0
\$\begingroup\$

I'm using an STM32F746G for exploring embedded audio DSP. My experiments sound great in headphones but when I connect the audio output to external amplifiers, the signal level is low and there's a lot of whine to don't hear in phones.

I'm using the same output jack (CN10). It's described in the manual both as a "line out" and a "headphone out." Reading about headphone levels it seems like they should be decent for line level although I thought I might need to pad a few dB.

Any tricks or tips to using that as a line out? Unique DAC config? Blocking caps?

Thanks.

\$\endgroup\$
  • \$\begingroup\$ Have you consulted the manual of that eval board st.com/content/ccc/resource/technical/document/user_manual/c6/… ? It says which audio codec is used. You should look that codec up and find its datasheet. \$\endgroup\$ – Marcus Müller Jun 5 '16 at 15:32
  • \$\begingroup\$ In my rich fantasy life I was going to fly above that and just use the BSP functions. In the ref manual I see cryptic mention of: "Each DAC channel output buffer can be enabled and disabled using the corresponding BOFFx bit in the DAC_CR register." Still looking for details... \$\endgroup\$ – Dennis Cronin Jun 5 '16 at 15:51
  • 1
    \$\begingroup\$ STM32F746G Discovery is the full name of the dev kit. Didn't realize there was ambiguity. Sorry. I see one reference in HAL code for setting the BOFFx of the DAC_CR but that routine never gets called. The chip is the Wolfson WM8994 apparently. \$\endgroup\$ – Dennis Cronin Jun 5 '16 at 16:02
  • \$\begingroup\$ No ambiguity; I was just overreading the fact that there was no F in the middle. \$\endgroup\$ – Marcus Müller Jun 5 '16 at 16:03
  • \$\begingroup\$ Is there any chance the headphone output is class D and the lack of an apparent signal there as a "line output" is due to lack of inductive/capacitive loading? I'm still wading through the datasheet for the DAC but it's pretty darn complex. Also, although it seems to match what I see in code, it doesn't, at a glance, match the reference manual for the STM32Fn eval boards. \$\endgroup\$ – Dennis Cronin Jun 5 '16 at 20:10
1
\$\begingroup\$

I think this just comes down to "no attention given to audio signal path" as there's a lot of apparent crosstalk between the input and output and I have to keep levels artificially low. Btw, no class D. Looked at signal with a scope. I don't think there's a panacea for this; just use another board if you're doing serious audio.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.