Single to Diff Output Amp with Single Supply

[mawyatt]

Here's another interesting bipolar transistor circuit that folks might find useful, inspired by:

https://www.eevblog.com/forum/beginners/simple-discrete-op-amp/

Often one needs to convert a Single Ended Signal to Differential with an active circuit, however this usually requires + and - supplies. The circuit shown below has a Class A/B Output that operates from a single + supply (the bias supply could be included with a different bias resistor).

The circuit operates with Q1 acting as a transconductor and current mirrored by Q2. Q3 and Q4 are Cascode devices and sink output currents thru R1 and R2 from the Vcc supply.

Note that Q2 inverts the current thru Q1 for small signal currents, for larger + input currents things become interesting. As the input current increases Q1's collector voltage rises and begins to turn off Q3 which has a fixed base voltage derived by bias diode connected transistors Q5 and Q6. This causes the current thru R1 to decrease and become ~zero for larger input + currents. Q2 continues to mirror Q1 and sinks current thru Cascode device Q4 to R2. So for large + input currents the output current is thru Q4.

When the input current goes - for large values, Q1 begins to cut off which also cuts off mirror transistor Q2. Now Cascode Q3 begins to supply the - input current and sinks this thru R1. So for large - input currents the output current is thru Q3.

This alternating +- current behavior produces a Class A/B like output current sinking capability, which theoretically is unlimited with ideal transistors. With small input currents both Q3 and Q4 conduct and this produces the Class A characteristic of the overall Class A/B type. The overall current gain is ~ unity and can achieve very wide bandwidths.

The circuit has many unique properties which are well beyond the scope of this introduction and we provided some notes for those curious. We've utilized variations of this circuit with fast SiGe devices that date back ~3 decades with origins from the late Barrie Gilbert in his MicroMixer.

Just a note on the LTspice schematic, V1 is not required for simulations, it's there to allow DC sweeps of the input to show details. Will discuss if needed.

Anyway, another interesting circuit with a few bipolar transistors that we hope a few folks will have fun with

Edit: Added DC input sweep by shorting across C1 and plotting Currents thru R1 and R2 vs Input Current thru Vin. For a much more in depth look see:

https://www.eevblog.com/forum/projects/interesting-amplifier-topology/

Best,

[magic]

Interesting idea, but your implementation has distinct crossover distortion even in simulation. While current transfer is exactly linear at least in theory, 50Ω signals are distorted and I believe this is due to mismatched and signal-varying input impedance of the circuit.

The old trick of adding 50Ω emitter resistors and biasing for r_e=50Ω (~520μA) should help, as it gives something closely approximating 50Ω termination regardless of signal magnitude or polarity.

edit
This is what I mean.

[mawyatt]

The crossover is only due to the weak bias, increasing Vbias (and/or reducing bias R3) increases the bias current in all the stages and "smooths out" the crossover. Also note the slight imbalance in the measured results, we didn't select matching devices.

One of the unknown "tricks" of this circuit is the 3rd harmonic "suck out" as shown in the hand notes. This is achieved by having the input waveform "see" an input impedance that allows the waveform instantaneous impedance modulation bounded within a range to create the desired properties. This gets very complex quickly and we had Post Docs and Grad Students working thru all the details and massive simulations required way back, and the SiGe BiCMOS chips we developed around this concept worked quite well indeed

However, with all this inherent complexity a simple version with general purpose transistors as shown can produce an interesting result and not overwhelm someone that wants to play around with it.

We posted some details of this circuit much earlier, but can't remember where

Edit: Found one reference here with lots more details and added to 1st post:

https://www.eevblog.com/forum/projects/interesting-amplifier-topology/

Best,