Maker Pro
Maker Pro

LM4876MX audio amplifier oscillation problem

HemyItay

Dec 4, 2022
2
Joined
Dec 4, 2022
Messages
2
I have oscillation problem (at ~5MHz!) with LM4876 "differential" audio amplifier. It seems to stem from stability issues at "zero crossing" level. One aspect of it is that driving it with high frequency input signal (few hundreds of KHz, removing the low pass feedback capacitor), the zero crossing distortion become very severe in addition to increased overall gain.
Any body any idea what that can be?
 

kellys_eye

Jun 25, 2010
5,296
Joined
Jun 25, 2010
Messages
5,296
What kind of construction method (layout) are you using? If trying to deliberately amplify signals in the 'few hundred kHz region you will need some pretty good 'screening' in terms of track layout etc.
 

Harald Kapp

Moderator
Moderator
Nov 17, 2011
12,884
Joined
Nov 17, 2011
Messages
12,884
driving it with high frequency input signal (few hundreds of KHz
The LM4876 is an audio amplifier, spec'd for 20 kHz, not hundreds of kHz.
removing the low pass feedback capacitor
Show us a schematic diagram of your circuit. There is no feedback capacitor in the application circuit.
1670236493362.png

Do you have the bypas capacitor in place?

Although the LM4876 is described as unity gain stable, construction and the resulting parasitci components (inductances of wires, capacitances of adjacent wiring) can make a huge difference and lead to oscillations.
 

HemyItay

Dec 4, 2022
2
Joined
Dec 4, 2022
Messages
2
The LM4876 is an audio amplifier, spec'd for 20 kHz, not hundreds of kHz.

Show us a schematic diagram of your circuit. There is no feedback capacitor in the application circuit.
View attachment 57194

Do you have the bypass capacitor in place?

Although the LM4876 is described as unity gain stable, construction and the resulting parasitic components (inductances of wires, capacitances of adjacent wiring) can make a huge difference and lead to oscillations.
Thanks for the response.
The bypass caps are in place.
I found the source of the severity of the oscillations. There were 1nF ESD caps to GND connected to each output pad (pads 5 and 8), leftover from previous design. W/o it the oscillation stopped but the signal was distorted at the "cross over idle" point. Further improvement was to place 500ohm resistor instead of these caps. This led to the addition of 5mA (2.5V/500ohm) quiescent "idle" current at each output pad. I assume the output stage is built of "totem pole" construction, leading to a sensitive "cross-over" working point conditions (too low "feed-through" current). With it the distortion were minimal are gone all together at the audio range frequencies.
 
Top