ESP32 Forum

Anyone try with 12mhz apll clock? How is jitter?

WiFive wrote:Anyone try with 12mhz apll clock? How is jitter?

Hi WiFive

1,2,8
12MHz
best wishes
rudi

Hmm I see.. and resulting jitter on i2s_bck ?

asap WiFive (not today) , i am on other place now.

do you see, allways the same buzzy waves, on pos and neg ( inverted )
edit: Like a pattern, watermark noise
edit: add a play on pic ( cut and paste the pos and neg )
with 32.5 MHz is comming a "compensation".

best wishes
rudi

Yes but we care about edges and how digital bck compares to jitter from bpll src https://esp32.com/viewtopic.php?f=17&t= ... t=20#p4727

That sounds bizarre. Have you had any luck with it? Maybe the community could take a look and give offer some assistance?

It took me 4 days, but I finally found it. Was a very stupid mistake in my network loop.
I'll commit my stuff to Github as soon as its at least a little stable, but for now Alexa thinks I'm not sending any audio for some reason. But at least she is accepting my authorization and json metadata! So yay, making progress.

dotthree wrote:What kind of instability problems are you having?

Since last evening I can actually post a pre-recorded audio file to Alexa, get an HTTP2 multipart MP3 answer back, decode it via MAD and play it via I2S amplifier.
There are some issues left that I really need to resolve before publishing the code:

• getting audio input via I2S input to work
• resolve SSL connection issues that sometimes cause very long connect times (minutes)
• handle authentication better

There is still much work to be done after that, but I'll publish it as soon as it does the bare minimum task. Pinky promise!

dotthree wrote:What's going on with your audio input? I've been trying to test and validate some audio stuff myself but seems that my I2S output isn't working, but my I2S input appear to be just fine at least on the scope and printing values.

Haven tried audio input yet, thats the next step. Output is working fine, you can check the web radio project for details. What exactly is your issue? Also, what hardware are you using for input?

dotthree wrote:Also what are you doing for authentication?

I'm using the refreshable grant authentication. Currently I manually get a new token via cURL and then hardcode it, which means the thing stops working after one hour. Shouldn't be too hard to react to to 401 headers I hope. I'll have to make a simple webapp to support the initial authentication grant.

@jakobson: I pushed some changes to the project. It has been refactored extensively and MP3 decoding is now in stereo. If you want to add your 24 Bit modification, the place to do it would be audio_renderer.c.

If you're still having trouble with git, you can also just send me a patch instead of a pull request.

Hi Buddy

Nice - I will give it at try over the weekend - Regarding Git - I just have to learn.

Have you notices performance increase after the idf 2.0 rc update. My DSP/Biquad runs 30-40% faster...

Thanks for the head up on the hack chat
/jørgen