Class D Amplifier PSRR testing standards

Class D Amplifier PSRR

Hi guys

I'm designing a Class D amplifier to be integrated in mobile applications, and I need to test the PSRR .

I've read some documents stating that the circuit should be tested for GSM EMI: square wave, Vpp=200mV, freq = 217 Hz.

Is this the standard?
I'm also simulating for battery charge variations.

Besides that, when I plot the output FFT how do I separate the signal components from supply EMI components? How can I get PSRR value?

I'm simulating using hspiceD and plotting results using Wave Viewer from Synopsis.

Thanks

Re: Class D Amplifier PSRR

fcfusion said:

... when I plot the output FFT how do I separate the signal components from supply EMI components? How can I get PSRR value?

Click to expand...

Run 2 (or 3) separate ac analyses: One with signal only (no EMI on pwr supply), another one (or 2) with your selected EMI configurations (the GSM one, or your battery charge variations) on the power supply, with signal input (AC) grounded.

PSRR value = (ac signal value on output) / (ac signal value on pwr supply) , usually expressed in dB (negative value, then, because it represents a rejection ratio).

Run 2 (or 3) separate ac analyses: One with signal only (no EMI on pwr supply), another one (or 2) with your selected EMI configurations (the GSM one, or your battery charge variations) on the power supply, with signal input (AC) grounded.

Click to expand...

In my case AC analysis are useless because the whole circuit is strongly non-linear. I belive that what you said is valid only for linear circuits. I need to extract th PSRR value from a transient simulation. That's the only way I will have accurate results.

You say "with signal input (AC) grounded". If I do that there will be no switching activity in my Class D which is completely different of having the circuit working normally.

PSRR value = (ac signal value on output) / (ac signal value on pwr supply) , usually expressed in dB (negative value, then, because it represents a rejection ratio).

Click to expand...

Like I said I cannot use AC simulations but this gave me an idea though:

PSRR = (signal_with_supply_noise - signal_without_supply_noise)/(supply_noise)=
PSRR = (supply_noise_at_output)/(supply_noise)

This way I think I can use transient results.


Thanks for your response man

fcfusion said:

In my case AC analysis are useless because the whole circuit is strongly non-linear. I belive that what you said is valid only for linear circuits. I need to extract th PSRR value from a transient simulation. That's the only way I will have accurate results.

You say "with signal input (AC) grounded". If I do that there will be no switching activity in my Class D which is completely different of having the circuit working normally.

Click to expand...

Right; I just forgot this is Class D, sorry! Don't know if a PAC analysis (with your non-autonomous switching scheme) could help? Usually, PAC analysis is used for a single frequency (sinus) operation.

erikl said:

PSRR value = (ac signal value on output) / (ac signal value on pwr supply) , usually expressed in dB (negative value, then, because it represents a rejection ratio).

Click to expand...

fcfusion said:

Like I said I cannot use AC simulations but this gave me an idea though:

PSRR = (signal_with_supply_noise - signal_without_supply_noise)/(supply_noise)=
PSRR = (supply_noise_at_output)/(supply_noise)

This way I think I can use transient results.

Click to expand...

Yeah, I'm sure this is ok for transient analyses, too.