I think the type of detector that shows a transistor as the detector as in post 5 is such that the transistor's base-emitter junction is being used as the detector diode. However, since current is sourced from the collector at about a 100:1 higher ratio to the B-E current (depending on the transistor's hfe) it simply raises the input impedance and lowers the output impedance of the detector. This might or might not be helpful though, depending on the design & bandwidth of the final IF transformer. At least the circuit in post 5 makes some use of the current in the collector circuit and that looks clever. Some of the transistors appear short on emitter stabilization.
For the circuit described here in post 26:
http://www.edn.com/design/analog/444...n-simple-diode
The diode is not really the detector and I think the diode could be removed (shorted out), the detection is happening at the B-E junction of the transistor, it is essentially the same circuit.
The benefit conferred by the series diode though is that if the drive voltage gets over about 7V to 8V peak, the diode prevents the transistor's B-E junction from zenering. The transistor has no base tie resistor which is never a good idea as the collector-base leakage is very variable.
Also, the method they used to help small signal recovery is not ideal; forward biasing the diode and the transistor's B-E junction from the power supply via a 1M resistor.
The better way to bias a detector diode is to establish a voltage reference, using another semiconductor junction with the same voltage drop and temperature coefficient as the detector diode.
This was already done by Sony in 1956. They also had perfected their IF transformers so that the detector loading was correct with respect to bandwidth considerations. I have attached the circuit (note NPN germaniums)
If you study the circuit carefully (the return point for the volume control and final IF output coil) you will see that Sony chose the base-emitter junction voltage of one of the IF transistors ( X2 ) to be the voltage reference to correctly set up the detector diode's DC conditions with a forward bias getting it to the threshold of conduction for low level signals. Not only that the AGC voltage developed with signal reduces the base current in X2. They also have additional agc diode for high level signal control.
X2's B-E voltage is also relatively independent of the battery voltage and this B-E junction is practically identical to the detector diode so the temperature tracking matches perfectly.
The circuit basically turns Sony's detector into a near zero voltage drop detector for low level signals out of their IF helping to eliminate square law distortion, but it doesn't diminish its large signal handling capacity either. In other words, Sony totally aced the detector/agc circuit in 1956.