[happyrabbit]

I started this project out of curiosity.   I currently own a 2A3 Set Amp. 
This design is a bit different from the TI reference layout.
I think it is a solid design..   I will have a couple boards next week. 

Design Spec:

TI TPA3118D2
10watts @ 8ohms
20db gain (60Kohm input impedance)    No input buffer.
LC Filter designed for a 6ohm load.
14.8V 4S LiPo PACK  (133wh from BatterySpace)
INPUT CAPS : 2824 Film
LC FILTER : Bourns 2100 + 2824 Film
POWER CAPS :  10mm + 2220 Film
4 Layer 2oz CU FR4
75mmx75mmx30mm



Updated the Subject to Reflect the actual board...

[wushuliu]

Huge fan of these chips. They are outstanding amps. Look forward to your posts.

Would you be willing to sell one of your boards?

[shadowlight]

Would be interested in a board also in case you decide to sell.  Need a winter project

[randytsuch]

Would be interested in a board also in case you decide to sell.  Need a winter project

+1

What changes did you make from the reference design?

[happyrabbit]

Differences...   I am sure TI engineers are much smarter than me so please forgive the diyness... 

*  Switched Mosfets need the Lowest ESR + ESL possible.... This is very different from TI's recommendations...
*the power caps are placed next to the TPA3118D2.   Inductance is not your friend.   
*I have pours on layers 1,3,4 tied to ground with a cap + ferrite bead in parallel. (layer 2 is gnd)
*The toroids are orientated parallel to the board.   The negative toroids are placed on the opposite side of the PCB.    Not sure this matters...  Mostly for packaging.
*I am concerned the LC Filter caps are not close enough to the battery gnd to be 100% effective.   I have no experience yet to know if this is an issue.  the 1st prototype will let me know.
*The TI reference design snubbers are closer to the outputs.       the 1st prototype will let me know
 
   

Dwight

[shadowlight]

still want a board if you help with all the newbie questions

[happyrabbit]

A simple housing for the board and battery...  135mmx135mmx105mm.  The RCA's enter from the sides.  Speaker Terminals, on/off, and charger port in the rear. 

[Folsom]

I'd be into it if there were some taps for more power capacitors.

Also why are there inpit caps on the grounds ?

[happyrabbit]

My omega 3xrs speakers only need a couple watts.   I think I have enough capacitance.  Look up the Panasonic OS-CON.  These caps can carry ripple current.    The 10mm size is impressive.   The rubycon st provides the low esr and low esl to the mosfets . 

 Differential inputs are biased to 3v.   Single ended inputs will still require both caps.

Dwight

[happyrabbit]

Salis,

I agree.   After looking at other similar designs, I need to expand my horizons...just a bit.    :
 
 Ver 1.1   - Added a pair of 10mm capacitors to the back plane  + Bourns 2200 toroids +  Jupiter HT Input Coupling Caps

Thanks for the insight

Dwight

[happyrabbit]

UPS finally found my place 

[happyrabbit]

 

[Folsom]

Well it looks great.

[happyrabbit]

Round 2...

Similar design goals but based on the TI TAS5611.  The main difference compared to the 3118/3116 is the bridges, gate drivers and the chip supply are separated from each other. 
Plus this chip allows up to 32.5V.  The Input Gain is fixed @ 20db.

   The board dimensions are 110 x 110 (mm).

[shadowlight]

Did you drop the plans for 3118 board?

[Folsom]

You should let us know how they sound.

Have you looked through the 3110/3116 thread? Class D is weak without a large cap backing it up. It's key to them having what the big dogs do. It's fine to have 100-220uf as close as possible, but you won't class A like bass without serious capacitence.

You may be able to avoid other compromises once you accept that.

Overly large caps don't work, at least not cheaper ones at like 15k. 

[Folsom]

C33-36 could be 470uf SMD FC's, and C57 should be like 8-10 1000uf's

Don't waste time with lm317 regulators either. Utterly ignore that stuff in data sheet. There's much better designs floating around DIY stuff.

[happyrabbit]

Hi Shadowlight,

Depending on the outcome.... yes.   TAS5611 is a better chip.  YMMV

Salis,

I only need  3 to 4 watts. I have enough capacitors for my application.    Many of the current designs do not bypass 3116 supply correctly.       Switching Mosfets and large AL capacitors don't play well together.    You need some MLCC X7R or  Film @ the chip.   Low ESR / ESL @ the supply is a common design goal for any DC/DC converter ( plus low inductance returns).   I am using these from Rubycon to bypass the TAS5611 :

http://www.rubycon.co.jp/en/catalog/e_pdfs/pmlcap/e_st.pdf

35V 2220 10uf

Not using LM3117.   They are PI filters based on ferrite beads.   Its hard to tell just looking @ the posted images.  The pi filter cleans up the incoming switching noise plus contains the noise for each half bridge.

http://katalog.we-online.de/pbs/datasheet/74279224401.pdf

I am using  a TI  Linear LDO for the vdd and gate supply :

http://www.ti.com/lit/ds/symlink/tps7a4700.pdf

Dwight

[Folsom]

I don't see the purpose in accomplishing "enough" when what it is much better is avaliable. If you can ave high grade bass for a difference in sub $5, why not?

The caps I said could be replaced are larger than I said you could use as a replacement (Ti=1000uf , me=470uf) at the chip. Also I didn't recommend removing the smaller caps that can be film, that are on the same legs for power in. The multiple cap recommendation is for the power section of the amp, not next to the chip.

My C numbers are from datasheet, not your PCB, if there was any confusion.

[happyrabbit]

Salis,

Here is your $5 upgrade 

I actually like the layout.    Thanks for the motivation 

Dwight

Per 1/2 bridge : 16mm Main Caps + (2) 10mm Caps + (2) 2220 SMD  w/ 2200 Bourns