Post Go back to editing

My PCB with the MAX98357 has just come back and the volume is very low. I have a 100K res to ground on GAIN_SLOT and it is driving a 32ohm small footprint speaker. I am driving mono ADPCM in. Any ideas to increase the volume? Thanks!

Category: Hardware
Product Number: MAX98357A

My PCB with the MAX98357 has just come back and the volume is very low. I have a 100K res to ground on GAIN_SLOT and it is driving a 32ohm small footprint speaker. I am driving mono ADPCM in and this is driving to +/- 32767 full-scale. Any ideas to increase the volume? Thanks!

  • Since you're using 100kΩ from GAIN_SLOT to ground, that should be set the internal gain to +15dB.  Even with a 5V supply, that is more than enough gain to clip the output signal.  A few questions to help me narrow this down:

    • I'm not familiar with ADPCM, but I looked this up online and it might work.  We'd need to confirm the clocks comply with the timing diagrams listed in the datasheet.  A scope shot (triggered on LRCLK) would help me figure out if the data input properly conforms.
    • Next, which part variant are you using?  (MAX98357A or MAX98357B)
    • What is the connection on SD_MODE?
    • If you supply a full-scale input, what is the RMS output voltage?  This will help be figure out what gain you are getting out of the system.

    Mike

  • I did a bit more research on ADPCM which I was not familiar with.  What I see online is that it is referred to as Adaptive Delta PCM.  It still isn't completely clear to me if this properly conform to the PCM standard.  Please refer to the timing diagrams and specifications for our supported digital audio formats and confirm your ADPCM conforms to these standards.

  • Thanks Mike, sorry the ADPCM is a red herring - I am decoding that with a processor and outputting 16 bit sample values to the MAX device and this part is all tried and tested.  I am using an I2S library running on the SAMD21 processor to output the values and I am using the MAX98357A.   SD_MODE seems to need to be high (I get nothing when it is low).  I'll need to make some measurements on RMS output voltage, although if I have a 3.3v supply and provide full-scale input values of 32767 would that not just clip at 3.3?  Many thanks for your help.

  • The MAX98357 has a built-in DC blocker and a FIR filter. I could be misunderstanding, but it sounds like you may be putting in a constant value.  Tell me more about the full-scale signal you're delivering.  Is it a tone inside of the passband?

    0dBV = 1VRMS.

    Based on max gain of 15dB with 0dBFS input would result in 17.1dBV or 7.16VRMS.  That's more than enough to clip the output even while the maximum operating supply voltage of 5.5VDC.

    If you deliver a sin wave inside the pass-band and still don't see the expected output level, we'll want to start looking at your DAI clocks/data on a scope.

  • It's a 1khz sawtooth from +ve full-scale to -ve and then ramp back up to +ve...it sounds perfect, it's just not as loud as I'd like. 

  • Just back from Holiday shutdown and am ready to assist you with this.  Depending on the shape of the saw tooth, I think I know what is happening.  If the saw tooth has a vertical transition between +full scale and -full scale, that is effectively a high bandwidth transition, and the output would not properly track from one extreme to the other.  If you could use a triangle wave rather than a sawtooth, it would be more likely to keep up.  Even a square wave would be better as it would have more time for the transition to reach full-scale.

    Any chance you could take a scope shot of the output?  Since the class-D output is PWM, you would need a class-D filter to resolve this to hearing frequency range to get an accurate reading.  I'm curious what this looks like.

    Mike

  • Hi Mike, thanks for this.  Its tricky to get a scope on it, but I think I might have a solution to help with my volume problem - rather than power VDD from the 3.3v that my uprocessor is using, I can power it straight from the 3.6v of the LIPO battery I have in my system, so the 98357 can suck as much current as it wants and the voltage will be higher.  I plan to mod the board when time allows to try this out.

Before You Switch


Switching languages will make ADI Explorer unavailable. Resume your session by switching back to English and reopening ADI Explorer.