Audio Innovations is one of those brands that, for anyone with even minimal experience in tube hi-fi, needs little introduction. Founded in Brighton, England, in 1984 by David Chessel and Peter Qvortrup, it became known for a very clear philosophy: simple circuits, carefully selected components, and a concrete sound, without unnecessary complications.

Differences between the 700 and 800 models

Among the various models produced by Audio Innovations, the 700 and the 800 are probably the most widespread and technically interesting. The 700 is an integrated amplifier, with volume control and multiple inputs, while the 800 is a pure power amplifier, with a single input and no control section.

Beyond these functional differences, the two units are essentially very similar, if not identical, from a circuit standpoint. The interface changes, but the substance remains the same.

Both deliver about 25W per channel and use a pair of EL34 tubes in push-pull configuration. The operation is heavily biased toward class A, transitioning toward AB as it approaches maximum power. Looking at the schematics, it is clear that the design philosophy is the same, although with some implementation differences.

IA700	IA800

The AI700 clearly derives from the 500, but with less clean circuit choices: the input stage is implemented with an SRPP followed by a paraphase splitter, a solution that is not ideal especially in the presence of global feedback. In addition, there is no real symmetry in driving the output stage, with two non-equivalent branches.

This setup makes the circuit somewhat more delicate and less tolerant compared to more linear architectures. Over time I have often encountered serious failures on these units, frequently accompanied by damaged output transformers. In many cases these are oscillations that start when conditions are no longer ideal: unbalanced tubes, worn components, or simply units that have not been serviced for years.

When this situation occurs, the circuit can enter self-oscillation, pushing the transformers into abnormal conditions and generating voltages high enough to break down internal insulation, leading to shorts to ground and cascading damage.

It is easy to blame the design or the feedback, but in practice the problem almost always comes from units left unattended for too long without maintenance. We are talking about amplifiers that are now decades old, and expecting stable and reliable operation without checking the condition of components and tubes is unrealistic. This does not mean the design is perfect, but that it requires more careful maintenance compared to other, more robust or less stressed amplifiers.

On this page I report some repair and restoration experiences on two units, an Audio Innovations 700 and an Audio Innovations 800. If you have one of these amplifiers that needs servicing or repair, you can contact me.

Audio Innovations 700 restoration

I was given an Audio Innovations 700 which, according to the owner, sounded very poor. It came from a sale of equipment belonging to a deceased person, so there was no reliable history of what had been done before.

Once opened, the situation was quite clear: the unit had suffered major failures and had been worked on multiple times, practically everywhere. This was not just a poorly executed repair, but a series of improvised interventions layered over time, carried out without any criteria. A truly poor condition.

The original output transformers were missing, replaced with very low-quality components clearly unsuitable for that circuit. The worst situation was in the power supply section. The small board mounted on the main capacitors had suffered heavy burning, and someone had tried to “improve” it by adding completely nonsensical modifications: two fluorescent lamp ballasts used as filter inductors, an improper solution because such components quickly saturate under DC and are not suitable for that purpose. A capacitor had also been added, with the idea of creating a separate LC filter for each channel.

The result was a mess of components thrown in without any logic, with high-voltage parts, over 400 volts, simply placed and held with adhesive tape. The main PCB was also in terrible condition: burns, holes, damaged traces and leakage points. In several areas, components were soldered poorly, with no care, and interventions that could only be described as makeshift and amateurish.

At that point I approached the work not as a simple repair, but as a complete restoration. I fully disassembled the unit and carried out a thorough cleaning of both the metalwork and the main PCB, removing years of dirt, residues and traces of previous “repairs”.

The output transformers were replaced with two new units, 1:1 clones of the originals, developed through careful reverse engineering. They are the same models I have been producing for years and are compatible with AI500, 700, 800 and Classic 25. They were installed in their original position, alongside the power transformer which, fortunately, was still perfectly functional.

For the power supply section I had to start almost from scratch. I scanned the original PCB, which was carbonized, and after several steps using Gimp, Inkscape and finally KiCad, I generated Gerber files to produce identical replicas. The new boards were then manufactured through an online service.

This board is compatible not only with the 500, but also with the 700 and 800, and is available as a spare part.

On the main PCB I removed everything that was damaged or compromised. Using a Dremel, I completely removed the carbonized areas, the “decay” of the board, which would otherwise have remained conductive. The excavated areas were then rebuilt with specific UV resin. I replaced all high-voltage electrolytic capacitors, the output tube octal sockets, and several damaged resistors. In some cases I also had to reconstruct broken traces. Several vias were completely clogged with dirt and carbon residues, so I mechanically reopened them with a Dremel to reinstall the new components.

At this point I reassembled the PCBs, reconnecting all the wiring.

The front control section still needed attention. The situation there was no better: the small PCB showed very poor soldering, damaged traces, and rough interventions. I had to replace the rotary input selector switch, restore several traces, and completely redo all wire joints. The flat cable connecting the rear inputs to this board was also damaged and required repair. In practice, whoever had worked on it before had touched every section of the amplifier, modifying or damaging it in some way.

After all this work, the amplifier returned to perfect operation. Interestingly, it was not necessary to replace the tube set it came with. Despite being used, they were still in good condition and, once the circuit was properly restored, they resumed working without any issues. This clearly shows that, in most cases, the real problem is not the tubes, but everything around them.

Instrumental measurements:

• Maximum power: about 25W
• Frequency response at 1W: 37Hz – 50kHz, -1dB
• Frequency response at 15W: 30Hz – 50kHz, -1dB
• THD: 0.16% @ 1W

Bandwidth graph

Spectrum analysis

Audio Innovations 800 repair, faulty power transformer

This Audio Innovations 800 was an interesting case, not so much for the complexity itself, but for what emerged during the repair. As often happens, beneath the surface of a “working” unit lies a very different story.

The amplifier arrived with a clear fault: shorted power transformer and fuse blowing immediately. After disassembly and measurements, I designed a replacement transformer, code 23S72, with the same electrical characteristics as the original.

The installation of the 23S72 requires no mechanical modifications. The original uses a 38mm core, practically unavailable in Europe. The new one uses a 40mm core, slightly larger by a few millimeters, but it fits perfectly in the original space without any need for modifications. This preserves both the structure and the appearance of the amplifier. The transformer is also compatible with the 700 model.

The amplifier had been sold as “perfectly working, never repaired”. In reality, as soon as it was opened, it was clear that someone had already worked on it. The EL34 cathode electrolytics had been replaced with Mundorf parts, which alone is enough to show the unit was not original.

The real issue, however, was something else: under one of the cathode resistors there was a clear burn mark. Upon inspection, I found 47 ohm resistors installed instead of the correct 470 ohm. The result was a completely incorrect bias, pushing the EL34 tubes well beyond their limits.

I restored everything to proper conditions, replacing the resistors with the correct values and changing a compromised capacitor.

After a complete overhaul and tube check, the amplifier was reassembled and tested. With used tubes it delivered about 25W, with a damping factor of 6.2. Operation returned to stable and consistent performance, as expected from this type of circuit.

Once restored, this unit returned to proper operation.

Audio Innovations 800 repair with faulty output transformer

Another case involved an Audio Innovations 800 with a noisy channel. The amplifier powered on, but one side produced abnormal noise. The issue was the output transformer, which was compromised.

Unwinding and internal analysis

I dismantled the transformer to analyze it and derive its winding scheme. Inside, I found a critical situation: the windings had been secured with masking tape, whose adhesive had completely deteriorated over time.

The enamel insulation of the wire was degraded. In some areas it came off together with the tape, leaving bare copper. This was not a sudden failure, but a slow degradation due to materials and time.

The result was a progressive short circuit between windings, without any obvious catastrophic event.

Reconstruction and tuning

Starting from the derived winding scheme, I built a new pair of output transformers, faithful clones of the originals. I replaced both to maintain channel symmetry.

After installation I carried out a complete set of measurements. Comparison with original transformer data shows nearly identical results.

Original	Clone
Square waves at 100Hz – 1kHz – 10kHz
Bandwidth graph

The reconstruction fully restored operation without altering the original behavior of the amplifier.

Spectrum analysis at 1W

Here the two installed transformers can be seen. The work was carried out maintaining dimensions, layout and finish as close as possible to the original.

Signal tube layout

Below is the layout of the signal tubes, useful as a reference for replacement. The EL34 output tubes are not indicated as they are immediately recognizable.

The ECC83 tubes in positions 2 and 4 are often unbalanced between their two internal sections. Measuring them with uTracer 3+ makes this behavior evident.

This is not a defect of the tube but a direct consequence of the circuit: the two sections operate under different conditions, so they tend to diverge over time. Even using perfectly matched tubes, the situation quickly returns to the same state. For this reason, there is no point in obsessing over perfect matching in these positions. As long as the tubes are efficient and the circuit operates correctly, there is no reason to replace them.

The Klimo Kent Silver are monoblock tube power amplifiers that left a clearly recognizable mark on the Italian hi-fi scene. Even today there is still a certain level of interest around these amplifiers, both from those who have owned them for years and from those who buy them used and then have to deal with neglected maintenance, aged components and repairs carried out in questionable ways. Their charm is also linked to the classic EL34 push-pull architecture, a very essential aesthetic and a name that, over time, has carved out its own reputation among tube amplifier enthusiasts.

For this very reason, when a piece of equipment like this arrives on the workbench, it makes no sense to simply replace a few failed components. You need to understand how it was built, what condition it arrived in, what design compromises were originally adopted and what damage may have been caused by previous unprofessional interventions. In this case I worked on the repair and restoration of a pair of Klimo Kent Silver that arrived in disastrous condition, heavily marked by tampering and improvised repairs.

This article is not an audiophile-style review. It is a technical analysis with practical observations, instrumental measurements and circuit considerations that may also be useful for those who own the same amplifier and want to understand whether it needs a serious overhaul, a full restoration or simply a properly executed check.

Initial condition and state of the equipment

The Klimo Kent Silver is a tube amplifier rated for 35 continuous watts, although, as we will see later, in practice the real power on 8 ohm is lower. Beyond the numbers on the specification sheet, the most important thing in a case like this was to assess the damage caused by previous repair attempts.

The initial inspection revealed a long list of problems: rough soldering, components soldered onto the stubs of others that had been badly removed, missing or damaged parts, makeshift fixings and various technical anomalies. In cases like this, the first job is not yet the actual repair, but understanding exactly what has been altered compared to the original configuration.

The goal of the restoration was not to turn the Kent Silver into something it is not, but to bring it back to an electrically correct, reliable condition consistent with its real capabilities. Even a device born with evident compromises can work well again, provided it is put back in order with method and competence. Here are some photos of the condition in which they arrived:

In short, the initial situation included cracked and glued resin tube sockets, deteriorated or exploded capacitors, components left loose and temporarily fixed with hot glue. This is the classic scenario where a bad repair ends up making the original defects of the equipment even worse.

On this subject, a small practical tip. Before disassembling equipment full of sterling wires, often all the same brown color, it is always advisable to mark the connections and photograph everything carefully. Once disconnected, putting them back correctly can turn into a puzzle. Marker pen marks on brown wires are often hard to see or completely invisible. Colored nail polish, on the other hand, is immediately visible. Besides being useful for temporarily locking trimmers and small screws without sealing them permanently, it is also very useful for this purpose. It is easy to find in many colors and costs very little, so it is a simple but very practical solution.

Repair and restoration work

After removing the boards, I started with a thorough brushing, focusing especially on the areas affected by soldering. This step was essential because, as I have also found in other Klimo equipment, the board cleaning during assembly was certainly not impeccable. Over time, flux residues tend to promote oxidation and corrosion of the copper.

I then carried out a complete washing of the boards, because in this case the general condition was particularly bad. I replaced the output tube sockets and all the electrolytic capacitors, except for the original blue capacitors which were still in good condition, as well as some resistors that I no longer considered reliable. Only after this work was it possible to reassemble everything with a minimum of order and technical consistency.

Output transformer and load impedance

Looking at the circuit, one of the first things you notice is that the amplifier only has two speaker terminals. The presence of an internal terminal block between the output transformer and the banana connectors might suggest the possibility of reconnecting the wires to obtain different impedances. However, it could also simply be a construction choice to simplify transformer wiring. In the absence of official technical documentation, it is safer to consider the secondary as fixed.

With 8 ohm applied to the secondary, the output transformer reflects approximately 8kohm on the primary. This means that by connecting speakers with different impedances, the impedance reflected to the tubes changes, and with it the available power, dynamic behavior, distortion and even the operating region between classe A, AB and conduction closer to classe B.

For this reason I strongly advise against using speakers with impedance lower than 8 ohm, especially on this amplifier and with this particular bias system, which I will discuss shortly. In practice the Kent Silver delivers about 25 watts on 8 ohm, while the declared 35 watts might only make sense on a 4 ohm load. As often happens in the commercial world, the number that is most convenient to print on the brochure does not always coincide with the most meaningful one in real use.

The bias system and its limits

The bias system used in this circuit is, in my opinion, one of the most questionable compromises of the entire project. The amplifier works in classe AB, but uses an auto-bias system, that is, cathode self-bias. The problem is that self-bias only makes sense in classe A, where the average current in the tubes remains essentially constant. In classe AB, on the other hand, the idle current is much lower than the average current drawn when the amplifier enters dynamic operation.

This means that in a classe AB amplifier with self-bias the operating point does not stay fixed, but continuously shifts following the signal. In other words, the bias becomes unstable. The cathode voltage rises when the output stage is driven, the operating point moves, and the cathode bypass capacitor inevitably introduces charge and discharge times that generate memory distortion.

In the video below I show exactly this behavior. The cathode voltage of one of the EL34 is about 32volt at idle and exceeds 40volt when increasing the input signal. In the video I raise and lower the signal on purpose to clearly show how the bias rises and falls together with the signal level.

This is also why I advised against using loads lower than 8 ohm. In that condition the operation moves even closer to classe B, the cathode voltage swing increases and distortion worsens. Not only that. The original capacitors I found, nominally rated at 100 volt, had exploded. It is very likely they also paid the price for this kind of stress.

Paradoxically it would have taken very little to obtain a more serious solution. If cost reduction was really necessary, it would have been better to adopt a simple fixed bias, maybe not adjustable, but still more sensible for this kind of configuration. Instead we end up with a compromise that saves a few components but unnecessarily worsens the behavior of the output stage.

Operating voltages and EL34 output tubes

There is often a belief circulating about these Klimo Kent amplifiers that the EL34 tubes are pushed beyond 500volt and that this explains their poor reliability reputation. In reality, measuring accurately, the situation is slightly different. The main smoothing capacitor, which indeed must be chosen for 500volt, during startup sees a peak of about 470volt, but then the voltage settles around 420-430volt once the tubes have warmed up.

The idle plate dissipation is about 25 watts per tube. It is not an absurd value, but it is still rather high. Personally I believe it would have been wiser to stay a bit lower, for example around 22 watts, to reduce the stress on the output tubes and improve long-term reliability margins. Part of the legend that tubes in this amplifier do not last long may also have originated from the EI EL34 tubes originally fitted in some units.

In my opinion these are overrated tubes and, in many cases, also fragile. Today you still see NOS matched quads for sale at completely irrational prices, with enthusiastic descriptions focusing on folkloric details like the getter, triple mica or other nonsense used to make them look special. The practical reality is much less romantic. It was a factory that, in the early 80s, remained to supply the residual demand for tubes that the market still required during the peak transistor boom. Built in the most approximate and spartan way possible, they are tubes that I have seen fail more than once without being abused. My advice remains to leave them where they are and not waste money thinking that just because they are NOS, they are also good quality.

Instrumental measurements

Let’s move on to the data. As already mentioned, RMS power on 8 ohm is about 25 watts. The damping factor is around 6. Below is the frequency response graph, with the ?1dB point around 15khz.

The graph clearly appears rather manipulated. Even without initially having the schematic, inside the circuit you can see several complex RC filters both around the feedback network and on the output tube grids. This suggests that, to bring the amplifier behavior within acceptable limits, it was necessary to correct non-negligible problems of the output transformer. The phase rotation is also quite significant.

Harmonic distortion at 1 watt is around 0.3%. In the spectrum you can clearly see components that suggest intermodulation products around the fundamental. This may depend both on transformer limitations and on the overall circuit behavior in the presence of feedback.

Square waves at 100Hz, 1khz and 10khz

Small personal reflections. How does it sound?

Sarcastic intent: the Klimo Kent Silver is an amplifier that offers an extraordinary sonic experience, characterized by authenticity, clarity and absolute involvement. The low harmonic distortion, standing at about 0.3% already at 1 watt, demonstrates its refined design and natural predisposition for extremely high-level musical reproduction.

With a constant power of 35 watts, the Klimo Kent Silver opens up a wide and deep soundstage, with millimetric precision and impressive three-dimensionality. The image is stable, sculpted, almost holographic. The most complex dynamics are reproduced with authority, cleanliness and absolute control.

The human voice emerges warm, transparent and precise. Instruments materialize in space with crystal-clear detail, reproducing every smallest interpretative nuance. Even with large-scale music the Kent Silver stands out, offering a credible, airy and emotionally rich reconstruction.

In summary, the Klimo Kent Silver is a surprising amplifier, capable of excelling with any musical genre and offering a deeply engaging listening experience.

In reality I was only able to listen to these amplifiers with the speakers available in the workshop, and what I can say for sure is that, once restored, their behavior was consistent with what emerged from the measurements and instrumental tests. The sarcastic tone of the fake review above is not directed at the Klimo Kent Silver itself. It is meant to mock a certain way of talking about audio equipment in some magazines, filled with rivers of evocative words but emptier than a beer barrel after Oktoberfest.

Every now and then someone defines these articles of mine as reviews, but that is not the point. What I do is analyze circuits, observe design choices, measure, verify, and share what comes out, for better or worse. When I find a well-made solution I say it. When I find a questionable compromise, I say that too.

The fake “How does it sound?” section was born exactly for this reason. While unsuccessfully searching for the schematic of these Klimo Kent amplifiers, I came across a text where someone even tried to put them on the same level as the GM20 by Graaf, with a final conclusion basically saying they were on par.

Frankly no. Here we are comparing equipment of very different levels. And this has nothing to do with personal taste or brand loyalty. Even without being a fan of OTL amplifiers, from a technical and construction point of view you cannot put the design attention of a GM20 on the same level as a Klimo Kent, which remains an amplifier built with evident compromises and a significantly lower budget.

Partial repairs and pointless modifications

I also had another pair of Klimo Kent Silver come in, again after passing through the hands of the village idiot. In one of the two amplifiers a power supply capacitor had failed, and the genius decided to replace only the failed one, leaving the other amplifier completely original. Result: the customer ended up with two amplifiers that were no longer identical, one with new components and the other with old components already near the limit. Moreover, the capacitor still present in the other amplifier was also close to the end of its life, so it was only a matter of time before it failed too. This is the classic example of work done without any logic: when working on a pair of monoblocks, especially on critical components like power supply electrolytics, you replace them in both amplifiers, not one yes and one no.

The masterpiece, however, was something else. On both amplifiers someone had decided to replace the signal wire that runs from the RCA connector to the board using solid copper wires about 3 mm thick, because according to him the signal was more “current-capable”.

Of course there was no shielding, and the wire ran less than a centimeter from the rectifier bridge and the entire power supply area, basically in the worst possible place, nicely positioned to pick up noise and interference.

We are at quite high levels of stupidity here, because the signal at that point is very low level and the only thing needed is a proper shielded cable, not a copper rod! Obviously I removed everything and installed two simple pieces of double-shielded Proel audio cable, which is exactly what is needed in that position.

Unfortunately these things happen because many enthusiasts get dragged into strange theories, trends, forums, improvised “gurus” and people who talk about current, miracle copper, cable directionality and other fantasies, without having the slightest idea how a circuit actually works. The result is that even good equipment gets ruined by useless or even harmful modifications. The advice I can give is simple: before modifying an amplifier, or having it modified, ask yourself whether the person doing it has really understood how that circuit works, or whether they are just applying some theory read on the internet. Because between improving a piece of equipment and ruining it, the distance is often much smaller than one might think.

Conclusions

In conclusion, the Klimo Kent Silver is an interesting amplifier, but it must be seen for what it really is. It is not an untouchable holy grail, and it is not junk either. It is a piece of equipment that can work well again, but it requires a properly executed overhaul, especially when it comes from years of use, inactivity or improvised repairs.

If you own a Klimo Kent Silver and notice instability, abnormal noises, output tubes that are overstressed, suspicious capacitors, worn sockets or simply do not know the real condition of your amplifier, the best thing to do is have it properly checked. On equipment like this, improvisation is the fastest way to make things worse.

Electrical schematic sent to me by a reader of the website

Download link

In this article we will explore the repair and upgrade journey of a pair of Audio Note Conquest amplifiers. I was entrusted with the task of restoring two of these units. Unfortunately, they had been subject to tampering that led to unwanted consequences, culminating in the failure of the coupling capacitors between the driver and the output stages and, subsequently, the burnout of one of the two power transformers. However, as often happens, every challenge offers an opportunity for growth, and that is exactly what happened in this case.

Warning! Would you like to fit KR 300B tubes in your Audio Note Conquest??? Do not do it, you could damage your amplifier! A modification to the unit is required to support these tubes. Read this article to learn more.

In addition to repairing the damage, we will explore an intriguing aspect, a small design flaw hidden in the driver stage, which affected both output power and sonic purity. Through a detailed analysis we will show how a careful revision of this section can lead to significant improvements in terms of sound performance. The goal of this article is to share a journey of repair, innovation, and optimization. By following a series of events that included replacing the power transformers, correcting past tampering, and introducing targeted circuit changes, we will lead you through a process that brought these amplifiers back to an optimal level of operation.

Let us start by looking at the original amplifier schematic…

The architecture is remarkably similar to that of the Sun Audio 300BE, with the variant of the output tubes in parallel configuration and a few small differences in component values. Just like in the Sun Audio, here the driver seems to almost want to “outdo” itself, pushing the second triode stage to operate on the edge of saturation. This happens because of a somewhat unusual choice, the removal of a simple capacitor in the signal path. This approach pushes the driver beyond its limits, to the point that it starts to distort even before it can bring the output tubes to maximum power.

It is true that putting a 300B tube on a chassis has a certain charm that attracts enthusiasts, and it is possible that this influenced the design decision. However, it seems that in some cases they chose to follow well known circuits without making the necessary changes. It is as if they opted for the easiest route. A driver like this would have required a much higher supply voltage for the second triode than for the first one, to compensate for the voltage loss under its cathode and allow capacitorless coupling. Some small changes to the power supply stage would have been necessary to implement this solution.

Modifying the driver circuit of this amplifier turned out to be surprisingly simple, it was enough to avoid excessive complications regarding capacitorless coupling between the two triode stages and add the capacitor where it was needed. This allowed the second triode stage to operate in a more normal way. With an adequate amount of plate voltage, it is now able to drive the 300B tubes steadily up to full power without generating significant distortion. In the following images I will show you how the distortion visible in the original driver is surprisingly evident compared to the hypothetical drawback introduced by an added capacitor. We will use LTspice.

In the first image we observe the cathode potential of the second triode of the 6SN7 (in green) compared with its grid in blue. As you can clearly see, the grid reaches positive values (higher than the cathode voltage), causing obvious distortion (click to enlarge).

In the same simulation, we examine the 300B grid signal and compare it with the cathode potential…

The observation shows that the signal reaching the 300B grid, besides being already distorted due to clipping in the driver triode, is still not sufficient to bring the 300Bs into saturation. As a result, these tubes are not delivering the maximum power they could potentially reach.

We will now examine the results of some slight modifications made to the circuit around the 6SN7 and what I was able to achieve. With the same input signal as in the first simulation, now, looking carefully, we can see that the signal reaching the grid of the second 6SN7 triode is still well away from saturation and does not distort…

And while the 6SN7 is still not in saturation, the 300Bs are right at the edge of their maximum deliverable power, with the rising crest of the sinewave almost touching the same cathode potential…

Now let us look, in this last image captured from the simulator, at the discrepancy between what reaches the loudspeaker through the two different circuits (the original circuit and the modified one), even though both have the exact same input signal level (original circuit in green, modified circuit in blue).

The discrepancy between the two is very evident, and anyone who still believes that such a large distortion compromise, aimed solely at eliminating one capacitor in the signal path, is acceptable, I am sorry, but they do not seem to fully understand the situation. In this simulation it should be noted that the original version, despite the considerable distortion, produced 15.44 watts RMS, while the modified version reached 19.55 watts. It is important to stress that these figures are theoretical values derived from the simulator, not taking into account any losses in the output transformer. In the concluding chapter of the article we will examine how the modified version, with the original output transformer, is able to deliver 12 to 14 watts before clipping related distortion becomes significant, and up to 17 watts under full clipping conditions. Theoretically, the original version should manage around 8 watts before significant distortion occurs, although this was not verified directly due to the unavailability of working units. However, it is likely to be the case. It is interesting to note that the Sun Audio (from which this schematic derives) was able to deliver 4 watts with a single output tube.

But now let us look at the work on the amplifiers…

Both units had undergone modifications of various extent. The output stage boards had been removed from their supports and mounted on springs, and in both cases there were clear signs of overheating under the cathode resistors of the output tubes. One of the two units, in particular (the one with the burnt transformer), showed much more evident signs of overheating than the other. I started the process by disassembling the circuit boards and the power transformers.

Unfortunately, the first obstacle I ran into was related to the British origin of the units, the power transformer had a construction specification in “38 column” format, a lamination stack configuration that is not common in Europe and is practically impossible to find. However, I was lucky that the transformer housing box was spacious enough to accommodate a “40 column” transformer. So I rewound two new transformers from scratch, on 40 column cores. In the photo below the old burnt transformer is next to the slightly larger replacement.

Continuing the circuit analysis, I was able to pinpoint the reason behind the overheating of the cathode resistors under the output tubes:

Both capacitors, made in “paper under oil” as some people call them :lol:, made of pure oxygen free copper, wound in an anti inductive way and rated as “audio grade”, both had leakage problems. One was in a dead short, while the other, although not shorted, had such significant leakage that it drove my insulation tester into saturation, even when set to the “charge” mode used to charge the capacitor before testing its leakage resistance.

The cause of this issue lies in the type of oil. In true “paper in oil” capacitors there used to be an oil that can no longer be used in the production of new capacitors due to current restrictions. The oil used in the past was specifically suitable for building these capacitors, but what is used today is decidedly different. Some esoteric brands even rely on vegetable oil, while other, more careful ones choose polypropylene in oil capacitors, because they recognize that oil alone does not offer the necessary reliability. Even if these capacitors sound like polypropylene, the main point is the idea of having oil in them. The fact is that modern ones, sometimes, may not be reliable, especially if they have aged 25 years like these. In the photo below, instead, we see a NOS “West Cap” paper in oil capacitor of avionics type (you can tell by the mounting flange) which should be between 60 and 70 years old and still shows insulation around 100Gohm (100,000 megaohm).

During disassembly, I also found a wire without insulation and with burn marks among those coming out of the damaged power transformer. The white wire is the center tap of the 3.15/3.15V winding that supplies the 6SN7 filament. It is likely that the wire was damaged due to friction caused by the springs, constant movement may have worn it down and brought it into contact with other live parts, eventually causing the short circuit. However, I consider it unlikely that this insulated wire was responsible for the capacitor failures. It is more likely that they were two separate problems that occurred at the same time.

The first step was to scrape away the burnt part of the fiberglass board and then secure what was left with a dedicated UV resin.

The visible result is rather unpleasant, also because the resin is green and the board is red, but at least the crumbling problem has been solved.

Another issue I found concerned the cathode bypass capacitors of the output tubes, which were rated for 35 volts and were of mediocre quality. Considering that in the circuit I expected a voltage above 40 volts, I decided to replace them with high quality capacitors rated for 100 volts, further bypassed with small, very well made polypropylene capacitors. The resistors, on the other hand, were perfect, these green ceramic glass types are very hard to burn.

Then I tested all the tubes…

Minor damage occurred during measurement on the curve tracer, promptly fixed with 3D resin printing…

First bench test of the new driver, powering the board with the stabilized anode supply described in this article…

Amplifiers rewired:

Instrument data: Genuine power delivery reaches 14 watts before the onset of significant clipping related distortion. The damping factor (DF) sits at 3, exceeding my expectations, also considering that this is a “zero feedback” amplifier.

Harmonic distortion 0.27% at 1 watt

Unfortunately, the output transformer shows unpleasant resonances that appear starting at 10 kHz. In addition, a trick was adopted to try to mask these resonances, namely connecting the primary in reverse, the start of the winding was connected to the tube plate, while the end of the winding was connected to the anode supply. This configuration was chosen to create a hidden capacitance between the output tube anode and the reference point to ground. In any case, the measured response is -0.5 dB at 20 Hz and -1 dB at 25 kHz.

It is almost amusing to see how on the internet you can find adventurous individuals who, in an attempt to improve output transformers, have chosen to replace the laminated core with a double C core. This, evidently, because the double C core is considered “prettier”. However, in light of the fact that there is a 0.5 dB drop at 20 Hz, one has to wonder what real benefit is obtained by changing the core.

The problems of this transformer, if we really want to “nitpick”, are present in the highest region of the frequency response, not in the lowest one.

Let us look at the square waves at 100 Hz – 1 kHz – 10 kHz

Here they are finished and ready to go back home!