Sziklai Pair and Quasi Complementary output stages
Sziklai versus Darlington
The Sziklai, like the Darlington is a very high current gain device.
After dabbling on a few pages about Darlington transistors, what about the pseudo Darlington, the compound feedback pair, the Super Transistor, the Sziklai stage. (in some of the images Sziklai is spelt Ziklai – apologies for this)
One needs to go back a few steps to understand the significance of the Sziklai design (after George Sziklai). Many years back the only transistors used in power amplifiers were Germanium types which had a few quirks, the most dodgy being thermal runaway.
Germanium and Silicon Bipolar Junction Transistors (BJT) have a corresponding 0.2V Vbe and 0.7V Vbe switch on voltage. Both transistor types, when heated have a drop in resistance between Collector and Emitter which in turn draws more current, which in turn means more power dissipation until the transistor Ic and Ptot is exceeded and the transistor is permanently damaged. In every case where a silicon or germanium transistor is used there should be considerable foresight given into the design to reduce the possibility of thermal runaway. A heatsink is almost always used in linear switchers and power circuits. But this is not enough. We’ll go there in a second.
Note: In a BJT the gain increases as the temperature rises. The Vbe drops at 2mV per degree Celsius. This needs to be meticulously controlled.
NPN and PNP designs – why NPN is more popular
We’ll stick to audio, RF and BJT switching circuits at this stage.
In the electronics world the objective is to have both NPN and PNP transistors evenly matched and switch at the highest possible frequency, this would then include radio transmission and reception.
- Drive voltages and therefore unnecessary power dissipation should be kept as low as possible.
- This is known as mobility and defines the speed at which an electron moves through a conductor drawn by an electric field. In semiconductors this refers to both hole and electron mobility known as carrier mobility. The average flow of electrons caused by this force field is known as drift velocity.
- Mobility calculation results in (cm/sec)/(V/cm) = cm²/(V.s)
- The greater the mobility the better the transistor specification.
At room temperature 26ºC or 80ºF or 300ºK the mobility for Silicon results in
- Si N type – 480 cm²/(V.s)
- Si P type – 1350 cm²/(V.s)
The PNP transistor which uses an N-type wafer (Base) sandwiched between two P types (Collector and Emitter) has higher mobility at the base than collector or emitter. The reciprocation is true of the NPN, the thicker wafers of N-type material having higher mobility than the P (almost three times as high as the P substrate).
Earlier manufacturing techniques made the PNP transistor more expensive than NPN. Those in the industry would recall the difference in prices between the 2N3055 (NPN) and 2N2955 (PNP)
Quasi Complentary designs
The first push pull transistor amplifier that I can recall were small Germanium PNP devices driven by a push pull transformer feeding the bases of each output transistor out of phase with each other. These may have been the AC or OC series transistors.
The first Silicon amplifiers used either NPN or PNP output stages with loudspeaker coupling through a capacitor. As these were single rail supplies the design often was fairly basic. These were almost all quasi complementary (same type output stages or either NPN or PNP).
The above shows the comparison between the Sziklai and Darlington transistor arrangements.
Below shows how they would be used:
The quasi complementary output stage above was very commonly used in the 70s due to the popularity and pricing of NPN Silicon transistors. As we now have three Vbe junctions to bias on we use three diodes and a variable resistor to set bias current. Bias currents on these designs were often higher than using the straight Sziklai pair as a complementary output stage.
Both the Sziklai and the Darlington circuits are used extensively in modern circuits but the Darlington is the more popular. There are different trains of thought here but many engineers prefer the Sziklai over the Darlington due to thermal issues. Power Darlingtons often come in a single package whilst Sziklai are discrete. Darlington transistors do need to have the bias servo transistor on the same heatsink as the driver and output transistors whilst Sziklai this is not a requirement.
As in any field we have purists and some are meticulous about which output topology is used, never combining Darlington with Sziklai or vice-versa. Darlington output stages are often referred to as being thermally unstable and because of less linearity are distortion prone.
Like the MOSFET and BJT debate a good design engineer would make their product sound as good as possible.
The amplifier drawing above uses a quasi-complementary output stage. Although listed along with bigger brother as being amongst the top twenty best audio amplifiers of all time the audio fraternity have managed to scrape some negative sentiment about the design. Although nearly 40 years old they are often seen as unstable and tend to self-destruct.
We added this to the article because of the quasi complementary Sziklai and Darlington output stages. In it’s time it was known for mammoth output power and lighting up home parties. Although we like to think this was not designed for pro-use the transformer and output transistors were adequately rated.
The one danger that owner’s faced was lack of speaker protection. Speaker fusing was advised and the spec was given in the owner’s manual.
Manuals are available on Hi-Fi Engine.
Next: Current Mirrors and Long Tailed Pairs (Part One)