In the landscape of commercial tube amplifiers, it is not uncommon to come across brands that, being incapable of designing truly valid equipment, inevitably end up disappearing within a few years. Before leaving the scene, however, these companies often manage to put a surprising number of poorly conceived amplifiers on the market, which then continue to circulate on the second-hand market, systematically disappointing those unlucky enough to buy them. I could mention AMC, with its tube amplifiers assembled according to construction criteria worthy of a VCR, where the tubes are even soldered directly onto the printed circuit board without the use of sockets; Gamma Acoustic, which proposes triodes connected as diodes in series with the power supply without any apparent technical logic; Music Angel, whose amplifiers concentrate 80 percent of the cost in the tubes, which are then mounted on circuits made with materials of questionable quality; and finally the Audiospace Galaxy 34. The latter, among all of them, is probably the least bad: it uses output transformers of at least acceptable quality and a well-made chassis. Unfortunately, inside it hides a circuit devoid of any real design coherence, which seems to have been conceived by someone used to thinking exclusively in transistor terms, even going so far as to try to use a triode as if it were a transistor.
Among all these failed brands, the Audiospace Galaxy 34 is the one that offers the greatest potential to be modified in a non-invasive and relatively simple way, in order to turn it into a decent amplifier. This article may spark the interest of some of my detractors, those who do not like modifying things. However, honestly, I do not care at all about the opinion of these individuals, and I let them grumble as much as they like.
I am updating this article seven months after the first version because I was given another unit to modify. In addition, I made a small improvement to the driver circuit compared to the initial version, a minimal intervention that nonetheless further optimizes the balance of the phase splitter.
Let us begin the story of this unit with the messages that were sent to me by two different owners:
- My amplifier distorts terribly from halfway up the volume onwards. The left channel seems weaker than the right. I am in Rome and the technician I took it to replaced the four power tubes (TAD E34L 22/27). Can you help me?
- Since the volume potentiometer of my Audiospace Galaxy 34 no longer responds to the remote control and I hear strange noises from the left channel, I would like you to modify it as I have read and check its operation, possibly repairing it.
The client entrusted me with his amplifier and, during an in-depth conversation, complained about a horrible sonic performance. In particular, he noted that by turning up the volume, the music turns into a confused mess, becoming practically unlistenable. To analyze the situation, I connected the amplifier to the test bench, using a wattmeter and an oscilloscope to identify any anomalies. Initially, on the load resistor, I did not detect any strange behavior. However, the damping factor measurement showed a DF (damping factor) value of 2 on a 40-watt amplifier. It became evident that such a low damping factor value can turn the amplifier into a source of extremely high distortion. This observation calls into question those who downplay the importance of the damping factor, demonstrating how it can significantly affect sound quality and debunking the arguments of those who consider this parameter insignificant.
I temporarily set the amplifier aside and said goodbye to the client, since I could not immediately tackle the job. After some time, as soon as I had the opportunity to get back to it, I started from the assumption that the amplifier was running at zero feedback and that the driver circuit might need some modifications, including the addition of an appropriate feedback mechanism to improve the overall operation of the amplifier. The first issue I encountered was a marked instability of the circuit when the volume was increased beyond a certain level. Under this condition, the circuit began to oscillate strongly at low frequency. In the attached video, I show this phenomenon, and observe carefully: I had not yet made any modifications, this is exactly how it behaves in the original configuration.
Turning the volume to maximum was necessary for the measurements, since the level of the signal injected by the function generator was handled directly by the measuring equipment. However, afterwards I set the volume to a stable level and proceeded with the acquisition of the frequency response and harmonic distortion.
Well, at minus 1 dB at 10 kHz with just 1 watt of power, it almost seems to outperform a 1950s Geloso PA amplifier. All right, in reality this measurement was not my priority. I captured it only for the readers. The next step was to decide to open the amplifier. I wanted to obtain the schematic, since my idea was to examine how they had biased the first input triode. I had in mind to split its cathode resistor in two and inject the feedback signal there. What follows is the schematic of the input stage, the triode of the 6SL7:
 |
 |
In all honesty, embarrassing. The yellow wire going to the 8 ohm terminal suggested the existence of some form of feedback, but its implementation was as horrible as one could imagine. The presence of a 220 ohm resistor between grid and cathode created a grid-at-zero-volt bias. Adding the 3.3uF electrolytic capacitor, the drive occurred in positive grid. To complete the mess, there was a snubber on the anode load resistor, adding further phase rotations and increasing distortion. In a non-offensive way, I can only conclude that whoever conceived this biasing method for a tube seemed to have no idea what they were doing, adopting an approach vaguely similar to what one would do with a BJT transistor. At this point, I decided to stop my analysis of the circuit, since it seemed to be a waste of time. Consequently, I made the decision to completely remove the driver circuit, with the aim of rebuilding something that could work more effectively.
I actually inverted the placement of the 6SN7 and 6SL7, using a triode of the 6SN7 as the input stage, the 6SL7 as a long-tail phase splitter, and the second triode of the 6SN7, with a small trick, I made it work as a current sink under the 6SL7.
Normally, a current sink is implemented with a pentode, but with some precautions it is also possible to use a triode. Of course, someone more technically inclined might suggest that it would have been better to use the 6SN7 as the phase splitter driving the output tubes and the 6SL7 as the input tube, and that is true. However, a single triode of the 6SL7 cannot handle all the current that can flow in a complete 6SN7. I also did not want to run the 6SN7 at excessively low currents, as that would have compromised its linearity. With this small compromise, aimed primarily at preserving the original tubes of the amplifier, in the end the circuit worked excellently.
The amplifier featured a selector that allowed the output tubes to be configured either in triode mode or ultralinear mode. This selector controlled two dedicated relays that handled the switching between the two configurations. However, beyond its apparently versatile functionality, I found some aspects that raised concerns.
First of all, there was the risk of arcing inside the relays, an issue that could compromise the stable operation of the amplifier over time. Secondly, personally, I am not a supporter of these circuital contraptions. The main reason lies in the fact that, by radically changing the gain of the output tube, it becomes extremely difficult to perform a precise tuning of the circuit and the feedback network.
In other words, if I optimize the setup in ultralinear mode, it might not work equally well in triode mode, and vice versa. These technical compromises often seem to be introduced only to attract inexperienced customers, who see these options as an added advantage without realizing that, in reality, they represent a compromise and a potential source of problems. Therefore, I made the decision to completely disconnect the relay and directly wire the leads from the output transformers to the tubes, fixing the configuration in ultralinear mode to ensure greater stability and optimal performance.
Already at the first power-up, even before implementing any form of feedback, I noticed a remarkable change in the frequency response of the transformers. This change manifested itself as a significant extension towards the higher frequencies. The difference was so evident that it immediately caught my attention, suggesting that the modifications made were already positively influencing the sonic characteristics of the amplifier. This encouraging sign strengthened my conviction that the modifications and optimizations would lead to notable results in the overall optimization of the system.
The old driver, apparently, was so inefficient that it negatively affected the proper operation of the circuit. Its limitations were so evident that, upon replacing it with the new setup, I noticed an immediate improvement in overall performance. In particular, with the volume set to maximum, no instability or oscillation occurred anymore, further highlighting how the new driver circuit had contributed to ensuring greater operational stability.
During these initial testing phases, another significant issue emerged related to the volume potentiometer, whose value was excessively high (250kohm). This episode represents a warning and an important lesson for all DIY enthusiasts. Often, the mistake is made of using potentiometers with very high values, such as 200k or even 500k, without realizing the potentially undesirable consequences.
The problem lies in the presence of a parasitic capacitance in the circuit located downstream of the potentiometer wiper. Combined with the capacitance of the shielded cables, as in this case, and the input capacitance of the tube, a low-pass filter is formed. The cutoff frequency of this filter varies depending on the volume setting. To demonstrate this issue, I created a video in which I inject a 10 kHz square wave into the amplifier with the volume set to maximum. Subsequently, by lowering the volume, the waveform undergoes a significant transformation, highlighting a notable increase in rise time.
This concrete example underscores the importance of carefully selecting the potentiometer value in order to avoid unwanted alterations in frequency response and to ensure optimal amplifier operation. The amplifier was equipped with a motorized potentiometer of Chinese manufacture, of a quality that I would define as low cost. Unfortunately, I could not replace it with a conventional motorized potentiometer, since the dimensions and layout of the chassis did not allow it.
Faced with this limitation, I decided to give up the remote-controlled volume functionality. In its place, I installed a simpler potentiometer, with a resistance of 100k, which solved the frequency response problem. Later, in the instrumental analyses, I will present a frequency response graph to highlight the severity of the impact that the original potentiometer caused.
For those who might be interested, it is possible to provide an alternative solution to linearize the frequency response when using a very high value potentiometer. The proposal involves implementing filters and capacitors, as illustrated in the attached image. However, I chose to avoid this particular solution.
The main reason why I avoided using it is that this circuit, besides being impractical to mount in the specific available position, did not satisfy me in terms of circuit cleanliness. Moreover, the low value potentiometer I used proved to be a better choice. Modern audio sources, such as DACs and CD players, all have low output impedance, and therefore do not present difficulties in driving even much lower loads. This option helped keep the circuit cleaner and fully met the needs of the system.
Let us now move on to the instrumental analyses. I pushed the output stages up to 45 watts, while the damping factor was measured at a value of 8. This choice aims to ensure optimal performance and clean sound without compromising the quality of the amplifier.
Below you will find the frequency response (at 1 watt) of the amplifier with the volume potentiometer at 100 percent. The frequency response was measured with the feedback already tuned, including compensation. One can observe minus 0.2 dB at 20 Hz and minus 1 dB at about 80 kHz.
The output transformers of this amplifier prove to be quite good, with overall performance that would even be excellent, were it not for a marked resonance that appears as one approaches 100 kHz. This flaw, however, can be effectively suppressed by compensating the feedback in the circuit. This characteristic, which makes amplifiers of this model particularly attractive, can be fully exploited when one finds a used unit at a favorable price, especially if the intention is to make modifications.
The modifications required to the circuit are limited to a few resistors, and the rewiring process is relatively simple. This makes the amplifier an ideal option for those who wish to perform customizations and improvements, fully exploiting the potential of the quality output transformers, while efficiently managing the resonance through feedback compensation. Now let us see what the frequency response looked like with the original 250k potentiometer set at half travel. Minus 1 dB at 7 kHz.
The square wave measurements highlight the excellent characteristics of the output transformers (100 Hz, 1 kHz, 10 kHz):
This is the harmonic distortion:
Modification of the BIAS adjustment
According to the amplifier manual, the bias adjustment should be set to 300 mV for each EL34 tube, corresponding to about 30 mA. With an anode voltage of 475 Vdc and 230 V on the mains, this results in a dissipation of about 14 watts per tube. However, EL34 tubes can sustain up to 25 watts of dissipation, so they are still well below their limits.
I noticed that the electrolytic capacitors of the amplifier are rated for 450 V, while in reality they operate at a higher voltage, which could reduce their lifespan. To solve this problem, I decided to increase the bias current to 45 mA, corresponding to 450 mV measured at the test points with a multimeter. This reduces the anode voltage to 450 V and brings the EL34 dissipation to 20 watts per tube.
I could have increased the current further, but I preferred not to do so in order to avoid overloading the power transformer, whose exact specifications I do not know. As a precaution, I ran the amplifier with the bias set to 45 mA for about an hour and monitored the temperature of the power transformer, finding that 50 degrees are acceptable.
Photo of a defective 6SL7 with only one section lit at twice the brightness.
Replacement driver tube set with NOS, 6SL7 Haltron and 6SN7 OTK.
These problems are quite common when buying something of Chinese origin. It is normal to find well-made cabinets in steel and solid milled aluminum, since such machining does not entail high costs in China. However, what is often found inside the chassis is absolute nothingness: senseless circuits and other low-quality components.
However, in the specific case of this unit, the presence of two fairly good output transformers opens the door to high-level results with a relatively limited expense. I invite readers who own one of these amplifiers to contact me if they are interested in having their unit modified. With a few targeted interventions, it is possible to obtain substantial improvements in overall performance.