I am trying to track down the cause of an increasing problem with some IGT= =20 electronic DC ballasts (model UB-12415) for F15T8 fluros. There are an increasing number that are being returned from the field which= =20 display the following symptoms. At switch on they are slightly below full brilliance, after a variable=20 period of time (between 3 -15 minutes) they suddenly become super bright=20 before shutting down. Once the transistors are allowed to cool they can be= =20 restarted by re-applying power and the cycle begins all over again. Sometimes the TIP41C pair of transistors begin to smoke, but not always. It seems most likely that the transformer is the problem and probably the=20 secondary windings, and yet after measuring the inductance between a=20 working device and a failing one there are only a few uH difference. One=20 suspect suggested was shorted turns, I suspect possible saturation of the=20 transformer. I've looked at the output under load of a fluro tube on a scope and there=20 is a nasty dip on the rising edge of the sine wave near the peak level=20 which gets deeper as the fluro goes into over brightness. A working=20 inverter produces a near perfect wave and at a much higher voltage approx=20 800v (hard to tell as the scope is ancient and can only measure everything= =20 on the 50v range and the reading goes off the scale). Frequency doesn't=20 change at about 19kHz. These are simple devices - 24v @ 620mA in, uses two TIP41's with each=20 transistor attached to it's own centre tapped winding - in other words=20 there are two CT windings one for each transistor on the same transformer,= =20 two chokes in the power feed to each, one 27nF cap for LC tuning and one=20 100nF on the secondary side. These ballasts are over 5 years old and are in gaming machines run 15 hours= =20 a day, in very hot environments and regularly fail - normally the=20 transistors short out and the 100nF dies. Replacing these does not solve=20 the above problem. The company refuses to purchase new ones and expect these to be repaired=20 until literally the PCB falls to pieces due to fibre glass failure.=20 From an academic point of view does any one have any thoughts on what coul= d=20 be the most likely cause? Points to note - the transformers are not available in small quantities and= =20 the company will not deal direct with Korea. I have no choice in the=20 oscilloscopes at my disposal at work.=20 Colin -- cdb, 13/02/2011 =20 -- colin@btech-online.co.uk --=20 http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist .