large caps on class D amp, what do they do?

[combatplayer]

https://www.banggood.com/Bluetooth-4_0-Digital-2_1-Class-D-HIFI-Power-Amplifier-Board-3-Channel-Amp-Module-p-1087441.html

this is the exact amp in question, i'm wondering about those two large caps sticking out next to the heatsinks. i've looked at the board layout and they seem to be part of some power filtering or something. now the question is, what do they do and what would be the effect of replacing them with caps of higher ratings etc? i will be using the amp with DC, i've speculated it might be for AC smoothing or some such.

[tooki]

They're for power supply smoothing. (You can see the voltage regulators right next to them.)

A class D amp (like any amp, I suppose) can have very large current peaks. With caps, you won't get dropouts when using a small power supply whose average power capability is more than enough, but which isn't nearly enough for the peaks.

I was playing around with some PAM8603 class-D amplifier boards, and using the peak capture mode of my Fluke 87, was able to capture current peaks of several amps, when the average was only a couple of hundred milliamps.

The caps on that board are already pretty beefy. Unless you actually find that the sound is getting distorted (crackles or dropouts) in actual use at high volume, there's nothing to be gained by upgrading the caps.

[mariush]

The specifications say maximum 22-26v DC in ... but I suspect they make a split power supply using two linear regulators after those capacitors, it looks to like there's a 7812 (fixed +12v) linear regulator and a 7912 (fixed -12v) regulator after the capacitors to make a  +/- 12v power supply.  I say looks like because I really can't tell what's written on those components from the pictures. 

Either way, based on the chip size and the number of leads, they look like fixed output linear regulators capable of maximum 1A - 1.5A current output.

They probably used 35v rated capacitors because 24v would be very close to 25v voltage rating, potentially damaging capacitors in time. If you plan to power the amplifier from a 18-19v laptop adapter, then it should be safe to use 25v rated capacitors.

There's also two diodes at the input, which means you can probably also simply connect a 18v AC transformer directly and the diodes would rectify to DC and then the capacitors would smooth out the voltage.

Yeah, you should be able to increase the capacitance by a bit without any problems but above some threshold, it won't make any difference. You usually need about 2200-3300uF per A of current, and if the linear regulators are capable of maximum 1..1.5A then even 4700uF should be sufficient... and since it's not a switching power supply running at high frequencies you don't need expensive low esr capacitors there, any decent 105c rated capacitor would do fine there.  The only risk there is that they'll overheat from those linear regulators there spreading heat in the circuit board and into capacitors through the leads.
 

[Zero999]

Yes they're decoupling capacitors which reduce the AC impedance of the power supply. You could replace them with a decent brand (no fake crap), with a low ESR and higher temperature and voltage rating.

I doubt it replacing them would make any difference to the sound quality but it's probably work it for reliability.

[combatplayer]

thanks guys, based on all your replies i'll just start out using the ones prefixed, maybe lifting them off the board for thermal reasons. later down the line i'm going to replace them with branded caps of similar or higher ratings. i'll also throw some small heatsinks on those regulators, just for the chance it'll improve something and because i have a fitting pair sitting around. though one thing;

based on the chip size and the number of leads, they look like fixed output linear regulators capable of maximum 1A - 1.5A current output.

this amp should offer peak outputs of 50 watts times two and 100w on a third channel, into 3 and 4 ohms, a real quick estimate would point out that this hits way above that 1 - 1.5a rating. would upgrading those regulators offer significant improvement?

[Zero999]

thanks guys, based on all your replies i'll just start out using the ones prefixed, maybe lifting them off the board for thermal reasons. later down the line i'm going to replace them with branded caps of similar or higher ratings. i'll also throw some small heatsinks on those regulators, just for the chance it'll improve something and because i have a fitting pair sitting around. though one thing;

based on the chip size and the number of leads, they look like fixed output linear regulators capable of maximum 1A - 1.5A current output.

this amp should offer peak outputs of 50 watts times two and 100w on a third channel, into 3 and 4 ohms, a real quick estimate would point out that this hits way above that 1 - 1.5a rating. would upgrading those regulators offer significant improvement?

I wouldn't have thought they'd use linear regulators for the power amplifier, which would be pointless. It would eliminate one of the biggest advantages of a class D amplifier: efficiency. If there are linear regulators on the board, they're most likely to power the logic and certainly won't be run anywhere near the maximum rated current.

Lastly, take the output power specification with a pinch of salt. There's no way it'll give the rated power without a huge amount of distortion. It'll probably be able to work at half the maximum rated power and still give respectable sound quality.

[mariush]

What I said was my best guess  since I can't read the text off the chips ... it was based on the chips being surface mount with small copper surface to radiate heat, and based on the chip package size. They could be 2-3 A linear regulators or something else.
If you want accurate answers, it would help to take a clear picture of the amplifier board or write down exactly what it says on the chips.

Those watt numbers could be unrelated to the actual power consumption and real audio watts. They could be PMPO watts or maximum watts if you allow up to 5-10% distortions, which is not something you'd desire from a music amplifier. 

So for example, let's say the board using an amplifier chip that can produce up to 100w with 10% distortions with 2-3 ohm speakers and with 25-26v (the maximum voltage) and enough current.  With 4 ohm or 8 ohm speakers the chip may output only 70-80w with up to 10% THD and with that high voltage.
But if you want less than 1% THD and that paired with slightly lower voltage  would limit the output to around 40-50w.
So the chinese guys say on the text 100w because technically it is correct and makes the amplifier look better (a lot more performance for the money) but in real world, you would get at most half that with very good quality.

Here's an example of a stereo amplifier chip that can do 2 x  120w with 6 ohm speakers , TDA7498 TR : https://www.digikey.com/product-detail/en/stmicroelectronics/TDA7498TR/497-11063-2-ND/2571064

So you could make an amplifier board with this chip and say on the sales page 120w but open the datasheet and read : http://www.st.com/content/ccc/resource/technical/document/datasheet/db/c5/04/74/fd/08/43/de/CD00244535.pdf/files/CD00244535.pdf/jcr:content/translations/en.CD00244535.pdf

Features
• 100 W + 100 W output power at THD = 10%  with RL = 6 ? and VCC = 36 V
• 80 W + 80 W output  power at THD = 10%  with RL = 8 ? and VCC  = 34 V
• Wide - range single - supply operation  (14 -39 V)

So you can see  both maximum output powers are for THD of 10% which is too much for listening to music, you want less than 1% (actually less than 0.1%) if you want a quality music amplifier.
You can also see that just by going with more common 8 ohm speakers, the maximum output power of this amplifier chip drops to 80w + 80w
And... in order to achieve these the voltage needs to be very close to the maximum input voltage (34v)

But scroll down in the datasheet and you can see some nice graphs .. go down to page 12, and you have there Figure 6 which says THD versus output power (at 6 ohm) ... and you can see there that the THD is below 0.1% as long as the output power is below around 50 watts per channel ... anything higher and the distortions increase dramatically... so you  don't want that.

Basically, your 100w+100w 6ohm amplifier chip that's 80w + 80w 8 ohm amplifier chip but which you only want to use at  maximum 40-50w + 40-50w at 8 ohm if you want quality.

This particular chip claims to be about 90% efficient, so assuming you design this amplifier for a total of 100w (let's say 50w+50w at 8 ohm) , you'd need a power supply that can provide up to around 110-120 watts of power... at let's say 38v (1v below maximum input voltage), that's 120w / 38v = ~ 4A of current. 

You'd use a 36v AC transformer (2x18v AC) , you'd rectify the AC voltage to DC using a bridge rectifier which gets you a peak voltage of 50v, and you'd use enough capacitance to get the minimum voltage above around 40v  at 4a  (  C = 4a / [2 x 60hz x (50v peak - 40v min) ] = 4 / 1200 = 0.0033333 Farads or 3333 uF minimum .. then use a linear regulator to stabilize this 40v .. 50v to 38v at 4A for the amplifier chip.
Or, you'd use a 30v AC transformer (2x15v AC), rectify it to get a peak voltage of around 41v, use about 12000-20000uF of capacitance and you have 36v..41v and then use a linear regulator to get ~ 35v output voltage.
Or you could just use a switching power supply with 24v DC output and just live with the fact that the amplifier board will only output around 35w per channel instead of 50w.


Sadly those amplifier boards you linked to don't mention the chip used and how it's configured (the gain and everything) and again, i can only guess what regulators that board uses... but if you figure that out you can look for datasheets and follow what i said above and determine what's realistically possible with those chips under the heatsinks.
 

[Zero999]

Yes, in the absence of any other information, all we can do is make educated guesses.

I did consider the capacitors could be DC blocking. It's possible the amplifier IC has four channels, two of which are configured single ended, for the left and right, and two bridged for the woofer. I doubt this is correct because of the positioning of the capacitors but wouldn't completely rule it out.

One thing to note about the power output, is class D amplifiers still have to go by Ohm's law. Unless there's a boost converter or transformer on the board, to increase the supply voltage or reduce the effective impedance of the load, seen by the amplifier, the maximum power output for a sine wave can be no greater than that given by following formula:

P = V²/(2R)

Where:
P = Maximum power out
R = The impedance of the load

Note that, the formula is for a bridge load, assuming no voltage drop across the power transistors. In real life there will be some voltage drop, which will lower power output.

[combatplayer]

thank you for your replies, i did some digging and it appears to be using a two TPA3116D2 chips, one for the two main channels and one bridged for the last sub channel. this is the datasheet i found http://pdf1.alldatasheet.com/datasheet-pdf/view/789612/TI1/TPA3116D2.html

the 100w sub channel does appear to be rated at 10% THD into 3 ohms but the 50w seems accurately reported at 1% THD into 4 ohms at 24v
i dont expect a full rating output, if it makes it halfway there i'll be satisfied, i'll be using 4 ohms for the sub and 6 for the sats, so it should be doable with a 24v ~10a dc supply.

i dont have the board on hand yet, i'm trying to spend the time it takes to deliver on a planning phase. i did find a closeup with one of the regulators in view, it is indeed a 1.5a positive regulator, a 78m05. the other i dont know, i didn't find a picture with it shown closely enough. however i also saw a tl074c op-amp which needs + - 18v, which is supported by the regulator i identified, i'll guess thats what they are powering. does this sound accurate?

edit: dont worry about just speculating, i did more or less come here to acquire educated opinions/guesses and to learn a bit along the way

[Zero999]

thank you for your replies, i did some digging and it appears to be using a two TPA3116D2 chips, one for the two main channels and one bridged for the last sub channel. this is the datasheet i found http://pdf1.alldatasheet.com/datasheet-pdf/view/789612/TI1/TPA3116D2.html

the 100w sub channel does appear to be rated at 10% THD into 3 ohms but the 50w seems accurately reported at 1% THD into 4 ohms at 24v
i dont expect a full rating output, if it makes it halfway there i'll be satisfied, i'll be using 4 ohms for the sub and 6 for the sats, so it should be doable with a 24v ~10a dc supply.

i dont have the board on hand yet, i'm trying to spend the time it takes to deliver on a planning phase. i did find a closeup with one of the regulators in view, it is indeed a 1.5a positive regulator, a 78m05. the other i dont know, i didn't find a picture with it shown closely enough. however i also saw a tl074c op-amp which needs + - 18v, which is supported by the regulator i identified, i'll guess thats what they are powering. does this sound accurate?

edit: dont worry about just speculating, i did more or less come here to acquire educated opinions/guesses and to learn a bit along the way

Good, the schematic will quite likely be very similar to the datasheet.

The regulator could be powering the TL074 but that would be a marginal design because the TL07x range of op-amps isn't specified to work down to 5V. There will be other logic circuits on the board which quite likely need 5V.

By the way all the channels will be bridged, as that's how the TPA3116D2 is supposed to be used. One of the channels on one of the chips is most likely not used.

[combatplayer]

There will be other logic circuits on the board which quite likely need 5V.

i believe the bluetooth board uses 5v