Transformers can blow in the blink of an eye

Recently my longtime trusted Eico 1030 regulated high voltage DC power supply succumbed to a catastrophic failure — a shorted transformer winding.  The power transformer has a separate winding that supplies screen voltage to the 6L6 pass tubes.  This is a half-wave circuit that was rectified by one original 1N4006 silicon diode.  Unfortunately, this old diode shorted and took the transformer winding with it, thereby ruining the transformer.

I use my HV power supplies on a daily basis for a number of tasks such as component testing (high voltage diodes), capacitor reforming, insulation testing, mocking up tube circuits, and many other purposes.  Fortunately, I kept a number of “backup” HV power supplies in case such a failure occurred.  (Sidenote: testing HV diodes on a DMM is a futile task.  The applied voltage is much too low to indicate whether the diode is shorted.  Many diodes will pass a DMM diode test but are, in fact, shorted when operated at the high voltages seen in the circuit.)

This transformer failure caused me to consider how unprotected most transformers are in vintage gear.  In many cases, a shorted cheap component (such as diode or filter capacitor) can ruin a very expensive or sentimental piece of equipment.  Typically, the only circuit protection that you see on most vintage gear is a line fuse, which does a poor job of protecting the transformer’s secondary windings, especially if the fuse is up-rated to accommodate inrush current at startup.

In rebuilding my backup Eico 1030, I took several steps that may help to protect it’s transformer.  I say “may” because fuses & circuit protection often do not blow on a consistent basis, so these steps should help but are not guaranteed.

I installed an ICL to limit inrush current at power-on.  I replaced the rectifier tubes with fresh silicon ultra-fast diodes because I wanted to reduce the heat inside the cabinet, promote stability, and reduce transformer load (2.26A saved by eliminating the rectifier tube heaters).

One of the cheapest circuit insurance you can buy is to place diodes in series.  If one diode shorts, you still have another one to continue proper operation of the circuit.  Likewise, series diodes sum their voltage ratings, so the circuit is more robust anyway.  For the price of a quality ultrafast diode from a reputable supplier — approx 20-cents each for a Fairchild UF4007 (1A, 1000v) — this protection is a bargain.  In each 5AR4 plate circuit, I used two UF4007 in series, or a total of four diodes in this circuit.  The bias supply (6X4) is a half-wave circuit, so I used two in series.  Likewise, I used two UF4007 in series in the screen supply, which was the winding that failed in my old power supply.

In addition to stacking the diodes, I installed fuses for each secondary winding. First, I fused the center-tap of the HV 440-0-440 winding with a 200ma fuse (200ma-250ma fuse should be reasonable in this circuit).  I fused the screen winding with a 62ma slo-blo fuse, which was the smallest that I had available.   I fused both 6.3vac front panel windings with 3A slo-blo fuses, which will accommodate the occasional overload and still remain within design parameters.

I have experimented with using PTC’s (resettable fuses) in each circuit, but PTC’s have three problems that make them less suitable for this type of circuit protection : (1) they do not trip with any operating current certainty (ex: a 150ma PTC may trip at 200ma or 600ma, you just have no way of knowing),  (2) they have a base resistance that may affect circuit operation, and the resistance increases as the PTC heats and approaches the upper range of its hold current design, (3) they trip slowly because the trip is based upon the PTC heating up.  Due to these characteristics, it would seem as though PTC’s would not have much practical use in protecting audio transformers.  In this power supply, though, I did leave a 40ma PTC in the screen supply, which helped to mitigate inrush current and possibly added some transformer protection.

Finally,  I decided to add protection for the current meter, which would always slam backwards if a connected load (such as a capacitor) discharged when the voltage switch was toggled off.  I replaced the SPST switch with a DPDT switch.  The first section of the DPDT switch replaced the original in the same manner.  I added a protection diode in series with the current meter to prevent current from reversing thru the current meter.  The second section of the DPDT switch is wired to dump the (+) terminal to ground through a 500 ohm high wattage wirewound resistor, which assures that any load attached to the power supply (such as a capacitor that was being reformed) is safely discharged as soon as the voltage switch is toggled off.