I was asked to repair a Phywe laboratory voltage source that exhibited small, periodic oscillations of its output voltage. This unit is a very simple passive supply: an autotransformer feeding a heavy-duty iron-core transformer, whose secondary goes to a full-bridge rectifier and a fuse. There are no capacitors and no active components. The transformer secondary is also available directly as an AC output.
The effect can be seen in the oscilloscope trace I photographed:
it shows the voltage measured across a 1 Ω power resistor connected to the DC output of the supply. The 100 Hz full-wave rectified ripple is visible, but superimposed on it is a strange, slow, periodically occurring decrease of voltage and current. This envelope modulation occurs roughly every few seconds.
Importantly, this effect is not visible when the same 1 Ω load is driven from the AC output of the supply.
Initially, I suspected some kind of thermal effect in the built-in bridge rectifier. To test this, I connected an external bridge rectifier to the AC output, and to my surprise I observed a similar effect (although somewhat less pronounced).
Next, I thought the cause might be some kind of transformer thermal protection, so I connected the external bridge to an entirely different source (my variac feeding a separate transformer). Yet again, the same slow modulation appeared.
Finally, I tried rectifying the AC voltage using only a single discrete diode (half-wave), and in that case the effect completely disappeared. The output remained stable with no slow oscillation.
This pattern leads me to wonder whether the phenomenon could be caused by some asymmetry or interaction specific to a full-bridge rectifier, but I have found no examples of such behaviour documented anywhere.
Can someone explain the origin of this effect? Why would a full-bridge rectifier repeatedly cause a slow periodic collapse of the DC voltage envelope under load, even with different transformers and different bridges, while a single-diode rectifier does not?
