Educate yourself by having a read of the paper you can download at this page - PSRR: The Real Story about Closed- and Open-Loop Class-D Amplifiers | EE Times
Figure 4 shows what's happening with an open-loop part such as STA508.
Figure 4 shows what's happening with an open-loop part such as STA508.
Thank you very much for providing such information, which gained my knowledge on this topic.. In this article, it's said that most consumer class D amps have feedback, does STA508 has this function? I didn't see this on its data sheet...... but I have measured its power supply, and output PWM, when the 8 ohm load is applied, the voltage of PSU and output PWM don't change....but the output sine swing reduces.......
Seems you're not quite following along yet...
Let's try again. The article in the part you mention was talking about amps, an amp is a completed product. The STA508 is only one (possible) component of an amp and by itself it doesn't have feedback. It could be used with feedback but then the feedback would be provided by additional components in the amplifier. Perhaps your schematics include such components but so far you've declined to share what those additional components are.
Let's try again. The article in the part you mention was talking about amps, an amp is a completed product. The STA508 is only one (possible) component of an amp and by itself it doesn't have feedback. It could be used with feedback but then the feedback would be provided by additional components in the amplifier. Perhaps your schematics include such components but so far you've declined to share what those additional components are.
I didn't add additional components to form a feedback loop around the STA508.... it works as a standalone part... but I did measure the PSU voltage(10V) and PWM(0V for low and 10V for high) output of the STA508, when the load is applied, both of the mentioned two voltage quantities don't change.... but the output of low pass snubber network shrinks .....
There you have your own answer then - without an externally added feedback network, the STA508 is operating open loop. Hence the plot of fig4 applies to your particular application. Notice how the noise that's on the power supply shows up in modulated form at the output?
Yeah, I'm supposed to see that the sideband of output signal is high.....
But can this explain that output signal shrinking when load applied?
But can this explain that output signal shrinking when load applied?
No, its answering the question you posed in post #18. It doesn't explain your particular instance of drop which doesn't seem to be related to power supply droop as you've observed none.
Thanks...... BTW do you also have several classic articles that introducing class D feedback ? Seem to be a good point for learning..
No, I've not come across any myself. I rather suspect that digital classD with feedback is vendors' own individual IP and they're not too keen on talking in detail about it. From examining the two FFTs in the article I've already linked, I can't see the point of feedback from the perspective of subjective sound quality.
Some component of the drop, not all is caused by PSU (including interconnecting wires). What I say is You can measure it.
Load is part of the outputfilter, an infinitive load could create huge voltages, only +15% is very low.
No, I get the same error message as you. I didn't try to follow up that link when I posted it, rather I got the paper directly as a pdf download (hence no link to share) from searching Yahoo on 'the real story about closed and open loop classD amplifier PSRR'. See if that search takes you directly to the pdf.
Thanks! I found this way. But not pleased by the article.
As the second sentence says BTL has an inherent cancelling effect without signal. This alone is the reason of the false PSRR measurement in BTL, ClassD is no different in this aspect from any other class. But a single ended ClassD can be measured perfectly with the old method. Nothing new here.
Additionally I have to say that the good old error feedforwarding can also suppress the effect of supply ripple. And many topologies implement inherent error feedforwarding (for example UCD, "sigma-delta", and some of the hysteresys based self-oscillating designs, but a simple PWM can also use error feedforward). The writer should know this since TI has several PWM controllers with error feedforwarding. I suspect the non-defined "closed-loop I2S amplifier" also contains this technique in addition to the feedback.
As the second sentence says BTL has an inherent cancelling effect without signal. This alone is the reason of the false PSRR measurement in BTL, ClassD is no different in this aspect from any other class. But a single ended ClassD can be measured perfectly with the old method. Nothing new here.
Additionally I have to say that the good old error feedforwarding can also suppress the effect of supply ripple. And many topologies implement inherent error feedforwarding (for example UCD, "sigma-delta", and some of the hysteresys based self-oscillating designs, but a simple PWM can also use error feedforward). The writer should know this since TI has several PWM controllers with error feedforwarding. I suspect the non-defined "closed-loop I2S amplifier" also contains this technique in addition to the feedback.
Which TI chips are using error feedforwarding? Do you have references as I'm curious about this.
I can't find the chip I saw, but there is a paper mentioning on Page 3:
http://www.ti.com/lit/an/slua119/slua119.pdf
http://www.ti.com/lit/an/slua119/slua119.pdf
- Status
- Not open for further replies.