiio EA13	Acronos RS-1
KT150/channel	1	1	No (4× EL34)	1
Topology	SE, UL/triode switchable, switchable NFB	SE triode	PP Class A	SE class A UL
Weight	26 kg net	18 kg	17.7 kg	14 kg
Consumption	177 W / 220 W	ND	220 W / 230 W	195 W
NFB	"low feedback" switchable ND	−39 dB (inconsistent)	ND	Exists; depth ND
DF	UL: 7–9 · TR: 15–17 HiFiNews · ≈ 6 Hi-Fi World	6	ND	≈ 50 (Et Sona)
Buzz reported	ND	ND	Yes	ND

Icon ST30SE — the most relevant precedent

Measured by Hi-Fi News in 2019: Zout 0.85–1.17 Ω in UL and 0.47–0.54 Ω in triode. Power 28 W in UL / 18 W in triode. THD 0.2–0.5 % at 1–5 W, 1 % at 17 W, 2 % at 30 W. Weighs 26 kg net. The UL connection has higher Zout than triode. The H/L/0 feedback switch allows removing or reducing NFB. It is a real precedent that DF depends on the loop.

Two publications, two figures for the same device. Hi-Fi World published DF ≈ 6 for the Icon ST30SE; HiFiNews published 7–9 in UL and 15–17 in triode. Both figures are true and describe the same device, but they depend on the measurement conditions (frequency, type of load, operating mode). This is important because it illustrates that DF is not a constant of the device, but a magnitude that depends on the method. Any DF figure —including that of the RS-1— must be read with that caution. The Hi-Fi World data point has been incorporated into the comparative table; the HiFiNews figure is kept because it is more detailed per mode.

Fiio EA13 — the hum precedent

Push-pull class A, 4× EL34, 18 W < 1 %, 17.7 kg, 21.1 L. It documents mechanical buzz from the power transformer despite OPTs encapsulated in resin and sand, a unibody chassis and anti-vibration feet. Direct precedent for §2.3.

Dared Saturn — inconsistent

SE triode with KT150. Declares: 25 W/channel, −39 dB global NFB, DF 6, THD 0.05 % at 4 W. Weight: 18 kg. The numbers do not add up: with −39 dB of NFB (factor ≈ 89.1), Zout should be reduced by that factor. With open Zout of 5–6 Ω, the final would be ~0.06 Ω (DF ≈ 130). Declaring DF 6 with −39 dB implies that one of the data points does not mean what it appears to. It does not serve as an empirical ceiling.

Lesson. No commercial KT150 SET reaches DF 50. Extremes: DF 7–9 (Icon UL) and 15–17 (Icon triode). The RS-1 is a technical anomaly, not the market norm. The Fezz Titania case (§4.4) reinforces the methodological warning.

Architecture hypotheses

Three live hypotheses to explain the RS-1's declared figures:

H1 — Deep global NFB with higher gain than estimated. With 100 mV sensitivity: THD 0.1–0.45 %. Does not reach. With 1000 mV, H1 revives (§4.4).
H2 — Cathode feedback + possible light global loop. Dedicated winding in the OPT feeding back to the KT150 cathode. Best fit: with CF ×5 + UL + 29 dB, THD at 0.025–0.089 %. Close, not exact.
H4 — Combination of UL + local NFB + global loop. Consistent with both figures, requires complexity not observed.

Discarded. "Pure buffer" (does not lower THD) and "no NFB" (contradicts the manufacturer).

Reading the tree. The structural argument discards the buffer. The calculation discards the standard global loop (with 100 mV). H2 remains the best fit and H4 as the alternative.

What §4.5 and the video add. DF is an OPT design goal, and the OPT is custom-made. Both data points reinforce H2/H4.

How to discriminate between hypotheses

Check	Discriminates	Cost
OPT inspection: third winding?	H2/H4 vs. H1	Low
Question to the manufacturer: "how much NFB, in dB, and of what type?"	H1 vs. H2 vs. H4	Low (email)
Zout vs. frequency curve on two taps	H1 / H2 / H4	Medium

The rest of the measurements that help discriminate are in the verification plan of §6.1.

Summary of chapter 5 in 3 points

Precedents: Icon ST30SE (measured): DF 7–9 UL, 15–17 triode HiFiNews · DF ≈ 6 Hi-Fi World. Fiio EA13: PP Class A, 17.7 kg, with documented buzz. Dared Saturn: −39 dB NFB with DF 6. Inconsistent.
Three hypotheses: H1 (global NFB), H2 (cathode feedback), H4 (combination). Discarded: pure buffer and no NFB.
With 100 mV: neither H1 nor H2 reaches 0.01 %. With 1000 mV, H1 revives. §4.5 and the video reinforce H2/H4. The model is sensitive to ra_UL and Raa: with ra_UL = 1.5–2 kΩ, H1 revives without CF.

6. Verification, conclusion and traceability

Key idea of chapter 6

In a report whose finding is "we do not know X, Y, Z", the verification plan is the conclusion. This chapter separates the two layers: 6.1 the action (16 measurements, priorities, protocols, quick sheet for owners), 6.2 the state (what we have learned, what CANNOT be concluded, ND traceability table).

6.1 Verification plan

Measurements that would close the report

#	Measurement / data point	Resolves	Kit	Priority
1	NFB depth and type (email)	Closes §4. Discriminates H1/H2/H4	—	CRITICAL
2	Zout vs. frequency (2-load method)	Verifies DF ≈ 50	Lab	CRITICAL
3	THD+N vs. power (full curve)	Verifies 0.01 % @ 18 W	Lab	CRITICAL
4	Real B+ and %UL	Class A ceiling and operating point	Owner	CRITICAL
5	Sensitivity 100 vs 1000 mV	If 1000 mV, H1 revives	—	CRITICAL
6	Listening at 10 cm from the transformer + Spectroid	Mechanical buzz	Owner	CRITICAL
7	OPT Lp + 20 Hz response at 1 W vs 18 W	Sizing of the custom OPT	Lab	CRITICAL
8	OPT: third winding? Real Raa	H2/H4 vs. H1; §4.3 model	Lab	HIGH
9	Comparative photo of the M1 and RS-1 OPTs	Specify how much has been redesigned	Owner	HIGH
10	Photo of the RF/EMI filter (reference)	Schaffner? Model	Owner	HIGH
11	Photo of the fuse holders (values)	Live+neutral or mains+B+?	Owner	HIGH
12	Measure real bias per KT150	Verifies 130 mA	Owner	HIGH
13	IR thermometer after 2 h	Thermal margin of the transformer	Owner	HIGH
14	Square wave 1 kHz and 10 kHz	Stability under NFB	Lab	HIGH
15	Direct loop gain measurement	Real NFB depth	Lab	HIGH
16	Real weight on a scale	Confirms or refutes the 14 kg	Owner	HIGH

Two kits, two profiles

Kit	Components	Indicative cost	Measurements covered
Owner (minimum)	1000 V DC multimeter, IR thermometer, smartphone with Spectroid/SpectrumView, two power resistors (8 Ω and 4 Ω) and some cable	~€50–80	4, 6, 9, 10, 11, 12, 13, 16
Laboratory	Focusrite Scarlett 2i2 (or similar), precision resistive loads, attenuator, oscilloscope, REW/ARTA	~€300–400	2, 3, 7, 8, 14, 15

Measurements 1 and 5 are email only: they cover the two most critical unknowns and cost nothing.

DF protocol

8 Ω and 4 Ω resistive load, sufficient power not to saturate.
Frequencies: 20 Hz, 100 Hz, 1 kHz, 10 kHz, 20 kHz.
Oscilloscope or True RMS. Two-load method (§4.2). Never in open circuit.
Note the signal level of each measurement, because Z_out can vary with excursion.
Repeat on both taps (4 Ω and 8 Ω) if the device has them.

Why this protocol and not another: the two-load method avoids the overvoltage risk of the On-Off method and is consistent with the practice the manufacturer himself declares in §4.5. See §4.2 for the full methodological justification.

Hum protocol

Listen at 10 cm from the power transformer after 1 h hot.
Press the chassis with the palm: if the buzz drops, chassis-transformer coupling.
Spectroid or SpectrumView app at 10 cm: 100 Hz peak = load buzz; 50 Hz peak = DC hum.
IR thermometer: transformer >65 °C or OPT >50 °C after 2 h = no margin.

Quick sheet for owners (10 minutes)

If you have the RS-1 at home and are not going to measure, these seven checks are the minimum. Designed to print or copy to your phone.

#	Check	How	Closes
1	Photo of the IEC C14 with filter reference	Photograph the socket inside or outside. Look for FN9222 / FN9280 / FN9244 silkscreen.	Filter brand and model.
2	Photo of fuse holders with values	Photograph the two caps. Check the engraved values.	Configuration and values.
3	Measure B+ at rest	1000 V DC multimeter, switched on (caution). Wait 15 min warm-up.	Real B+.
4	Measure real bias per valve	Drop across each KT150 cathode resistor (typically 10 Ω). Expected 1.30 V.	Real bias.
5	Measure real weight on a scale	Compare with declared 14 kg.	Real weight.
6	Listening + Spectroid / SpectrumView app	Phone at 10 cm from the transformer after 1 h hot. Caution: do not touch the chassis with hot valves.	Transformer hum.
7	20 Hz response at 1 W vs 18 W	Sound card + REW/ARTA + 8 Ω resistive load.	Sizing of the custom OPT.

Safety before touching: never open the chassis plugged in; wait 10 min after switching off and discharge with 100 kΩ / 5 W down to < 20 V; do not measure in open circuit at the output (destroys the OPT); do not touch the KT150 pins while powered (450–500 V DC); do not touch the chassis if just switched off.

6.2 Conclusion and traceability

Synthesis of the report

The central finding. Under the §4.3 model, the THD 0.01 % at 18 W demands 49–54 dB of loop gain; the declared DF of ~50 corresponds to 31.7 dB as an impedance ratio. The available loop with 100 mV sensitivity is 21–29 dB. Neither H1 (0.1–0.45 %) nor H2 with cathode feedback (0.025–0.089 %) reaches 0.01 % without additional assumptions. If the sensitivity were 1000 mV, the loop would rise to 41–49 dB and H1 would revive.

Class A ceiling. 25 W with 130 mA bias require B+ = 384.6 V theoretical and ~478 V practical. The 450–500 V zone is not optimistic, it is what is needed.

Lineage and transformers. The RS-1 shares five exact parameters with the M1 (14 kg, 45×26×11 cm, 18 W @ 0.01 %, 100 kΩ, 100 mV). The manufacturer's video confirms that the OPTs are custom-made, different from those of the M1. The coincidence of weight suggests a moderate redesign in mass, but does not prove it: same weight does not imply same internal distribution. The 18 W @ 0.01 % is probably a family design goal, not an independent measurement of the RS-1.

Manufacturer's response. DF is calculated in the OPT design from 12 factors. The DF ≈ 50 figure is an average measured at 1 kHz with fixed resistive loads. It confirms the report's methodological suspicion, reinforces H2/H4 and leaves the question about NFB in dB open.

Hum risk. The power transformer can vibrate at 100 Hz due to rectifier peaks. The 1.5 mm steel chassis + solid wood panels inherited from the M1 can mitigate, but was not redesigned for the KT150. The verification is binary and without instrumentation.

Ethical closing. This report is not a measurement of the RS-1. The critical figures remain DECLARED until a specific unit is measured with a two-load protocol and a THD vs. power curve, and the hum is verified.

6.3 What CANNOT be concluded from this report

On topology. It cannot be concluded that the RS-1 has no NFB (it does), that it is triode or pure pentode (it is UL), nor that it has no RF/EMI filter (it does). All three facts are confirmed by the manufacturer.
On the declared figures. DF ≈ 50 might not hold at 20 Hz with reactive load; THD 0.01 % at 18 W is a single point, not a curve. The declared figures are not necessarily false or exaggerated, only unverified.
On the 100 mV sensitivity. Its nature is not determined. If it were an erratum (1000 mV), the entire hypothesis tree changes.
On NFB. The depth and type are not determined: ND. The manufacturer's email adds DF context, not loop context.
On the cathode follower. It is not the cause of DF ≈ 50. That figure is decided in the output.
On the class A ceiling. It does not imply that the 25 W is false. It only bounds the required minimum B+.
On the 18 W @ 0.01 %. It is probably a family design goal inherited from the M1, not a measurement of the RS-1.
On the transformers. The video confirms that the OPTs are custom-made, but not how much they have been redesigned. The chassis is probably inherited from the M1, but it is not explicitly confirmed. The OPT is not the same as the M1 (the video rules it out), but it might not be substantially larger.
On sizing. Weighing 14 kg does not imply being undersized. The volumetric density (1.09 g/cm³) is the highest in the comparison group. The kg/W coincidence with the Jazz 90 is not significant.
On hum and sound. It cannot be concluded that the RS-1 hums or does not hum. It cannot be concluded that it sounds better or worse than other SEs.

Traceability — what remains ND

Block	Unknowns
Loop model (6 ND)	NFB depth in dB ND · NFB type ND · KT150 ra in UL ND · Real Raa ND · %UL ND · Phase margin and real loop ND
Custom OPT (5 ND)	Specifications (gap, N, Raa, Lp) ND · Primary inductance ND · Individual weight ND · Core section and air gap ND · 20 Hz response at 1 W vs 18 W ND
M1 lineage (3 ND)	Confirmation M1 chassis = RS-1 chassis PROBABLE · Precision of the OPT redesign ND · Origin of the 18 W @ 0.01 % PROBABLE inherited from M1
Electrical and acoustic verification (12 ND)	Real B+ ND · Real bias per valve ND · Sensitivity 100 vs 1000 mV ND · Full THD curve ND · HV bank and filtering ND · Bias drift at 500 h ND · Real temperature of the OPTs ND · Real weight on a scale ND · Mechanical hum from the transformer ND, risk PROBABLE · Individual function of the 8 trimmers ND · Brand and model of the RF/EMI filter ND (probably Schaffner) · Configuration and fuse holder values ND · Presence of NTC or delay relay ND

Summary of chapter 6 in 3 points

16 measurements, 7 CRITICAL. Two kits: owner (~€50–80) and laboratory (~€300–400). The question to the manufacturer about NFB is the cheapest and most discriminating. Quick sheet of 7 checks for owners.
Central finding: the model estimates 49–54 dB for THD; 31.7 dB for DF 50. Neither H1 nor H2 reaches exactly 0.01 % with 100 mV. With 1000 mV, H1 revives.
Ethical closing: this report is not a measurement. The critical figures remain DECLARED. The comparative photograph of the M1 and RS-1 OPTs closes §1.2 and §2.2 with a simple photo.

Interactive calculators — SPL and Zout

Two tools to interpret the report with your own numbers: one estimates the required power according to sensitivity, distance and target SPL; the other quantifies how much a high Zout colours the response over a realistic load.

SPL and required power calculator

Formulas: SPL_1W@d = sens − 20·log10(d) and P = 10^((target SPL − SPL_1W@d − factor)/10). Results per channel.

Sensitivity (dB/W/m) 89.0 dB/W/m

Listening distance 3.0 m

Target SPL (peak) 95 dB

Acoustic correction factor

Enable JavaScript to see the calculator result.

Methodological note. The reference powers are declared nominal, not P@1 % THD. Sensitivity is a calculation hypothesis; the owner's reference speaker is 89 dB/W/m at 3 m.

Response variation calculator by Zout

Amplifier Zout (Ω) 2.0 Ω

Speaker Zmin (Ω) 4.9 Ω

Speaker Znom (Ω) 6.0 Ω

Enable JavaScript to see the calculator result.

What this calculator measures. It returns the difference between the attenuation at Zmin and at Znom with the load modelled as resistive: it is a pedagogical approximation, not a simulation of a real speaker. The RS-1's Zout is ND: the declared DF ≈ 50 would give 0.16 Ω, while the measured SE KT150 precedents are between 0.5 and 1.2 Ω. Compare both extremes.

GLOSSARY — 11 TECHNICAL TERMS FROM THE REPORT

Glossary

Eleven terms that appear in §2.2, §2.3, §4.3 and the tables. For the reader arriving without prior electronics.

UL (ultralinear). Operating mode of a beam tetrode or pentode in which the screen grid is not connected to a fixed B+ (pentode) nor to the anode (triode), but to an intermediate tap of the OPT primary (typically 20–50 % of the turns). Result: more power than triode, less THD and Zout than pentode. §3.2

NFB (negative feedback). Design technique that subtracts a fraction of the output signal from the input. Effects: lowers THD, lowers Zout (raises DF), widens bandwidth, flattens response. The depth is measured in dB of loop gain (β·A). §4

CF (cathode feedback). Negative feedback local to the cathode of the output valve. It can be through cathode degeneration (undecoupled resistor) or through a dedicated winding in the OPT. Lowers Zout and THD by a factor of 3–5× without needing a deep global loop. It is the signature of hypothesis H2 in the report. §5

r_a (internal plate resistance). Dynamic equivalent resistance of the valve anode seen from the circuit. It determines how much Zout the valve contributes to the whole. In UL it is lower than in pentode; the KT150 datasheet only gives the pentode value (≈ 3 kΩ), hence the uncertainty in §4.3. §4.3

R_aa (reflected load impedance). Impedance that the output transformer reflects to the primary, seen by the valve. It relates to the speaker impedance by R_aa ≈ n² · Z_speaker. The two values (3 kΩ and 5 kΩ) correspond to two different n and explain the double hypothesis of the report. §4.3

Gap (air gap). Physical cut introduced in the magnetic core of the output transformer to prevent saturation by the SE quiescent DC current. Without a gap, the core would saturate with the 130 mA of the KT150. In exchange, it reduces the magnetising inductance and demands a larger core. It is the reason why an SE weighs more per watt than a PP. §2.2

Custom OPT. Output transformer commissioned specifically for the RS-1, with specifications different from those of the M1. According to the manufacturer's video, gap, number of turns N, Raa and working voltage are adjusted. The identical total weight suggests a moderate redesign in mass: same platform, probably different winding and air gap. §1.2

Hum. Noise at 50 Hz (or 60 Hz on the American mains) due to asymmetric saturation of the power transformer core caused by DC on the mains or by ground loops. It is eliminated with a DC blocker. Through speakers: continuous low-pitched hum. §2.3

Buzz. Noise at 100 Hz (or 120 Hz) due to the current peaks from the rectifier. It is not due to DC on the mains, but to load stress. The DC blocker does not eliminate it. Through speakers: sharper, modulated hum. §2.3

Lorentz (force). Force that appears on a conductor through which current flows inside a magnetic field. In the power transformer, the rectifier current peaks generate periodic forces on the windings that make them vibrate at 100 Hz. §2.3

Magnetostriction. Property of ferromagnetic materials to change dimensions slightly when magnetised. In the transformer core, the magnetisation cycle at 50/100 Hz causes a mechanical vibration at 100 Hz that adds to the Lorentz one. It is the second of the two buzz motors. §2.3

APPENDIX A — TECHNICAL QUESTIONNAIRE FOR JUAN CALATAYUD (JC VALVULAR)

Appendix A — Technical questionnaire

Questions to close the report's NDs. We start from four confirmed facts: NFB in all Acronos, UL in the valve models, IEC C14 RF/EMI filter in all, and custom OPTs in the RS-1 (the latter confirmed on video). Within each block, the questions are ordered from most to least important.

Block 1 — Feedback

NFB depth.
How much NFB does the RS-1 have, in dB? What is the estimated loop gain?
NFB type.
Global loop from the OPT secondary, or also local NFB (cathode feedback to the KT150 cathode)? If there is cathode feedback, does the custom OPT have a dedicated winding?
100 mV sensitivity.
Is it for a power stage or for an integrated with the potentiometer at maximum gain? If it is for a stage, it implies 43 dB closed, anomalously high. Or is it an erratum (1000 mV)?
NFB contribution to DF and THD.
What fraction of the declared DF is due to NFB and what fraction to the custom OPT design? And of the THD?
Compensation and stability.
Frequency compensation network? Phase and gain margins? Behaviour with 1 kHz and 10 kHz square wave?

Block 2 — Operating point and mode

Real B+ and class A ceiling.
What is the real B+? With 130 mA bias, declaring 25 W in pure class A requires B+ = 384.6 V theoretical and ~478 V practical. If it is lower, the 25 W includes excursion outside class A.
Full distortion curve.
The two declared points, at what frequency and load are they measured? Can the THD vs. power curve be published? Is the 18 W @ 0.01 % a measurement of the RS-1 or inherited from the M1? Does it comply with IEC 60268-3 (bandwidth and filter)?
%UL.
What percentage of the primary tap is used for the KT150 screen?
Bias and trimmers.
Is the 130 mA bias single or is there individual adjustment? What function does each of the 8 trimmers have?

Block 3 — Power supply, transformers and compatibility

Custom output transformers.
The video confirms that they are different from the M1. Can you specify how? Weight of each OPT? Primary inductance? Raa? Gap? Core section? Is the same core section and window as the M1 OPT maintained or has it changed?
Power transformer hum.
Has the mechanical hum at 100 Hz been verified? Is the transformer mounted on silentblocks?
OPT primary inductance and real Raa.
It adjusts the §4.3 model and the OPT sizing.
Lineage with the M1.
Is the RS-1 chassis the same as the M1's? Have only the OPTs and the bias circuit been changed, or is there more?
Power supply filtering and RF/EMI filter.
How many filtering stages (C, LC, RC)? Total capacitance of the HV bank? Regarding the filter: brand and model? Common mode, differential, or both?
Threaded fuse holders.
Do they protect live+neutral, mains+B+, or one per channel? Values?
Noise and compatibility.
Residual noise with input shorted? Has it been tested with 89 dB or lower speakers?
Operational data and support.
Net weight without packaging, consumption at rest and at full power? Is a schematic with test points provided?

APPENDIX B — MEASURING NFB WITHOUT OPENING THE CHASSIS

Appendix B — Measuring NFB without opening the chassis

The most critical ND of the report is the NFB depth in dB. There is an indirect method that does not require opening the amplifier, and that any technician with a bench can perform.

Physical principle

In an NFB loop, the ratio between the output signal and the signal fed back to the input is fixed by the feedback network (β). If a signal is injected at the OPT secondary with the amplifier switched off, and the attenuation at the input is measured, the β of the network is obtained. With the gain A known by another route, β·A is calculated.

Protocol

Switch off and unplug. Wait 10 min.
Access the positive speaker terminal (red). Do not open the chassis.
Inject a signal with a generator or calibrated sound card. 1 kHz. 1 V_rms.
Measure the signal at the corresponding RCA input. Amplifier switched off, passive network only.
Note the attenuation: 20·log10(V_out/V_in) in dB.
Repeat at 100 Hz, 1 kHz, 10 kHz to see the network curve.

Interpretation

Measured attenuation (β)	Probable topology	Coherence with §4.3
< −50 dB	Shallow global loop or no loop	H1 impossible
−30 to −50 dB	Typical global loop	H1 possible only with low ra_UL and/or 1000 mV
−15 to −30 dB	Deep loop + possible CF	H2 or H4 coherent
> −15 dB	Very deep loop	H1 revives without additional assumptions

The measurement does not distinguish between global loop and cathode feedback: both appear as network attenuation. The distinction requires seeing the internal topology or asking the manufacturer.

Second measurement — closed-loop gain

Switch on the RS-1, let it warm up for 15 min.
Connect an 8 Ω resistive load.
Inject 1 kHz at the input (10–100 mV).
Measure V_out and V_in.
A_cl = V_out / V_in.
With A_cl ≈ A / (1 + βA) and β known, solve for A and β·A. 2–3 iterations are enough.

REFERENCES AND CREDITS

References and credits

Technical sources

Fezz Audio · Titania MK2 report (nauscopio, V3.0.3). Sister publication that develops in its §5.3 the methodological framework for DF measurement (On-Off method vs two-load method, absence of fixed conditions in IEC 60268-3) and cites this Acronos RS-1 report in its Appendix D. From this document have been taken: the description of the two measurement methods, the car analogy, the Hi-Fi World figure for the Icon ST30SE (DF ≈ 6) and the pedagogical formulation of the catalogue/real distinction. (2026-10)
JC Valvular — direct contact: NFB in all Acronos, UL in the valve models, IEC C14 RF/EMI filter in all. (2026-10)
JC Valvular · public video of Juan Calatayud — declares that the RS-1 output transformers are different from those of the M1 and that they were commissioned custom-made with other specifications. Add video URL and timestamp (e.g. min 12:34). (2026) PENDING URL
JC Valvular · response on DF — email from Juan Calatayud (§4.5): DF calculated in the OPT design from 12 factors; DF ≈ 50 measured at 1 kHz with resistive loads. (2026-10)
JC Valvular — Acronos RS-1 commercial spec sheet: 25 W max., 18 W @ 0.01 %, 25 W @ 0.5 %, 130 mA bias, 195 W consumption, 20 Hz–40 kHz, 100 kΩ, 100 mV, 14 kg, 45×26×11 cm. (2026-10)
JC Valvular · Acronos M1 Pura Clase A — official spec sheet of the predecessor (jcvalvular.com): 20 W max., 18 W @ 0.01 % at 1 kHz, 100 kΩ, 100 mV, 14 kg, 45×26×11 cm, KT120 in class A ultralinear. 2023-01
Et Sona — RS-1 review (et-sona.com/post/jc-valvular-acronos). Genealogy with the M1, valve architecture, DF ≈ 50 and chassis construction details. (2026-10)
Tung-Sol KT150 — official datasheet (tungsol.com). 70 W plate, 850 V pent. / 650 V triode, heater 1.75–2.0 A. Pentode A1 point: 400 V, 150–180 mA, 20 W with THD = 14 %. (2026-10)
Philips/NXP EC86 — official datasheet (frank.pocnet.net). µ = 68, gm = 14 mA/V, ra ≈ 4.86 kΩ. Max plate 220 V. Equivalent 6CM4. (2026-10)
Genalex / New Sensor — KT120 (60 W) and KT88 (42 W) datasheets.
Schaffner — IEC C14 filter datasheets: FN9222, FN9280, FN9244. Leakage 0.2–0.5 mA; attenuation 20–60 dB between 150 kHz and 10 MHz. (2026-10)
HiFiNews — Icon Audio ST30SE review, HFN 2019.11. Zout 0.85–1.17 Ω UL and 0.47–0.54 Ω triode; 28 W UL / 18 W triode. Weight 26 kg net / 28 kg packed. Consumption 177 W idle / 220 W full power. (2026-10)
Hi-Fi World — Icon Audio ST30SE review. DF ≈ 6 figure for the same device that HiFiNews measures at 7–9 (UL) and 15–17 (triode). It illustrates that even two publications of the same device can give different figures depending on measurement conditions. (2026-10)
Dared — Saturn Signature datasheet: 25 W/channel, NFB −39 dB, DF 6, THD 0.05 % at 4 W, weight 18 kg. Internally inconsistent.
Fiio · EA13 — commercial spec sheet and user reports. PP Class A, 4× EL34, 220 W idle / 230 W full, 17.7 kg, 21.1 L. Documented mechanical buzz at 100 Hz. (2026-10)
Fezz Audio · PA MK2 manual (60 W): 4× KT88, DF > 40, THD < 0.05 %. (2026-10)
HiFiNews · Fezz Titania PA Evolution (January 2025): DF = 1, Zout 7.4–8.0 Ω over 20 Hz – 20 kHz. (2026-10)
IEC 60268-3 — measurement methods for audio amplifiers: measurement bandwidth and filter for THD. It also describes the general procedure for measuring electrical parameters of amplifiers by the voltage-drop method, without fixing specific conditions (frequency, bandwidth, type of load) to express DF as a catalogue figure. (2026-10)
Reference system — Crossover A v2.4 MR-A-01: 6 Ω nom, Zmin 4.9 Ω, 89 dB/W/m, listening at 3 m.
Methodology. Formulas: Pmax = B+·Ibias/2 (Morgan Jones, Valve Amplifiers); Lp ≈ Raa/(2·π·fc); A = µ·RL/(RL+ra); Zout,cl = Zout,ol/(1+βA); ΔV ≈ I/(f·C); Lp ∝ N²·μef·A/l.
Nauscopio comparative reports (2026-10) — Acronos P52, P52 vs Cayin Jazz 90, Fezz Audio Titania MK2.
Edition V5.4.0 (2026-10). Refactored with the nauscopio V5.10 FINAL GSC + SOCIAL template. Changes from V5.3: JSON-LD with author.split name/alternateName and url https://nauscopio.wordpress.com/about/, about → Thing with @id, isPartOf, isAccessibleForFree, ISO 8601 with +02:00 zone, top-level ImageObject with width 1600 / height 925 / caption, publisher.logo updated to S1600 with dimensions, mainEntityOfPage with @id. First image migrated to S1600 with wrapper for lightbox, loading="eager" and 1600×925 dimensions, alt with og:image note.
→
. aria-pressed theme toggle to false. Added #s6-3 "What CANNOT be concluded" with anchor and TOC. Badge and footer updated to V5.4.0. GSC+SOCIAL checklist block at the start (commented). The technical contents, calculations, tables, glossary, appendices and references of V5.3 are preserved in full.

Image credits

General view of the Acronos RS-1: JC Valvular / Et Sona.
Figures 1 and 2 (Acronos-RS-1-pcboards.jpg, Acronos-RS-1-rear.jpg): JC Valvular.