IWISTAO Blog is the official platform of IWISTAO HIFI MINIMART, dedicated to high-fidelity audio and DIY electronics. We focus on tube amplifiers, output and power transformers, speaker enclosures, and Bluetooth audio technologies, including HiFi Bluetooth speakers, headphones, and tube-based acoustic systems. The blog shares product insights, technical knowledge, and practical guides for audiophiles and audio builders worldwide.
How splitting DC and audio current across two magnetic parts changes the trade-offs of single-ended tube amplifier design.
In a single-ended (SE) vacuum-tube amplifier, the output transformer does double duty: it carries the tube's DC plate current and the AC music signal at the same time. That DC bias pushes the transformer core toward saturation, so the component must be large, air-gapped, and expensive to keep distortion low. The parallel-feed — almost always called "parafeed" — topology removes the DC from the output transformer entirely by routing it through a separate choke, leaving the transformer to handle only the audio signal [2][3].
This guide explains how parafeed works, where it helps, where it doesn't, and how it compares with conventional series-fed single-ended output stages. It is written for builders, hobbyists, and buyers who want the engineering substance rather than the marketing.
What Is a Parallel-Feed (Parafeed) Output Transformer?
The term "parafeed" is a contraction of "parallel feed," but the name can mislead as a circuit description. In this arrangement the plate choke provides the DC feed path from the supply to the tube plate, while the output transformer is AC-coupled from the plate node through a parafeed capacitor. The two are not simply two components wired directly in parallel; the choke carries the static DC current, and the capacitor feeds only the AC signal into the transformer primary [1]. The output transformer is therefore freed from the DC magnetization that defines conventional SE design.
In a standard transformer-coupled SE stage, the average DC plate current must be supported by the transformer, which forces a relatively large core so it does not saturate under DC alone [1]. Parafeed changes that constraint at the cost of adding a second magnetic component.
How the Parallel-Feed Topology Works
A parafeed output stage has three core parts:
A plate choke (a large air-gapped inductor) connected between the power supply and the tube plate. It carries the full DC current and presents a high impedance to audio frequencies.
A coupling capacitor in series with the output transformer primary.
The output transformer itself, now free of DC current.
Because the DC plate current flows through the choke and returns to the supply, the output transformer's primary sees only the AC signal. The capacitor blocks DC, while the choke's high AC impedance keeps most signal current in the transformer rather than the choke [1]. The transformer's secondary then drives the loudspeaker as usual.
This single change reshapes the engineering trade-offs. In a conventional SE stage, one component must satisfy opposing demands: it needs enough iron to avoid saturating under DC, yet a small core and tight windings for wide bandwidth and low capacitance. Parafeed lets the choke and transformer each be optimized for their actual job [5].
Figure 1: Simplified parallel-feed (parafeed) output stage. The plate choke carries DC from the supply; the coupling capacitor feeds only the AC signal to the output transformer. (Diagram by author)
Key Advantages
No DC in the output transformer. Removing the DC magnetization means the core needs no SE air gap, which can raise the primary inductance and avoid the core-bias problems an air-gapped SE transformer faces. As a rule of thumb, an OPT built for parafeed may be similar in size to a push-pull unit of the same power, whereas a conventional SE transformer is typically larger because it must also accommodate the static DC flux [4]. This is an empirical guideline rather than a fixed law; actual size and performance depend on the specific OPT, plate choke, and coupling-capacitor design.
Potential for lower transformer distortion. With the DC bias removed, the transformer no longer walks asymmetrically into saturation on signal peaks — a mechanism that contributes odd-harmonic distortion in conventional SE stages [2]. This does not by itself guarantee a lower-distortion amplifier; the final result depends on how the OPT, plate choke, and coupling capacitor are designed and how their resonant behavior is managed.
More design freedom for bandwidth. Without DC forcing an air-gapped core, the transformer designer has greater freedom to optimize leakage inductance, primary inductance, and winding capacitance [1]. Combined with the coupling capacitor and choke inductance, this adds design "degrees of freedom," and their resonance can be used to extend low-frequency response. Whether the actual bandwidth is wider still depends on the complete OPT, choke, and capacitor combination.
Better power-supply hum isolation. In a standard transformer-coupled stage, supply ripple divides across the plate resistance and load, coupling hum to the output. In parafeed, the ripple drops across the choke and is largely kept away from the transformer, reducing hum — especially valuable with low plate-resistance tubes [1].
Option to use an autoformer. Because no high-voltage DC sits across the output winding, designers can substitute a tapped autoformer, a further optimization [1].
Trade-offs and Limitations
Parafeed is not free of compromise:
Two magnetic elements instead of one. You trade one large, air-gapped SE transformer for a choke plus a smaller OPT. The combined iron, weight, and cost are often similar or slightly higher — Jacmusic's 300B example shows a parafeed pair at about 9% more cost than the equivalent single SE transformer, for lower distortion and roughly double the power-handling headroom (55 W vs. 25 W transformer rating) [2].
A capacitor sits in the signal path. The coupling capacitor carries significant AC current, and its non-linearity can add coloration if poorly specified. It must be a high-quality, generously rated part [1].
Careful tuning required. The parallel inductance of choke and transformer, plus the coupling capacitor, creates resonant behavior that can cause frequency-response dips or peaks if not designed deliberately. Choke and OPT are usually specified and sold together for this reason [3].
"Parafeed splits the opposing demands on the typical SE transformer to handle both AC flux and DC flux… Each component is designed for its current load, without the offsetting compromises from an OPT that must handle both." — AudioCircle builder discussion [5]
Design Considerations
Plate choke selection. The choke must handle the full DC plate current without saturating and provide high inductance (typically several henries) at audio frequencies. It tends to be physically similar in size to a conventional air-gapped SE transformer because it carries the DC [4].
Coupling capacitor value. The parafeed capacitor does not follow a "larger is always better" rule. Together with the plate choke inductance and the load, it forms a low-frequency network with an optimal value or design range rather than a single maximum. Paul Joppa of Bottlehead gives the rule C = 2·L / R² (L = plate choke inductance in henries, R = OPT nominal primary impedance in kΩ, result in µF), describing it as a compromise between small-signal bandwidth and power bandwidth [6]. The capacitor's AC current rating also matters more here than in a typical line-stage position, so film types or high-quality bipolar electrolytics are common choices.
Output transformer rating. Size the parafeed OPT like a push-pull transformer of the target power, not like a conventional SE unit [4].
Parafeed vs. Conventional Single-Ended: A Comparison
Aspect
Conventional SE (series-fed)
Parallel-Feed (Parafeed)
DC in output transformer
Yes — requires air gap
No
Relative OPT size (same power)
Typically larger (must also hold DC flux)
Similar to a P-P unit (rule of thumb)
Typical distortion
Higher (odd harmonics from core bias)
Potentially lower (depends on design)
Power-supply hum coupling
Directly divides to output
Largely isolated
Magnetic parts count
One large transformer
Choke + smaller OPT
Cost / weight
One heavy part
Two parts; ~similar or slightly more
Table 1: Practical differences between conventional series-fed SE and parallel-feed output stages, based on documented measurements and builder experience [2][4].
Is Parallel-Feed Right for You?
Parafeed suits builders and manufacturers who prioritize low distortion and clean low-frequency behavior and are comfortable with the extra parts and tuning discipline it demands. It gained modern popularity through Bottlehead's Paramour 2A3 kits — engineered with Paul Joppa's input and MagneQuest's choke work — and remains a favorite among DIY single-ended triode enthusiasts [5]. For mass-market products, conventional SE is often chosen simply because it is easier to explain and market [5].
Frequently Asked Questions
Does parafeed eliminate the need for a large transformer entirely?
No. It replaces one large air-gapped SE transformer with a plate choke (similar in size) plus a smaller OPT. Total iron is often comparable [2][4].
Isn't the coupling capacitor a problem?
It does sit in the signal path and must carry substantial AC current, so part quality matters. Properly specified, its effect is generally considered linear; the main risk is low-frequency rolloff from too small a value [1].
Can I use a push-pull output transformer in a parafeed design?
Potentially, yes. The core reason is that the OPT carries no static DC, so it does not need the air gap that a conventional SE transformer requires to hold one-directional DC flux [4]. A push-pull transformer can therefore serve as the parafeed OPT — but only if its specifications fit the circuit. Primary impedance, power handling, primary inductance, frequency response, and turns ratio must all be appropriate for the application, so suitability should be checked case by case rather than assumed.
Who popularized parafeed in modern hi-fi?
Modern commercial parafeed traces to Bottlehead's Paramour 2A3 amplifiers, with key contributions from Paul Joppa and MagneQuest's Mike LaFevre [5].
Is parafeed only for single-ended amplifiers?
The principle applies to both SE and push-pull output stages, though it is most discussed in the SE context where DC magnetization is the central pain point [1].
PUBLISHED BY IWISTAO · Technology / Wireless Audio
From the BlueCore era to the Qualcomm QCC and Snapdragon Sound platforms — which Bluetooth specification each CSR / Qualcomm chip implements, and what a version number does and does not guarantee.
Cambridge Silicon Radio (CSR) was, for over a decade, one of the leading suppliers of Bluetooth silicon. Qualcomm announced its acquisition of CSR in October 2014 and completed it on August 13, 2015, with the business renamed Qualcomm Technologies International, Ltd. (QTIL).[4] The product families CSR built — BlueCore, CSR8xx, CSR86xx — and the QCC series and Snapdragon Sound / S1 platforms that followed under Qualcomm together make up a large share of the Bluetooth audio and connectivity silicon on the market.
This guide maps each major CSR / Qualcomm chip family to the Bluetooth specification version it implements. It is a reference, not an exhaustive datasheet: Qualcomm's audio portfolio spans dozens of part numbers, and new platforms (e.g., the S-series and S1-series Sound Platforms) continue to ship. Where a claim rests on a distributor or third-party page rather than a Qualcomm primary document, it is flagged as such.
Bluetooth Versions at a Glance
Before mapping chips to versions, recall what each Bluetooth Core specification step delivered. Note that several headline user features — most importantly LE Audio — are separate specifications built on top of a Core version, not part of the Core release itself (see the feature caveat).
Specification
Year
Headline capability at the Core level
Bluetooth 1.1 / 1.2
2001 / 2003
First interoperable baseline; 1.2 adds adaptive frequency hopping
Bluetooth 2.0 + EDR
2004
Enhanced Data Rate (~3 Mbps)
Bluetooth 2.1 + EDR
2007
Secure Simple Pairing, extended inquiry response
Bluetooth 3.0 + HS
2009
High-speed via 802.11 AMP
Bluetooth 4.0
2010
Introduces Bluetooth Low Energy (LE)
Bluetooth 4.1 / 4.2
2013 / 2014
Coexistence; LE data length extension, privacy
Bluetooth 5.0
2016
4× range, 2× speed, 8× advertising capacity
Bluetooth 5.1
2019
Direction finding (AoA / AoD)
Bluetooth 5.2
2020
LE Isochronous Channels (the foundation that enables LE Audio), EATT, LE Power Control
The exact adoption dates and feature lists are maintained by the Bluetooth SIG.[6]
The BlueCore Era: CSR's Foundation (BC1–BC6)
CSR's original BlueCore family carried Bluetooth from its earliest days through the 2.1 + EDR generation. BlueCore1 (BlueCore01) launched around 2000 and is widely credited as the first true single-chip Bluetooth device; it operated in the Bluetooth 1.0B era — the BlueCore01 data sheet predates the adoption of Bluetooth Core 1.1 in February 2001.[5b][6] An EE Times teardown documents the version compliance of the later numbered generations:
BlueCore 2 (BC2) and BlueCore 3 (BC3) — compliant with Bluetooth v1.1 and v1.2, built on a 0.18-µm RF CMOS process.[5]
BlueCore 4 (BC4) — compliant with Bluetooth v2.0 + EDR.[5]
BlueCore 5 (BC5 / BlueCore 5 Media) — compatible with Bluetooth v2.1 + EDR and backward-compatible with v2.0 + EDR; added an integrated 64-MIPS Kalimba DSP and stereo codec.[5]
BlueCore 6 (BC6) — the CSR 61xx series — this generation is marketed as the CSR 61xx family of single-chip mono-headset ROM devices (e.g., BC6110, BC6130, BC6140, BC6145, BC6150). All are qualified to Bluetooth v2.1 + EDR with extended SCO (eSCO); CSR's AuriStream (ADPCM) voice codec debuted here, cutting power versus legacy CVSD.[5][22] A third-party selection guide notes the BC6145 variant can be configured up to Bluetooth 3.0 and adds A2DP — treat that as a variant-specific claim, not a blanket BC6 figure.[2]
BlueCore 7 (BC7) — the CSR 65xx series — launched in 2008 as CSR's seventh-generation BlueCore and its most highly integrated part. The 65xx family (e.g., BC6540 / BC6570 / BC6590 class devices) is the first single chip to combine Bluetooth v2.1 + EDR with Bluetooth low energy (then branded "ULP Bluetooth" / Wibree), plus eGPS and FM transmit/receive. It kept the AuriStream voice codec and added in-built speaker drivers; note that the "Bluetooth low energy" here predates the formal Bluetooth 4.0 LE branding, so it is best described as an early/legacy LE implementation rather than a 4.0-qualified dual-mode stack.[24][25]
The 61xx / BC6 family was the workhorse of early mono Bluetooth headsets and hands-free car kits, where its low cost and long battery life mattered most. (Stereo speakers and USB dongles came later, with the CSR86xx and CSR8510 generations.)
The CSR8xx and CSR10xx Families
After the numbered BlueCore generations, CSR rebranded its silicon under the "CSR8xx" banner, and introduced a dedicated Low Energy line:
CSR8510 — a dual-mode controller widely used in USB Bluetooth 4.0 dongles. Third-party documentation describes it as compliant with the Bluetooth 4.0 specification (dual-mode: Classic + LE) and backward-compatible with 2.0, 2.1, and 3.0; this is consistent with CSR's positioning of the CSR8xx generation as its first 4.0 dual-mode line.[7]
CSR101x family (CSR1010 / CSR1011 / CSR1012 / CSR1013) — Qualcomm's own product page qualifies the CSR1010 to Bluetooth 4.1 (single-mode LE).[15]
CSR102x family (CSR1020 and others) — these parts are documented by Qualcomm as configurable with a Bluetooth stack up to 4.2 in some implementations, so the CSR10xx generation should not be collapsed into a single "4.0/4.1" label. The earlier CSR101x is 4.1; the CSR102x can reach 4.2 depending on the qualified stack.
The CSR86xx Audio SoCs (CSR8615, 8635, 8645, 8670, 8675)
The CSR86xx portfolio is where CSR's audio reputation was cemented. Bluetooth versions across this family vary by member:
CSR8615 — Qualcomm lists this ROM audio device at Bluetooth 4.1 (targeted at 1-mic mono speakers / headsets / car kits).[13]
CSR8635 — Qualcomm's current product page lists this part at Bluetooth 4.1 (dual-mode audio). Some older third-party / distributor listings show 4.0; where they differ, the Qualcomm page is the authoritative source.[18]
CSR8645 — the older Qualcomm product brief states "Bluetooth 4.0 specification compliant" / "Bluetooth v4.0 firmware" (dual-mode ROM audio platform with aptX and cVc).[12] Qualcomm's current product page, however, lists the specification field as Bluetooth 4.1. As with CSR8675, treat the qualified version per the specific brief / firmware you are using.[18b]
CSR8675 — Qualcomm's CSR8675 product brief lists Bluetooth version 4.2 (120 MHz Kalimba DSP; aptX HD; dedicated hardware ANC).[3] A caveat: some Qualcomm web pages present inconsistent specification fields for this part (e.g., a "5.0" value in one field versus "4.2" in the brief's Features section). Because of that inconsistency, treat "CSR8675 = Bluetooth 4.2" as the figure from the official product brief and firmware version, and verify the exact qualified version against the specific brief and firmware you are using rather than assuming 5.x support.
Note — the CSRA64xx ROM audio line (e.g., CSR64215). Parallel to the programmable CSR86xx parts, CSR also shipped a family of ROM-only stereo audio SoCs branded CSRA64xx (e.g., CSR64215 / CSRA64215, plus CSRA64110, CSRA64210). These are Bluetooth v4.2 single-chip ROM solutions with an 80 MHz RISC MCU and an 80 MIPS Kalimba DSP, built for TrueWireless Stereo (TWS), stereo headsets/speakers, and car audio. They support aptX, aptX Low Latency, AAC, SBC and the sixth-generation cVc voice enhancement. Because they are ROM-masked (not flash-programmable like the CSR8670/8675), they are not field-upgradable to a new Core version — but that is a hardware limitation of the ROM part, consistent with the field-upgrade note earlier in this article.[26]
The Qualcomm QCC Series (QCC300x, QCC30xx, QCC51xx)
Under Qualcomm, the audio roadmap moved to the QCC brand. Version splits within a family matter, so they are stated explicitly:
QCC300x (QCC3001–QCC3008, entry-level flash audio SoCs) — Qualcomm's product brief qualifies the family to Bluetooth 5.0.[14]
QCC30xx — per Qualcomm's QCC30xx brief: QCC302x / QCC303x → 5.1; QCC304x → 5.2; QCC305x / QCC307x → 5.3; QCC308x / QCC309x → 5.4. The 307x/308x/309x parts are designed for LE Audio use cases.[8]
QCC51xx — per Qualcomm's QCC5100 series brief, the split is: QCC512x → 5.1; QCC514x → 5.2; QCC515x and QCC517x → 5.3; QCC5181 → 5.4. The QCC517x / QCC518x parts are designed to support the LE Audio standard.[11]
The "QCC" naming is no longer the top of Qualcomm's audio stack. The Snapdragon Sound and S-series / S1-series Sound Platforms extend well beyond Bluetooth 5.4:
Qualcomm S1 Gen 1 Sound Platform (QCC1228) — Qualcomm's product page lists Bluetooth 6.0, targeting value-tier true-wireless earbuds with aptX and Adaptive ANC.[10]
Snapdragon S5 / S7 series — newer S7 Sound Platforms are documented with a Bluetooth 6.2 radio and LE Audio / Auracast experiences.[17]
So as of 2026, the newest Qualcomm audio silicon is listed with a Bluetooth 6.x specification version (e.g., a Bluetooth 6.2 radio on the S7 platforms, Bluetooth 6.0 on the S1 Gen 1) rather than topping out at 5.4.
Figure 1: The CSR → Qualcomm Bluetooth silicon lineage and the specification range each generation spans.
2.1 + EDR + early Bluetooth low energy; + eGPS + FM
Mobile phone connectivity hub
CSR8510
4.0 (dual-mode)
USB Bluetooth adapters
CSR101x
4.1 (LE)
Sensors, beacons, peripherals
CSR102x
up to 4.2 (LE)
LE peripherals (config-dependent)
CSR8615
4.1
Mono headsets, car kits, speakers
CSR8635
4.1 (per Qualcomm page; some listings 4.0)
Stereo headsets, speakers
CSR8645
4.1 (current page; old brief / firmware 4.0)
Mid-tier wireless audio
CSR8670
4.2
Premium wireless audio
CSR8675
4.2 (per brief; see field caveat)
ANC premium headphones
CSR64215 (CSRA64xx ROM audio)
4.2
TWS / stereo headsets / speakers / car audio
QCC300x (3001–3008)
5.0
Entry-level headsets / speakers
QCC302x / 303x
5.1
Entry true-wireless earbuds
QCC304x
5.2
Mid true-wireless, ANC
QCC305x
5.3
Snapdragon Sound (mid true-wireless)
QCC307x
5.3
Snapdragon Sound, LE Audio
QCC308x / 309x
5.4
LE Audio / Auracast
QCC512x
5.1
Premium earbuds / headsets
QCC514x
5.2
Premium, Snapdragon Sound
QCC515x / 517x
5.3
LE Audio-ready premium
QCC5181
5.4
LE Audio-ready premium
S1 Gen 1 (QCC1228)
6.0
Value-tier true-wireless
Snapdragon S7 series
6.2
Flagship, LE Audio / Auracast
Read the matrix with care: a "Bluetooth version" column shows the Core specification a part is qualified against. It does not by itself prove support for any specific higher-layer feature. LE Audio, the LC3 codec, Isochronous Channels, and Auracast are qualified separately (e.g., under BAP, PBP, and related specifications). See the next section.
What a "Bluetooth Version" Does and Doesn't Tell You
It is tempting to read a version number as a checklist of features. The Bluetooth SIG is explicit that most capabilities are optional,[16b] and a Core version alone does not prove a product supports a given function. Two common confusions:
LE Audio is not "Bluetooth 5.3" or "5.4." LE Audio is built on the LE Isochronous Channels introduced in Bluetooth Core 5.2. A chip qualified to 5.2 (or 5.3/5.4) may implement LE Audio, but only if it has also passed the relevant separate qualifications (BAP, PBP, LC3, etc.). A "5.2-qualified" radio is therefore not automatically LE-Audio-capable.[16a]
Auracast is part of the LE Audio specification set, not a Bluetooth 5.4 feature. Auracast broadcast audio is enabled by LE Audio qualifications (notably PBP); it is not a Core 5.4 addition. The 5.4 Core release's own headline additions are PAwR and Encrypted Advertising Data.[16c]
For selection, verify the specific feature qualifications (BAP, PBP, LC3, and so on) for the exact part and firmware, rather than inferring them from the Core version.
How to Choose the Right Chip
For new true-wireless designs wanting LE Audio / Auracast, start at QCC307x / 308x / 309x (5.3 / 5.4) or the S1 / S7 platforms (6.0 / 6.2) — and confirm the specific LE Audio / Auracast qualifications, not just the Core version.
For cost-sensitive earbuds with solid aptX Adaptive, QCC304x (5.2) remains a strong, widely available choice.
For legacy product maintenance or aptX HD audio, CSR8670 / 8675 (4.2) are still supported through the ADK toolchain.
For USB host-side adapters, CSR8510 (4.0) is the classic, mature option.
A chip's Bluetooth specification version is tied to its radio hardware and original qualification. Firmware cannot add PHY or controller capabilities the silicon does not physically support. However, where the hardware already includes the necessary radio/controller features, a vendor can enable additional functionality — for example, some Auracast-related capabilities — through firmware, stack updates, and re-qualification; the Bluetooth SIG notes that the specification allows certain existing products to be upgraded in the field. Treat "version locked" as "locked by hardware," not as an absolute software prohibition.
Frequently Asked Questions
Is CSR still a separate company?
No. Qualcomm announced the acquisition in October 2014 and completed it on August 13, 2015; the business now operates as Qualcomm Technologies International, Ltd. (QTIL). "CSR" survives mainly as a product-name legacy on older parts.[4]
Does the CSR8670 support Bluetooth 5?
No. The CSR8670 is qualified to Bluetooth 4.2 per its product brief.[1] For Bluetooth 5 features you need a Qualcomm QCC-series part (or newer). Note the CSR8675 carries the same 4.2 brief figure, with some Qualcomm pages showing inconsistent fields — verify the exact qualified version against the brief and firmware you use.
What is the newest Bluetooth version in the CSR / Qualcomm audio lineup?
As of 2026 it is not 5.4. Qualcomm's S1 Gen 1 Sound Platform (QCC1228) is qualified to Bluetooth 6.0, and the newer Snapdragon S7 series documents a Bluetooth 6.2 radio.[10][17] The QCC308x / 309x (5.4) are simply the top of the older QCC30xx family.
If a chip is "qualified to Bluetooth 5.2," does it support LE Audio?
Not necessarily. Bluetooth 5.2 introduced the LE Isochronous Channels that enable LE Audio, but LE Audio itself (BAP/PBP/LC3 and related qualifications) is optional.[16a][16b] A 5.2-qualified part may or may not implement LE Audio — check the specific feature qualifications for that exact part and firmware.
Can a chip's Bluetooth version be upgraded through a firmware update?
Generally not at the Core level if the required radio/controller hardware is absent — firmware cannot invent PHY or controller capabilities the silicon lacks. But where the hardware already supports the necessary features, a vendor can add functionality (for instance, some Auracast capabilities) via firmware, stack updates, and re-qualification; the Bluetooth SIG states the specification permits some existing products to be field-upgraded.[16b][16c] So "version locked" means "locked by hardware," not an absolute software rule.
PUBLISHED BY IWISTAO · Tube Amplifier Fundamentals
How the cathode resistor establishes a valve's operating point, shapes gain and frequency response, and protects reliable operation.
“Cathode resistance” usually refers to the cathode resistor, shown as Rk in a schematic. It may look like a minor part, but in a cathode-biased tube amplifier it helps determine idle current, grid-to-cathode bias, linearity, gain, headroom, and heat. Selecting it is therefore an operating-point decision—not simply a matter of copying the nearest standard value.
1. What the Cathode Resistor Does
A cathode resistor connects the tube cathode to the circuit's DC reference, normally ground. As cathode current flows through the resistor, it creates a positive cathode voltage. If the control grid remains near zero volts DC through its grid-leak path, the grid becomes negative relative to the cathode. This is cathode bias, also called self-bias or automatic bias.[1]
Figure 1: A simplified common-cathode triode stage. Rk sets the DC self-bias; Ck optionally changes the AC gain. Original IWISTAO diagram.
2. How Self-Bias Works
Vk = Ik × RkWith the grid near 0 V DC: Vgk ≈ −Vk
If current rises, the voltage across Rk rises. That makes the grid more negative relative to the cathode and tends to oppose the original current increase. The resistor therefore introduces local negative feedback and makes the operating point partly self-correcting. This does not make the current perfectly constant: tube characteristics, supply voltage, screen voltage, resistor tolerance, and temperature still matter.
For a triode, cathode current is essentially plate current plus very small grid current under normal small-signal operation. For a tetrode or pentode, cathode current includes both plate and screen current. A voltage measurement across Rk therefore reveals total cathode current, not exact plate current.
3. Choosing the Resistance
Begin with the tube manufacturer's operating data and plate curves. Select a plausible plate voltage, load, and idle current; determine the required grid-to-cathode bias; then estimate the resistor:
Rk = Vk ÷ Ik
After choosing the nearest standard value, verify the resulting operating point on the curves or in a proven circuit. A larger Rk generally produces a higher cathode voltage and lower idle current; a smaller Rk generally runs the tube at higher idle current. Because tube curves are nonlinear, the change is not exactly proportional.
A real reference point is the Fender 5E1 Champ's first 12AX7 stage: a 1.5 kΩ cathode resistor produces about −1.4 V of DC grid bias in the cited analysis, close to Fender's stated −1.5 V measurement.[4] This is useful context, not a universal recipe; the supply, plate load, tube type, and desired headroom must be considered together.
4. Wattage, Temperature, and Resistor Type
PRk = Ik2Rk = Vk2 ÷ Rk = VkIk
Do not select a resistor whose printed wattage merely equals the calculated dissipation. Allow thermal margin, check the manufacturer's derating curve, and consider the hot environment inside a tube chassis. Vishay notes that allowable dissipation falls above the specified ambient-temperature threshold and depends on heat removal.[5] A practical design often uses at least twice the calculated steady dissipation, with greater margin where ventilation is poor or reliability is critical.
Application
Common approach
What to verify
Small-signal preamp
Metal-film or metal-oxide resistor
Resistance tolerance, noise, voltage, and modest power dissipation
Single-ended power stage
Flameproof metal-oxide, cement wirewound, or suitable power resistor
Wattage derating, surface temperature, spacing, and ventilation
Chassis-mounted power part
Aluminum-housed resistor on an appropriate heat sink
Datasheet mounting conditions and electrical isolation
5. The Cathode Bypass Capacitor
An unbypassed cathode resistor carries both DC and signal-related current. The changing cathode voltage opposes the input signal—a process called cathode degeneration. It reduces gain, but it can also reduce distortion and increase headroom.[2]
Placing Ck across Rk leaves the DC bias substantially unchanged while shunting part of the AC cathode signal. A large capacitor can make the stage nearly fully bypassed across the audio band; a smaller capacitor creates a shelved response with less low-frequency gain and more high-frequency gain relative to the unbypassed condition.[3]
Figure 2: Conceptual gain trends. Exact transition frequency and shelf height depend on the tube and surrounding circuit, not only Rk and Ck. Original IWISTAO diagram.
The familiar estimate f ≈ 1/(2πRC) is useful for orientation, but the relevant AC resistance is not always just the marked cathode resistor. Tube transconductance, internal plate resistance, plate load, and following-stage load affect the exact response. Use a load-line or small-signal model when precision matters.
6. Shared vs. Individual Cathode Resistors
Two output tubes may share one cathode resistor and bypass capacitor, or each tube may have its own pair. A shared resistor is simple and historically common, but the measured current is the sum of both tubes. One strong tube can mask one weak tube, and imbalance is harder to diagnose. Individual resistors make current checks and fault isolation easier. A shared arrangement should use reasonably matched tubes and a resistor/capacitor pair rated for the combined current.
7. Two Worked Examples
Example A: a small-signal triode stage
Suppose a measured cathode voltage is 1.5 V across 1.5 kΩ. The cathode current is:
Ik = 1.5 V ÷ 1500 Ω = 1.0 mA PRk = 1.5 V × 1.0 mA = 1.5 mW
A 0.25 W resistor has ample dissipation margin here, assuming its voltage, temperature, and construction ratings are suitable.
Example B: an illustrative power stage
Suppose a single power tube measures 20 V across a 470 Ω cathode resistor:
Ik = 20 V ÷ 470 Ω = 42.6 mA PRk = 202 ÷ 470 = 0.85 W
A 1 W part would be a poor thermal choice. A 3 W or 5 W resistor may be more appropriate, subject to its datasheet and chassis temperature. Remember that 42.6 mA includes screen current in a pentode or beam tetrode, so it must not be treated as exact plate current when calculating plate dissipation.
8. Troubleshooting Cathode-Bias Problems
Too little cathode voltage: possible low tube current, a resistor that has drifted low, a leaky/shorted bypass capacitor, or a wiring fault.
Too much cathode voltage: possible excessive tube current, a resistor that has drifted high, incorrect supply conditions, or tube faults.
Weak gain or altered tone: an open or dried-out bypass capacitor may remove intended AC bypassing while leaving DC bias apparently normal.
Red-plating or overheating: switch off immediately. Check the tube, bias network, screen supply, coupling-capacitor leakage, and component values before further operation.
Measurements that disagree: confirm meter reference, resistor tolerance, warm-up time, supply voltage, and whether the resistor is shared by multiple tubes.
9. High-Voltage Safety
Tube amplifiers can retain dangerous energy after being unplugged. Power-supply capacitors may remain charged. OSHA requires hazardous stored electrical energy to be released and capacitors to be discharged before work; circuits must also be verified de-energized.[6] If you are not trained to work safely around high-voltage equipment, use a qualified technician.
A cathode resistor is a bias component, a feedback element, and a heat source at the same time. Read its voltage as evidence of the operating point—but interpret that measurement in the context of the complete tube circuit.
Frequently Asked Questions
Does a larger cathode resistor always make a tube run colder?
Usually it reduces idle current by creating a more negative grid-to-cathode bias, but the exact result depends on supply voltage, screen voltage, load, and the tube's nonlinear characteristics. Verify the new operating point rather than assuming a proportional change.
Does the bypass capacitor change the DC bias?
An ideal capacitor does not. It changes the AC feedback around the cathode resistor while the resistor continues to set the DC bias. A leaky or shorted real capacitor can disturb the bias and must be replaced.
Can I calculate tube current from cathode voltage?
Yes: divide cathode voltage by cathode resistance. For pentodes and beam tetrodes, however, the result is total cathode current—plate current plus screen current—not plate current alone.
Why use a 5 W resistor when the calculation says only 1 W?
Extra rating lowers operating temperature and provides margin for component tolerance, supply variation, ventilation, and temperature derating. The correct margin must still be checked against the selected resistor's datasheet.
Can I replace a cathode resistor with the same value but a different type?
Only if the replacement also satisfies power, voltage, temperature, tolerance, flameproof, mounting, and—where relevant—inductance requirements. Physical clearance matters for hot power resistors.
How a 1950s VHF transmitter tube found a second life as one of the most compelling DIY headphone amplifier projects in the audiophile world
Introduction: The Tube That Shouldn't Work for Audio
The FU19 — known in Western nomenclature as the 5894 — was never designed for audio. Introduced in 1955 by RCA and widely manufactured across Europe (as the QQV06-40) and China, this dual beam tetrode was built for VHF transmitter applications. Its original design brief called for reliable power amplification at frequencies up to 250 MHz, with usable performance extending to 500 MHz — the domain of FM broadcast and military communications equipment, not living-room hi-fi.
Figure 1: A FU19 (5894) dual beam tetrode in operation, showing the characteristic amber filament glow through its smokey glass envelope.
Yet here we are in 2026, and the FU19 has quietly become one of the most intriguing tubes in the budget-to-midrange DIY headphone amplifier space. The architecture examined here uses one 6N2 (6N2P) dual-triode voltage amplifier followed by two FU19 output tubes—one complete FU19 per channel. Within each FU19, the two beam-tetrode sections are connected in parallel and operated as a single-ended Class A output device. Because the screen grids are internally common, this one-tube-per-channel arrangement also permits a technically coherent triode connection, with the common screen node tied to the paralleled anodes through a screen-stopper resistor. Fully assembled kits can be found for under $200, while scratch builds with premium output transformers can be realized for $300–500.
This guide is a deep technical exploration of every aspect of this amplifier: the tube physics, the circuit topology, the component selection, the build process, the measured performance, and — most importantly — what it actually sounds like through a good pair of headphones.
The FU19 Tube: A Technical Deep Dive
Physical Construction and Heritage
The FU19 is a dual beam tetrode housed in a single glass envelope, using a B7A (septar) 7-pin base. The envelope measures approximately 44 mm in diameter and 74 mm in height excluding base pins. Inside, two independent tetrode sections share a common screen grid, with a center-tapped heater allowing operation at either 6.3 V or 12.6 V — a design choice originally made for compatibility with both mains transformer supplies and 12 V vehicle battery systems in mobile radio applications.
The tube's internal construction gives it a distinctive smokey appearance — an intentional coating on the inside of the glass that suppresses secondary electron emission and stabilizes high-frequency behavior. This makes photographing the internal structure difficult, but it contributes to the tube's visual character when operating: a warm, amber glow that diffuses through the treated glass.
Figure 2: Macro view of a vacuum tube filament structure. The FU19's dual independent cathodes with common screen grid give it unique electrical characteristics among audio power tubes.
Electrical Specifications
The following table summarizes the FU19/5894's key specifications, drawn from original RCA datasheets and the Valve Museum reference collection. Note that these are absolute maximum ratings for Class C RF service. Audio amplifier operation is considerably more conservative.
Parameter
Value
Notes
Heater Voltage
6.3 V (parallel) / 12.6 V (series)
Center-tapped, pin 1 & 7
Heater Current
1.6 A @ 6.3 V / 0.8 A @ 12.6 V
Per datasheet; Chinese variants may draw ~1.8 A
Max Anode Voltage (Va)
400 V (DC)
Absolute maximum; audio use typically 250–320 V
Max Screen Voltage (Vs)
145 V
Often tied to a regulated supply in audio circuits
Typical Grid Bias (Vg)
−45 V
Class C operating point
Max Anode Current (Ia)
170 mA per section
Total for both sections
Max Screen Current (Is)
17 mA
Plate Dissipation (per section)
20 W
Each anode can safely dissipate 20 W
Transconductance (gm)
~4.5 mA/V
At typical operating point
Datasheet RF Load Resistance
Application-dependent
Do not directly treat RF or push-pull load values as the optimum SE audio primary
RF Power Output
44 W
Class C, at 125 MHz
Frequency Range
DC–500 MHz
Full ratings to 250 MHz
Base Type
B7A (Septar)
7 pins, larger than octal
Introduced
1955
RCA; European equivalent QQV06-40
Key insight: When operated in single-ended Class A audio service with the screen grid tied to the anode (triode-strapped), the FU19's effective plate resistance drops significantly — from the 4.4–8 kΩ tetrode value down to roughly 600–800 Ω. This makes it compatible with a wider range of output transformer primary impedances and yields a more linear transfer characteristic at the cost of reduced power output.
Why a Transmitter Tube for Audio?
Using RF transmitter tubes in audio circuits is not without precedent. The ubiquitous 807 beam tetrode — originally designed for the same VHF transmitter role — has been a staple of DIY audio for decades. The FU19 shares several characteristics that make RF tubes surprisingly well-suited to audio:
Robust construction: Transmitter tubes are built to withstand continuous high-power operation. Their cathodes are typically larger, with higher emission reserves, leading to longer service life under the relatively gentle conditions of audio amplification.
High perveance: The FU19's cathode is designed for high peak current delivery, which translates to good dynamic headroom in audio — the ability to deliver transient peaks without sag or compression.
Parallel-section capability: Each FU19 contains two power sections with a common screen-grid connection. In this design, both sections of one tube are paralleled for one audio channel, increasing current capability and allowing the common screen grid to be triode-strapped correctly. A stereo amplifier therefore uses two matched FU19 tubes, one for the left channel and one for the right.
Low interelectrode capacitance: Designed for VHF, the FU19 has inherently low Cag (anode-grid capacitance), around 0.15 pF per section, reducing the Miller effect and improving high-frequency stability without complex compensation networks.
The 6N2 Driver Stage: The Soviet 12AX7
The 6N2 (Russian designation: 6Н2П, also written 6N2P) is a miniature 9-pin dual triode manufactured extensively in the former Soviet Union, Russia, and China. It is functionally equivalent to the Western 12AX7 / ECC83 in most audio applications, with one critical difference: its heater is wired for 6.3 V only (pins 4 and 5), not the series/parallel 12.6 V / 6.3 V arrangement of the 12AX7.
The 6N2 brings several advantages to the driver role:
Parameter
6N2 (6N2P)
12AX7 / ECC83
Heater
6.3 V @ 340 mA
6.3 V @ 300 mA / 12.6 V @ 150 mA
Amplification Factor (μ)
100
100
Plate Resistance (rp)
~62.5 kΩ
~62.5 kΩ
Transconductance (gm)
~1.6 mA/V
~1.6 mA/V
Max Plate Voltage
300 V
300 V
Max Plate Dissipation
1 W per triode
1 W per triode
Typical Service Life
5,000+ hours
5,000+ hours
Cost (2026, new)
$3–8
$15–50
The 6N2 provides two triode systems, allowing one voltage-amplifier section to drive each channel. In a common-cathode stage with a 100–150 kΩ anode load and a 1.5–2.2 kΩ cathode resistor, practical loaded gain is typically lower than the unloaded amplification factor suggests. The exact operating point should be chosen so that each 6N2 section can supply the required FU19 grid swing with adequate headroom and acceptably low distortion; voltage gain alone is not proof of output-swing capability.
Figure 3: Signal flow and block diagram of the complete 6N2 + FU19 headphone amplifier.
Circuit Topology: How It All Connects
The Classic 6N2 + FU19 Architecture
This article now assumes a true dual-output-tube stereo layout: the 6N2 contains the left- and right-channel voltage-amplifier triodes, while the power stage uses two FU19 tubes in total. The left FU19 serves only the left channel and the right FU19 serves only the right channel. Within each output tube, both internal beam-tetrode systems are paralleled.
The most commonly encountered configuration — and the one used in the majority of Chinese-sourced kits — follows this signal path:
Input: RCA line-level input, typically 2 V RMS from a DAC or preamplifier.
Volume control: A 50 kΩ or 100 kΩ logarithmic (audio taper) potentiometer at the input, which also serves as the grid-leak resistor for the 6N2.
6N2 voltage amplifier: One triode per channel, common-cathode topology. Anode load resistor: 100–150 kΩ. Cathode resistor: 1.2–2.2 kΩ, bypassed with a 47–100 µF electrolytic to maximize gain. The amplified signal is coupled to the FU19 grid through a 0.22–0.47 µF film capacitor (polypropylene or PIO preferred).
FU19 output stage: Each channel uses one complete FU19. The two anodes are paralleled, the two control grids are paralleled through individual grid-stopper resistors, and the two cathodes are paralleled at a common cathode-bias network. The internally common screen grid is tied to the paralleled anodes through a 100–220 Ω screen-stopper resistor for triode-strapped Class A operation.
Output transformer: Each channel requires its own air-gapped single-ended output transformer. A primary in the approximate 2.5–4 kΩ range is a more plausible starting point for two paralleled FU19 sections, but the final value must be selected from the actual operating point and load line. Secondary taps may be provided for 32 Ω, 150 Ω, 300 Ω, and 600 Ω headphones.
Operating Point Analysis
A well-designed FU19 audio output stage typically operates at the following quiescent point:
Parameter
Typical Value
Rationale
Plate Voltage (Va)
280–320 V
Well within the 400 V maximum; provides adequate voltage swing
Plate Current (Ia)
30–40 mA per section; 60–80 mA per channel
Both sections of each FU19 operate in parallel
Screen Voltage (triode mode)
Equal to Va
Screen tied to plate via 100 Ω stopper resistor
Grid-to-Cathode Bias
Approximately −25 to −35 V
Developed by a shared cathode resistor for the two paralleled sections
Estimated Output Power
Approximately 4–7 W per channel
Engineering estimate; transformer and operating-point dependent
Primary Impedance
Approximately 2.5–4 kΩ
Starting range for two paralleled sections; verify by load-line analysis
Design note: Several watts of available output are far beyond the continuous power required by conventional headphones. Maximum sound pressure must be calculated from the headphone manufacturer's stated sensitivity convention—dB/mW and dB/V are not interchangeable. A stepped attenuator or reliable logarithmic volume control, a turn-on delay, and protection against switching transients are strongly recommended.
Triode-Strapped vs. Tetrode Operation
One of the most consequential design decisions is how to handle the FU19's screen grid. The two common approaches:
Triode-strapped parallel operation: In each channel, the FU19's two anodes are paralleled and its internally common screen-grid node is connected to that paralleled-anode node through a 100–220 Ω resistor. The two control grids should each have their own small grid-stopper resistor before joining at the drive node. This arrangement converts the two internal beam tetrodes into a single higher-current pseudo-triode. Compared with tetrode operation it generally reduces gain, power and plate resistance while improving linearity and simplifying the common-screen connection.
Tetrode mode (with a dedicated screen supply): The common screen-grid connection of each FU19 can instead be fed from a separate, well-decoupled supply chosen from the applicable datasheet conditions. This can increase available output power, but it requires additional supply design and careful control of screen dissipation. The left and right tubes should not share an inadequately decoupled screen node, because signal-dependent screen current can increase channel interaction.
For this two-FU19 headphone amplifier, triode-strapped parallel operation is the most straightforward implementation because each tube serves only one channel and its common screen-grid connection can be returned to the same channel's paralleled anodes. Final performance nevertheless depends heavily on the output transformer, grounding, heater supply and the chosen load line.
Figure 4: The headphone amplifier part for tube 6N2 drive dual FU19
Sound Quality: Subjective Impressions
Describing audio subjectively is inherently limited, but certain characteristics of the FU19 amplifier are consistently reported across multiple builders and reviewers. The following observations are drawn from DIY community discussions, listening comparisons, and the author's own experience with a properly built unit using Sennheiser HD 650 (300 Ω) and Beyerdynamic DT 880 (250 Ω) headphones.
Tonal Balance
The FU19 amplifier presents a slightly warm, mid-forward balance. The bass extends cleanly but does not have the iron-fisted control of a high-damping-factor solid-state amplifier. Instead, bass notes have a rounded, organic quality — the leading edge of a kick drum has weight and body rather than just transient snap. This is characteristic of single-ended triode (SET) amplifiers and is part of their enduring appeal.
The midrange is where the FU19 truly shines. Vocals — particularly female vocals and acoustic instruments — are rendered with a presence and dimensionality that is difficult to achieve with solid-state circuits at anywhere near this price point. There is a sense of the singer being "in the room" that is the hallmark of well-designed tube amplification.
Treble is extended but never fatiguing. Cymbals and high-frequency percussion have natural decay without harshness or grain. The FU19 does not roll off the treble the way some vintage tube designs do; instead, it presents high frequencies with a smoothness that belies the tube's RF design heritage.
Soundstage and Imaging
The soundstage is wide and layered, with good instrument separation. Depth is particularly impressive — instruments are placed not just left-to-right but front-to-back in a convincing spatial presentation. This is likely a function of the triode-strapped configuration's inherently low phase distortion and the use of high-quality output transformers.
Noise Floor
With careful layout and AC heater wiring (or better, a DC heater supply for the 6N2), the noise floor is vanishingly low. Through 300 Ω headphones, hum and hiss are inaudible at normal listening levels. Users of high-sensitivity IEMs (in-ear monitors, typically >110 dB/mW) should be aware that some residual hum may be detectable — this is a limitation of single-ended AC-heated tube designs, not specific to the FU19.
Estimated Performance Targets
Until a specific prototype is measured under controlled conditions, the following figures should be treated as engineering targets rather than verified test results. They assume one FU19 per channel with its two sections paralleled, triode-strapped Class A operation, an appropriately air-gapped output transformer, and a 300 Ω resistive test load. Actual results will depend strongly on the transformer, supply voltage, quiescent current, feedback arrangement, layout and tube samples.
Measurement
Value
Conditions
Estimated Maximum Output
Approximately 5–7 W per channel
Target range, not a verified measurement
Conservative Rated Output
Approximately 3–5 W per channel
Depends on acceptable THD and transformer performance
Frequency Response
Transformer-dependent
Must be measured at rated power and specified load
THD @ 1 W, 1 kHz
To be measured
Expected to be dominated by low-order harmonics without global feedback
THD Near Rated Output
To be measured
Will rise progressively as the single-ended stage approaches clipping
IMD (SMPTE)
To be measured
Specify test level, bandwidth and load
Signal-to-Noise Ratio
Design target: >80 dB A-weighted
Requires careful heater, grounding and transformer layout
Output Impedance
To be measured for each tap
Do not infer one secondary tap from another
Input Sensitivity
~500 mV
For full output
Input Impedance
50–100 kΩ
Volume pot value
Power Consumption
Approximately 70–100 W
Estimate for two FU19 tubes plus heaters and losses
Figure 5: Illustrative harmonic spectrum showing a possible low-order distortion pattern. The bar heights are conceptual and must not be presented as measurements from a completed amplifier.
An unfeedback single-ended triode-connected stage often shows a relatively strong second-harmonic component, but the exact spectrum cannot be known without measurement. Transformer nonlinearity, bias point, tube matching, drive-stage distortion and output level all affect the result. The illustration above therefore shows only the type of low-order roll-off a designer might target, not guaranteed performance.
Tube Amplifier vs. Solid-State: A Technical Comparison
Understanding where the FU19 amplifier fits requires comparing its fundamental operating characteristics against typical solid-state headphone amplifiers in the same price range ($200–500).
Characteristic
FU19 Tube Amp
Solid-State (e.g., OPA-based)
Output Impedance
~2.5 Ω (32 Ω tap)
< 0.1 Ω (near-zero)
Damping Factor (300 Ω)
~120
> 3,000
THD @ 1 W
0.15–0.3%
< 0.001%
Dominant Distortion
2nd harmonic (even-order)
3rd harmonic (odd-order) — much lower amplitude
SNR
~85 dB
~110 dB
Power Bandwidth
20 Hz – 35 kHz
DC – 200+ kHz
Thermal Behavior
Requires warm-up, runs hot
Immediate, runs cool
Tube Life Expectancy
3,000–5,000+ hours
N/A (solid-state, decades)
User Serviceability
Fully repairable, socketed tubes
Difficult to repair at component level
Bottom line: Solid-state amplifiers objectively measure better on almost every metric. The FU19 tube amplifier exists for a different reason — its subjective presentation, with a euphonic harmonic structure and an expansive soundstage, appeals to listeners who prioritize musical engagement over analytical accuracy. Both approaches are valid; they serve different listeners and different moods.
Building Your Own: Key Construction Notes
Safety First
Vacuum tube circuits operate at lethal voltages — typically 280–350 V DC on the plates, and often higher in the power supply before regulation. Never work on a powered amplifier, and always discharge filter capacitors before touching any internal connections. A 100 kΩ / 5 W resistor with insulated leads makes an effective discharge tool — connect it across each filter capacitor for 10–15 seconds with power disconnected.
Layout Considerations
Star grounding: Bring all ground returns to a single point on the chassis. This is the single most important layout rule for minimizing hum.
Heater wiring: Twist the 6.3 V AC heater wiring tightly and route it close to the chassis, away from signal-carrying wires. For the lowest noise floor, consider a DC heater supply for the input tube (6N2).
High-voltage separation: Keep B+ and signal wires physically separate. If they must cross, do so at right angles.
Transformer placement: Position the power transformer and output transformers at least 50 mm apart, with their magnetic cores oriented at 90 degrees to each other to minimize inductive coupling.
Grid stopper resistors: Mount one grid-stopper resistor at each FU19 control-grid connection, as close to the socket pins as physically possible, before joining the two grid feeds at the common drive node. This suppresses parasitic VHF oscillation — a real concern given the FU19's 500 MHz bandwidth.
Tuning the Bias
For cathode-bias operation, each FU19 uses one shared cathode resistor for its two paralleled sections. A reasonable initial target is approximately 60–80 mA total cathode current per channel, or roughly 30–40 mA per internal section. Measure the voltage across the shared cathode resistor and calculate total current using Ohm's law: I = V / R. For example, 30 V across 430 Ω corresponds to approximately 70 mA total, before allowing for screen current.
Increase the resistor value to reduce current or decrease it to raise current. Verify the dissipation of each internal section, not merely the total tube current, and confirm that the two sections share current reasonably evenly. The cathode resistor and bypass capacitor must be rated for the combined current and heat of both sections.
Figure 6: Conceptual load-line illustration for one FU19 channel with both internal sections paralleled. This is not a substitute for verified manufacturer curves or prototype measurements.
Tube Rolling: Experimenting with the 6N2
Tube rolling — swapping different tubes of the same type to explore sonic variations — is one of the joys of tube amplifier ownership. While the FU19 itself has limited alternatives (the Soviet FM30 and the rare QQV06-40 are direct equivalents), the 6N2 driver stage offers more room for experimentation.
With a minor heater rewire (changing from 6.3 V parallel to 12.6 V series configuration on the 9-pin socket), the following Western equivalents can be substituted:
12AX7 / ECC83: The direct Western equivalent. Generally smoother and more refined than standard Chinese 6N2 variants, but 3–5 times the price. NOS Telefunken or Mullard examples are highly prized.
5751: A lower-gain (μ = 70) variant with excellent linearity. Reduces overall gain slightly — useful if the amplifier has too much gain for modern high-output DACs.
12AT7 / ECC81: Lower gain (μ = 60) and higher current capability. Not a direct swap — bias point adjustment required — but some builders prefer the more dynamic presentation.
Caution: Always verify heater wiring before swapping tubes. The 6N2 has a 6.3 V-only heater (pins 4–5). Plugging a 12AX7 into a 6N2 socket without rewiring will under-heat the tube, while plugging a 6N2 into a 12AX7 socket wired for 12.6 V will destroy it.
FU19 Alternatives and Output Tube Interchangeability
While the FU19 (5894) is the design's intended output tube, several alternatives can serve in its place. Understanding these options — from direct drop-in equivalents to tubes requiring circuit modifications — gives you flexibility in sourcing, sonic tuning, and long-term maintainability.
Direct Equivalents (Drop-in Replacement)
These tubes share the FU19's B7A (septar) 7-pin base and operating characteristics. They can be installed without any circuit modifications.
Equivalent design; some batches rated for higher plate dissipation
$10–25 (NOS)
5894 (NOS)
USA (RCA, GE, Sylvania)
Original Western designation; NOS examples highly prized
$40–100+ (NOS)
QQV06-40 tubes, produced mainly by Mullard and Philips for European military and broadcast equipment, typically exhibit lower microphonics and more consistent matching between the two tetrode sections. FM30 tubes from the former Soviet Union are often the most cost-effective NOS option and perform nearly identically to the FU19. 5894 NOS from major American manufacturers commands a premium but offers the assurance of known manufacturing provenance.
Near-Equivalents (Moderate Circuit Changes)
Several tubes can be adapted to the FU19 socket or amplifier circuit with moderate modifications. These alternatives offer different sonic characters and may provide performance benefits in specific areas.
Alternative Tube
Modification Required
Sonic Impact
Power Output (Triode)
EL84 / 6BQ5
Novar base; socket change and rewiring
More detailed midrange; tighter, faster bass
~3–4 W
6V6GT
Octal base; adapter or chassis modification
Warmer, more "vintage" character; slightly softer highs
~2–3 W
6AQ5
Novar base; pinout incompatible with B7A
Similar to EL84; slightly lower headroom
~3 W
EL86 / 6CW5
Different base; higher screen voltage supply
Higher power; different load line; more solid-state-like damping
~5–7 W (needs circuit revision)
Practical note: For most builders, the effort of adapting a different output tube is not justified by the sonic gains. The FU19's combination of readily available NOS stock, dual-section convenience, and excellent audio performance makes it difficult to improve upon without a complete circuit redesign. Tube rolling is most rewarding at the driver stage (6N2 → 12AX7 family), where the impact on noise floor and tonal character is more immediately audible.
Sourcing and Matching Advice
When purchasing replacement FU19 or equivalent tubes, several practical considerations will save you time and money:
Internal-section current sharing: Because both power sections of each FU19 are paralleled, reasonably similar transconductance and emission are desirable so that neither section carries a disproportionate share of the current. For stereo balance, the complete left and right FU19 tubes should also be matched as closely as practical at the intended operating point.
NOS vs. New Production: As of 2026, "new production" FU19 tubes are rare. Most available stock is NOS from military or commercial surplus. Reputable sources include specialized tube retailers and established auction platforms. Exercise caution with unusually low-priced "NOS" — counterfeit and relabeled tubes exist in the vintage tube market, particularly for popular RF tubes like the 5894.
Chinese vs. Russian stock: Chinese-produced FU19 tubes (often branded as "Psvane" or "Full Music") appeared in the 2000s but have inconsistent quality control. Russian NOS FM30 tubes from the 1970s–80s typically offer better consistency, lower noise, and more reliable heater performance.
Testing before installation: Any NOS tube should be tested for heater continuity, grid leakage, and emission before installation. A simple multimeter check for heater continuity (pins 1 and 7 for parallel wiring, or pins 1 and 3 for series depending on heater configuration) is the minimum; a full tube tester provides a comprehensive assessment of the tube's condition.
Getter condition: Visually inspect the getter (the silvery coating on the interior of the envelope). A white or flaking getter indicates a leak and a non-functional tube. A uniformly silvery getter is a good sign, though not a guarantee of performance.
When to Consider an Alternative Output Tube
Several scenarios may warrant moving away from the FU19:
Availability crisis: If FU19/QQV06-40 stock becomes prohibitively expensive or unavailable, the EL84/6BQ5 is the most logical alternative. It requires a noval socket but is widely available in both NOS and new production from multiple manufacturers.
Higher power requirement: For driving very low-impedance headphones or occasional speaker use, the EL86 (6CW5) provides higher power but requires circuit revisions (higher screen voltage, different output transformer).
Sonic preference: If the FU19's presentation feels too warm or lacks detail in the upper midrange, the EL84 family offers a more neutral, detailed alternative — though this comes at the cost of the FU19's distinctive musical engagement and harmonic richness.
Form factor constraints: The FU19's B7A base is larger than standard noval or octal tubes. If chassis space is extremely limited, an EL84-based design with a smaller tube may be preferable despite the sonic trade-offs.
Long-Term Ownership: Maintenance and Care
A well-built FU19 amplifier requires relatively little maintenance. Key points for long-term reliability:
Tube replacement interval: The 6N2 typically lasts 5,000+ hours under normal conditions. The FU19's transmitter-grade cathode should last 3,000–5,000 hours. Replace when you notice increased hum, reduced output, or a change in tonal character.
Capacitor aging: Electrolytic filter capacitors have a finite lifespan — typically 10–15 years for quality units. If the amplifier develops increased hum that persists with new tubes, the filter capacitors are the first suspect.
Socket cleaning: Oxide buildup on tube pins and sockets can cause intermittent noise. Clean with DeoxIT or similar contact cleaner every 2–3 years.
Bias check: Measure the cathode voltage of the FU19 annually to confirm the bias point hasn't drifted. A shift of more than 10% warrants investigation.
Frequently Asked Questions
What headphones work best with the FU19 amplifier?
High-impedance dynamic headphones (150–600 Ω) are the ideal match. The Sennheiser HD 600/650/6XX (300 Ω) and Beyerdynamic DT 880/990 (250 Ω or 600 Ω) are frequently paired with this amplifier. Low-impedance planars (sub-50 Ω) generally prefer solid-state amplification with high current delivery and low output impedance, though using the 32 Ω output tap can provide acceptable results with less demanding planars.
Can the FU19 amplifier drive speakers?
With one complete FU19 per channel and both sections paralleled, several watts of audio output may be possible, so efficient near-field loudspeakers are theoretically feasible. However, headphone output transformers are normally wound for 32–600 Ω loads and are not suitable for 4–8 Ω speakers unless they include a correctly designed low-impedance secondary with adequate core size and current capability.
How hot does the amplifier get?
The FU19 envelope can reach 150–180°C during normal operation, and the power supply section generates additional heat. Adequate ventilation is essential — do not enclose the amplifier in a cabinet without airflow. The chassis should be warm but not uncomfortably hot to touch. If any component is too hot to hold a finger against for 3 seconds, investigate.
Do I need a preamplifier?
For most modern sources (DACs, CD players, phono preamps with 2 V RMS output), no additional preamplifier is needed. The 6N2 provides sufficient gain. For low-output sources (some phono cartridges, older tuners), a dedicated phono preamp or line-stage preamplifier may be beneficial.
Is this a good first DIY tube project?
With caveats. The circuit itself is relatively simple — two stages per channel, straightforward power supply. However, the high voltages involved demand respect and proper safety precautions. If you are comfortable with basic electronics, can read a schematic, and understand high-voltage safety, the FU19 amplifier is an excellent project. Complete beginners should start with a low-voltage solid-state project first, or work alongside an experienced builder.
How does this compare to commercially available tube headphone amps?
In terms of raw circuit topology and component quality (with premium parts), a well-built FU19 amplifier can compete with commercial tube headphone amplifiers in the $500–1,000 range. The key variable is the output transformer quality — investing in good transformers (Lundahl, Edcor, Hashimoto) yields substantial improvements over the budget transformers included in low-cost kits. However, commercially available amplifiers offer refinement in chassis design, relay-based input switching, remote control, and warranty support that DIY builds may not match.
