Related articles:

• How to make a Chinese amplifier sound good
• What negative feedback is
• What damping is

Introduction

This article is a double feature: on one side I propose some simple but effective modifications to a small Chinese single-ended amplifier based on 6L6G and ECC83 tubes, and on the other I use this very unit as a test bench to tackle one of the most discussed, and most often misunderstood, topics in the world of tube high fidelity: negative feedback.

I chose this inexpensive amplifier as a base to demonstrate, with instrumental measurements and real audio recordings that you can listen to on headphones, what actually happens when negative feedback is switched in or out of a properly designed circuit. The result is clear: when applied with the right amount, negative feedback does not degrade the signal at all. On the contrary, it faithfully preserves the original content, introducing only minimal variations. Conversely, in zero-feedback mode, the changes in sound are obvious, especially in the low end and overall response, to the point where the difference is clearly recognizable by ear.

This does not mean that tube technology is useless or outdated. Quite the opposite: the beauty of the tube world is precisely that you can experiment, build, and shape your own sound according to personal taste, choosing one tube rather than another, one capacitor instead of another, and chasing those nuances that make every project unique. However, it is important to put an end to certain false myths that keep getting repeated without any objective verification, such as the idea that negative feedback would “erase” information while zero feedback would preserve it. In this test, with the same circuit and the same test conditions, it becomes clear that negative feedback tends to preserve the integrity of the signal better, while the zero-feedback configuration introduces additional content that is not present in the original.

By using low amounts of negative feedback (without ever overdoing it) and playing with the choice of tubes, transformers, and components, you can achieve listening experiences that are truly interesting, fun both from a didactic and a musical point of view. It is fertile ground for those who love experimenting, understanding, and shaping sound according to their own tastes.

However, I am convinced these experiments would work even better if they were approached with the awareness of deliberately altering the sound in a pleasing way, to achieve a “guilty-pleasure” effect that is personal and even creative, rather than being stuck with the belief of chasing absolute fidelity, while in reality you are modifying the signal without realizing it. Knowing how to distinguish the pursuit of fidelity from the pursuit of sonic character is essential. Only then does DIY become a form of conscious expression, and not an illusion fueled by distorted technical myths. This article aims to be a practical and transparent contribution, dedicated to those who listen with their ears and measure with their brain, not with prejudice.

---

I would like to take this opportunity to remind those who are looking for low-cost tube amplifiers that it is preferable to choose one of these Chinese little boxes available on the market rather than one of the many questionable, poorly built solutions offered by Facebook tinkerers, hobbyists, or incompetent individuals you can find around.

The purpose of this article

Over the years, I have always shared with enthusiasm the experience and knowledge I have gained in building amplifiers and audio transformers. I have often expressed critical opinions about some common practices in the field, but I also understand that, given my direct involvement, someone might think I am simply promoting my personal ideas.

One of my long-running battles has been to clarify the truth about negative feedback, often unfairly blamed for bad sound, various issues, and even catastrophes. On the contrary, I have always highlighted the importance of amplification with an adequate damping factor to enjoy pleasant listening, free from annoying, disturbing bass.

While browsing the web and listening to podcasts, I came across people trying to refute my statements, perhaps in response to the doubts raised by my articles. It is well known that it is often easier to convince people with superficial arguments than with in-depth analysis. Many sellers and TV-style pitchmen ride the wave of popular credulity, claiming that “zero feedback” amplifiers are absolutely superior. They also claim that damping is not that important, and that annoying bass is just an excuse to sell special loudspeakers (often very expensive) marketed as the only solution to a problem they themselves created. It is the classic case of selling the disease first and then the cure.

Some critics have even gone as far as calling my transformers, which are larger and more expensive, useless, while promoting cheap, lower-quality products that promise miracles at bargain prices. I believe it is essential to keep a critical mindset and not be influenced by extremist ideas without solid technical foundations. Every choice should be weighed according to your needs and your personal sonic preferences.

It is natural for the reader to feel disoriented when faced with contrasting opinions, especially when there is no chance to compare solutions directly. In the world of tube amplifiers, where personal taste can vary enormously, there is no single universal answer. What one person likes may not please another at all. Many factors influence the listening experience, such as the environment, individual preferences, and system matching with other audio components. Despite the frustration of not having absolute certainties, it is important to keep an open and curious mindset, carefully evaluating your needs, tastes, and budget.

Unfortunately, enjoying something does not always guarantee the highest sound quality. Often you are satisfied simply because you have not had the chance to hear something better. It can happen that one day you come across a higher-quality system, and only then you realize that previously you had never truly listened in an optimal way.

I am reminded of the experience of a customer of mine who, after years of listening to a Beatles record, had the opportunity to hear it through a higher-quality system. Only then did he discover details and sonic nuances he had never perceived before, even though he had already used prestigious amplifiers. This example clearly shows that there are quality levels that can be reached only with truly high-end equipment.

Sadly, some dealers push enthusiasts into a continuous cycle of buying and replacing equipment, fueling constant dissatisfaction. Many audiophiles even evaluate an amplifier more for its resale value than for its actual sound quality. This is a wrong and counterproductive approach. I have seen people replace excellent-quality amplifiers with products that have a disappointing sonic impact but are highly rated by the specialist press. I believe that those who behave this way have not really understood what high-quality audio means, and might not even be able to tell the sound of a fine amplifier from that of a simple intercom.

It is essential to pursue a personal listening experience, do research, compare options, and, if possible, rely on industry professionals who have deep and impartial expertise. Only in this way can you make an informed choice and obtain an amplifier that delivers a truly satisfying listening experience, beyond commercial trends.

I find it sad that some people buy audio gear just to show off their wealth. Often what gets advertised as excellent ends up disappointing expectations. There are expensive, famous products that deliver mediocre performance, just as there are cheaper, lesser-known products capable of excellent results.

This article aims to be an honest contribution, an invitation to personal reflection. It is intended for DIY builders as a fun didactic experiment that still offers a satisfying sonic result. I do not promise perfect sound, but I do guarantee an enjoyable and instructive experience. My goal remains to offer an open and transparent view of the audio world, so that everyone can discover and appreciate sound according to their own preferences and possibilities.

The Nobsound 6p3p + 6n1 (6L6G + ECC83)

The basis of this experiment is a small Chinese amplifier, with virtually no real name. You can easily find it on various online shopping sites, available both assembled and as a kit. It is easily recognizable from the photos, sold under the Nobsound brand, under other names, or even unbranded. It is probably one of the cheapest options available on the market.

This is the original circuit schematic:

Unfortunately, the original output transformers of this amplifier have an impedance of 3500 ohm, which is not suitable for use with a 6L6GB (or equivalent) tube. During bench tests I was getting limited power, about 3.5 watt, with a very high distortion level: in practice, one half-wave was reproduced well while the other one was heavily compromised.

To improve the situation, I disassembled the original transformers. Since I could not replace them with larger transformers built according to my method (because there was not enough space), I chose to rewind the original transformers using the same bobbin and laminations of identical size, but I replaced the original simple silicon-steel laminations with grain-oriented (GO) laminations. I therefore built transformers with an impedance of 4500 ohm, ideal for the 6L6GB in single-ended configuration, with another specific goal in mind.

In several previous articles I have already pointed out that many zero-feedback amplifiers, including some from prestigious brands, use output transformers with insufficient primary inductance or cores close to saturation. This approach is used to mask the most obvious defects caused by the absence of negative feedback, especially those famous “boomy” and annoying bass frequencies. It is clear that, if you choose the zero-feedback route and categorically refuse negative feedback, those problematic bass issues must be eliminated somehow. The most common solutions are filtering the amplifier input, using intentionally bandwidth-limited output transformers, or pairing the amplifier with loudspeakers that are naturally weak in the low frequencies (for example single-driver speakers or open-baffle speakers).

Here is the schematic, click to enlarge, or click here to download the high-definition version

As you can see, I included a switch that allows the negative feedback (NFB) signal to be disabled at will. I replaced the 6N1 tube with an ECC83, a necessary choice to achieve an adequate damping factor while keeping the output stage an easy load to drive. The original rectifier, a 5Z4, was also replaced with a GZ34, which is essential to recover usable voltage: the stock Chinese transformer, in fact, tended to have more voltage sag than expected.

Note: this schematic is intended for tubes of the 6L6 / 6L6G / 6L6GA / 6L6GB / 5881 type, and also the 6p3p. The 6L6GC and 6N3C-e are excluded, as they require a higher screen-grid voltage. However, it is possible to adapt the amplifier to those tubes by changing the value of the 12k resistor in series with the screen grid to a lower value, until the desired operating point is achieved. Always verify plate and dissipation values under operating conditions, because screen current can vary significantly from tube to tube. I am not indicating a precise value because the idle current of the screen grids is often unpredictable and may deviate from datasheet values. Let’s now move on to the assembly steps:

For anyone curious about how I managed to solder tin onto stainless steel, I recommend watching this YouTube video: https://youtu.be/SHxo5tcNNMg. The procedure shown is identical to the one I used, with the only difference being that I used ready-made ferric chloride instead of preparing the solution from scratch. Without this galvanic treatment (which I carried out with the help of my modified power supply), soldering on stainless would be virtually impossible. Of course, a flat-tip soldering iron (screwdriver type) of at least 150 watt is also required.

Let’s now move on to the usual instrumental measurements. I want to clarify a point that is often a source of confusion: some people claim that, in a circuit without negative feedback (zero feedback), total harmonic distortion (THD) increases progressively from minimum power up to maximum power. Conversely, according to the same theory, in a feedback circuit THD would be higher at very low volume, then decrease at mid volumes, and rise again as clipping is approached.

This behavior, more similar to what you would expect from a class B output stage than from a single-ended amp, is sometimes used as an argument against using negative feedback in systems intended for low-volume listening. The idea is that, at low power levels, a feedback circuit would distort more than one without feedback.

To verify this claim in a concrete way, let’s look at the THD results measured in the two configurations (zero feedback and negative feedback enabled), keeping the rest of the circuit unchanged and comparing several power levels, from 0.1 watt up to just below clipping. We will compare the values to understand what actually happens.

With negative feedback enabled	Zero Feedback
0.1 watt RMS – (0.18%)	0.1 watt RMS – 0.85%
0.5 watt RMS – 0.47%	0.5 watt RMS – 2.45%
1 watt RMS – 0.73%	1 watt RMS – 3.6%
3 watt RMS – 1.31%	3 watt RMS – 6.2%
Just below clipping (5 watt) – 2.35%	Just below clipping (4 watt) – 8.4%

As you can see, distortion increases progressively from a minimum value up to the clipping threshold, both in the zero-feedback configuration and in the feedback configuration. However, distortion is always higher in the version without negative feedback, even at the minimum power level measured.

For transparency, I left the schematic and the necessary references to replicate the test. Anyone can build the circuit and repeat the measurements, using transformers made by me or by any supplier of their choice. In this case it is a classic 22×30 core with a 4500 ohm primary and 4 and 8 ohm secondaries, without any particular ambitions.

The maximum output power of the circuit with negative feedback enabled is about 6 watt, with a noticeable distortion threshold starting around 5 watt. The damping factor is 0.3 in zero-feedback mode and reaches 9 with negative feedback enabled. The bandwidth, in the zero-feedback configuration, is 40 Hz – 40 kHz at -1 dB. It should be noted that the transformer was specifically designed with a low-frequency roll-off, a mandatory choice given the small core size. With negative feedback enabled, the frequency response extends from 10 Hz (at -0.6 dB) up to 40 kHz (-1 dB), offering a noticeably wider overall response. Below I show the comparative plots… and some extra content.

Bandwidth with negative feedback (resistive load)	Bandwidth with zero feedback (resistive load)
Bandwidth with negative feedback (reactive load) DF = 9	Bandwidth with zero feedback (reactive load) DF = 0.3

Now let’s move on to the extra content. I realized that many readers do not fully understand the meaning of the plots related to behavior with a reactive load. For more details, I refer you to my article dedicated to building a reactive load.

In general, there is little awareness of what a low damping factor actually implies. Beyond the usual descriptions, such as the speaker cone that, once pushed, keeps moving by inertia instead of faithfully following the signal, there is a lack of concrete understanding of the effects on the amplifier’s real-world behavior during use.

Everyone seems to focus exclusively on harmonic distortion and its components, while few people bother to observe what happens when a low-damping amplifier actually drives a loudspeaker. For this reason, I made two short clips showing the oscilloscope screen, with the amplifier connected to a speaker and driven by a function generator.

Warning! The following videos contain steady sine tones. It is strongly recommended to turn the volume down to minimum before playback, especially if you are using headphones. These sounds may be annoying or even harmful to your hearing, as well as to headphones, computer speakers, or smartphone speakers. Audio is nevertheless essential to correctly understand the phenomenon being analyzed. I assume no responsibility for any damage to audio devices or hearing resulting from watching these videos. Proceed with caution.

Let’s start by analyzing the situation without negative feedback (Zero Feedback). On the oscilloscope screen two traces are visible: the lower trace represents the signal generated by the function generator, while the upper one shows the signal actually present at the speaker terminals.

As you can clearly see, while the amplitude of the lower trace (the generator signal) remains constant, the upper one, that is the amplifier output signal, varies significantly as the frequency changes. It is behavior that resembles a tone control built into the amplifier, with the difference that in this case the points of maximum and minimum response are determined by the loudspeaker’s characteristics.

What you see in the plots for the reactive load is exactly this phenomenon: a frequency response altered by low damping. To those who only talk about harmonic distortion, I ask a question: this is not harmonic distortion, but it is still a distortion. How should we classify it? Now let’s see what happens in the same situation, but with negative feedback enabled and a damping factor of 9.

In this configuration, with negative feedback enabled, you can see that the output signal amplitude shows only minimal, almost negligible variations. As I have stated in other contexts, I believe a damping factor of 5 is optimal for small-power amplifiers such as single-ended designs. In higher-power equipment you can safely go up to 10. Exceeding these values, in my opinion, rarely brings benefits in this kind of application: too much negative feedback can reduce subjective listening pleasure and make the result feel less natural.

Now let’s move on to another extra: a direct comparison between the sound of the feedback circuit and that of the zero-feedback configuration. How did I do this? I connected an attenuator directly to the speaker terminals and sent the attenuated signal to the line input of an old laptop, which I normally use to drive a laser engraver.

I apologize for the recording quality: the laptop is almost 10 years old and I paid 40 €, so some background noise from its sound card is inevitable. Despite that, the sonic differences between the two configurations are still audible.

It should be said that a more accurate demonstration would have required a binaural microphone, capable of capturing the distortion physically introduced by the loudspeaker, and not only the electrical component. Finally, I add that, since I used transformers with a small core and a primary calculated with modest inductance, the amplifier does not show the typical boomy and uncontrolled bass even in zero-feedback mode.

To begin, I chose a royalty-free audio track titled “At First Sight” by FiftySounds, downloaded from EpidemicSound.com. You can listen to the original file at the provided link.

Please note: to fully catch the differences and sonic details, it is essential to listen with headphones.

I played the track from my media center, connected directly to the Nobsound, and I recorded the signal taken from the speaker terminals in both configurations (zero feedback and negative feedback enabled), trying to keep the level as comparable as possible between the two tests.

Then, using an audio editing program (Audacity), I overlaid a segment of the original recording with the same passage filtered through the amplifier. Thanks to the software’s “solo” function, during playback I can quickly switch from one version to the other, allowing you to clearly hear the differences in real time. Let’s start with the comparison between the original track and the amplifier output in zero-feedback mode.

In the audio view, the upper track is the original file, while the lower one is the recording of the amplifier in zero-feedback configuration. By ear, you can clearly notice a loss of body in the low frequencies and a certain dullness in the highs, along with an emphasis in the midrange. Also, even if the electrical recording already gives you an idea, it is in the real room that you can perceive how the zero-feedback playback sounds “dirtier”.

Now let’s move on to the recording with negative feedback enabled. At this point, all the skeptics who believe the usual urban legends will expect an unpleasant sound: metallic, harsh, like broken glass, nails on a chalkboard, or that infamous fork-on-a-plate screech. Let’s see (and hear) whether that is really the case.

As in the previous test, the upper track is the original audio, while the lower one is the recording of the signal after passing through the amplifier, this time with negative feedback enabled.

And here is a concrete demonstration: in this case, with proper use of negative feedback, the signal is not degraded in any appreciable way. On the contrary, the zero-feedback version sounds worse. Those who claim to prefer zero feedback, in reality, are chasing a certain type of distortion, and they do not always admit it.

The recording with negative feedback is so faithful to the original that it is almost indistinguishable, if it were not for a slight background noise and a small loss of brilliance, which is also present in the zero-feedback version. These limitations are due exclusively to the built-in, far-from-professional sound card of the old laptop used for recording. To wrap up, I am providing a download of the two full recordings. In the future, if I have the chance to repeat similar experiments with more serious equipment (see binaural microphone), you can be sure I will do it. Click here to download the zip with the two files in flac format