TPA3116D2 Amp

[Destroyer OS]

I've not yet seen anyone post up a theory of how more caps on the supply makes for better bass. If anyone has one I'd dearly love to read it. The people I have seen advocating more caps have based their suggestions on experiments - i.e. practice. I myself have done that experiment and found more caps leads to better bass so I'm with Destroyer OS on this.

Indeed. When I get bored I just pull out the soldering iron and parts. Then I try things I've speculated on, or have been talked about.

Sometimes what I do is a huge waste of time, sometimes I can never go back.

 

[Destroyer OS]

The capacitors on the amp board are for decoupling. These should be as close to the chip as possible. The principal requirements are very high ripple current handling capacity and very low ESR. It should not be their role to be bulk caps to ensure that the power supply can supply the requested current, that is the role of the power supply. If you indiscriminately add capacitance after a switch mode psu you will remove it's capability to react to current changes, and so you need to add a lot more capacitance to make up for that. So large bulk capacitors on the amp board is definitely not a good idea, that is reserved for the power supply instead but that requires that you make a lay-out with as wide Vcc routings as possible.

I think you might be wrong on this, at least partially.

If the large capacitors for VCC were purely for decoupling, I think the ebay modules would have smaller capacitance like the TI eval board if the value was THAT specific.

My recommendations are not to do away with any of the smaller filter caps. I'm just suggesting that you can have them close, small, and then have a seperate power input area with more caps. By using many you'll negate some of the effects of them not being in the inner sanctum of componets.

I'll point out some of the prefered boards on here have two VCC caps nowhere near the intersanctum, and they are larger than the eval board's.

Furthermore, any good cap is better than almost any SMPS that people on here use. I don't think anyone is worried about hurting the feeligs of their SMPS when it's duties for large current change adaptation are taken over by PSU caps. I'll point out, again, the eval board and other boards technically have caps big enough on them that the SMPS current duty is altered to begin with. Perhaps that's why these units require a lot more overall VCC capacitance, somewhere.

But don't take my word on any of it, try it and find out.

 

[DUG]

Good to know.

I do strongly suggest a ferrite bead output filter though but naturally that requires speaker cables to be as short as possible. In 3312 (or 3412) size you can get those in as 10A rated with 4mOhm resistance.

I used the 10uH from this type:

http://www.vishay.com/docs/34104/lp25cz01.pdf

But I am open to suggestions if better are available.

P/N ?

 

[rsavas]

DUG,

Are you not confusing a inductor with a ferrite bead. TI says that you can use a ferrite bead in place of a inductor. This is a cost saving feature and has to be evaluated if it would meet EMI emission standards. Know one in this discussion seem to be concerned of EMI emissions nor have taken the time to measure such a specification.

Not sure if I read why to use a ferrite bead? This is to be in place of or in addition to the specified 10uH inductor?
Below is what TI has to say about filter selection criteria.

FERRITE BEAD FILTER CONSIDERATIONS
Using the Advanced Emissions Suppression Technology in the TPA3116D2 amplifier it is possible to design a
high efficiency class-D audio amplifier while minimizing interference to surrounding circuits. It is also possible to
accomplish this with only a low-cost ferrite bead filter. In this case it is necessary to carefully select the ferrite
bead used in the filter. One important aspect of the ferrite bead selection is the type of material used in the ferrite
bead. Not all ferrite material is alike, so it is important to select a material that is effective in the 10 to 100 MHz
range which is key to the operation of the class-D amplifier. Many of the specifications regulating consumer
electronics have emissions limits as low as 30 MHz. It is important to use the ferrite bead filter to block radiation
in the 30 MHz and above range from appearing on the speaker wires and the power supply lines which are good
antennas for these signals. The impedance of the ferrite bead can be used along with a small capacitor with a
value in the range of 1000 pF to reduce the frequency spectrum of the signal to an acceptable level. For best
performance, the resonant frequency of the ferrite bead/ capacitor filter should be less than 10 MHz.
Also, it is important that the ferrite bead is large enough to maintain its impedance at the peak currents expected
for the amplifier. Some ferrite bead manufacturers specify the bead impedance at a variety of current levels. In
this case it is possible to make sure the ferrite bead maintains an adequate amount of impedance at the peak
current the amplifier will see. If these specifications are not available, it is also possible to estimate the bead
current handling capability by measuring the resonant frequency of the filter output at low power and at maximum
power. A change of resonant frequency of less than fifty percent under this condition is desirable. Examples of
ferrite beads which have been tested and work well with the TPA3130D2 can be seen in the TPA3130D2EVM
user guide SLOU341.
A high quality ceramic capacitor is also needed for the ferrite bead filter. A low ESR capacitor with good
temperature and voltage characteristics will work best.
Additional EMC improvements may be obtained by adding snubber networks from each of the class-D outputs to
ground. Suggested values for a simple RC series snubber network would be 18 Ω in series with a 330 pF
capacitor although design of the snubber network is specific to every application and must be designed taking
into account the parasitic reactance of the printed circuit board as well as the audio amp. Take care to evaluate
the stress on the component in the snubber network especially if the amp is running at high PVCC. Also, make
sure the layout of the snubber network is tight and returns directly to the GND pins on the IC.

 

[DUG]

DUG,

Are you not confusing a inductor with a ferrite bead.

...

No confusion...I would use an inductor and not a bead...Loading the FET's into a bead/cap combination is not something I believe in doing.

(Worked 12 yrs doing EMC/EMI testing...still doing a little.)

 

[radiosmuck]

What does tone control have to do with current topic???

There is plenty of documented evidence that all well designed amplifiers sound the same. Assuming this to be correct, then the only way of adjusting the recorded sound to ones own taste is with tone controls.
Don't knock tone controls, they are still using banks of them in the recording industry.

 

[Lo_Tse]

Gentlemen,

I just have a crazy idea . Since there are quite a number of PCB designs popped up lately in this thread, and everyone seems to enjoy immensely in engaging "serious discussion and critiquing". Why don't those of us that are interested in this TPA3116 based-amp, each donates a small sum (5-10$ may be) to set up a fund. This fund will then be used to make boards of different designs. Then different amps will be built by the designers of the board. After all the amps are built, whoever participated in the venture could take turn listen to the different amps and do their own evaluation. We do not have to argue which amp is the "best" so there will be no"fights". Just express our opinion and provide feedback. We can also throw in the commercial amps e.g. the red, blue and green boards for comparison too.

Let's "walk the walk" instead of just "talk the talk". I do not think it will take that much money to produce 4-5 designs of PCBs. Besides, this will be a great intellectual exercise for the designers and an opportunity to verify their design principles.

Regards,

 

[Lo_Tse]

There is plenty of documented evidence that all well designed amplifiers sound the same. Assuming this to be correct, then the only way of adjusting the recorded sound to ones own taste is with tone controls.
Don't knock tone controls, they are still using banks of them in the recording industry.

Or one can use DSP with your speaker.

Regards,

 

[rsavas]

Lo_Tse,

I presented my design, I am willing to offer this design to the DIY community, gerber/fab data and documentation.
If we get people all wanting something different, it does not work, you have to decide upon a form factor and features.
My basic criteria is the same functionality as the TI eval pcb and then add features and packaging to that.
Very few have commented on how the design I presented can be improved upon. I like feedback as far as functionality and design goes. I want it to be a best effort and very little compromise.
If I fab something, I want to be certain that it fits my requirements at least.
I just now figured that i'd like to add a summer/lpf to the design as another option, so i can use two of them and make a 2.1 system, so off I go adding a dual opamp and some more passives.

So if anyone is willing to contribute to my efforts, we can start a new thread and focus on making the design go into production.

 

[blossom]

If someone could make a 4 channel amp that has same footprint as a miniDSP 2x4 board and could stack on it with standoffs that would be attractive. Stock miniDSP class D amp is only 10 watts/ch. I am buying two separate boards now and they won't be stackable. Or maybe a 2 channel PBTL (stereo) that uses 2 amps for 100 w/ch at 2 ohms?

This is a great idea. Even greater if there is the option to run either 4 channel or 2.1 with one of the chips in PBTL.

 

[blossom]

Problem is, as with any design, if one was to be designed, to your specifications, how many people would buy one?

Probably directly related to the amount of people that have those minidsp's.

 

[kp93300]

REd YJ board and Trend Audio 10.1 comparison

The red YJ is the standard board but the TRend is slightly modded .
Both of this are powered with the 12 V smps power brick that came with the Trend.
Both of this are used to drive the Markaudio 12P fullranger .

I notice the sound of the YJ is rather 2 dimensional and lack bass and treble refinement
Vocal is also better in the Trend.


Has anyone mod the board for better sound?


thanks


kp93300

 

[abraxalito]

I do strongly suggest a ferrite bead output filter though but naturally that requires speaker cables to be as short as possible. In 3312 (or 3412) size you can get those in as 10A rated with 4mOhm resistance.

Which manufacturer is that please? I just went over to TDK (since they have the best part searching interface known to man for these parts) and it turned out that the largest size was 1812 and that didn't have a very good current spec at all (like around 1A). Whereas the much smaller ones (0603) have very low resistance (down to 7mohm) and great current specs (6A). Two of their highest rated 0603s in parallel would be good for 12A but presumably the impedance at RF is going to be rather on the low side (26ohm each @ 100MHz)..

 

[xrk971]

Lo_Tse,

I presented my design, I am willing to offer this design to the DIY community, gerber/fab data and documentation.
If we get people all wanting something different, it does not work, you have to decide upon a form factor and features.
My basic criteria is the same functionality as the TI eval pcb and then add features and packaging to that.
Very few have commented on how the design I presented can be improved upon. I like feedback as far as functionality and design goes. I want it to be a best effort and very little compromise.
If I fab something, I want to be certain that it fits my requirements at least.
I just now figured that i'd like to add a summer/lpf to the design as another option, so i can use two of them and make a 2.1 system, so off I go adding a dual opamp and some more passives.

So if anyone is willing to contribute to my efforts, we can start a new thread and focus on making the design go into production.

Nice circuit, Rasavas. It's got all the bells and whistles. I did not see selectable mod frequency - is that something people want?

 

[Lo_Tse]

Lo_Tse,

I presented my design, I am willing to offer this design to the DIY community, gerber/fab data and documentation.
If we get people all wanting something different, it does not work, you have to decide upon a form factor and features.
My basic criteria is the same functionality as the TI eval pcb and then add features and packaging to that.
Very few have commented on how the design I presented can be improved upon. I like feedback as far as functionality and design goes. I want it to be a best effort and very little compromise.
If I fab something, I want to be certain that it fits my requirements at least.
I just now figured that i'd like to add a summer/lpf to the design as another option, so i can use two of them and make a 2.1 system, so off I go adding a dual opamp and some more passives.

So if anyone is willing to contribute to my efforts, we can start a new thread and focus on making the design go into production.

I am glad that someone commented on the idea instead of calling me crazy.

My idea is that the group will fund a few different designs. The reason is to see how different designs performs. Obviously, the designer of the amp has to provide his best effort. The design does not have to be everyone's ideal board. I am sure on paper, every designer will think their design will perform well. However, the final verdict has to come from a prototype. That's when the fund comes in, to provide some form of financial help to the designer to move the project along.

Personally, I am just curious to see how much performance can be squeezed out from this chip. I have no other motives behind this idea. Paying a small sum to get the answer worth my while. Of course, other people in this group might think otherwise. I wish I have the capability to fund the whole idea.

Regards

 

[rsavas]

I did not see selectable mod frequency - is that something people want?

I have made MODSEL.AM(2:0) selectable using jumpers.

My idea is that the group will fund a few different designs.

Please clarify what you mean by different designs?
You can have one pcb and try all sorts of different components to do your evaluations.

I am sure on paper, every designer will think their design will perform well. However, the final verdict has to come from a prototype.

I have already made a TPA3100 based design. I do not think that a TPA3116 will sound any much better. It might mitigate some RFI issues I am getting, so want to try out AM avoidance circuitry. Others have already expressed there comparative opinions. Yes I do agree proof is in the pudding for any design.

Personally, I am just curious to see how much performance can be squeezed out from this chip.

performance?

That's when the fund comes in, to provide some form of financial help to the designer to move the project along.

Hum, so how would that work out? The only costs that I would have, out of pocket would be pcb fab and comp costs, the same as everyone else, wanting to build one. Where the savings can be realized, are in the one time (NRE) pcb setup fees.
I will clarify, that I will give my implementation away for free, for none commercial use or profit.

 

[mijijim]

Lo_Tse,

I presented my design, I am willing to offer this design to the DIY community, gerber/fab data and documentation.
If we get people all wanting something different, it does not work, you have to decide upon a form factor and features.
My basic criteria is the same functionality as the TI eval pcb and then add features and packaging to that.
Very few have commented on how the design I presented can be improved upon. I like feedback as far as functionality and design goes. I want it to be a best effort and very little compromise.
If I fab something, I want to be certain that it fits my requirements at least.
I just now figured that i'd like to add a summer/lpf to the design as another option, so i can use two of them and make a 2.1 system, so off I go adding a dual opamp and some more passives.

So if anyone is willing to contribute to my efforts, we can start a new thread and focus on making the design go into production.

I would be interested in a board like this, doubly so if the fine pitch soldering were done

 

[kp93300]

I am interested in the board if the TPA 3116 chip is already soldered on the board.
I have no skill and equipment to soldered the chip.

 

[rsavas]

I would be interested in a board like this, doubly so if the fine pitch soldering were done

How about I give you instructions instead
Only soldering one comp leaves doubt, as to, if the ic and the rest have been soldered correctly. It is time consuming hand soldering, thus is why it is done automatically in a production environment. I can not compete against the Asia mfg's, so it is DI all Y or nothing.

Sorry
Rick

 

[wushuliu]

I am glad that someone commented on the idea instead of calling me crazy.

My idea is that the group will fund a few different designs. The reason is to see how different designs performs. Obviously, the designer of the amp has to provide his best effort. The design does not have to be everyone's ideal board. I am sure on paper, every designer will think their design will perform well. However, the final verdict has to come from a prototype. That's when the fund comes in, to provide some form of financial help to the designer to move the project along.

Personally, I am just curious to see how much performance can be squeezed out from this chip. I have no other motives behind this idea. Paying a small sum to get the answer worth my while. Of course, other people in this group might think otherwise. I wish I have the capability to fund the whole idea.

Regards

Agreed. I would definitely contribute some money for such a venture. I have too many interests as it is or I would be learning how to go about this myself. Frustrating to see so much potential but not have the time or skill to prusue further.