The Audion Sterling Stereo MK1 is a tube integrated amplifier that has been widely discussed in the hi-fi world, often coveted for its minimalist approach and for the choice of EL34 tubes in a single ended circuit. In this article I document a real repair and, above all, an in-depth instrumental analysis of the circuit, with measurements and technical considerations regarding feedback, stability, bandwidth, square-wave response, and the construction quality of the transformers. The goal is not to produce a commercial review, but to understand what really happens on the workbench when a “mythologized” device meets the reality of components, tolerances, and the reactive loads of a loudspeaker.

If you are looking for reliable information about the Audion Sterling MK1, EL34 single ended, ultralinear operation, damping factor, and high-frequency oscillation issues, here you will find a concrete case: a unit that arrived in the lab in critical condition, with anomalous symptoms and a history of burnt tweeters. We will see what had already been attempted, what I found during diagnosis, and why certain circuit choices, when combined with sub-optimal transformers, can lead to unpredictable behavior.

First, I want to report the translation of what is said about the Audion Sterling MK1…

The Audion Sterling Stereo MK1 EL34 model stands out for its affordable price while offering high-quality sonic performance. Produced by Audion, it represents a cost-effective option within the amplifier range. The minimalist design reflects traditional characteristics of tube sound. The amplifier lacks a remote control and a dedicated turntable input, but this does not detract from its essence.

The slim chassis, made of laminated material, gives the amplifier a sober and elegant appearance. Attention to detail is evident, with some parts assembled without the use of screws or rivets. The beauty of this amplifier lies in its simplicity. The front panel features an anodized surface, while the other body panels are painted gray.

Inside the unit are small output transformers of non-Chinese origin, as indicated in the original specifications. This concept extends to all components of the amplifier, with exclusive use of elements sourced from America and Europe. Care for detail is also evident in the arrangement of the five input selector switches, which gives the amplifier a unique appearance. There is also a tape output and the possibility of lifting the ground.

At the heart of the system is the circuit, powered by EL34 tubes without negative feedback. The driver stages use NOS Jan Philips 5687 tubes, and the entire amplification chain operates in class A. The limited output power of 12 watts per channel requires high-sensitivity loudspeakers for optimal performance.

The Audion Sterling Stereo MK1 EL34 is distinguished by its sonic delicacy. This means that clarity of reproduction is prioritized over sheer power. Despite the ability to increase volume, slight softness in the high frequencies and mild clipping in the midrange can be noticed at higher levels. Exceeding a certain volume threshold may not enhance the listening experience, but instead make the amplifier seem strained and struggling.

In terms of listening comfort, an increase in gain can lead to a variation in tonal balance in the low frequencies. However, this characteristic should not be considered a flaw, but rather an aspect that contributes to a sense of “loudness.” The amplifier’s dynamics are modest but respond quickly and are engaging. Low-frequency impact is notable, while mid frequencies show a clear and expressive attack. High frequencies offer pleasant auditory transparency. While lacking the typical tube warmth, the amplifier maintains a musically engaging balance and impressive detail rendition. For example, soprano voices are reproduced neutrally. Background noise is minimal, while high frequencies are slightly emphasized, improving clarity and naturalness of the overall sound. In conclusion, the Audion Sterling Stereo MK1 EL34 presents itself as an amplifier that combines design simplicity with sonic excellence, offering a delicate and engaging audio performance.

This description, as often happens, emphasizes design intentions and listening impressions, but leaves some fundamental points uncovered: what really happens in terms of stability with a real load, how much safety margin exists in the gain chain, how coherent the claim “without negative feedback” is with what is actually seen in the circuit, and how much the quality of the output transformers matters in a single ended amplifier operating in class A with an EL34. These are precisely the issues that emerged in my analysis.

An example of the Sterling MK1 was brought to me partially disassembled because someone before me had tried to work on it to understand its problems. The owner reported having had to replace the tweeters of his speakers several times over the years, as they would occasionally burn out without a clear reason. The previous technician had noticed that one of the two channels distorted at low frequencies, but lacked the specific skills and instruments needed to investigate further. So, hoping for a solution, he handed it over to me.

When a client tells a story like this, the first thing that interests me is not “how good it sounds,” but what conditions can cause an amplifier to abnormally stress a tweeter. A tweeter typically burns out due to excessive high-frequency power, prolonged clipping that generates high-order harmonics, or ultrasonic oscillations that are not audible but heat the voice coil. In a tube amplifier, especially if there is feedback and the output transformer introduces significant phase rotation, the hypothesis of oscillation at very high frequencies can never be excluded. For this reason, in addition to “classic” measurements of power and THD, I immediately set up tests with square waves, resistive loads, and observation of behavior well beyond the audio band.

This amplifier is known for its enthusiastic reviews and its place of honor on many audiophiles’ wish lists. However, enthusiasm generated by marketing promises does not always coincide with practical reality. In my examination of the Audion Sterling MK1, I discovered details that raise questions about its true performance.

In particular, when a design focuses heavily on a “minimalist” image and on a few eye-catching specifications, it is easy for certain compromises to remain hidden until real measurements are taken. A single ended EL34 circuit can be excellent, but it is extremely dependent on the output transformer, proper management of overall gain, and any applied feedback. If even one of these elements is out of balance, the amplifier can become nervous, sensitive to disturbances, and in some conditions even dangerous for loudspeakers.

One of the first aspects I investigated was the issue of feedback. Contrary to what is stated, during circuit analysis I noted the presence of excessively heavy feedback. This is not a problem in itself, but it is important to understand its implications on sound and behavior, especially if this excess of feedback is such that it brings the circuit to the point of instability, possibly even being the cause of tweeter damage in certain situations.

Here it is worth clarifying one point: feedback is not inherently “good” or “bad.” If properly implemented, it reduces distortion and output impedance, improves load control, and makes the amplifier more linear. Problems arise when feedback is heavy but the feedback loop is not adequately compensated and the transformer introduces significant resonances and phase rotations. In that case, feedback that should stabilize can become the trigger for instability. This is a typical scenario in which an amplifier may seem fine with one speaker and problematic with another, because the real load is not a pure resistance but a complex combination of impedance, phase, and resonances.

The schematic of the unit, found on the internet and containing an error (the anchoring resistor of the second triode is missing)

Even the schematic that “circulates” online is part of the problem, because many people take what they find at face value without verifying that it actually corresponds to the unit on the bench. In this case there is also an error, and when components that define an operating point or stabilize a stage are missing, interpretations become misleading. For this reason, when I carry out serious analyses, the priority is always to reconstruct reality: trace connections, check values, and reason about the circuit as it is built, not as it is described.

One of the most surprising moments of my analysis was when I identified a transformer with a serious construction defect: an open core with an air gap of at least 3 millimeters. This defect has direct implications for amplifier performance and raised questions about construction quality and quality control in general. PI could joke by saying that this transformer sounds “transparent,” the gap is so wide that light passes through it!

In a single ended output transformer, the air gap is a functional element, because it prevents core saturation due to the DC bias current of the output tube. But “air gap” does not mean “random slot.” If it is too large, primary inductance drops, losses increase, low-frequency response worsens, and distortion at low frequencies rises rapidly. If the lamination assembly is uneven, or if the E and I laminations do not mate correctly, additional losses and resonances are introduced. In practice, you get a transformer that does not work properly.

I do not intend to undermine the reputation of this product or its supporters, but I believe it is essential to look beyond marketing and enthusiastic reviews. The truth behind any product must be evaluated objectively, based on concrete facts and tangible evidence. My first intervention was to dismantle the open transformer to fix it…

In cases like this, the point is not “to create controversy,” but to refocus attention on what matters: measurements, construction, and repeatability. An amplifier can be fascinating, it can have a particular character, but if it shows instability or has transformers operating out of specification, it becomes a device that requires attention and targeted intervention. And when burnt tweeters are involved, the issue must be treated seriously, because the risk is not theoretical.

Unfortunately, it was not as simple as I had hoped. In reality, all the laminations were glued together as a single block due to the resin with which they had been impregnated. I had to completely open the transformer, separate the laminations one by one, since the “I” group did not align with the “E” group, reposition them correctly, replace the air gap, and close the transformer again.

This step deserves some additional detail, because it clearly explains how a “mechanical defect” immediately becomes an “electrical defect.” If laminations are glued irregularly or the stack is not properly clamped, the magnetic circuit is not uniform. In an output transformer, this means more stray flux, more losses, more distortion, and a greater tendency to trigger resonances.

Afterwards, I performed a quick inspection of the components on the board, replacing some deteriorated capacitors and a resistor that had been compromised by excessive heat. The resistor was replaced with one of the same value but with higher dissipation capability. I also examined the tubes, replacing a very worn ECC88, while keeping the other components unchanged.

When a device arrives already “partially tampered with,” the rule is to first restore basic reliability. Capacitors out of specification and thermally stressed resistors can shift operating points, increase noise, and make measurements inconsistent. Even a very worn tube can introduce imbalance, microphonics, noise, and sensitivity to disturbances. At this stage the goal was not “optimization,” but bringing the circuit back to healthy conditions in order to measure properly and attribute symptoms to real causes, without unnecessary variables.

I performed the first power-up with a variac without encountering problems, then I checked how the repaired transformer behaved at low frequencies. In the image below, even if the trace is only partially visible, you can see the two transformers delivering 20Hz. It is quite amusing that the one producing the ugly distortion is the transformer that was still intact. The explanation is very simple: I used a thinner air gap than the original, the thin-lamination core is evidently of good quality and does not saturate, so the transformer I repaired now works better than its twin, which, having a thicker air gap, has more losses and therefore goes into distortion earlier than the repaired one.

This is one of those cases where measurement immediately overturns gut intuition. One would expect the repaired transformer to be the weaker one, and instead it is the other that shows distortion first at 20Hz. The reason, as explained, lies in the compromise between DC current, air gap, lamination quality, and transformer design. A thicker air gap can “handle more DC current” but lowers inductance, increases losses, and brings low-frequency distortion earlier. If feedback then attempts to correct what the transformer physically cannot deliver, the amplifier may react with even stranger behavior, especially near its limits.

Subsequently, I performed an analysis of the amplifier’s power delivery. Unfortunately, this amplifier seems to have been influenced by the fashion of single ended ultralinear operation, a term that often raises exaggerated expectations, ignoring the fact that this technology was originally conceived for push-pull applications. Further details on this topic can be found in a dedicated article here. It is worth noting, however, that the damping factor of this amplifier reaches a value of 18, which gives an idea of the excessive amount of feedback applied to the circuit.

Such a high damping factor in a single ended amplifier with an output transformer does not come “for free.” It means that output impedance has been reduced through heavy feedback, and this requires a stable and well-compensated loop. In theory, a high damping factor can provide tighter bass control, but in practice it can also mask problems, because the circuit attempts to correct what is introduced by the transformer, with the risk of emphasizing resonances and instabilities outside the audio band. Moreover, when an amplifier becomes very sensitive, it takes little to push it to the limit, which directly connects to the issue of tweeters and possible ultrasonic phenomena.

Despite this considerable feedback, the resulting sine wave remains visibly asymmetric. Output power under deep clipping conditions corresponds to the manufacturer’s declared value of 9 watts, but a more accurate evaluation shows that power before clipping in this single ended ultralinear circuit stops at about 6 watts or 6.5 watts at most. Despite the significant amount of feedback applied to the circuit, it is surprisingly sensitive, reaching clipping with an input signal of just 160 millivolts peak-to-peak. This sensitivity is such that disturbances can be observed entering the signal simply by bringing a hand close to the volume knob. Two stages to drive an EL34 are too many, the ECC88 alone would probably have been sufficient.

This abnormal sensitivity is an important signal. An amplifier that clips with such a low input not only reduces dynamic margin with modern sources, but also becomes more vulnerable to noise, capacitive disturbances, and unwanted coupling. It is something that can also be perceived “by ear” as nervousness or ease of clipping, but on the bench it is even more evident because the numbers are clear. In addition, in a circuit with more stages than necessary, each stage adds phase shift, parasitic capacitances, and opportunities to trigger oscillations, especially when feedback closes the loop and makes everything more critical.

However, the main issues with this amplifier emerged when I conducted tests on square-wave behavior and generated the bandwidth plot, which are visible in the following images:

Square-wave tests are not a “lab obsession,” but a very direct way to understand how stable an amplifier is and how well the output transformer behaves. If you see overshoot, ringing, and pronounced resonances, it means there is energy bouncing between inductances and capacitances, and if feedback is involved, that energy can become a serious problem. The bandwidth plot then immediately shows whether the device is linear, whether it has out-of-band peaks, and whether phase behavior is moving in a dangerous way.

10kHz square wave:

Frequency response

The behavior of this amplifier shows some odd characteristics when examining square waves and analyzing the bandwidth plot. At low frequency, at 20Hz, a slight increase of +1dB relative to 100Hz is observed, a phenomenon that, while unusual, does not appear particularly problematic. However, the real concerns emerge in the high-frequency region.

That slight low-frequency rise can have various explanations, for example coupling between the feedback network and transformer response, or a resonance effect linked to capacitor and inductor values in the signal path. On its own it would not be dramatic. The problem is that when the same system shows significant resonances at high frequency, it means the circuit is also “playing” outside the audio band, and this is exactly the territory where a tweeter can suffer without the listener immediately noticing.

A resonance appears at 60kHz, followed by a second at 90kHz, with a clear phase inversion just slightly after 90kHz, highlighted by the abrupt change in direction of the blue line at the bottom of the graph. These resonances are likely related to the output transformers and seem to be amplified by the strong feedback present in the circuit. In fact, in the presence of a 180-degree phase rotation, negative feedback becomes positive, leading the circuit to instability. This situation is further confirmed by the absence of the typical compensation capacitor in parallel with the negative feedback resistor in the circuit, as shown in the diagram above.

When phase rotates and feedback “changes sign,” the amplifier is no longer correcting, it is self-feeding an error. At that point it takes very little to trigger oscillations. The most insidious aspect is that these oscillations can be intermittent, dependent on the cable, the loudspeaker, the input selector, the volume position, and even the nearby electrical environment around the potentiometer. The result is an amplifier that may seem normal in some conditions and become aggressive, unstable, or “sizzly” in others, without the owner being able to identify a logical cause.

I performed two tests with the addition of a compensation capacitor, but the circuit immediately began to oscillate violently at radio frequencies, an unequivocal sign that it is at the limit of stability in the presence of a phase rotation exceeding 180 degrees. This situation is critical. Under these conditions, when the amplifier is connected to reactive loads such as loudspeakers, it is possible that during audio signal modulation, momentary oscillations occur that could interact with the music. This situation may not manifest with all speakers, but the owner’s experience, having already replaced tweeters four times, is clear evidence of the stability problems of this unit.

And here the initial story of burnt tweeters becomes perfectly coherent: if an amplifier oscillates at radio frequency or in the ultrasonic range, that content still ends up in the transformer, the cable, and the crossover. Even if you do not hear it as “sound,” the tweeter converts it into heat. Additionally, the crossover and the tweeter’s impedance can create even more unfavorable conditions, because they introduce further phase rotations and impedance peaks. It is a classic scenario where an amplifier appears “ok” on the bench with a resistive load, but becomes dangerous with real loads, especially at sustained volume or with certain recordings rich in transients.

For completeness, I also share the total harmonic distortion (THD) plot, which is around 0.6% at 1 watt of power. This is a further indicator of the error resulting from adopting ultralinear connection in a single ended circuit. In a circuit with so much feedback, one would expect THD no higher than 0.1% at 1 watt, but the observed value is significantly higher.

This figure is interesting for another reason as well: when THD remains relatively high despite heavy feedback, it often means that the main nonlinearity is not “easy to correct” with feedback. In other words, distortion may be generated in points where the loop cannot truly impose linearity, or it may be linked to magnetic phenomena of the transformer, such as saturation, hysteresis, or losses, which cannot be tamed simply by increasing feedback. It is a warning sign because it suggests that we are asking the circuit to compensate for a physical limit that cannot be overcome.

In conclusion, I would like to emphasize that this article aims to provide a technical, instrumental, and scientific analysis of the Sterling MK1 amplifier circuit, without any intention of being a review in the commercial sense of the word. It is important to clarify that the objective here is to provide objective data based on measurements, not to express personal judgments.

The key point for the reader is to understand that a device can be appreciated for aesthetics, simplicity, and “sonic signature,” but this does not eliminate the need for technical robustness. Technical robustness means stability with real loads, sensible gain margin, coherent transformers, and control of out-of-band resonances. If one of these pillars is missing, problems can remain silent for years and then erupt in the form of failures, stressed components, and in the worst case, damaged loudspeakers.

I do not intend to offend or diminish those who may be enthusiasts or appreciate this device. I report what I measured and observed, and the critical issues that emerged from the technical analysis of the circuit. Every enthusiast or owner of this amplifier has the right to evaluate this information objectively.

For me, the most important thing is always transparency of method: what tests I performed, what the instruments show, and which conclusions are justified by the data. It is also the reason why I publish images of square waves, bandwidth, and spectrum. These are repeatable pieces of information, and anyone with adequate test equipment can compare them, confirm them, or discuss them on a technical basis.

The unit was returned to the technician who brought it to me, and I shared all the observations and measurements reported in this article. It will be up to the amplifier’s owner, with whom I am not in direct contact, to make an informed decision on how to proceed. However, personally, I believe that resolving the issues identified may require a significant modification of the circuit.

When critical issues are related to feedback loop stability and transformer resonances, it is often not enough to “change a couple of components.” A comprehensive view is required: redistribute gain, define proper compensation, and, if necessary, intervene in the operating philosophy of the output stage. In a single ended design, every choice heavily affects the result, and compatibility with loudspeakers cannot be left to chance.

Among possible modifications, I believe that eliminating the ultralinear connection and opting for a purely triode or purely pentode configuration could be a solution worth considering. In addition, reducing overall circuit gain and lowering the amount of feedback could help improve the amplifier’s performance.

A purely triode configuration can improve intrinsic linearity and often makes it easier to achieve stability with less feedback, at the cost of power. A purely pentode configuration, if well designed and with reasonable, well-compensated feedback, can provide more power and better control, but it requires disciplined design and suitable transformers. In both cases, however, the direction is the same: avoid “hybrid” solutions adopted for fashion, and aim for a measurable, stable, and repeatable technical balance.

It is important to emphasize that if the amplifier’s owner decides to make modifications, I will update this article, and that I am willing to collaborate and provide assistance for a possible upgrade to other people as well.

If you own an Audion Sterling MK1 and want to verify whether your unit is stable and safe with your loudspeakers, the correct approach is always the same: complete instrumental measurements, bandwidth verification beyond 20kHz, square-wave testing, checking for any ultrasonic oscillations, and evaluation of the output transformers. Only in this way can you understand whether the device is healthy, whether it can be optimized with targeted interventions, or whether a deeper revision is needed to eliminate the issues at their root.