TPA3116D2 Amp

Jump to Latest
#2,721
AudioLapDance said:
http://www.diyaudio.com/forums/class-d/237086-tpa3116d2-amp-144.html#post3797499
Click to expand...

"Remember, this is a digital chip working at very high frequencies, it needs digital power-supply bypassing.

Pardon my ignorance, I do not understand what do you mean by " a digital chip". I thought TPA3116 is a class D switching amp chip. An analogue signal is input to the amp. I looked up the block diagram on page 4 of the data sheet and do not see any ADC at the input of the amp and DAC at the output of the amp. Is the amp really working in the digital domain?

"Through hole for gain and other user selectable parameters is fine, nay great for DIY but NOT power-supply!"

This comment may be applicable to the specific board discussed at that particular post but is it also applicable to the YBDZ board?

Those OrangeDrop looking caps are pretty but they physically and electrically CANNOT address digital high frequencies.


Would you mind explain what is "digital high frequencies" and elaborate on the subject? Definitely, capacitors made of different materials will have different properties/charateristics e.g. leakage. Since capacitative reactance is inversely proportional to frequency, for a fixed capacitor value, higher the frequency the less effective it is as a filter. To improve its high frequency filtering ability, small value has to be used. This applies generally to capacitor, is it not?

And see how the main PS caps are sideways! The long leads to the ground plane would swamp any low-ESR caps you're using?!?! This designer is trying to seduce the DIYer with audiophoolery, NOT follow proper digital design procedures."

Again, This comment may be applicable to the specific board discussed at that particular post but is it also applicable to the YBDZ board?

Thank you in advance for educating this newbie.

Regards,
 
#2,722
This class D amp has an analog input that goes to at least one comparator and then to FET drivers and then the FET's themselves. (basically)

From the comparator part on the signal is either a "1" of a "0", just the timing makes music.

The devices in this "digital" part should switch from 0 to 1 and from 1 to 0 with as little transition time as possible.

This fast transition time helps retain the PWM or class D signal timing and for the FET's the reduced transition time equates to reduced linear on time and therefore reduced power dissipation.

For devices to switch fast there needs to be a very stiff supply for a fast changing load.

This is where you will see at least 2 capacitors ( of different values and types ) in parallel for the supply bypass.

ESR and ESL of the supply are what we try to reduce in class D amps.

Hope that helps.
 
#2,723
DUG said:
This class D amp has an analog input that goes to at least one comparator and then to FET drivers and then the FET's themselves. (basically)

From the comparator part on the signal is either a "1" of a "0", just the timing makes music.

The devices in this "digital" part should switch from 0 to 1 and from 1 to 0 with as little transition time as possible.

This fast transition time helps retain the PWM or class D signal timing and for the FET's the reduced transition time equates to reduced linear on time and therefore reduced power dissipation.

For devices to switch fast there needs to be a very stiff supply for a fast changing load.

This is where you will see at least 2 capacitors ( of different values and types ) in parallel for the supply bypass.

ESR and ESL of the supply are what we try to reduce in class D amps.

Hope that helps.
Click to expand...


DUG,

Thanks! So this is still an analog amp per se, basically controlled by a fast "on and off switch" and the signal is not manipulated in the digital domain.

Regards,
 
#2,724
Lo_Tse said:
DUG,

Thanks! So this is still an analog amp per se, basically controlled by a fast "on and off switch" and the signal is not manipulated in the digital domain.

Regards,
Click to expand...

I would not call it an analog amp.

It is class D "all in one chip".

The digital reference refers to power supply bypassing principles.

Look in the TPA3116 data sheet schematic...3 caps on each supply feed.
 
#2,726
You could generate a class D (PWM) signal using a microprocessor.

Use the on-board ADC to obtain the level of the audio signal at that instant in time and adjust a PWM output according to a formula (or even a look-up table)

This would be a digitally processed class D amp.

I could also say that some self oscillating class D amps could be labeled as an analog amp.

TPA3116 is neither digital nor analog...just class D.

IMHO, of course.

🙂
 
#2,728
There's art in choosing the optimum core material. I think that inductors sound different due to different losses. Ferrite is lowest loss but the normally fitted ferrites are rather undersized and made from very cheap core material (too high losses). When I eventually get one of these amps to try I'll experiment with hand-wound ferrite cores made from quality material with relatively large gaps to ensure no saturation.
 
#2,729
abraxalito said:
There's art in choosing the optimum core material. I think that inductors sound different due to different losses. Ferrite is lowest loss but the normally fitted ferrites are rather undersized and made from very cheap core material (too high losses). When I eventually get one of these amps to try I'll experiment with hand-wound ferrite cores made from quality material with relatively large gaps to ensure no saturation.
Click to expand...

What about using regular old air core crossover coils as close to the amp as possible?
 
#2,732
abraxalito said:
Not a great idea for a couple of reasons. Firstly they leak flux like nobody's business, second they'd probably have greater losses (skin and proximity effect) than a well chosen ferrite core. Can't see an upside other than impeccable linearity.
Click to expand...

That's what I figured.

Hey 'Braxy, since we're moving towards good output inductor choice and you've already dropped some hints, could you say more about these two points: flux and losses?

Thanks from all of us,
Cheers,
Jeff
 
#2,733
Flux we want contained, hence a magnetically shielded inductor is called for. Losses come in two forms - copper and core. An air cored inductor uses more wire than any other kind so has more copper losses than others, but has zero core loss (because the core is just air).

Copper losses are fairly complex to estimate for an inductor at high frequencies (such as here) because skin and proximity effects (causing the current to concentrate in a narrow band at the surface of a wire) need numerical methods to quantify them.
 
#2,736
wushuliu said:
How would an inductor like a coilcraft ser2915l measure up? It says it has low loss and super low dcr.
Click to expand...

Low DCR and loss are important but:

Look at the inductance vs current graph.

Should not drop more than 10% at max current. (IMHO)

Look at the self resonance frequency. This is where the inductor starts to turn capacitive. Should be much higher than switching frequency.

If this information is not available then pass on that part. (IMHO)

🙂
 
#2,737
wushuliu said:
How would an inductor like a coilcraft ser2915l measure up? It says it has low loss and super low dcr.
Click to expand...

In my (fairly limited) estimation, that inductor is as close to perfect for this application as its possible to get. Very low losses at the frequency of interest. A bit pricey at $3.60/100 qty from Mouser, but not prohibitively so.

<edit> Coilcraft do send out samples so perhaps someone could wrangle a set from them and see how they sound...😛
 
Last edited:
You must log in or register to reply here.
Top Bottom