Mechanical noise in amplifier circuit

[hpt]

Hi all,

While testing my latest 3-stage design into a 9R dummy load, I noticed at high powers (50W plus) that the amplifier circuit itself is generating a sound which tracks the frequency as well as amplitude of the signal being handled. The sound is evident only for fairly high powers and is fairly low in volume. The sound is evident for a wide range of frequency, ie. hundreds up to thousands of Hz.

I wonder where this sound comes from and how to eliminate it? I suspect it must be from mechanical vibration in an inductive coil, but I do not have any inductors in my circuit - perhaps the wire-wound output resistors are the source of the noise? Strangely it did not appear to be coming from the wire-wound resistors in the dummy load - only from the amplifier PCB itself!

The toroidal transformer is quiet, and the amplifier performs well at all tests, ie. slew, overshoot, linearity, gain, freq. response, output impedance etc. stability seems fine with no oscillation.

Cheers,

HPT

[Alexontherocks]

Could it be a faulty cap?

Inviato dal mio GT-I5800 usando Tapatalk

[AndrewT]

I suspect semiconductor packages are vibrating in sympathy with the signal they are passing. I think the ClassAB pulses in the driver and/or output stage may vibrate more than the devices passing ClassA signals.

[hpt]

I certainly cannot imagine the mechanism by which a transistor could vibrate in sympathy with the signal, save perhaps by magnetic force against the heatsink? I would have suspected this force to be truly insignificant - even more so because naturally the heatsink is aluminium, not ferrous. The transistors are mounted against silicon rubber pads and bolted down tightly.

Is this a common phenomenon? A design flaw? What is the exact mechanism by which this occurs? Does this phenomena have a particular name? I cannot find any references to this occurance on the internet despite what I search; nor have I come across references in any of my literature.

Alex - I did suspect perhaps the caps in the Zobel networks, but I could not locate the noise to one or the other cap.

I could not locate the sound with my ears; however if it is generated by all output devices against the heatsink, then it would make sense that the sound source is difficult to locate beyond simply being in the vicinity of the amp board.

[DF96]

Probably a capacitor. Some dielectrics (especially ceramics) are piezoelectric.

[hpt]

Originally Posted by DF96

Probably a capacitor. Some dielectrics (especially ceramics) are piezoelectric.

A good case for MKTs or polys?

I have 0.047uF monolithic ceramics in the Zobel network, and 39pF ceramics for compensation.

I will O/C the Zobel networks and see if this changes anything!

(EDIT: I have found a reference which suggests the higher the dialectric constant, the greater is the piezoelectric effect for such materials.)

[DF96]

High value small size ceramics are best used for RF decoupling and kept well clear of audio signals. Low value ceramics are fine.

[hpt]

Thanks DF96, this seems to be the most likely source of the mechanical noise.

I suspect my Zobel caps are quite high as far as ceramic caps are concerned. I shall replace these with MKT if neccesary. First to remove Zobel and confirm it's reponsibility.

Thanks for all the replies.

[quasi]

Are the ouput devices TO3 by any chance?

[gmphadte]

Take a flexible tube and put it to your ear and find out which component does it.