Current source rectifier confusion

The LM334 is not only an adjustable current source, but it is also a rectifier.

Right on the front page of the TI data sheet are the words:

“Reverse applied voltages of up to 20V will draw only a few dozen microamperes of current, allowing the devices to act as both a rectifier and current source in AC applications.”

Which is absolutely brilliant if you are trying to save a diode drop as you can replace two of the diodes in a bridge rectifier with LM334s and you have a current-regulated output bridge rectifier.

Particularly as the minimum operating voltage when delivering 1 – 5mA is 1.0V – barely more than a 1N4001.

Perfect for low-current use 🙂

But what if you have not got any LM334s, or you need more that 20V or blocking?

Can the classic discrete one transistor constant(ish) current source do something similar?
– see the top diagram, where the same circuit is connected both ways around to an ac voltage source. The 50Ω resistor sets the current and the 500Ω resistor is the load (load current left).

So I modelled it in LTspice and it seems to work perfectly.

A bit too perfectly it seems to me, as the graph is showing no reverse current in bases (not shown in the plot) and I am convinced the reverse-connected base-emitter junction should conducting when the circuit is reverse biased.

Sadly, I do not understood Spice at all, so simply examining the model parameters is beyond me without more education.

So my first questions to more knowledgeable folk is:

Are there transistors in the LYspice library that have V_EBO modelled, and some that do not?

Or is is modelled here and I am just not understanding?

I did try modelling this circuit (right) using the LTspice model of the ZTX1048A – which is one of Diodes’ (was Zetex) ‘matrix’ transistors that retain good gain and saturation when operated with collector and emitter swapped, making this a good test of whether LTspice can model the reverse characteristic.

The voltage source varies between 0V and 20V, and there are three constant current circuits (from the left):

The normal constant current circuit, one with the same transistor in the reversed-collector-emitter configuration, and then the right one with the first circuit completely reverse-biased.

The 500Ω load currents for the left hand two are shown here (left) in blue and red (applied voltage in yellow), and both constant(ish) current sources are behaving well (conventional transistor connection in red). Trying the same plot with more conventional npns (no diagram) indicated awful performance in the collector-emitter swapped configuration.

This leads me to believe that LTspice is modelling reverse-connected transistors well (at least for the ZTX1048A and the other one I tried).

The right hand all-reversed connection revealed this base current plot (purple, right), which has no reversed emitter-base breakdown that I can see, just some leakage (so something is being modelled, and I realise this is not V_EBO as the collector is connected!).