Skip to content
September 13, 2019 / consort3

Darlington Part 5

The Darlington Part 5 is an even more cost effective amp than previous efforts. Please read parts 1-4.

A single switch-mode power supply powers a Darlington amplifier with a rail splitter. The rail splitter can be low power since only the op-amp driver current needs to be sourced or sunk. Adding an output capacitor enables this feature and acts as protection for the speaker coil if a fault develops.

The amp. has been designed around the Meanwell LRS150-48 which is a good combination of performance and price. Do not be tempted by cheaper clones. The voltage regulation solves the quiescent current variation issue, while the current limit protects the amp from overload. If you buy the PSU from TME, the price is about half the cost of a 150VA toroidal transformer.

The voltages are defined by bias chain R4-R7. The virtual ground point is defined at 24V by R1 and R2.

The TIP142 should be a Fairchild or ON device not the ST one.  I have simulated the rail splitter but relied on previous work on the amplifier.  The thermistors must be in thermal contact with the output device to avoid thermal runaway. The quiescent current is set to 35mA by monitoring the voltage at one of the emitter resistors and tweaking R22 downwards from 100k. The D2 can be a BZX84C12L rather than that indicated on the diagram. I have inadvertently omitted local decoupling on the 48V supply,  a 47uF 63V capacitor and a 0.1uF should be added. The other half of the NE£5532 could be used for a stereo pair. Update: R19 should be between Q4 base and the junction of R22 and R10/Q2C. Update 2: The junction of R14 and C7 should go to Q3,4 collectors not to U3 output.

Project name is Leitmotif2


I thought I would create a more cost effective version of the Darlington part 5 amplifier by using a Meanwell 36V 100W supply and using TIP122/7 darlingtons rather than the TIP142 type. I thought I would model the circuit in Ltspice since I found some models of the devices in the /examples/Educational /PAsystem/audioamp filechain. I chose the LT 1468-2 as a substitute for the NE5534 and the LT1057 as a substitute for the NE5532. The models are good enough to get an idea of what is going on.


Modelled crossover perturbation on portion of a 10kHz sine wave as R5 and R6 are varied from 100 to 240 ohms. Note that R5 and R6 are for modelling convenience. Note the value of C2 shown gives too much bandwidth for audio. A more appropriate value would be 15pF.


A real amplifier needs the networks R8,R9,R20 and R10,R11,R21 as shown in the upper diagram otherwise you will get thermal runaway. To confirm this you could run a Spice directive .temp 0 25 50 100 which moves the simulator from 0 to 100 degrees and look at the plot of R2 current. It gets rather too high at 100 degrees!


The green trace is at 0 degrees It shows Class B hence has crossover distortion. The blue trace is 25 degrees with overlap, Class AB, giving low crossover distortion. The red trace is 50 degrees showing a non switching class B and finally the turquoise trace at 100 degrees shows Class A on its way to thermal runaway if there is no temperature or power limiting. All with the same design just the temperature varying. Fortunately the simple thermistor bias circuit keeps the bias current stable if the temp varies.

I am grateful to Tiefbassuebertr of the Diyaudio forum who mentioned a paper by Rieder (Abacus’ amplifier designer) which said that the common emitter output used here gives better isolation to the rest of the circuit with a reactive load than the usual emitter follower. It also nearly swings the rails.

The amplifier has some similarities with the Blomley amp in that it splits the upper  and lower half drivers but using a different technique.

If you use the 2STW100/200 darlingtons and the LRS350-48 you could get some serious power.

There is an amp of this type posted on Diyaudio using paralleled 2SA1930 devices. The opamp is a AD825. The netlist runs on Microcap. I think you would need thermistors to avoid a thermal runaway situation but you might be OK in a domestic environment. The modelled amp has a bandwidth of 1.8MHz which shows how good the technique is. The schematic also shows how to parallel output devices, although the thermal compensation becomes more difficult.

I was unaware of this Jim Williams paper on boosted op-amps until I found it on Michael Chua’s excellent Ampslab website

I found this high voltage rail to rail boosted op-amp by Jung on Wayne Kirkwoods Pro audio design forum



The above is an authorised stereo version of the 36V amp hitherto presented as a simulation circuit. Note that the Darlington emitter resistors can be 4 one ohm resistors in parallel (hence the 0.25)

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: