My first Class D plans (schematics)

[RealForce]

Before I start making the schematic I want something to ask. Some questions might be noob but that's why I am asking them.

I want to build Class D amplifier with oscillator LM566. The input signal will be passed throe TL072 Op Amp. Then this signal will be amplified little and send to LM319 or LM311 comparator.
Now I don't know what IC to use for separating High and Low signals.
I don't want to use any of IR or IRS components. Can someone help me? I want to make half bridge amp.

And the last question what is all about that feedback? Do I need feedback with separate oscillator like LM566 or this feedback is only for self oscillating circuits that don't have separate oscillator.
And what is all about that dead time? How should I implement that dead time? I was reading about this but I can't get it So I would be happy if someone can explain these questions to me.

Thanks.

[jej]

Hi RealForce,
Why don't use IR products ?
My first class D amplifier used separates components. One for sawtooth generator, one high speed comparator and one driver. But layout is very sensitive! You can trust me
Poor result
With IRS2092, the only one problem is the requirement of many supply but we obtain very good results

[RealForce]

Originally Posted by jej

Hi RealForce,
Why don't use IR products ?
My first class D amplifier used separates components. One for sawtooth generator, one high speed comparator and one driver. But layout is very sensitive! You can trust me
Poor result
With IRS2092, the only one problem is the requirement of many supply but we obtain very good results

So you are telling me that there is no way to make good class D amp with discrete components?

[jej]

I simply say It's easier with appropriate IC! Your layout is more elaborate with several components. Don't forget you work at high frequency. Leakage inductance, parasitic capacitances become prevailling factors. With only one component including dead time, feedback, driver... you reduce length of connections and then don't use RC networks to reduce this effects.
Of course you can perform a amp with dicrete components but (it's my opinion and my experience) unlike class AB Amp you will have several designs before obtaining good amplifier

[savu]

Originally Posted by jej

Hi RealForce,
Why don't use IR products ?
My first class D amplifier used separates components. One for sawtooth generator, one high speed comparator and one driver. But layout is very sensitive! You can trust me
Poor result
With IRS2092, the only one problem is the requirement of many supply but we obtain very good results

hello Jet,

stupid question:

what did you mean to say by:"With IRS2092, the only one problem is the requirement of many supply but we obtain very good results "

regards,

[jej]

Hello Savu,
IRS2092 need 5 supplies: ±5V for digital part, 15V for analog part and ±50V for power.
For better results this supplies must be separated. This is more expensive,It's the only one problem!
regards

[savu]

Originally Posted by jej

Hello Savu,
IRS2092 need 5 supplies: ±5V for digital part, 15V for analog part and ±50V for power.
For better results this supplies must be separated. This is more expensive,It's the only one problem!
regards

Ok ...

I understand the part with the +15V and the +/-50V, but where exactly do you use +/-5V on it?

From my point of view it's not that hard concidering that you have to add 3 more windings on your SMPS or trafo, and some rectifiers, stabilizers, etc

Take in concideration that the multi voltage part doesn't interfeer with the audio comutation.

An UCD type amp is prone to generate more parasitics if you don't understand PCB importance. But in this case you get a nice tight and low parts count PCB for the audio part witch gets you out of the woods.

The SMPS or trafo is more easy to modify and it won't damage the audio quality and amp randament.

best regards,

[RealForce]

Personally I really dislike "ONE CHIP CLASS D AMPLIFIERS".

[savu]

Originally Posted by RealForce

Personally I really dislike "ONE CHIP CLASS D AMPLIFIERS".

Hello ...

any particular reason for the above ?

regards,

[ChocoHolic]

Originally Posted by RealForce

So you are telling me that there is no way to make good class D amp with discrete components?

Well, there in posting #50 I had posted a schematic with discrete level shifter.
1kW Gen2
Of course you can also do the gate drivers in discrete....

You may also dig for the older UcD circuits, their discrete level shifters and gate drivers are proven circuit parts, which can basically be adopted for any class D amp.

The drawback with lower integration:
Resulting mess is an overproportional function of the component count.
- optical
- signal integrity
- human errors & debugging

And of course you will also find a negative impact when calculating your mtbf or fit rate.