Updated note: this is a historical article. Today I would no longer recommend buying this amplifier with the specific intention of modifying it, because the starting point is very poor and, to obtain a serious result, practically everything that really matters has to be replaced: the circuit, the output transformers and part of the power supply. The final result of this modification sounded good and the measurements were valid, but with some compromises. With the experience gained over the years, I now generally consider it more sensible to design a new amplifier from scratch rather than trying to save a cheap unit that was badly conceived from the beginning. The article remains useful, however, to show how much transformers, circuit design and proper engineering affect the real performance of a tube amplifier.

This Music Angel amplifier belongs to the category of modifiable Chinese amplifiers. Unlike other kit amplifiers, it is probably not worth buying one specifically in order to modify it. However, if you find one used at a good price, or if you already bought one and put it aside because you did not like it, it can become an interesting basis for a conscious DIY experiment.

Original condition

(Note: this first part of the article is very old, so the measurements were taken with rather poor instruments.) I began by checking the performance of the original unit and, as I expected, it was very poor. Without going through too many figures, it is enough to say that the original output transformers were already at -3dB at 50Hz. The upper bandwidth seemed very extended, apparently up to 60/70khz, but only apparently. In reality, from about 8/9khz upwards, these transformers showed a resonance with a very marked phase rotation and strong resulting distortion. The damping factor was below 2. The spectrum analyses and square waves shown below highlight the problem very clearly.

Spectrum at 1 Watt

Spectrum just before clipping

1khz square wave

10khz square wave

I then completely dismantled the unit and also found several rusty spots inside.

These are the original transformers: the power transformer and one of the output transformers.

The Music Angel power transformer tends to sag under load. In fact, although the amplifier uses EL34 tubes, in its original form it runs them with a very limited bias current. For this reason I decided to use 6V6GT tubes, which are better suited to the transformer available in this unit. Using larger tubes would not have made much sense, because it would not have been possible to drive them properly to power. The output transformers are my SE5K6-UNI, which, without their covers, have the same dimensions as the original transformers. By reusing the two “L” brackets, they can be mounted back onto the chassis using the original covers.

To build a simple circuit, also using tubes that are often ignored by snobbish audiophiles, I chose PC88 or EC88 tubes as drivers.

The 6V6 tubes operate as pentodes, with negative feedback that is absolutely necessary to obtain a correct and controlled sound. Since there was no room to fit a filter choke, I used a solid-state capacitance multiplier. As always, the premium schematic is available for purchase.

Let us look at the build made by “C.R.”.

Measurements:
Power: 3.5Watt RMS per channel before clipping
THD @ 1 Watt: 0.98%
Damping factor DF: 20
Bandwidth: 10Hz – 42khz -1db

Spectrum analysis

Bandwidth on resistive load

Bandwidth on reactive load

Square waves at 100hz, 1khz and 10khz

1khz triangular wave

How does it sound?

I bought this Mini X3 at a very low price so that I could have it modified. I confirm that, in its original form, the sound quality of this amplifier is anything but good. The construction is fairly simple and I will leave Stefano to describe the circuit details. I will only say that, once switched on, the listening test left me speechless: I had never heard such an inexpensive amplifier sound so good. I even like it more than much more complex and expensive units. 3.5W of sonic pleasure. Excellent circuit and excellent output transformers, as always. Stefano has outdone himself once again. “R.C.”

Below is the white build made by “S.Z.”.

In recent years there has been a real rediscovery of valves originally developed for the television industry. Many electronic tubes designed to operate in the horizontal deflection stages of color televisions from the 1960s and 1970s actually have extremely interesting electrical characteristics for audio use as well. The PL508 is one of them. Although it is almost unknown in the traditional hi-fi world, this valve offers a very unusual combination of ruggedness, current capability and low internal resistance, making it ideal for building powerful, dynamic valve amplifiers with a very distinctive sonic character.

This project was born from the desire to build a different kind of push pull amplifier, avoiding the usual EL34, KT88 or 6550 valves that are now seen everywhere. The idea was to use a forgotten but technically excellent valve, operating it in triode connection to obtain a more natural, linear and refined sound. The final result is a stereo push pull amplifier of about 20W RMS per channel, capable of combining excellent measured performance with a very musical and surprisingly modern tonal balance.

The PL508 was designed to operate as a line output valve in color television receivers from the 1960s, practically the smaller sister of the better known PL519 and PL504. It had to drive the convergence coils required by the early shadow mask picture tubes and the frame coils used on wide neck tubes. It is therefore an extremely rugged valve, designed to withstand very high pulse voltages and heavy working currents for thousands of continuous hours.

The PL508 can withstand positive pulses of up to about 2,500V and its heater was designed for 300mA series connection, typical of universal televisions without a mains transformer. Although its original purpose was completely unrelated to audio, this valve was already classified by the manufacturer also for AF use and shows very interesting characteristics when used in hi-fi amplifiers.

I was asked for a premium schematic and, since the person had an open mind and was not interested in yet another already seen amplifier, I decided to use one of the many forgotten valves that have excellent electrical characteristics but are systematically ignored by DIY builders simply because they are not well known.

In terms of size and dissipation, the PL508 vaguely resembles a 6V6, but electrically it is a completely different valve. It has a much lower internal resistance, a far greater current capability and an internal structure designed to work under very demanding conditions. In triode connection it can dissipate about 15 watts, showing very linear and predictable curves. In an audio amplifier, these qualities translate into better control, dynamics and drive capability.

One of the most interesting aspects of this valve is precisely its behavior in triode connection. Many television valves, once triode connected, become difficult to manage or retain characteristics that are too “hard” for refined hi-fi use. The PL508, on the other hand, behaves surprisingly well and makes it possible to obtain an amplifier with a very balanced character, controlled bass, present mids and an extremely extended high range.

The circuit developed is therefore a push pull amplifier using triode connected PL508 valves, capable of delivering about 20 watts RMS. The goal was to keep the circuit relatively simple, but without pointless compromises, using a proper input stage able to drive the output valves correctly.

The input stage consists of a long tail cascode phase splitter built with two ECC81 / 12AT7 valves. This solution provides excellent symmetry and good voltage swing capability without resorting to overly complicated circuits or excessive global feedback. It is followed by a buffer stage made with the two sections of an ECC82 / 12AU7, necessary to correctly drive the input capacitance of the output valves and maintain stability even at high frequency.

The amplifier uses a total of 10 valves. The output valves work with fixed bias, adjustable by trimmer, a solution that allows the operating point to be optimized and the PL508 characteristics to be used to their full potential. The power supply section is deliberately classic but robust, with a bridge rectifier followed by the indispensable CLC filter, ensuring clean anode voltage and very low ripple.

Below you can see the premium schematic of the project.

First photos of the unfinished build made by “S.C.”

From the first bench tests, the amplifier already showed very promising behavior. The measured power was 19.7W RMS undistorted, perfectly in line with the initial expectations. The most interesting aspect, however, is not so much the maximum power, but the ease with which the circuit maintains cleanliness and stability even close to clipping.

Later, S.C. brought me the completed build so that I could carry out the final measurements and the last optimizations around the global feedback network. At this stage the work focused mainly on the balance between frequency extension, stability and damping factor, trying to preserve the musical character of the amplifier without turning it into something sterile or too “laboratory like”.

After several months, the project was finally completed with its definitive enclosure.

The final appearance deliberately recalls classic custom-built valve amplifiers, leaving the valves and transformers clearly visible. The result is a device with a very personal look, far removed from the usual standardized commercial kits.

The measured performance figures are:
Maximum undistorted power: 19.7 Watt RMS
Overall THD @ 1 watt: 0.16%
Bandwidth @ 1 watt: 10Hz – 110khz -1dB
Damping factor DF: 5.0

The measured performance is definitely remarkable, considering that this is a valve push pull amplifier using triode connected television output valves. The extremely wide bandwidth demonstrates the excellent quality of the output transformers and the overall stability of the circuit. The damping factor is also higher than that of many traditional amplifiers, allowing better control of the loudspeaker, especially in the low range.

In listening tests, the amplifier shows a very fast and dynamic character, with dry and well controlled bass, detailed mids and a very airy high range that never becomes aggressive. The triode connected PL508 valves retain part of the “strength” typical of television valves, while maintaining an extremely pleasant musical presentation.

The Triodino 4 is a stereo single-ended amplifier based on 300B tubes, created to overcome the limitations of the old Triodino 3. After numerous requests for an upgrade from users, I decided to revisit that project, which has been circulating online for years, and bring it to a more complete and mature level.

The Triodino 3 has been built and modified in countless ways, but always remained tied to an extremely simple circuit. This simplicity, while making it accessible, also introduces several practical and sonic limitations. With the Triodino 4, I wanted to preserve the character of the 300B while removing the most evident compromises, improving input sensitivity, power supply, bias management, and speaker control, without distorting the single-ended philosophy.

In essence, the Triodino 3 is just a 300B with its driver stage and little else. It works, of course, but it remains more of a sketch than a truly complete amplifier design. The 300B is an important, expensive, and prestigious tube, and in my opinion it makes little sense to use it in a deliberately minimal circuit where much better results could be achieved.

Main limitations of the Triodino 3:

1. It requires a preamplifier, because the input needs about 6Vpp of swing to be properly driven. Many modern sources, such as DACs, CD players, or phono preamps, have lower output levels, so the power stage may not reach full output without an additional gain stage.
2. The AC filament supply of the 300B, even when carefully balanced, almost always leaves some residual hum at the speaker. With low-sensitivity speakers it may go unnoticed, but with high-efficiency systems it becomes easily audible.
3. The self-bias configuration does not offer the same cleanliness and stability that can be achieved with a well-designed fixed bias. The cathode bypass capacitor becomes a critical component, indirectly influenced by the dynamic behavior of the stage, and the large cathode resistor dissipates unnecessary heat inside the chassis.
4. The damping developed by this zero feedback circuit is, optimistically speaking, around a factor of 2. This means the Triodino 3 can struggle with low-frequency control, especially with certain speakers. The result can be bloated bass, poorly controlled and not always faithful to the musical content.

I therefore took what I consider a draft and completed the work, arriving at a circuit that addresses all the main aspects. I did not want to cut corners on the important parts, because using a 300B in a stripped-down design makes no sense. The Triodino 4 is therefore a project without unnecessary compromises, designed to achieve a mature, quiet, stable, and truly usable single-ended 300B amplifier even with modern sources.

Below, in the premium schematics section, you can find the final stage, power supply, and servobias circuit diagrams.

The first major difference compared to the Triodino 3 is the input stage. In my design I used a modified mu-follower, consisting of a high-quality octal signal pentode loaded by a triode. This solution provides significantly better input sensitivity for real-world use, without forcing the user to add a preamplifier just to properly drive the power stage.

The Triodino 4 can be connected directly to an audio source such as a CD player, a DAC, or a phono preamp, without inserting additional active stages in the signal path. The fewer elements in the signal path, the better, as long as the circuit already provides the required gain. In this sense, the design can be considered a minimal integrated power amplifier, with its own volume control and sensitivity compatible with modern sources.

The 300B operates with fixed bias. Bias adjustment is handled by a servobias circuit that senses the output tube current through a small resistor placed under the cathode and automatically regulates the negative grid voltage. This circuit is not part of the audio signal path, it only works with DC voltages and does not interfere with the amplifier’s sound quality.

The advantage is very practical: the operating point of the 300B remains correct both at cold start and after full warm-up. In addition, when tubes are replaced, the user does not need to deal with a tester, screwdriver, trimmers, and potentiometers to adjust the bias. The circuit automatically brings and keeps the tube in the correct operating condition.

Compared to the self-bias solution used in the Triodino 3, the sonic advantage is significant. There is no reactive cathode capacitor in the output stage, no large resistor dissipating unnecessary heat, and no reliance on the quality of a bypass capacitor for such a delicate part of the circuit behavior. The benefits of a well-implemented fixed bias are well known to those experienced with these circuits.

The filament supply of the 300B is implemented in DC, filtered with a CRC cell, in order to drastically reduce residual hum at the speaker. In a single-ended amplifier with directly heated tubes, this aspect is fundamental, especially when used with high-efficiency speakers. For even better performance, a CLC cell could be used to achieve an even cleaner filament supply. The requested design includes the CRC solution, but small variations remain possible.

Since many people are still skeptical about things I have been saying for years, I included a circuit with switchable negative feedback. This way, those who want to listen with feedback can do so, while those who prefer to try a zero feedback configuration can disable it and judge with their own ears what actually works better.

The feedback has been set at a low level, without making the circuit overly dependent on the feedback loop. Many feedback-based amplifiers become excessively sensitive, noisy, or difficult to control when the feedback is removed. In this case, the circuit remains usable in both conditions: it requires about 3Vpp, that is 1.1Vrms, with feedback engaged, and about 2Vpp, that is 0.7Vrms, without NFB.

The goal is also educational: I want people to directly verify what I have been stating for years. A well-designed circuit does not sound worse simply because it uses a reasonable amount of feedback. On the contrary, it often results in tighter low frequencies, better damping, and more correct behavior with real speakers. Those who do not trust this can disable it and use the amplifier in zero feedback. I am quite confident that after trying both conditions, many will not go back. Also because my transformers are not like others.

But that is not all. The old Triodino 3 used a simple bridge rectifier. Some more dedicated builders modified it by using a tube rectifier, but I wanted to go further. To push audio quality to the maximum, I implemented a tube-based voltage regulator.

The regulator consists of a 6080 or 6AS7 used as a series element, driven by an ECC83. The 6080 is a tube widely used in audio, but originally designed specifically for this purpose: voltage regulation. Historical documentation confirms this: Dual power triode, ruggedized 6AS7G. Intended for use as series voltage regulator.

A practical note for anyone building this project: the ECC83 used in the regulator operates in a nearly static circuit. It does not carry the audio signal, and there is no need to use an expensive NOS ECC83 to achieve good performance. A good modern production ECC83 is more than sufficient. The money saved is better spent on high-quality capacitors, where the benefit is more tangible.

The regulated high voltage brings several advantages. The amplifier operates with a more stable voltage, less dependent on mains fluctuations and dynamic current demands. The audible result is greater clarity, better micro-detail, a more stable soundstage, and improved three-dimensionality. In a single-ended 300B amplifier, these aspects make a real difference, as the circuit is highly revealing and every power supply choice is audible.

Assembly by “R.”

“R” brought the amplifier to me for standard measurements, general verification, and final tuning. This kind of check is important because a project like this cannot be evaluated only by looking at the schematic. Wiring, component layout, grounding paths, transformer placement, and distributed capacitances can significantly influence the final result.

You can admire the FullMusic 300B tubes with mesh plates. For those unfamiliar, and who might think this is a red plate condition, it is not. Mesh 300Bs have a plate made as a metallic grid. The red glow you see is not the plate overheating, but the filament light passing through the mesh structure.

A small technical note: with the FullMusic tubes, the amplifier’s damping factor reached 6.0, while with the setup below, using Electro Harmonix 300B tubes, it settled at 4.4. Most likely, the FullMusic 300Bs are built with the plate closer to the cathode, resulting in lower internal resistance and therefore higher damping factor.

This is interesting because it shows how the output tube can influence not only the tonal character, but also measurable electrical parameters of the amplifier. When discussing 300Bs, people often focus only on brands, trends, or subjective preferences, but the internal geometry of the tube can measurably change the behavior of the output stage.

First build photos by “C.”

I had the opportunity to work on “R”‘s build for final tuning. During testing, I solved a startup stress issue affecting the 6080 by modifying the servobias. Now the 300Bs start in forced cutoff for about 30 seconds, giving time for the indirectly heated tubes to warm up. After this delay, the servobias gradually brings the output tubes into operation, avoiding unnecessary transients and improper startup conditions.

The issue arose because the 300Bs are directly heated tubes and start within seconds, while the other tubes require more time to stabilize. This mismatch could cause noise, hum, and unwanted overcurrent in the 6080 during startup. With the implemented modification, the amplifier starts in a more controlled and orderly way.

I also corrected a few wiring errors in “R”‘s build. In the output stage schematic, I indicated the color coding of the output transformer primary wires to avoid mistakes that could turn negative feedback into positive feedback. I also added a single resistor in the regulator schematic, improving overall circuit behavior.

Here are the measured results:

Power: 8.3Watt RMS per channel
Damping factor: 4.44
THD @ 1Watt: 0.38%
Bandwidth: 10Hz / 20kHz -1dB

I must say that “R”‘s build is not among the best in terms of layout and wiring. It is possible that distributed capacitances and some practical implementation choices slightly penalized the circuit performance, especially bandwidth. Nevertheless, the result is far from poor, in fact considering the type of build the measurements are more than respectable.

A single-ended 300B amplifier should not be judged only by maximum power. Here we have an amplifier delivering 8.3Watt RMS per channel, but with already low distortion at 1Watt and a damping factor higher than typically expected from a zero feedback 300B circuit. This means that, with suitable speakers, the amplifier can provide a much more controlled and credible listening experience than many deliberately simplistic single-ended designs built around a romanticized minimal circuit concept.

Harmonic spectrum

The harmonic spectrum shows an orderly behavior consistent with the nature of the circuit. Distortion is not absent, because a single-ended triode is not designed to chase extreme numbers like a solid-state amplifier, but the harmonic distribution is clean and does not show concerning artifacts. The THD value at 1Watt confirms that the circuit operates linearly in the most commonly used listening range.

Bandwidth on resistive load

The bandwidth on resistive load extends from 10Hz to 20kHz within -1dB. This is a very good result for a single-ended 300B, especially considering that the output transformer is always the most challenging component to design properly in this type of amplifier. The low-frequency response shows that the transformer is not undersized and that the circuit maintains good extension without artificially boosting the bass.

And on reactive load

The behavior on reactive load is even more interesting, because a real speaker is never a pure resistor. Many amplifiers that appear correct on a resistive bench load change character when driving a complex load. Here the circuit maintains a stable and controlled behavior, showing that the design was not created just to look good under ideal measurement conditions.

Square waves at 100Hz, 1kHz and 10kHz

Square waves provide an immediate view of the circuit’s transient response. At 100Hz, low-frequency transformer behavior is evaluated, at 1kHz the general cleanliness of the response is observed, while at 10kHz phase issues, overshoot, instability, or high-frequency limitations become evident. In this case, the behavior is consistent with a well-designed circuit, without anomalies or signs of instability.

Some readers expected to compare measurements with and without NFB, but unfortunately “R”, who now “knows”, chose not to implement the switchable feedback. It is a pity, because it would have been interesting to directly show the instrumental difference between the two conditions. Nevertheless, the circuit has been designed to operate in both modes, leaving the builder the freedom to choose.

So, beyond the measurements, how does this Triodino 4 sound? Below is a comment from Cristian, who named his unit Afrodite.

Hi Stefano

I am listening to the Triodino 4, which I named Afrodite.

I must say, the result is quite different from other 300B single-ended amplifiers I have listened to so far. Vocals remain the strong point of these tubes and the highs have a very fine texture.

What is surprising is the bass, which, with a damping factor only slightly above 4, is particularly well controlled and has impressive punch.

The driver tubes used are 6SJ7 Ken-Rad NOS with metal envelope and a 6SN7 GTA NOS Philips. The output tubes are ordinary E.H.

Great idea the servobias, which works very well, although I think it will not appeal to those who “must” adjust the bias every fifteen minutes 🙂

The result is very good in my opinion.

Cristian

This feedback perfectly confirms the purpose of the project. The Triodino 4 does not aim to erase the character of the 300B, but to allow it to operate under better conditions. Vocals remain the highlight, highs retain their fine and natural texture typical of good triodes, but the bass is no longer left uncontrolled. The higher damping factor, more stable power supply, and servo-assisted fixed bias result in a more authoritative, controlled, and mature sound.

In conclusion, the Triodino 4 is the answer for those who love the 300B but do not want to settle for the usual minimal circuit built around an expensive tube. It is a more complete, more refined, and more modern design, while remaining faithful to the single-ended philosophy. It is not an amplifier for those who want just a couple of components and plenty of forum mythology to feel satisfied, but for those who truly want to hear a 300B used properly, within a circuit worthy of the tube it employs.