I included three stage boost charging routing in my low frequency inverter design, the transformer does not hum during the bulk and absorption stages, but at the beginning of float stage the transformer starts to hums with varying degree of intensity.
The charging is set to starts at 12Amps and continue until the VRLA battery voltage gets to 28.8V, then 28.8V is maintained until the charging current drops to 5Amps, at which charge voltage is reduced to 27.6V (Its at this point that humming starts). The hum continues until charging current drops to about 2Amps.
The transformer inductance is 14H and charging frequency is 16Khz.
The controller is PIC16F690. current limiting is done per pulse in ISR routine, while the ADC is read in main() function.
The hum does not disrupt any process, but I will like to understand why it occur, thanks in advance for your valuable contributions.
When the charger gets into low-current mode, it uses one of several possible mechanisms to reduce its operating current: operates in short bursts, uses PWM, or something similar. Humming means that you get some frequency component in the audible range. 16kHz is audible to some people, too, but short bursts of 16kHz might be more hearable - for example, the bursts occur at several hundred Hz or even slower. To reduce the noise, you could use some glue on the transformer, preventing its components from small movements produced by the alternate current. You can also alter the control algorithm in order to reduce the noise by moving it to less hearable frequency; but gluing the transformer parts together is a reasonable practice - also it will protect people with sensitive ears and pets, who can hear 16kHz pretty well.
Thank you Yuli Starodubtsev for taking time to answer my question.
I have tried the charging routine at 25kHz before coming down to 16Khz, the problem was that the microcontroller was not able to limit the current to 12Amps as wanted, probably due to the speed of the ADC.
I will try adding more glue to the transformer windings, am thinking of using Silicon gasket maker, as evo stik adhesive may not be suitable. Do you have any recommendation in type of glue that may be suitable.
Chinedu, I am not a glue expert, but I think it is safer to use something that you can remove, so I would not use epoxy. Even a hot melt glue will probably make a difference. Some sort of silicone should be good. Can be also a core vibration, if the core has more than one component in it (like E-core).
Alright, I will apply more glue to the suspected winding, if the vibrations continue, then the charging algorithm will be adjusted further.
I will update the findings subsequently.
Thank you once again.
Update:
The Transformer is made of E-I Core.
After application of coil varnish and injection of silicon gum to the coil, the vibration did not stop, so I dog deeper. It turned out that the vibration only occur when the transformer is mounted in the inverter casing.
The casing had a metal plate with about 3mm separation from the transformer core, during charging, for some reasons(still unknown to me), the plate vibrates in tune with the core magnetic flux in such a way that it produces loud sounds, but does not produce same volume of sound during inverting operation.
I monitored the charging pulses using a 4 channel Hantek PC oscilloscope and found that as Yuli suggested, the charging pulses changes from continuous stream to short burst of pulses when charging voltage drops from 28.8V to 27.6V.
As charging progresses at 27.6V, the frequency of the bursts reduces and so does the vibration sound.
By modifying the casing to remove the clearance between the base plate and the transformer core, the sound was eliminated completely.
In the cause of the investigation I used a different transformer with lower inductance to compare, I found that the bursts of pulses does not appear with lower primary inductance until the charging current has reduced considerably, and the vibration sound did not occur with this transformer.
While studying the difference between the second transformer and the first, I found that the orientation of the core relative to the mounting base affects the vibration of the base metal plate. Mounting the transformer with the center core parallel to the base reduces the the sound to almost inaudible level, but when the center of the core (Center leg of the E) is perpendicular to the base the sound becomes quite loud. I believe this is due to the level of rapid magnetization and demagnetization of the metal plate under the transformer.
I think the presence of sound is rather a detectability of the sound. the metal plate works as classical loudspeaker diaphragm, but either continuous stream frequency is too high for human ears, or the metal plate cannot vibrate at higher frequency due to its mass and other mechanical reasons. If you inject some audio signal into the feedback loop via large enough ceramic capacitor, you might actually hear the music :)