IRs2092 oscillating, any suggestions ?

Just got my latest revision of pcb through for my 2092 class d amplifier.

The previous version worked fine and ran off +/- 35 volts.

For the new version I have upped the voltage to +/- 45 volts and scaled the VB charge resistor accordingly. Also increased a bit the vaa and vss supply resistors, scaled them to match the 10mA supply requirement.

For some reason the new version has a low level 1KHz-ish oscilation. The audio plays fine it just has this extra tone on top of it. The tone is fairly low level, with oscillation i usually expect a high level but it isnt.

Clearly I am 99% of the way there with the chip running.
Just need a little help with this last hurdle.

Other than the mods mentioned above the components used are as per the datasheet. This new pcb layout is much tighter than the previous one with plenty of decoupling. Seems odd the older slack layout worked and the new tighter one doesnt.

Tried different inductors. Tried extra decoupling. A scope on some of the 2092 pins makes it better or worse.
 
It may be an interference. Does the audible freq. change if you modify switching freq.? Is there any other high freq. source nearby (other channel, PC, SMPS, unintentionally oscillating linear power supply or preamp)?

There does seem to be a lot of noise around especially on the supplies.
Despite having very thick tracks the HF seems to get on the ground too.

If I put a scope probe on VAA it cancels out the tone.
Will look into it a bit more tonight.
 
Had a look tonight and the problem is lack of decoupling on the VCC rail.
Added another 47uF elect. and now the oscillation on the output has totally gone.

Oddly that is the same capacitor I had problems with on my first design taken from the datasheet. The datasheet says it should be 10uF, so I changed it to 47uF to stop glitches on the VCC rail.

I can only guess the cheap 12 volt regulator is not really doing its job very well despite having decoupling on the 12 volts out.


[IMGHTTPDEAD]http://www.ckp-railways.talktalk.net/100UF.jpg[/IMGHTTPDEAD]
 
Last edited:
top as in capacitor for opening top fet, goes from vcc(there is only one vcc) to diode and then to that cap... you don't have it on picture
So, 470uF from VCC to COM, and 47uF from VB to VS

You have to be careful with the upper voltage capacitor as it is charged from the VCC rail when VS goes low.
So the VCC capacitor has to be bigger to stop glitches on the vcc rail.

I guess what was happening with mine was the vcc undervoltage check was just on the border and causing a high frequency clicking on the output.
 
I can only guess the cheap 12 volt regulator is not really doing its job very well despite having decoupling on the 12 volts out.

I quess this also. You could have measure it before replacing the capacitor. Different kind of regulator needs different kind of output capacitor to be stable. Some needs electrolyte cap, some needs low ESR chip capacitor very close to the output.