SSM2602 Noise

Hello, bdoan,

This is super difficult for me to determine what is wrong or different. 

I noticed the detail that you just swapped the part onto the same board that the WM8731 was installed and working. So that is an important detail and I did not miss that point.

What is making this difficult is because there are so many variables so more info and tests need to be done. It would be good to look at the noise on a scope and see if it is also on the power supply?

So the question to figure out is where is the noise coming from? Where is it seen and not seen?

So many switching power supplies these days can have noise on it. So that is a possibility.

You mentioned no audio input,... Have you actually grounded the input? I do not have your schematic so I am not sure if this is possible without shorting an output. 

Where is the noise seen? There are four outputs. Two are a headphone output and two are just line level outputs. Is it on all of them?

Then is it seen on the VMID pin?

Is there too much gain in the system? Is the gain the same as the WM part? 

You mentioned the registers are "almost" the same. So this will require more study to see what is different and could this be part of the problem?

So this is my initial first thoughts on collecting some more data to get an idea of what is going on and where? 

PCB layout can also cause issues but I will put that on the back-burner since the layout worked for the WM part. Still, this is an assumption. 

Dave T

Thanks

I have found, first and foremost, that the input impedance of the SSM2602 is much higher than that of the WM8731. 

This exposes a (fairly) high impedance node with a 1” trace. 

I will look at terminating the line input to the Vmid terminal to load it. 

Hello bdoan,

The inputs will already be terminated to VMID internally. I am not sure exactly what you are using or not using and which inputs are the problem. But, look at the datasheet typical application for some hints.

You will note that the VMID pin is there mostly for decoupling. 

The input is capacitively coupled to the output of a unity gain buffer.

The issue seems to be stability.  If the DC level of the buffer is offset from the 1.65V (Vmid) the noise is not present but when the DC levels on either side of the cap are the same, the whistle appears.

I have tried various methods of supply decoupling and isolation with no change in performance.

The noise is present with 0db gain and at increased gain levels.  

So many switching power supplies these days can have noise on it. So that is a possibility.

I have tried removing the LM2830 and substituting a LDO, no change.

You mentioned no audio input,... Have you actually grounded the input? I do not have your schematic so I am not sure if this is possible without shorting an output. 

I have tried cap coupling the LLineIn to ground, no change.

Where is the noise seen? There are four outputs. Two are a headphone output and two are just line level outputs. Is it on all of them?

I only have access to the LHP out

Then is it seen on the VMID pin?

It is not seen on Vmid , but touching Vmid with the scope probe momentarily disturbs the sound.

Is there too much gain in the system? Is the gain the same as the WM part? 
The whistle is present from gain = 0db to gain =+24db

You mentioned the registers are "almost" the same. So this will require more study to see what is different and could this be part of the problem?

The AGC registers are not present on the WM8731, but those registers default to AGC functions OFF


Are you are liberty to have a phone conversation?

Hello bdoan,

Right now it is a tough time of year to take a phone call and I still would need the info I am asking for. 

Please send over a schematic. 

Can you sample the whistle and send a short sound file?

Bunch more questions.

Is the gain, 0 to _24dB actually that amount of gain as measured and not from just front panel markings? 

Touching the VMID with a scope probe can certainly cause a quick audible disturbance due to the capacitance of the scope probe. 

When you cap coupled the input to ground, did you disconnect the cap that is being fed from the input buffer amp?

When you say the DC level on either side of the cap is the same... what is that voltage?

You also say you only have access to the LHP out but can you see the whistle with a scope and then you could look at the other outputs. 

Take a screenshot of the scope showing the whistle. 

Dave T

The whistle is not distinctly visible on the scope as it is down near the noise floor. I can send a recording to help identify it. 

If I mute the line inputs in software, the noise is not present so I know that it is involved in the line inputs. 

Also, since my decoupling scheme is not exactly the same as suggested in the example, I suspect that it is a stability problem. 

What values of ferrite beads are recommended?

My ground plane is a solid plane on layer 2 of a 4 layer board, not a 2 layer like the demo board. 

Is it absolutely necessary to split the analog and digital grounds if there is a solid plane?

Once again, the current layout works fine for the WM8731