This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM4875 Audio power amp, spurious oscillations/noise

Other Parts Discussed in Thread: LM4875

I am using one of these chips to drive some small speakers in an FM audio receiver.

I notice a burst of oscillation at the zero crossing points when I drive the LM4875 with a sine wave. I can hear the bursts as a roughness in the audio. The real problem comes when the receiver is quietened, carrier up but with no modulation and the zero crossings from noise cause bursts of oscillations with a correspondingly poor SNR. When there is no audio input, I can see a 12 MHz waveform about 200mVpp across the speakers.

I have the vol control at max, 3.6V supply, 68u and 100n decouplers right up on the chip.

If I replace the speakers with an 8Ohm pot, same problem. Change the supply voltage, freq of oscillations change a bit (about 5 MHz).

Speakers are about 35 Ohms each. Happens at all drive levels but sounds worse at low volumes.

Data sheet does not suggest it but should I terminate the speaker output at HFs with a cap and low value resistor in parallel with the LS? 

Help appreciated, this product is ramping up for production.

  • Hi Julian,

    Can you share your schematic for us to review? I've sent you a friend request so that you can send me the schematic in private.

    Do you have the scope capture of the waveforms?

    reg,

    Paul Chen

    Applications Engineer

    Dallas TX USA

     

  • Hello Paul, thanks for your reply. 3480.TI audio.doc4034.66 MHz Rx V6.pdf

    I hope you can read these, looks like rubbish on this screen.

  • Just to be clear, the 1uH series inductors and 22p shunt caps on the speaker leads are to keep the RF out of the audio stages. I removed them and it makes no difference.

  • Here's a video of the scope as i wind the amplitude up and down. I checked the input to the amp, definitely no spikes here. Sorry wont let me upload, basically spikes or bursts get worse as the audio level DECREASES, then breaks into continuous oscillation when no audio. Tried feeding chip with an external audio source (so its not been driven by the receiver) problem still there.

    Also noticed that when I take the 22pF caps off, it gets better marginally. The speakers are connected by 1 ft lengths of twin screened lead, perhaps it doesnt like driving capacitive loads, if not what can i do about it. Cant afford to put any series R with the speakers as we are pushed to get the volume as it is.

    Julian.

  • Hi Julian,

    Removing the 22pF cap improves but you also mention that when both 22pF and the L are removed it made no difference. Does it imply that removing only 22pF cap will decrease the ringing marginally? I wonder if there is some glitch at zero crossing that hit resonance freq of the LC that causes the ringing. How about the Vdd supply, does it look clean during operation?

    Typically op amps will run into stability issues when driving capacitive loads. Does the ringing get better when shorter cables are used? Is it possible to experiment removing the 22pF cap, replace the L with 10R resistor and drive a 32R load to see if the capacitive loading is causing the amp to oscillate?

    You mention that it breaks into oscillation when there is no audio. Is the input clean without any high freq noise?

    Does it happen to the EVM driving the same cables and load?

    reg,

    Paul.

     

  • Thanks Paul,

    I tried today with both the L and C removed, the problem is still there, always now with an ext sig gen. The ringing is more pronounced on one side of the speaker than the other. I am using a single cell LiPo battery at 3.3 to 4.2V. When I run off an external power source and increase the voltage to above about 4 Vdc, the spikes pretty well go away.

    I tried 4R7s in place of the inductors, didnt make any difference really.

    What I notice now, using the sig generator, I cant really hear the spikes, but I can see them on the scope. However, when I use my receiver, i think it must be picking up a harmonic of the oscillation and demodulating it, after all, one would not expect to hear a burst of 5 MHz on top of a 400 Hz sine wave?.

    I can see 100mV of ripple on the power supply, I think its the internal resistance of the LiPo when its down on charge a bit. The 68u cap isnt really big enough but we havent got the space for anything much bigger in a tant smt.

    There is definitely a stability problem, I can touch the speaker leads and get 5 MHz ripple, move them or turn the batt volts up, and the RF goes away.

    I have some amp boards made up somewhere, with just the power amp chip on, I'll think I'll revisit these, as I dont remember it happening when I first tried out the chip to see if it was suitable.

  • Actually I missed one of your questions, there is a fair bit of 455kHz ripple on top of the audio and when there is no speech, comes out of the Philips FM Rx chip.

  • Hi everybody. I don't want to create next thread with the same problem:

    I'm trying to use this chip to my dsb modulator (because has dc gain regulation and bridge mode). Fortunately I started tests first :)
    I have to confirm this problem. This chip can oscillate when DC Gain voltage is close to 1V (so it's still not in mute or standby mode).
    My application is quite typical, on big copper plate in manhatan style.
    drive.google.com/open
    Frequency of oscillations (it around few MHz!) depending on headphones, cable lenght, my hands are closer or not, ect. No matter what is on input - can be shorted to ground (by capacitor of course). As I checked later, the same problem is with small speaker (~10 Ohms resistance) on short cable. I was trying to add VCC blocking capacitors, ect, but only the red marked components can help.
    As I checked the bandwith of internal amplifiers is very very wide, I tested this chip with my frequency generator (DDS, sine) and the gain is still significant to 1MHz at least. It's probably too much and because of that, with some circumstances (even not inductive load) can oscillate, what is bad, of course.