Hi,
I have always wondered, does the supply voltage affect sound quality.
Let's see how this comes about. At lower at mid volumes, does it make sense to lower the supply voltage to increase the PWM %.
Example does at lower volume a original signal required a duty cycle of 1%. If the voltage is reduced 4×. The Duty cycle would increase to 4%.Amplitude reduces by 75%
So in theory, it should be better. But can you hear it?
Oon
I have always wondered, does the supply voltage affect sound quality.
Let's see how this comes about. At lower at mid volumes, does it make sense to lower the supply voltage to increase the PWM %.
Example does at lower volume a original signal required a duty cycle of 1%. If the voltage is reduced 4×. The Duty cycle would increase to 4%.Amplitude reduces by 75%
So in theory, it should be better. But can you hear it?
Oon
(1 x n) vs (4 x 1/4n) is the same thing ?
I see what your getting at though.
The resolution would be 4 times better but at the cost of 4 times the volume.
I see what your getting at though.
The resolution would be 4 times better but at the cost of 4 times the volume.
Looking at the data sheet for the tpa2116d2, it actually shows lower THD at 24V vs 12V. I’m my limited experience with class d amps, they’ve always sounded better with a higher supply voltage, especially in the bottom end.
Hi,
I think I probably should have been a little clearer in what I meant. Does it sound better at lower volume with a lower voltage supply? Versus a lower volume at higher supply? The pulse amplitude would definitely be lower amplitude.
Looking at most THD graphs. The THD tend to be higher at low levels versus high levels.
Oon
I think I probably should have been a little clearer in what I meant. Does it sound better at lower volume with a lower voltage supply? Versus a lower volume at higher supply? The pulse amplitude would definitely be lower amplitude.
Looking at most THD graphs. The THD tend to be higher at low levels versus high levels.
Oon
NO, unless you reach clipping.
NO . What Nigel said.
OK
NO problem unless you try to surpass 100% duty cycle, in which case it will clip.
You are confusing supply voltage with signal/audio voltage, not the same.
As in Clas A or AB amplifiers: having 20V or 40V rails is the same if you put out 1W RMS (low volume) and never reach rails.
None will be "better" than the other.
Although starting as pulses, both at the SMPS and Class D amplifier, BOTH are INTEGRATED into plain DC or "slowly varying DC" a.k.a."Audio" .
"Slowly varying" compared to clock switching frequency, of course.
I think you are talking about Audio output now.
If we are talking about a certain Audio level, (say 1W) , why would amplitude decrease by 75%?
It does not work the way you imagine.
Nothing different, so no difference to be heard.
NO
See the above comparison to Class A or AB
I assume both your "lower volume" are the same of course, say 1W output.
Otherwise the "comparison" is meaningless.
NO
For, say, 1W output what matters is the integrated Audio voltage recovered after the output filter.
Supply voltage is irrelevant as long as you don´t try to go above 100% duty cycle, in which case it will clip.
I am referring to the PWM signal BEFORE the low pass filter. The amplitude of the of the PWM signal is actually your supply voltage.
So if you were powering it at 24V and you wanted 240mV. You would would be running the PWM at 1%. It is one of the challenges of early Class D design because switching frequency is already very high at a few hundred kHz. A 1% duty cycle, means a pulse of only a few tens of nanoseconds. Reducing the supply voltage to 6V for example will mean a 4% duty cycle for a 6V PWM. This will decrease the amplitude which will also ultimately show an improvement after the low pass filter since the amplitude is less.
There us an innovative approach to this problem by Infineon MA12070 in another thread using multilevel switch to reduce the amplitude level of the PWM. That approach is even better being able to switch from 0 to 1/2Vcc and 1/2Vcc to Vcc depending on output voltage required.
There is an audio manufacturer who actually came out with pretty smart solution of keeping the PWM constant regardless of volume level and changing the voltage of the PWM instead.
Oon
So if you were powering it at 24V and you wanted 240mV. You would would be running the PWM at 1%. It is one of the challenges of early Class D design because switching frequency is already very high at a few hundred kHz. A 1% duty cycle, means a pulse of only a few tens of nanoseconds. Reducing the supply voltage to 6V for example will mean a 4% duty cycle for a 6V PWM. This will decrease the amplitude which will also ultimately show an improvement after the low pass filter since the amplitude is less.
There us an innovative approach to this problem by Infineon MA12070 in another thread using multilevel switch to reduce the amplitude level of the PWM. That approach is even better being able to switch from 0 to 1/2Vcc and 1/2Vcc to Vcc depending on output voltage required.
There is an audio manufacturer who actually came out with pretty smart solution of keeping the PWM constant regardless of volume level and changing the voltage of the PWM instead.
Oon
However, I would like to add that this is all theoretical in many aspects. Lowering the voltage will also impact the other circuitry with the chip itself too, such as the op amp etc. And may actually make things worse.
There are other ways the chip might also compensate for that, including feedback as well as driving the PWM out of sync in a BTL configuration etc.
So I am curious if anybody actually tried it to see how it sounds.
Oon
There are other ways the chip might also compensate for that, including feedback as well as driving the PWM out of sync in a BTL configuration etc.
So I am curious if anybody actually tried it to see how it sounds.
Oon
Digital amps come as modules and require a min supply Voltage, so maybe a 2:1 range is possible.
If you permanently need only half the voltage, then you can do that, but you have to boost the input voltage by 6dB to get full scale output for full resolution.
But what happens if your input signal is above half scale?
you have to connect the higher voltage before the signal reaches above half, this is done in class H or G analog power amps to reduce the dissipation.
It wouldn't bring power saving with a classD amp, but can induce additional switching noises in low frequency, not a good idea
If you permanently need only half the voltage, then you can do that, but you have to boost the input voltage by 6dB to get full scale output for full resolution.
But what happens if your input signal is above half scale?
you have to connect the higher voltage before the signal reaches above half, this is done in class H or G analog power amps to reduce the dissipation.
It wouldn't bring power saving with a classD amp, but can induce additional switching noises in low frequency, not a good idea
Last edited:
Ok
yes.
So far so good.
Most definitely NO
You are mixing how a Power Supply works, supposing you want to get a low DC voltage from a higher one, and how a Class D amp works, completely different.
In a Class D amp zero output means exact 50% duty cycle, the *easiest* condition possible, not 1% and below.
Yes, so?
You will never reach 1% duty cycle, either way, unless signal peak *tries* to reach peak value, one rail or the other.
The opposite of a low power output.
As I thought, you are applying single polarity power supply analysis to symmetrical rails Class D amplification, absolutely different situation which you can never compare.
Wrong analysis.
Not sure what are you talking about but not what we are talking here.
WHY would he change rail voltages and keep pulse width cionstant?
That´s putting the cart before the horses.
And anyway, why pulse width modulate AT ALL?
Simply apply that "audio modulated voltage" straight to speaker, period.
IF there is a DC component just use a capacitor coupled output.
Last edited:
Good point about 50% duty cycle. I keep thinking of it as how it might be implementated in a split supply rail.
However the purpose of this is to reduce the amplitude of the square wave into the output filter. The output filter is only 2 pole. So quite a bit of the square wave is still present at the output.
Reducing the amplitude would also reduce the amplitude of the square wave.
I think there is a misconception here about what I am suggesting.
The supply voltage is not being determined by signal level. Rather it is determined by Volume setting. If you already determined that based on the specified input level, you won't exceed the clipping level. Would it be better to reduce the supply voltage.
In other words a buck converter powering your Class D amplifier based on your VOLUME control. So the voltage supply is actually constant.
Looking at TPA3116 datasheet, the THD at low power 10mW is about 10X that at 10W. The lower the power the higher the distortion. So there must be some mechanism. 10mW is not exactly very low, it is 283mV on an 8 ohm speaker. If you are running high efficiency stuff, this is just a softer passage in music.
Looking at the data sheet of TPA3116. you will see that distortion level changes with frequency, power output etc. And the distortion vs power profile changes with different frequencies changes as well Some increasing, some decreasing.
However it would be worthwhile to note that the biggest change is affected by speaker impedance with 8 ohm showing a lot less than 4 ohm.
Oon
Some have digital input, change pulse-code in pwm and drive the transistors directly.
But even the analog ones have a time discreet 1bit +V/-V output and the output filter recreates the analog.
Last edited: