Introducing the 100W RMS Amplifier, a straightforward power amp meticulously crafted to offer a cost-effective and relatively uncomplicated assembly process. This amplifier boasts superior musical quality when compared to the standard STK module amplifiers commonly found in the majority of mass-produced stereo receivers available today. The genesis of this project emerged from my own need for a 100-watt per channel amplifier, driven by a desire to avoid substantial expenditure. Consequently, I fashioned the design around components readily accessible in my workshop.
While this design adheres to well-established principles, it is plausible that similar commercial counterparts exist. As far as my knowledge extends, it is not an exact replica of any existing commercial unit, and I am unaware of any patents associated with its underlying topology. To seasoned builders, I acknowledge that there is room for numerous enhancements and refinements. However, the primary objective was to maintain simplicity, ensuring that anyone with the capability to create a circuit board and the patience to execute a meticulous job could successfully undertake this project.
The input stage employs an LF351 op-amp, delivering the majority of the open-loop gain while also stabilizing the quiescent DC voltage. This stage is followed by a level shift stage that references the voltage swing to the negative rail. The transconductance stage utilizes a Darlington configuration to enhance high-frequency linearity. The 2SC2344, on its own, exhibits a relatively large collector-base capacitance that varies with voltage. To address this, the MPSA42 presents a low impedance and features a C(ob) of only a few picofarads, effectively countered by the 33pF pole-splitting capacitor. Power to this stage is supplied by the 2SA1011 active load (current source), approximately 20 mA in magnitude, with the current to the stage being constrained to around 70 mA at worst-case scenarios by the 2N3094.
The output stage consists of a full complementary Darlington configuration with paralleled outputs. While it is feasible to utilize a single output stage with 8-ohm, easy-to-drive loads, this is not recommended. The use of parallel devices enhances the amplifier’s capacity to drive reactive loads, augments the damping factor, and mitigates the maximum current demanded from each transistor during peak performance. It’s crucial to note that the gain of a power transistor decreases as current levels rise.
The compensation scheme involves two poles and one zero. The op-amp’s pole and the pole created by the 33pF capacitor, along with the 470-ohm bias resistor of the MPSA42, dominate the compensation. (The 33pF value gets multiplied by the stage gain.) Furthermore, a 22pF feedback capacitor introduces lead compensation, sourced from the output of the transconductance stage rather than directly from the output. This approach avoids introducing phase lag associated with the output transistors into the high-frequency feedback loop, resulting in a closed-loop pole that constrains the high-frequency response. Both compensation capacitors must be of type 1 ceramic (NPO) or silver mica, featuring a zero voltage coefficient.
The amplifier’s design is optimized for operation with a +/- 55-volt unregulated supply, which drops to approximately +/- 48 volts under full load conditions when two channels are driven. It relies on a 40-0-40-volt, 5-amp toroidal transformer, a bridge rectifier, and 10,000uF of filter capacitance per side. If a standard EI transformer is employed, a 6-amp rated unit should be used. With this power supply setup, the amplifier delivers a continuous output of 100 watts per channel when driving 8-ohm resistive loads without clipping. The dynamic headroom is approximately 1.5 dB. For increased headroom, unloaded voltages up to +/- 62 volts can be employed without requiring circuit modifications.
By the way, the schematic is in Postscript.
Without any modifications, this amplifier can efficiently drive 4-ohm speaker systems without engaging current limiting. The short-circuit current limit is calibrated to approximately 4.5 amps at its peak, adequately handling typical speaker loads. It’s worth noting that as the output voltage swing approaches the rail, it naturally generates higher peak currents.
However, if you plan to operate it with high-end speakers featuring impedance minima as low as half an ohm or those that maintain a reactive profile throughout most of the audio spectrum (e.g., 0.5 +j3.2 ohms), you might already possess a more suitable amplifier for such demanding applications. When employing the higher-power Motorola power transistors, this amplifier can effectively manage a 2-ohm resistive load without encountering significant issues, except for heat dissipation concerns.
In my experience, I have not detected any slew-induced distortion in this amplifier when subjected to a band-limited (22KHz) signal from a CD player. While it’s conceivable that fervent audiophiles could subject it to rigorous testing with a TTL square wave combined with a 19KHz stereo pilot tone and elevate the volume to the extreme, it’s essential to recognize that such tests may generate spurious signals throughout the frequency spectrum. Nevertheless, the question remains, who truly listens to such artifacts in normal listening conditions?
Possible Modifications: (What if I want mo’ power???)
The Toshiba output transistors, specifically the 2SD424/2SB554 pair, should not be employed in applications where supply voltages exceed +/-60 volts. If you intend to operate the amplifier at higher voltage levels, consider using additional transistors in parallel or opt for the 250-watt Motorola pairs (MJ15024/MJ15025). In scenarios where very low impedances are anticipated, it may be beneficial to raise the bias in the transconductance stage to provide enhanced base drive to the output Darlingtons, or you can introduce an additional current gain stage. It’s important to note that power transistors with higher Beta (and faster characteristics) tend to struggle with reactive loads. Therefore, avoid substituting high-fT components unless their second-breakdown capability is confirmed.
For the input stage, the NE5532 op-amp can be effectively utilized. In instances where multiple op-amps are employed, operating from the +/-15-volt shunt regulators for purposes like balanced inputs, anti-slew Bessel filters, etc., you may need to consider reducing the 2.7K dropping resistors to around 1.8K ohms to maintain proper regulation. Typically, the 2.7K resistors support up to 4 LF351 type op-amps when sourced from the regulator. As an example, I employed a quad 347 for balanced inputs in a DJ setup to mitigate hum-related issues.
The output transistors, along with the thermal compensator (2SC1567), should be affixed to a shared heat sink. An ideal choice is a finned heat sink measuring 5 inches in height, 8 inches in width, and featuring 1.25-inch fins, which should suffice for one channel. This arrangement can also enhance the aesthetics of the case if you incorporate them into the sides. For most typical applications, this level of cooling should prove adequate. The selection of the 2SC1567 for the output bias regulator was made due to its insulation capabilities. Note that the ECG version may necessitate additional mounting hardware. TO-3 hardware for the outputs is readily available and cost-effective.
The driver transistors and voltage amplifiers, such as the 2SC3344/2SA1011 pairs, also require heat sinking. Individual TO-220 heat sinks on the circuit board will serve the purpose, as the voltage amplifiers dissipate approximately 1.4 watts each. A single piece of 1/8-inch thick, 1-inch wide, and 4-inch long angle aluminum can suffice for all four transistors on each channel. However, it is crucial to orient them to facilitate natural convection, and ensure the transistors are insulated.
Maintain a clear separation between the input grounds and other components, and consolidate them at a single designated point. Neglecting to do so can lead to high distortion levels, often reaching around 5% or even causing oscillations.
To set the bias of the output stage, aim for approximately 25 milliamps in the output transistors. This value may take some time to stabilize, and it’s advisable to monitor it over an hour or so during the initial setup. To measure it, assess the voltage across the emitter resistor and apply Ohm’s law. This method enables you to check the current distribution among the parallel output transistors simultaneously and make adjustments if there’s a significant discrepancy. When using components from the same date code, the variation between them should not exceed 10% after the warm-up period. If you intend to employ a higher idle current, exceeding 50 milliamps per side, you should increase the value of the emitter resistors, but be meticulous in the setup process.
Never directly connect something like this to your power source! Even a seemingly minor mistake could lead to catastrophic consequences, including the risk of a house fire or instant damage to expensive transistors. While a variac might seem like a solution in theory, it’s important to note that the amplifier could potentially latch onto the rail if the supply voltage drops too low. Instead, I strongly recommend using a ballast resistor, specifically a 60 to 100 watt light bulb, placed in series with the AC mains.
Here’s how it works: When you initially power up the amplifier, you’ll observe a bright flash as the capacitors charge. Subsequently, the light dims significantly as the idle supply current stabilizes at its nominal low level. At this point, the amplifier should operate normally at lower volume levels. However, if the amplifier starts drawing excessive current for any reason, the lightbulb will brighten, increase its resistance, and thereby restrict the power supplied to the circuit.
Typically, if issues arise, they can be traced back to either a wiring error (which you can identify with a Digital Multimeter or DMM) or oscillation (which can be detected using an oscilloscope or an RF power measuring device). If you notice the bulb fluctuating between dim and bright repeatedly, it’s an indication that the amplifier is marginally stable, and you should thoroughly review the grounding layout. Additionally, you may need to make adjustments to the compensation capacitor values if you’ve made significant changes to the circuit. However, please note that in my experience, the amplifier remains stable in its current configuration.
The schematic is available in PostScript format for convenient printing. Transistor emitters are denoted by the label “e,” as arrows on the transistor symbols were omitted due to convenience, and this configuration has been effectively used for over a year.
If you encounter difficulty in sourcing components, you can explore MCM (1-800-543-4330), as they stock all the required transistors. For a stereo version of this amplifier, the total cost should fall within the range of $150 to $250, contingent on any cost-effective deals you might find for the case, transformer, and heatsinks. However, if you end up paying the full “list” price for all components, the total build cost is likely to approach $1000.
It’s important to acknowledge that the information provided here is offered as-is, without any warranties, whether express or implied. The author assumes no responsibility for the technical accuracy of the information provided herein or for any utilization or misutilization of said information.
Furthermore, it’s worth noting that the equipment detailed in this article was conceived, constructed, and tested during the author’s personal time and utilizing their own personal resources.