Noise from sgtl5000

Not open for further replies.


Well-known member
I am making a plugin board for the teensy 3.1/3.2 with the sgtl5000 codec on it. Also have some pots and switches, OLED screen and so on. Will at some point be good for audio effects/stomp box.
But right now I am facing some troubles.
I get some strange noise from the output of the codec. It is not possible to attach mp3s here so I add a link to the working unit(standard teensy +audio shield) and the sound from my card.

The layout on the board is not perfect regarding ground. The datasheet recommend a star connection on the analog side of the codec. But as far as I can see, that has not been done on the audio shield either.
The sound is recorded from the line out.
The noise is somewhat following the activity on the teensy.. its not constant at least.
Anyone seen (or heard) similar things?
I found it...
For some reason it worked to put a 1Meg i parallell over the 100n cap on the VAG pin on the codec. Maybe because the grounding was not perfect. Don't know..'
But it you load it more ( <600k) it will not work!
Can anyone speak to the applicability of omjanger's mod of the Teensy Audio Board for the 4.0 version of the Audio Board?
Thank you.
This mod was on a board I made myself. And probably an issue of some layout stuff. Don't think it will help anything on the teensy audio board..
Thank you for the quick reply. However, I find your reply rather "odd", for the following reason:
I am following an example from a company called SparkFun, and they reference your noise modification as being applicable to the Teensy 3.2 Audio board.
I will now follow-up with SparkFun, to see what they might have to say on this matter.
Thanks again.
Yes, I have seen that instruction. I actually think they read this post and just put in the mod thinking it would benefit on the standard teensy audio board. I don't think it will. but hey, just ask=)
[For others that may be following this conversation,] I received a reply from SparkFun:
"Hello Paul,
I've thumbed through the post you mentioned. So his fix was for an adapter board he made. Not necessarily the one we sell. However, they use the same codec chip. His fix was meant for that codec chip on the same line that we mentioned. Doing this fix on the adapter board we sell should reproduce the same results he posts.
Unfortunately, I can't guarantee that you'll see exactly the same behaviors. This post was made 4 years ago."

My (current) conclusion:
I'll keep this mod "in my back pocket" as a possible solution should I run into "noise trouble" as I work on my project (programmable guitar effects pedal).

Thanks to the folks at SparkFun, and especially to omjanger.
you're welcome=) as guitar effects usually needs a huge dynamic range I have ended up with using the AK4558 codec. That has lower noise and works great. The SGTL5000 is a good chip, but is best suited for handling line level signals.
You obviously have much more experience with this topic than I do. I'm an Electrical Engineer, but most of my experience is related to radio signal processing.
I am curious about your previous statements pertaining to "level" and "dynamic range", however.
I recognize that the signals from typical guitar pickups are rather small, and often require a lot of gain prior to any further processing.
[In the radio world we have similar issues, and often require our "front ends" to be "low noise".]
But, to my way of thinking, the "dynamic range" of the signal is independent of whether or not gain has been applied (assuming a linear amplifier, of course).
My intent is to insert some (hopefully low noise) amplification prior to applying the signal to the ADC/signal processor, taking it up to "line level" as/if necessary.
I need to read up on the differences between the AK4558 and the SGTL5000 ... perhaps that will clear my confusion.
Does the AK4558 support a larger digital word width (that could explain your use of the term "dynamic range")?
Thanks again.
P.S. Not only am I a beginner with Audio DSP, I'm also a beginner with music ... so my "technical performance requirements" are likely not as demanding as yours :)
Not open for further replies.