A discrete class D with HIP4082

[Joku]

Here's a discrete class D amp utilizing HIP 4082 driver. It is a basic PWM design with a triangle wave generator and a comparator. It should be able to use Vcc of up to 80V, and with parts such as IRF540N it should have enough current output even for low impedance loads. Thoughts?

[nigelwright7557]

What is the function of the bottom right LT1011 ?

If you want to keep the ALI and BLI in exact phase you could use a 4017 with one XOR gate inverting and the other buffering.

Why is there no feedback ?

[nigelwright7557]

Looks like a ramp generator.

[Joku]

Yeah, the bottom LT1011 is ramp generator. Good idea about using a 4017 to keep it exactly in phase, but is the difference noticeable in any way?

There's no feedback because a full bridge does quite well without it, and it would add complexity and potentially cause problems. Do you think it would be needed?

[Pafi]

Joku!

There is no pull-up resistor on output of U$3.

Triangle is better carrier waveform than sawtooth (what you have implemented). (Especially in closed loop designs.)

5 V is not enough for IRF540N.

R4,R6,R7 can be too high, check datasheet page 6!

Carrier waveform is proportional to supply voltage (IC1 output), what means every disturbance in this voltage can be heard on output.

[Joku]

Hi Pafi!

I just used 7805 on the schematic because there was no 7812 on Eagle, when building it I will use it instead. Anyways, is the 7812 output stable enough to make disturbances in supply voltage small enough?

This is an open loop design because it doesn't have feedback, so is triangle wave still better?

Also, I got the sawtooth generator schematic from a friend who has succesfully used it, apparently it doesn't need a pull-up resistor. The LT1011 datasheet page 6 was about oscillation problems, why would R4, R6 and R7 cause them?

[Pafi]

is the 7812 output stable enough to make disturbances in supply voltage small enough?

I dont know. It depends on some other factors too (supply input, decoupling capacitors, layout, number of channels, proximity, required SNR, luck...).

This is an open loop design because it doesn't have feedback, so is triangle wave still better?

A little.

Also, I got the sawtooth generator schematic from a friend who has succesfully used it, apparently it doesn't need a pull-up resistor.

OK, eventually it can work, however it's not too elegant. It's harder to analize this way.

why would R4, R6 and R7 cause them?

It's my fault, I ignored the positive feedback. But still the high impedances at switching freq. are not good, because they make the circuit more sensitive and more difficult to measure.

Also on the input a low pass filter is advisable, and a shunt resistor in the input to discharge the coupling capacitor in case of absence of signal source.

[marce]

No contigous ground, split ground copper pour, signal crossing the split.
All the above are going to give EMC aqnd signal integrity problems.