I2S line connects only when probe is on

Thread Starter

italiansoccer19

Joined Apr 19, 2021
16
Hi all,

I have a setup for a mono/stereo I2S audio line, (BitClock + LeftRightClock + DataClock)

Now I switch mics from left to right and right to left and stereo. It all works when a probe is set on the BitClock pin. If there is no probe on the pin, the mic doesnt switch reliably.

Somehow connecting a probe to the BitClock line fixes the issue. I have tried many software settings, including forcing a transmit of data by filling the transmit register with a dud value.

It seems to get it going, but inconsistently. But this inconsistency goes away when I have a probe connecting to the BitClock.

Any suggestions? I currently have a 22pF capacitors on all 3 clock lines.

Thanks in advance,

I
 

MisterBill2

Joined Jan 23, 2018
9,308
Usually when connecting a probe improves a circuit function it is either that the circuit has lost a connection and is otherwise floating, or else that the small additional capacitance corrects a timing fault. So my first suggestion is to check the integrity of all the connections in that line between the control source and the controlled device, since one of those connections may have excessive resistance.
The second suggestion is less good, wich is to connect a 1 megohm resistor across the points that the probe was connected across.
 

Thread Starter

italiansoccer19

Joined Apr 19, 2021
16
I attached a resistor to the end of Bitclock to ground, but it only made the audio worse and noisy.

I find only a touch of the probe on the BitClock line does the trick. Seem like the audio is stuck, then a touch on the BitClock gets it going again.

Thanks !
 

Yaakov

Joined Jan 27, 2019
3,523
I attached a resistor to the end of Bitclock to ground, but it only made the audio worse and noisy.

I find only a touch of the probe on the BitClock line does the trick. Seem like the audio is stuck, then a touch on the BitClock gets it going again.

Thanks !
If it's not the DC then try adding ~20pF to the capacitance in parallel, the probe isn't magic, it's doing something between that pin and the other side of the probe.
 

Thread Starter

italiansoccer19

Joined Apr 19, 2021
16
If it's not the DC then try adding ~20pF to the capacitance in parallel, the probe isn't magic, it's doing something between that pin and the other side of the probe.
Sorry I made a mistake and attached the 22pf to LRclock instead of Bitclock.

When on Bit Clock it just gives me noisy signal.

I have a 1Mohm and 22pf in parallel from BitClock to ground.
 
Last edited:
Do you have a rough overview of how things are connected, are both I2S devices on the same board?

The 10X probe has an internal capacitance of the probe leads, maybe this is providing a return path for the signal? I'm wondering if there might be a grounding/earthing issue.
 

Thread Starter

italiansoccer19

Joined Apr 19, 2021
16
There is a modem that controls the I2S line. It is the master ; it is clocking at 48kHz i believe. Then there is the Left/Right clock also from the modem. The data line would be connected to the micro/SAMD51/Cortex-M4. So the modem comes on first in master mode, and the cortex clocks the data out as a slave

Again, all works fine when the prob e 10 x is attached !
 

MisterBill2

Joined Jan 23, 2018
9,308
Since the problem is not solved, but the system can be started by just a touch, I am going to suggest slapping the modem once or twice and seeing if that starts things working. A gentle slap used to be a valid diagnostic tool when servicing Sony and Superscope cassette tape decks. The slap would vibrate the switches and if that produced a signal in the output then a shot of contact cleaner was in order. That is the one simple diagnostic tool when it seems that nothing else provides a solution. Cheap and easy besides.
 

Yaakov

Joined Jan 27, 2019
3,523
Since the problem is not solved, but the system can be started by just a touch, I am going to suggest slapping the modem once or twice and seeing if that starts things working. A gentle slap used to be a valid diagnostic tool when servicing Sony and Superscope cassette tape decks. The slap would vibrate the switches and if that produced a signal in the output then a shot of contact cleaner was in order. That is the one simple diagnostic tool when it seems that nothing else provides a solution. Cheap and easy besides.
“Percussive Maintenance”
 

Thread Starter

italiansoccer19

Joined Apr 19, 2021
16
Bumping the modem doesn't do it - also a 10Mohm and 10pf in parallel just create the noise like the 1x probe. 10x probe still works like magic !

Attached is the Bitclock line that the probe hangs on to that fixes everything
 

Attachments

Last edited:
Top