You could be right of course, but I used to commute on the GEML, (after the initial problems) and frequently on the 306s, - occasionally on the 307s. The main problem that I remember with the 306s was the repeated failures to transition from series to parallel with the clack clack of the contactors and the jerking of the motor cars. That fault would have been an issue when the were operating on DC.
Many of those journeys were on 306s from Chelmsford to Ilford or Shenfield to Chelmsford, thereby passing through the Mountnessing changeover point. I was on a Saturday up class 306 when we were delayed after Brentwood eventually crawling past a smouldering class 305 (No. 518) on the up fasts, - that would have been in the early '70s.
I'll just retract my original assertion about the converted units not blowing up their transformers, as I have now spent the afternoon reading the
final report into the various misadventures of these early EMUs.
It seems that there were five major transformer failures affecting these particular units shortly after conversion to AC operation caused by the automatic power control (APC) voltage changeover switch not operating and the 6.25kV AC circuits being exposed to 25kV AC. The first four resulted in explosions (the fourth also caused a fire) but the fifth was not so destructive only because the air-blast breaker (ABB) opened in time. The other four were so destructive that they set up a sustained short circuit to the rectifier frame which was only dealt with once the line breakers had operated, but by that time the transformers were toast.
1950s-60s rectifiers used the Mercury Arc process, with a tank of mercury. This should surely be apparent that large mercury-filled tanks, which overall in an emu weigh several tons, are more practical in lineside buildings (where they performed flawlessly) rather than underneath a rolling rail vehicle, all the while needing to maintain the critical gap from the electrodes to the mercury liquid surface. Sure, it ought to work ... and the moment solid state silicon rectifiers came along (1970?), they started replacement. Incidentally, when were the AC unit rectifiers replaced? And were they all?
As you say, mercury arc rectifiers were the state of the art at the time, but technology was moving quickly forwards. But the problems connected with their use in railway traction was not the amount of mercury required nor it's sloshing around, but how to maintain the correct temperature differentials between anode and cathode and what happened if you fail to keep them all excited when there were interruptions to the power supply. They were certainly finnicky things to set-up and operate and could easily cause faults which would affect reliability, particularly if you lost excitation.
For the Cl303s this was achieved through modification of the cooling arrangements and they proved to work tolerably well, but the GEC Cl305s were a harder nut to crack and the final report recommends that they be quickly changed to silicon rectifiers. The Cl302s which also had mercury arc rectifiers seemed to have worked fine straight out of the box with the only explosion being a capacitor divider on the APC equipment which blew the mounting plate into the guard's compartment below and injured the guard. However, this was found to be a one-off as no repeats occurred (although they did fit stronger mounting plates together with an explosion diaphragm).
Same generation but not the same complexity, as these units retained their DC control equipment so the transformer would only have had to cope with the dual supply voltage, without also having a tap changer to vary the secondary voltage going to the motors. However the transformer and rectifier were retrofitted, which is often a recipe for unreliability.
I've no idea whether the presence of the tap changer was a major contributor either to the explosion problems or to general unreliability.
No bearing at all. Tap-changers were damaged due to arcs inside the transformer, but they were not the cause.
Did all the early AC locos and units use Mercury Arc rectifiers? The 84s certainly did and were a disaster; the 83s had them and also spent a long period out of service, despite being built by the normally dependable English Electric.
The AM4s might have been a bit rubbish, but I don't recall them ever being unreliable, let alone blowing up.
It can't just have been the 25/6.25 changeovers that caused the problems, as I don't thing the 83s and 84s ever ran under anything but 25kV
AM4s, together with the AM8s and the converted AM6s and AM7s, never had mercury arc rectifiers. The AM4s, AM6s and AM7s were fitted with germanium rectifiers while the AM8s and subsequent EMU builds had silicon diode rectifiers. Although built for dual-voltage operation, all the electrification along the WCML through the Midlands and North-West was energised at 25kV AC, so the only time that AM4s ever operated on 6.25kV was when seconded to the Eastern Region to cover for the side-lined GEC AM5s.
The voltage changeovers were largely responsible for setting the environment in which these failures could occur and some were directly to blame. The design of the modified APC voltage changeover switch meant that it would only operate if the ABB was open (or the pantograph lowered), as it should do automatically at a neutral section. For various reasons this didn't always happen, and the five major transformer failures affecting the converted DC units I mentioned above were all directly attributable to this. All five of these units had just passed the changeover point at Shenfield and, for various reasons, the ABB failed to open, meaning that the APC voltage changeover switch didn't operate and they entered the 25kV AC section still set for 6.25kV AC. In this, almost every type of dual-voltage unit suffered a failure of some sort resulting in the 6.25kV AC circuits being exposed to four times the designed voltage, but not all suffered catastrophic damage.
The reason for the modification to the APC voltage changeover switch was that it could be hoodwinked into selecting the wrong setting by induced currents or by fault conditions in the OLE. The investigations in Glasgow with the AEI equipped Cl303s identified a number of incidents where this happened. Experiments showed that an unearthed but otherwise dead section of OLE next to an energised line at 25kV AC would have in it an induced voltage of 5kV which was well within the range that the APC voltage changeover switch would be looking for to identify a line at 6.25kV AC. Also if there was an OLE fault the falling voltage gradient from the end of the section towards the fault could cause the train to misidentify the supply voltage. In each case when the power was restored the train would be exposed to 25kV AC while still set for 6.25kV AC. Normally this would cause the Primary Overload Relay to trip and open the ABB, but by that time the transformer would have already been exposed to strong electro-magnetic forces which could cause shifting of the windings, abrasion of the insulation and even arcs caused by short circuits.
Admittedly some of the problems were not caused by the voltage changeover but by running on the 6.25kV AC system. Not being an electrical engineer I can't explain the reason, but these fault conditions caused by things like pan bounce, ABB "chopping" and the associated knock-on effect on the mercury arc rectifiers meant that the transient over-voltages caused were 4 times higher on 6.25kV AC than they would be on 25kV AC.