vs1003 decoder + TPA3004D2 - noise problem...

Designing hardware that uses VLSI Solution's devices as slave codecs such as an external MP3 decoder chip for a host microcontroller.
Post Reply
Michael_K
User
Posts: 7
Joined: Thu 2011-10-20 12:39

vs1003 decoder + TPA3004D2 - noise problem...

Post by Michael_K » Mon 2011-10-31 16:59

Hello, support team...

I've made the prototype board for our project with vs1003 decoder and TPA3004 D-class amplifier
(and many other components). And I've got a strong white noise issue with this board.

I've tried to find the reason of this. So i've cut the traces between decoder and amp.
- decoder loaded directly on headphones sounds fine.
- amplifier with shorted inputs makes no audible noise at all.
- decoder with audio shutted down (SCI_VOL = 0xFFFF) + amplifier -> the noise is almost inaudible
- but if I set the SCI_VOL to any other value the strong noise appears again.

I've read the forum topics about noise issue.
But still can not reach any significant improvement.
(I tried to add RC-filters, separate the power paths, add really good and big decoupling capacitors etc. )

Now I'm ready to make the second iteration of this board design (as I've debug the firmware).
I would like to separate the analog ground from the main board ground-plane.

1. Can you advise me how to improve the noise performanse?
2. What the right way to connect analog ground plane to the main ground (close to decoder or to amplifier, may be via small inductance etc.)?

The prototype schematic and board layout are attached (the common ground plane polygon is hidden for clear view)
vs1003_sch.jpg
vs1003_sch.jpg (119.41 KiB) Viewed 4243 times
vs1003_layout.jpg
vs1003_layout.jpg (315.82 KiB) Viewed 4243 times

User avatar
Panu
VLSI Staff. Currently on holiday.
Posts: 2736
Joined: Tue 2010-06-22 13:43

Re: vs1003 decoder + TPA3004D2 - noise problem...

Post by Panu » Tue 2011-11-01 14:01

Hi!

Usually the noise with D class amplifiers comes from the sigma-delta noise. That's remedied by adding an RC filter (the "3rd stage Sigma-Delta reconstruction filter" in the schematic at: viewtopic.php?f=9&t=69&p=1062 ). Your schematic does not have this RC filter, you mention you have tried some RC filtering, hopefully you have also tried it the way it's in the viewtopic.php?f=9&t=69&p=1062 schematic. Note that the RC filter marked "High Frequency Grounding Path" does not eliminate the Sigma-Delta noise (you have this in your schematic and it's also required).

Other noise may come from other sources. The most usual source is digital noise from power suppy, which happens when digital buses are active, SD cards are accessed, etc..., but the Sigma-Delta noise is the most common problem, and it happens because the D-class amplifier also works by sampling the signal.

Sigma-Delta amplifiers should be connected differentially to achiebe best sound quality (that's OK in your schematic). There's a good VS1000 to D-class amplifier connection in our "USB Hifi Recorder" board (discontinued). Can you get any ideas from there:
viewtopic.php?f=9&t=95&p=1409&hilit=usb+recorder#p1409

Sincerely,
Panu
Info: Line In and Line Out, VS1000 User interface, Overlay howto, Latest VSIDE, MCU Howto, Youtube
Panu-Kristian Poiksalo, VLSI Solution Oy

Michael_K
User
Posts: 7
Joined: Thu 2011-10-20 12:39

Re: vs1003 decoder + TPA3004D2 - noise problem...

Post by Michael_K » Tue 2011-11-01 14:30

I tried to add RC-filter, but with no significant effect.
In fact, noise "moved" to lower frequencies, but still was annoying.
(And as i can understand the TPA3004 has 250kHz "sampling rate",
so i'm wonder if sampling has such a noticeable effect on audio frequencies?)

As I can see appnotes do not cover the layout or ground separation.
What do you think about these questions?
What the right way to connect decoder and amplifiers grounds, for example?

And your comment cause another question:
Do you think it's better (for noise performance) to use analog amplifier instead of class-D amplifier?
(It probably would be the simplest solution)

User avatar
Panu
VLSI Staff. Currently on holiday.
Posts: 2736
Joined: Tue 2010-06-22 13:43

Re: vs1003 decoder + TPA3004D2 - noise problem...

Post by Panu » Tue 2011-11-01 17:00

Hi!

Your response shows that you have noise also in the audio frequencies, and not just the sigma-delta noise... :cry:
(And as i can understand the TPA3004 has 250kHz "sampling rate",
so i'm wonder if sampling has such a noticeable effect on audio frequencies?)
Yes, the effect is very noticeable, because sigma-delta noise is from 100kHz to up to 6 MHz, and because of the 250kHz sampling rate, all the noise will alias down to audio frequencies if it's not removed before the amp.

But your problem clearly is "real" noise, not just sigma-delta noise.

The ground layout... ok... the official recommendation is to have single ground plane for the analog and digital. But if there is noise problem, then it can be helpful to split the ground to two ground planes that meet only under the IC. There's some discussion at viewtopic.php?f=11&t=105&p=473#p473 .
Do you think it's better (for noise performance) to use analog amplifier instead of class-D amplifier?
(It probably would be the simplest solution)
Not really, I quite like D class amplifiers. The only problem with D class amplifiers is that they make the sigma-delta noise hearable. But in your system there seems to be other noise problems. Have you tried to clean the operating voltages somehow?

To be honest, I am not familiar with the D class amplifier you are using. Maybe it's some kind of problem with how to use this amplifier. Have you used it in other projects?

PS. if you send your board to us, we'd be happy to take a look at it!

-Panu
Info: Line In and Line Out, VS1000 User interface, Overlay howto, Latest VSIDE, MCU Howto, Youtube
Panu-Kristian Poiksalo, VLSI Solution Oy

Michael_K
User
Posts: 7
Joined: Thu 2011-10-20 12:39

Re: vs1003 decoder + TPA3004D2 - noise problem...

Post by Michael_K » Wed 2011-11-02 12:16

Panu, thank you for your answers.
Panu wrote:sigma-delta noise is from 100kHz to up to 6 MHz, and because of the 250kHz sampling rate, all the noise will alias down to audio frequencies
Ok.
Panu wrote:you have noise also in the audio frequencies, and not just the sigma-delta noise...
Yes, i'ts very likely that the noise is external because the device is rather complex.
There are several switched-mode converters, 7" TFT display, RF-circuits etc. On a single board. With singl ground polygon.

So I think that sigma-delta sampling is _not_the_only_ source of noise.

I don't think it's a noise from source voltage - audio part has dedicated voltage regulator and I've tried to add really big and good capacitors.
Panu wrote:To be honest, I am not familiar with the D class amplifier you are using.
Well. It's the first my project with this amplifier also... Amplifier schematic mostly taken from its datasheet.
Panu wrote:if you send your board to us, we'd be happy to take a look at it!
Ok. Now I'm going to make some "experiments".
If you could take a look I'll send you the layout before making the PCB. Thanks alot.

Post Reply