The Shindo Montrachet is probably one of the best-known amplifiers ever produced by Shindo Laboratory. Built around a push-pull EL34 output stage, it perfectly embodies Ken Shindo’s design philosophy: the absence of global negative feedback, extensive use of vintage components, and a sonic approach aimed more at character and tonal coloration than at laboratory-grade measurements.
As with much of Shindo’s production, the Montrachet was built using a large number of surplus components and design solutions that would be considered rather unusual today. This undoubtedly contributes to the amplifier’s charm and unique sonic personality, but it also introduces a number of peculiarities that are worth understanding when servicing these amplifiers.
The unit that arrived at SB-LAB exhibited a behavior that its owner had known for many years: a background hum that tended to appear under certain operating conditions and had always been regarded as one of the amplifier’s peculiar characteristics. Later, without any particular warning signs, the amplifier suddenly stopped working and became completely silent.
As always, the first step was an internal inspection of the amplifier. During the preliminary checks it immediately became apparent that a 500mA slow-blow fuse had been installed in series with the high-voltage supply. The correct value was not documented anywhere, and the only indication available consisted of a handwritten note apparently added at a later date.
Considering the presence of more than 330uF of total capacitance following the CLC filter, the 500mA rating appeared rather conservative. After verifying the absence of short circuits and carefully inspecting the entire power supply, I observed that the 500mA fuse tended to fail after two or three consecutive power cycles. For this reason it was replaced with an 800mA slow-blow fuse, a solution that proved completely stable throughout all subsequent testing.
Inspection of the capacitors did not reveal any significant issues. Only a pair of electrolytic capacitors used as cathode bypass capacitors for the EL34 output tubes showed characteristics that were no longer optimal. Since these components were not directly related to the main fault, they were replaced as part of routine preventive maintenance.
All tubes were then tested using a curve tracer. The EL34s, EF86s and ECL82 were still performing perfectly. The only anomaly involved an EL84 that exhibited an unusual heater short circuit. The tube was replaced with a tested and fully functional equivalent.
Since no official schematics were available, it was necessary to proceed by tracing the circuit directly. Fortunately, the overall architecture is reasonably easy to understand. The unbalanced input signal is applied to a pair of transformers used as phase splitters, generating the balanced signal required by the following stages. Two completely independent volume controls follow, one for each channel.
This arrangement may appear unusual from a modern perspective, but it fits perfectly within the philosophy of the amplifier. In a circuit characterized by relatively wide component tolerances and the absence of global negative feedback, achieving perfect channel matching is not always straightforward, if not impossible. The presence of separate controls allows the user to directly adjust the final balance according to personal preference and the music being reproduced.
Following the volume controls, a pair of EF86 tubes per channel directly drives the EL34 output stage. The four EL34s operate with automatic bias through separate cathode resistors.
Particularly interesting is the management of the auxiliary power supplies. An EL84 is used as a dropping element to generate the screen-grid voltage for the EL34s, while an ECL82 operates as a true voltage regulator complete with reference and feedback circuitry dedicated to supplying the EF86 stages.
It is worth noting that, in this specific application, the quality of the EL84 and ECL82 has an extremely limited influence on the overall sonic performance of the amplifier. For this reason there is no particular need to seek expensive NOS examples. It is sufficient that these tubes operate correctly and meet the electrical requirements of the circuit.
Regarding the hum issue, the investigation eventually focused on the rectifier bridge. As a first step, the diodes were replaced in order to rule out possible faults. Subsequent measurements revealed a significant improvement after introducing a damping resistor immediately after the bridge and before the first filter capacitor.
This is a solution that often proves beneficial in solid-state power supplies, particularly when large filter capacitances are involved. In this specific case, the final result was achieved using a 3.3ohm resistor together with 1nF capacitors connected across the bridge diodes.
The modification completely eliminated the noise generated by the power supply, although several characteristics remain that are part of the amplifier’s very identity. A certain level of background noise is inevitably associated with the extensive use of carbon composition resistors. Furthermore, the chassis is not connected to protective earth, which makes the amplifier somewhat sensitive to contact with the front-panel controls. During testing it was possible to observe that the position of the volume controls significantly affects the behavior of the circuit, with a greater tendency for noise pickup at the extremes of their rotation and generally more stable operation around their middle positions.
Examining the circuit, it is plausible that this behavior is related to the high value of the potentiometers connected directly to the EF86 control grids and to the absence of a ground reference for the control bodies themselves. Once the repair had been completed, the usual laboratory measurements were performed. These tests required additional checks and numerous confirmations to rule out any remaining faults, but all components proved fully functional and all tubes operated within correct parameters.
The technical data published by the manufacturer are extremely limited, and the various versions of the Montrachet differ considerably from one another. For this specific version, the only reference available indicated an output power of approximately 20W, a figure that measurements effectively confirmed. This version does not employ ultralinear operation. The screen grids of the four EL34s are connected together and supplied by the circuit built around the EL84, a configuration that causes the output tubes to operate as pure pentodes. The amplifier is therefore capable of reaching approximately 20W at maximum distortion.
The genuinely clean output power available before clearly audible distortion appears is approximately 6W RMS. The measured damping factor is around 0.5. From a frequency-response standpoint, this is a classic zero-feedback design.
The low-frequency roll-off does not appear to be an oversight but rather a deliberate design choice intended to achieve a specific tonal balance while avoiding an excessively dominant bass response. This characteristic may originate from the output transformers or from the relatively modest EL34 cathode bypass capacitors, rated at 22uF. The measured response is approximately 30Hz-30kHz at -3dB and roughly 80Hz-15kHz at -1dB. Despite this, during laboratory testing the Montrachet proved fully capable of driving the acoustic suspension loudspeakers normally used for evaluation.
The measured harmonic distortion at 1W is approximately 1.4%.
Naturally, some may point out that there are amplifiers capable of delivering substantially better measured performance. Zero-feedback amplifiers such as this one were never designed with the goal of maximizing objective accuracy. Instead, the particular coloration generated by the circuit forms an integral part of their sonic identity. For this reason, measurements should always be interpreted within the context of the design philosophy that guided the creation of the amplifier.
In the absence of the original schematics, the amplifier could only be evaluated through direct circuit tracing and extensive laboratory testing. At the conclusion of the repair, no anomalies were found that would suggest the amplifier was operating differently from its original design intent. All components were found to be functional and the measured performance, while far removed from what would be expected of a more modern design, appears entirely consistent with the design philosophy and distinctive sonic approach that characterize this model.