The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> One question on class-d audio design
https://designers-guide.org/forum/YaBB.pl?num=1215151528

Message started by dandelion on Jul 3^rd, 2008, 11:05pm

Title: One question on class-d audio design
Post by dandelion on Jul 3^rd, 2008, 11:05pm

Hi,
I have a question on the class-d audio design. In the silicon test, I found the output will be shut down periodicly when the supply voltage is high or the input signal level is high.

After FIB, I found the bounce of the digital VDD is high and it falsely triggered the auxiliary overheat protection block and then it output wrong alarm signal. So, output power stage is wrongly shutted down.

What I am confused is, in the system simulations, the di/dt noise is simulated and the bording wire of package is also counted into the L*di/dt noise evaluations. But obviously, the bounce is larger than the simulation. I wonder why the big discrepancy between the simulation and actual silicon test results? Is there anything I missed in the designing?

PLs. shed some light on it.

Thanks a lot!

Title: Re: One question on class-d audio design
Post by ywguo on Jul 6^th, 2008, 12:58am

Hi dandelion,

What do you hear when that class-D has such a fault?

Yawei

Title: Re: One question on class-d audio design
Post by Visjnoe on Jul 6^th, 2008, 9:52am

Dear,

1. What value for the bondwire inductance did you use during simulations?

2. Have you run system simulations including all logic running from DVDD?

3. What was the rise time of any digital (switching) input you used during simulation?

4. Did you include a bondpad + package model in your simulations?

Regards

Peter

Title: Re: One question on class-d audio design
Post by dandelion on Jul 7^th, 2008, 2:51am

ywguo wrote on Jul 6^th, 2008, 12:58am:

Hi dandelion,

What do you hear when that class-D has such a fault?

Yawei

Hi yawei,
The periodicly shutdown can be monitored in oscillope. The voice is not comfirtable.

Title: Re: One question on class-d audio design
Post by dandelion on Jul 7^th, 2008, 2:57am

Visjnoe wrote on Jul 6^th, 2008, 9:52am:

Dear,

1. What value for the bondwire inductance did you use during simulations?

[i]The inductance is 2nH for my MSOP package. I believe it is enough fot simulations.[/i]

2. Have you run system simulations including all logic running from DVDD?

3. What was the rise time of any digital (switching) input you used during simulation?

4. Did you include a bondpad + package model in your simulations?

For,2,3,4, they are all considered in the simulations.
Regards

Peter

I am thinking if it is the effect of the supply voltage pumping? I am not very clear this effect. Hope to get some comments from you.

Thanks a lot

Title: Re: One question on class-d audio design
Post by Visjnoe on Jul 7^th, 2008, 8:12am

Dear,

1. What do you mean by supply voltage pumping?

2. 2nH seems a good value for the bondwire inductance.
Would increasing it to 5nH let you replicate measurements?

3. Did you use an ideal voltage source for the supply in your simulations?
Or did you use some finite impedance?

Regards

Peter

Title: Re: One question on class-d audio design
Post by dandelion on Jul 7^th, 2008, 6:36pm

Visjnoe wrote on Jul 7^th, 2008, 8:12am:

Hi Oeter,
Thanks for the help.
1.About the supply voltage pump, pls. refer the link
http://www.diyaudio.com/forums/showthread/t-50861.html

2.Yes, I tried the 5nH inductance, but it seemed no much effect

3.Yes, I uesd the ideal supply voltage in my simulations. Maybe U should add some finite impedance for my system verifications. I wonder how do you describe the supply voltage in system simulations?

Thanks a lot

Title: Re: One question on class-d audio design
Post by Visjnoe on Jul 7^th, 2008, 11:55pm

Dear,

1. If the DVDD is generated on-chip using a regulator, include this
regulator in your simulation. This way, you will have a finite impedance
(varying with frequency) instead of zero (ideal voltage source).

2. If DVDD is applied externally to your chip, you will at least have a
bondwire towards your chip + bondpad parasitics + package parasitics.
Maybe you already modelled the 'idea'l supply voltage this way.

Other than (1) and (2) I don't have any suggestions at the moment.

Regards

Peter

Title: Re: One question on class-d audio design
Post by sck236 on Jul 9^th, 2008, 7:47am

dandelion wrote on Jul 7^th, 2008, 6:36pm:

Visjnoe wrote on Jul 7^th, 2008, 8:12am:

To see if it is supply voltage pumping, try to add a small resistor(10ohm) between the VDD and the ground. The reverse current will be sunken by the small resistor.

Or you can add a huge capacitor between the VDD and the ground to minimize the rise in the VDD caused by the reverse current.

Monitor the signal after the LC filter and the pumped VDD. If the VDD is pumped up when the signal is negative, it should be due to the reverse current.

questions?
Are you using a single ended load not BTL?
Are you using the same regulator for the both of the digital VDD and the power stage?

Title: Re: One question on class-d audio design
Post by dandelion on Jul 9^th, 2008, 10:05pm

sck236 wrote on Jul 9^th, 2008, 7:47am:

Thanks sck236 for the reply. It is helpful to me. I will try your proposal.

We use the BTL instead of single ended.

We have no regulator on the chip. The digital VDD/GND and analog VDD/GND is tied far at the pad,