The "thump" is caused by a saturation, when the current starts at infavorable phase. The magnetic flux is an integrate of the voltage across the coil, so certain AC voltage forces it to vary with a given peak-peak value. And because it is an integral, it has its peak at voltage zero crossing. Because tge flux needs a current for it to form, when the circuit is open, there is zero flux. Normally the flux density is symmetrical around zero, so only half of its peak-peak span is well below saturation. But when you complete the circuit e.g. at voltage zero cross, the flux is zero at the time it should have its maximum (e.g. negative). When it is forced to swing its peak-peak level from that point till the next voltage zero cross, the whole peak-peak swing attempts to become the flux in the core. Because normally it is half of it, it means double flux than normally. Because these magnetic components use to be designed to really work on the edge of saturation (to have minimum turns, so shortest wire, so the lowest resistive losses possible), the full peak-peak leads to core saturation. And that leads to excessive field around the ballast (causing vibration to nearby steel components) plus tge magnetostriction movements within the core itself. Both lead to that noise. After few cycless the losses even the things out and "move" the flux swing symmetrically around zero, so there is no such saturation anymore.
These effects are normal and should cause any problems whatsoever. It does cause current spike, but it is for so short time the thermal inertia has no issue absorbing the related peak losdes without even any observable impact at all. The only problem (mainly with big transformers) is the high current spike when the core saturates falsely tripping the overcurrent protections, hence tge use of various soft start (starting at lower voltage, so the peak-peak "fits" within the saturation) or starting phasing methods (timing the connection so it happens at the point where the flux would be crossing zero).
The difference between your two cases could be just slightly different timing of the starting components, or with the "loud" ballast tgere is something either lose or not that rigid, so the external flux pulse makes it to vibrate more. There only the "something lose" could be a problem, not because the noise, but because being lose in the first place. With lighting ballasts it is just taken into account for the circuit protection design (using slow blow fuse/breaker, rated for double the arc current in case of a series reactor)
|