25 W Class
Amplifier (a la JLH)
In 1969 John Linsley Hood published a small Class A amplifier. The amplifier gained a good reputation for its performance despite the limited output power. The original article can befound here. The schematic is shown below.
I will not go into detail here, but Nostalgia is configured in the same manner. However, there are two main differences. Nostalgia is using only FET transistors, and it is also using a split supply. The latter avoids the output capacitor and the FET input makes it possible to avoid the capacitor on the input as well. The FET part is inspired by Nelson Pass, who in 2005 had a new look at the original JLH amplifier shown above. He removed the input stage and replaced the other transistors with MOSFETs, resultating in the PLH amplifier, shown below. The article can be found here. Here Nelson Pass has introduced a potentiometer (P2) that together with the bootstrap capacitor (C2) can vary the amplifier from operating in push-pull to single ended. In the article you can also study the distortion figures for the various potentiometer settings and the Nelson Pass preferences.
In the Nostalgia amplifier I have not used a potentiometer, but a switch to change between aSingle ended and a Push-pull configuration. The schematic of Nostalgia is shown below.
The input stage
The input stage is shown in the figure below. I am using the P-channel JFET J175 from Fairchild as the amplifying device. The IDSS-variation is quite large. I have used transistors with an IDSS of about 20 mA.
From the characteristic below you can see that our 1.8 kohm resistor (R14) will give about ID =1.5 mA of drain current for VGS = 2.7 V as the quiescent point for J5 (assuming that the output is at 0 V DC), since ID = VGS/R14. The potentiometer P6 is used to set the working point for M18 (and M11). The gain in the first stage is given by the transconductance of the JFET and the resistors R4 and P6. The transconductance as a function of the drain current is shown in the figure at left. The gain is given by
Push-pull/Single ended operation
The phase splitter is M11, see the schematic below.
The signal at the source of this MOSFET is in phase with the signal on the gate, and the signal at the drain of this MOSFET is phase inverted. This means that the power MOSFETs M16 and M18 will sink and source the output current. The coupling of C20 at the upper end of R9 implies that the two transistors are operated in push pull mode, i.e. when the current in M16 increases, the current in M18 decreases, and vice versa. However, when C20 is connected to the lower end of R9, M16 is almost coupled as a current generator and M18 must deliver most of the current variation, we can say that the output is operated single ended. The gain is lower for the single ended case. Neglecting the effect of C20 and R10, this is about:
Here RLoad is the speaker load (assumed 8 ohms here) and gm is the transconductance of M18 (about 5 S at an idle current of 1.5 A). In fact the gain A2 is somewhat higher because of the bootstrapping (C20 and R10): A2 ≈ 23 dB. When operated in push-pull, the gain is about 3 dB higher: A2 ≈ 26 dB. The resulting open loop gain is thus A ≈ 44 dB in Single ended mode and A≈ 47 dB in Push-pull mode. The open loop bandwidth is about 40 kHz and 30 kHz for Single ended and Push-pull, respectively. The resulting closed loop gain with R4 = 120 ohm is G ≈ 23dB with a bandwidth extending 700 kHz.
The distortion is higher in Single ended operation with more second harmonic distortion, while in Push pull the distortion is lower, while the amplifier now can deliver more current to the load. The output impedance is also higher for the Single ended mode than for Push pull. The operation between the Push pull and Single ended operation is done with a switch on the amplifier back plane (In the PLH amplifier this operation is done with the potentiometer P2).
I started out with an ordinary unregulated power supply with additional RC-filtering, see theschematic below.
However, this was not sufficient for the negative leg (M11 and M18 is operating in common gate for signals on the negative leg), causing a ripple at the amplifier output that was annoying. So one more RC section was added at the negative rail with another one ohm resistor and 22000μf capacitor. A better solution would probably be to use a capacitance multiplier instead of all the RC-filtering. I have used a 2x18 V 500 VA transformer (TR in the figure above) common for the two channels. However the rectifier and filter capacitors are separate for the two channels. The voltage is about 24 V on the positive leg and about 23 V on the negative leg. This means that the amplifier can deliver a maximum of about 25 W to an 8 ohms load. The rectifier should withstand 35 A at least. The power resistors used are 5 W wirewounds types. The filter capacitors should be rated to more than 25 V.
The printed circuit board (PCB) measures about 99x52 mm, see the component placement here. The layout is shown here.
With the exception of the power resistors, 0.6 W metal filmoxide
resistors with 1 % tolerance is used. The power resistors R17 and R19
are ordinary wirewound 3 W types. If you want to change the gain,
increase R4 to reduce the gain, and vice versa. The output transistors
used for M16 and M18 is IRFP240 from International Rectifier (IR). If
you are looking for a replacement for these, be sure they are vertical
MOSFETs. These power transistors are mounted directly to the heat sink.
The phase splitter is IRF 610, that also come from IR. This is also a
vertical MOSFET, but this transistor is not mounted on the heatsink. I
have not tried any substitution for this transistor. I have mounted the
PCB directly to the heatsink.
The outputs from the transformer are carried to the rectifiers as shown in the power supplys chematics. I have used the principle of star ground here; that is, I use a common connection between the filter capacitors (C4, C5, C8, C9) to the mid point from the transformer. This point is then connected to a star point where the loudspeaker minus also is connected. The GNDS terminal on the PCB is also connected to this star point. The same is the GND terminal on the PCB. I have also connected the GNDF terminal on the PCB to the GNDS terminal on the PCB itself. The signal ground (SG) on the PCB with the Input is connected to the phono socket on the back of the amplifier with a screened cable. I have used a chassis connection from the starground. From the loudspeaker plus output I have used a single leader to the OUT on the PCB. You may try other grounding schemes, but if some sort of instability or noise should occur, the probability is high that the reason is bad wiring (e.g. earth loops).
I used a variable DC voltage generator first time the amplifier was started up. When the power supply voltage was increased, both the output offset voltage and the idle current was adjusted by means of the potentiometer P6 and P23. The idle current was set to about 1.6 A. It is advisable to run this amplifier with a high idle current, the higher the better. However the heatsinks must be large, and heat sinks with a minimum capacity of 0.35K/W is recommended. The idle current and offset was readjusted when the ordinary power supply was connected. The quiescent current is quite stable when the heat sinks reach their working temperature.
|Some measuring results
|Bandwidth (Small Signal)
|THD at 10 V RMS 1 kHz
||2H: -49, 3H: -58, 4H: -70, 5H: -79 dB
||2H: -67, 3H: -68 dB
Knut Harald Nygaard