Produced between the late 1970s and the early 1980s, approximately from 1977 to 1982, the Luxman MQ3600 is one of the most representative tube amplifiers ever built by the Japanese manufacturer Luxman. This iconic unit, capable of delivering 70 watts RMS per channel, was developed in a period when Luxman was pursuing the best possible balance between sonic refinement, technical design, and originality of engineering. What makes the MQ3600 truly unique is the use of the extremely rare 8045G output tubes, manufactured by NEC specifically for Luxman. The 8045G are essentially KT88 tubes with the screen grid internally connected to the anode, in practice KT88s forced to operate in triode mode. This made them very distinctive devices, but today they are unfortunately no longer available, and their replacement necessarily requires circuit modifications.

Features and recommended modifications

Luxman itself already provided guidelines at the time for replacing exhausted 8045G tubes. The basic modification consists in connecting the screen grid pin to the anode pin through a 330ohm resistor, allowing the use of the much more common KT88 or 6550. Furthermore, since the MQ3600 output transformers are equipped with taps for ultralinear operation, it is also possible, if desired, to configure the new tubes in ultralinear mode instead of triode, achieving higher output power at the expense of a slight tonal change.

What to do if the 8045G tubes in your Luxman MQ-3600 are exhausted

The unit shown in the photos arrived faulty: one of the 8045G output tubes exhibited the characteristic white “getter”, a clear sign of loss of internal vacuum. As already mentioned, the 8045G were tubes developed by NEC to Luxman specifications and today they are completely unobtainable, except occasionally as used salvage tubes. In these cases, adapting the circuit for more readily available tubes becomes unavoidable. The standard modification, already recommended by Luxman at the time, consists in connecting the screen grid to the anode through a 330ohm resistor, allowing the use of standard KT88 or 6550. Since the output transformers are equipped with ultralinear taps, it is also possible to connect the new output tubes in that mode, obtaining higher output power compared to triode operation, albeit with a slight tonal change.

The amplifier also showed significant damage caused by an internal short circuit in one of the four output tubes. This fault resulted in several resistors around the sockets being burned, multiple internal wiring runs being heat damaged, and the bias adjustment trimmers being destroyed. In addition, the filter choke used to discharge the large smoothing capacitors, located right next to the faulty tube, had burned out. The choke, completely soaked in tar, was cleaned and subsequently rewound to restore full functionality.

In addition to the modification under the output tube sockets, Luxman also recommended adding two 22k grid stopper resistors at the signal input. Furthermore, three bypass capacitors were specified at points B2, B3, and B4 on the printed circuit board. In these positions, I installed 470nF polypropylene capacitors of suitable quality for audio use. I also bypassed the two cathode electrolytics of the 6AQ8/ECC85, replacing them with a 220nF polypropylene capacitor to improve high frequency behavior. Finally, I removed the old flat Nichicon capacitors, components notoriously prone to degradation in Luxman units of that era. In every sample examined so far, they were found to be out of specification or compromised.

During inspection, I tested the preamplifier tubes with the uTracer and found that one of the two 6042G tubes was partially worn out: one section was still functional, while the other was no longer usable. I therefore replaced them with a matched pair of 6CG7, which are electrically compatible tubes.

At this point it is useful to clarify a common misconception found online: it is true that 6CG7 tubes can be used in place of 6042G, but they are not perfectly equivalent. The curves obtained with the uTracer clearly demonstrate this: the 6042G curves are more widely spaced, indicating a higher mu, amplification factor, compared to the 6CG7. The substitution therefore results in a slight reduction in overall gain and in the amount of negative feedback, with a consequent change in the amplifier’s tonal character.

Ultralinear modification

One of the most common and sensible modifications to the Luxman MQ-3600 concerns the replacement of the now rare 8045G output tubes with the more readily available KT88 or 6550. Luxman itself recommended this substitution at the time as an official alternative. The KT88 can be used while maintaining triode connection, replicating the behavior of the 8045G, or one can opt for an ultralinear configuration by using the dedicated taps already present on the original output transformers. In the latter case, the circuit benefits from increased output power, rising from about 60 watts with the 8045G to about 80 watts with KT88 or 6550 in ultralinear operation. The damping factor is slightly reduced, from 16 to 13, but harmonic distortion behavior remains essentially unchanged, as also confirmed by official Luxman graphs.

Among the modifications recommended by the manufacturer is also the addition of three decoupling capacitors on the supply lines feeding the ECC85 and 6240G/6CG7 tubes, to improve stability and reduce interaction between stages. Personally, I always replace the flat gray Nippon Chemicon capacitors, which are often out of specification both in capacitance and insulation. These components systematically degrade over time and fall well below nominal values. Another necessary intervention concerns the bias voltage generation circuit. In the original configuration, adjustment is often near the end of the trimmer range, with idle currents too high for KT88 tubes, making it necessary to modify the resistor values in the divider. Finally, Luxman also suggested adding two grid stopper resistors at the inputs of the output tubes to prevent high frequency instability phenomena, especially when KT88 or 6550 are installed in ultralinear configuration.

Luxman MQ-3600: technical recovery of a severely damaged unit

Before anyone feels the need to raise objections, often more ideological than technical, it is important to clarify a fundamental point: the decision whether to rewind or replace faulty transformers does not belong to the technician, but to the owner of the equipment. My task is to offer technically sound and realistic solutions, not to fuel purist attitudes that in many cases clash with the real availability of components.

It is worth remembering that the original transformers have not been in production for decades, and those that can be found usually come from other dismantled units, which in turn end up irreversibly mutilated. Defending originality at all costs, in these cases, simply means shifting the problem from one unit to another.

Those who truly care about music, not just collecting for its own sake, know very well that a proper electrical restoration is more important than fetishistic formal integrity. In this case, the sonic result is far from a compromise. The replacement transformers deliver high level instrumental and tonal performance, making the chosen solution entirely reasonable.

For the sake of completeness, the original faulty transformers were not discarded but preserved for possible future philological reconstructions, for those who may truly wish to invest time and resources in that direction.

This particular MQ3600 had clearly passed through the hands of one of those “creative repairers” who, despite limited competence, always manage to leave a mark, unfortunately an irreversible one. It is therefore not surprising that it was purchased by weight as scrap metal, given that its condition was closer to junk than to a vintage audio component.

The situation was bleak: carbonized sockets, blackened resistors, cheap PCB-mounted trimmers, burned wiring, and transformer leads as thick as a finger, all generously seasoned with oil and grease everywhere. The first operation was unavoidable: complete removal of the transformers and thorough cleaning of the chassis, to restore at least a minimum level of dignity before beginning the actual reconstruction.

I then proceeded to remove the burned sockets and the PCB-mounted trimmers, evidently installed by someone who believed they could play an adequate role in a context like this. The overall impression was of a unit that had undergone more than one repair attempt, each unfortunately worse than the previous one, until, presumably, the limits of technical capability were reached and the work was abandoned.

Once chassis cleaning was complete, I tested the transformers. To my surprise, the power transformer and filter choke were found to be perfectly intact, despite the overall condition of the amplifier. One of the two output transformers, however, was unfortunately shorted.

In agreement with the customer, a pragmatic approach was chosen, opting for replacement rather than rewinding, in order to keep costs under control while ensuring high sound quality. I therefore designed an output transformer with a 3600 ohm primary, sized for push-pull operation with KT88 connected as triodes. It is worth remembering that the original 8045G tubes used by Luxman were true triodes, not pentodes connected in ultralinear mode. For this reason, the restoration was carried out following the original circuit concept and maintaining design consistency.

Restoration with original NEC 8045G tubes, when possible, it is done

On a rather rare occasion, a customer brought me two Luxman MQ3600 units, expressing the intention to restore one of them using its original NEC 8045G output tubes, while the other would undergo the classic KT88 conversion. The customer owned a fair number of used 8045G tubes, both those installed in the amplifiers and others previously removed and kept in storage. He therefore asked me to test them all to evaluate the possibility of selecting a usable matched quartet.

I carried out a complete analysis of all available tubes using the uTracer, plotting the characteristic curves of each one. At the end of the selection process, I identified two very well matched pairs, effectively an almost perfect quartet, sufficient to bring one of the two amplifiers back into operation while preserving its original configuration.

The first of the two MQ3600 units was, as often happens, already tampered with in the past through questionable interventions. I performed the usual complete overhaul, starting with verification of all signal tubes, which in this case were found to be fully functional. I then proceeded with the installation of the selected 8045G tubes, completing the bias and balance adjustments.

As with all units of that era, the power transformer was rated for 220V nominal. When powered from today’s 230V mains, difficulty arises in correctly setting the bias, because voltages end up higher than in the original design. For this reason, it is always necessary to perform a small modification to the bias circuit, in order to bring the adjustment range back into a safe and usable window.

The usual flat Nichicon capacitors, almost always degraded, were also replaced, and polypropylene bypass capacitors were added on the supply rails of the various stages to optimize stability and frequency response.

Once the overhaul was completed, I carried out the instrumental measurements, which yielded results surprisingly above average, especially considering that the amplifier remained in its original configuration with 8045G tubes. The delivered power was clearly higher than that of many MQ3600 units converted to KT88 in ultralinear mode, with the added benefit of a higher damping factor and significantly lower harmonic distortion. This provides further confirmation of the intrinsic value of this design when restored to proper operating conditions.

Measured performance with original NEC 8045G tubes

After the complete overhaul and careful adjustment of the bias circuit, I performed the usual instrumental measurements on the unit equipped with original NEC 8045G output tubes. The bias was set to 75mA per tube, corresponding to an anode dissipation of approximately 38W, well below the declared maximum limit of 45W for this tube type, ensuring reliability and long service life.

The results obtained were decidedly positive:

• Output power: 72W RMS per channel
• Damping factor (DF): 16
• Bandwidth (original transformers): from 15Hz to 40kHz at –1dB
• Harmonic distortion (THD) at 1W: 0.14%

These data confirm the high potential of the MQ3600 even in its original configuration, provided that it is carefully overhauled and properly adjusted. The low distortion at low power levels and the very wide bandwidth testify to the excellent quality of the original transformers and the soundness of the overall design.

THD

Bandwidth

Square waves at 100Hz, 1Khz, 10Khz

Do you have an MQ-3600, or another tube amplifier, that needs servicing?

If you own a Luxman MQ-3600 or another tube amplifier in need of serious and competent servicing, I can help you restore it to original performance, or even beyond. Whether it is a simple maintenance job, a conversion, or a deep intervention like the one described in this article, every project is approached with craftsmanship, professional instrumentation, and close attention to sonic results. You can contact me at any time through the contact form on the website, describing the model and condition of your equipment. I will get back to you as soon as possible to evaluate the best solution together.