Line in protection SGTL5000 ESD / EMI causing SGTL5000 to stop

Status
Not open for further replies.
F

FredT

Member
Hi, I'm looking for wisdom/advice. I'm using the the SGTL5000 with teensy 3.2. I'm only using Line In input on the SGTL5000 and just passing the data via serial USB. The teensy is then connected to linux host.
I have been chasing an issue where the SGTL5000 seems to get in a bad state. Teensy still get data but it is null.
Current theory is that sometimes when connecting/disconnecting various audio source to "line in" there is a surge and trigger the bad state. We can also see that the USB port on the linux host is reseting the USB port due to EMI.

Example from DMESG:
usb usb4-port1: disabled by hub (EMI?), re-enabling...
usb 4-1: USB disconnect, device number 5
usb 4-1: new full-speed USB device number 6 using ohci-pci
usb 4-1: New USB device found, idVendor=16c0, idProduct=0483
usb 4-1: New USB device strings: Mfr=1, Product=2, SerialNumber=3

The teensy never seems affected. I confirmed twice that performing a SGTL5000 reset from teensy recover the SGTL5000 and start wording again.

I'm starting to wonder if I should put TVS to protect from ESD/EMI such as http://www.ti.com/lit/ds/symlink/tpd1e10b06-q1.pdf on the line input.

Anybody have suggestion or experience to share in that matter?

Thanks
 
Do you have a scope to show what's happening? That can be rather useful.

Aside from the TVS, you can try ferrite beads (they are easily removable). You could also try adding a couple of resistors -- something like 220 ohm series and 22k to ground.
 
Update, I'm now able to repro the problem. This is clearly a transient when connecting to another device. I can repro by wiggling the 3.5mm mini jack audio connector when connecting to a device(like re-insertion multiple time but halway in the connector.) The scope is showing 15V+ peak during insertion. Now I don't know if this could be due to in-rush current due to different ground reference or just the nature of ESD. I will try adding TVS and see if that helps.

About adding Resistor in series and parallel, wouldn't this affect the audio signal? The input impedance of the line input of the SGTL5000 is 29k according to the spec.
 
Anything you put in will have some effect on the audio, it's just that the effect can be vastly different depending on what's added. The resistors are very useful when used in relatively high speed digital lines to deal with reflections. Whether that's the case here or not is not possible to know for sure until you try them. And even if the spec sheet says 29k, I'd still put in something, like 47k - 68k. Real resistors may behave slightly differently than those implemented with semiconductors in ICs.

You obviously want to find and fix the root cause, whatever that may be -- I would probably connect device grounds together to see if that makes any difference. If that doesn't help, then try to power everything from the same source if possible to eliminate differences there, too. Methinks you should not be seeing 15V+ peaks.
 
I just started back to look at this issue. I'm using a custom audioboard with SGTL5000 but still using a retail teensy 3.2. The audio circuit is following closely the audio board reference schematic.. I can't repro the issue using the audio shield so this can't be ESD or EMI since the audio shield does not have ESD protection. The only thing that I can think of, is either Analog/Digital/Supply layout ground problem or a component that is not to spec?
I did a new PCB that follow the star GND configuration where the Analog Ground of the SGTL5000 connects to the main ground at one point. The issue is still happening.
I have reduced my circuit to bare minimum. only SGTL5000 power and line in + Teensy and still happening.
I'm pulling my hair on this one.

You can see the schematic here: https://drive.google.com/file/d/1LW9mBs5LZYzr2Vv14stKto-nC7_saOrM/view?usp=sharing
 
Looks like adding 100 Ohm resistor in series on the line in prevent the problem. So far I have not observed side effect on the audio signal. The SGTL5000 has input impedance of 29k so the impact should be minimal. Perhaps it is limiting inrush current?
Any suggestions?

Thanks
 
FredT said:
Looks like adding 100 Ohm resistor in series on the line in prevent the problem. So far I have not observed side effect on the audio signal. The SGTL5000 has input impedance of 29k so the impact should be minimal. Perhaps it is limiting inrush current?
Any suggestions?

Thanks
Click to expand...

Well it reduce the occurrence but can still happen.
 
Increase the series resistor and add a 47k - 68k resistor to ground at the 3.5mm mini-jack connector. If you have a fast enough scope, try checking for reflections on the line, both with, and without the series resistors.

What is the source of the the audio signal?
 
I believe I figured out the problem. Hopefully it will help others that may have "forgotten" to use basic PCB layout rules...

First of all, I realized that I blindly use the schematic of the teensy audio board to create my PCB/layout. I clearly oversight correct positioning of the decoupling capacitor.
On top of that, the SGTL5000 was powered by 2 different 3.3 supply line for different pin. So in a nutshell, I did not properly routed things in star configuration. Once I cut the additional 3.3V line and connected to a common 3.3V source for all SGTL5000 pins the problem could no longer be reproduced.

So in conclusion... take the time to do proper layout!

Redid the PCB routing....
 
Guess you may have found the root cause. Of course that was in my 2nd post...

While breadboards have their own issues, I always do everything on them first to iron out possible issues.
 
Yep you did bring that up. At the time I was not seeing it...

Although this is working fine on the official audio shield and now my board, there are still significant voltage spikes when connecting the input to an audio source. That is happening on both board on the 3.3 and even 1.8. it is not reseting the sgtl5000 anymore but I will probably add TVS anyway for additional protection.
 
Did you try the "47k - 68k resistor to ground at the 3.5mm mini-jack connector" ? (before the caps, between 2,3 and analog GND) If you worry that it will have some impact on the audio, you could try something a bit higher.

What is providing the input signal? Do you see the spikes when you connect just an audio cable with nothing connected to the other end?
 
Status
Not open for further replies.
Back
Top