Phase shift in class D amplifiers. How it affects the sound?

Jump to Latest
#81
jan.didden said:
]

Maximum output power for Ri = RL is only true if you can freely chose RL. With audio where RL is fixed at 4 or 8 ohms, max power occurs with minimum Ri. The smaller Ri is with respect to RL, the more power is transferred to the speaker.

Jan
Click to expand...

The interesting question, with class D amplifier: Ri is Z (w), so that we no longer have a simple resistor divider as class A and class AB amplifiers.
 
#84
Last edited:
#86
#87
jan.didden said:
But there are lots of power amps that start to roll of at 30, 40, 50kHz and that have much larger phase shift than the Benchmark, which is really the exception here. What is the phase shift of an nCore at 20kHz, anyone knows this?

Jan
Click to expand...

Phase shift in class D amplifiers. How it affects the sound?

Hypex NC400 phase shift, about 16º at 20 kHz

index.php
 
Last edited:
#88
Last edited:
#90
Indeed, the folds in the human outer ear (pinna) introduces minute phase shifts of order 1-2usec. Our brain uses these phase shifts, which are asymmetric depending on orientation of sound relative to pinna (asymemtric), to determine directionality. This directionality does not depend on stereo reception. Try using one ear and see if you can hear direction from above or below or front or back. These shifts correspond to frequencies up to 300kHz. Hence, amps should be designed to be flat phase up to 300kHz or best spatial imaging.

Here is paper documenting phase shift perception in human hearing:
http://www.dtic.mil/dtic/tr/fulltext/u2/600151.pdf

So we hear frequencies up to 20kHz, but use phase shift information up to 300kHz. Using info and "hearing" sound are different. The directionality in itself, is not music/sound. Sort of like encoded metadata in a jpg.
 
Last edited:
#92
xrk971 said:
Indeed, the folds in the human outer ear (pinna) introduces minute phase shifts of order 1-2usec. Our brain uses these phase shifts, which are asymmetric depending on orientation of sound relative to pinna (asymemtric), to determine directionality. This directionality does not depend on stereo reception. Try using one ear and see if you can hear direction from above or below or front or back. These shifts correspond to frequencies up to 300kHz. Hence, amps should be designed to be flat phase up to 300kHz or best spatial imaging.
Click to expand...

Useful only if you have speakers that can reproduce 300khz and source recordings that contain this information.

So we hear frequencies up to 20kHz, but use phase shift information up to 300kHz. Using info and "hearing" sound are different. The directionality in itself, is not music/sound. Sort of like encoded metadata in a jpg.
Click to expand...

Ever had a hearing test? According to my audiologist most people don't even get close to 20khz ... my personal hearing runs from 40hz to about 13khz ... apparently that's better than most people my age. The average in his practice is something like 50 to 14k.
 
#93
This subject is very interesting.
Especially because we´ve devolloped our own class D amp which has 0 dgr. phase shift troughout the audioband.
"We" are four guys who decided to do some cool electronics, end will be referred to as we.
We have used ICEPower 1200AS2 as one of our references in the devellopment process just as well as nCore 500.

Our experience with the pretty large phase shift compared to 0 dgr´s is, that you obtain pretty much better 3D with no phase shift.
Also the sound seems more homogeneous in some sense.
It is a bit more straight forward to listen to.
That said the ICE 1200AS2 is an awesome amp, with both a lot of power and subtle detail. It is a full bridge design, which will not be able to compede on data with half bridge designs.
And an important note is, that i.e. Hypex nCore 500 is only 14 dB gain against 26 dB og the Hypex.
Our own module is 26 dB gain, 25µV noise, and a calculated Zout of only 9 µOhms. we need to do Kelvin measurements on that later on.
You can read more in our survey thread : Survey on class D project
 
#95
maty tinman said:
Hopefully this week amirm can measure the phase after changing a small fuse that burned out of the Nord with PURIFI 1ET400A modules. And a picture of inside.

Can you share your phase measurements of Icepower 1200, Hypex NC500 and yours?

BOSC 0º at 30 kHz. How many kHz have you reached with 0º ?
Click to expand...

We did not measure 1200AS2, because ICEPower products always measures as they say they do.
I´ll se if I cand find the phase measurements, or maybe we will have to do them again, så be patient please 😀
 
#96
Hi maty why are you so hooked on that BOSC graph, i mean you look selective only on the phase curve and get impressed by that, but look at its amplitude domain and notice its distorted in not being a flat linear curve but has a overshoot up higher before it rolls off as a rock probably with higher order slope than is normal for analog amps and thats why phase turn in deg at 20kHz point get so low.

Are you aware in physics for nomal natural sounds then phase pattern follow amplitude domain called a minimum phase device or feature, we can probably break that pattern with modern compensation or fir filters using DSP but why would that be good other than get a stand out sound that don't belong to this world.

Up the bandwidth will enhance phase, but 0º at 20kHz is probably impossible when at same amplitude is flat linear, for example a 1st order roll off at 1mHz perform about -1,2º at 20kHz and the 300kHz bandwidth xrk971 ask for is at about -3,8º at 20kHz.

In past thought you were looking for if some class D amps breaks being a minimum phase device and that i could see was interesting point, but as time goes it looks you think get phase at 0º @20kHz by any means is key to perfect performance and that sounds honestly as a noncense argument, also because real world at 20kHz point we tons of decades down because a microphone cascaded a tweeter cascaded our ears will roll off at 20kHz ways down.

BTW that BOSC overshoot you can virtual set for any other amp using your player Jriver to copy that hot boost up there.
 
Last edited:
#99
Last edited:
#100
A very cheap class D from China.

Archimago's Musings: MEASUREMENT: Yeeco TI TPA3116 Class D Amplifier (aka Nobsound NS-01G, Douk Audio F900S, Mogu F900S)

Yeeco_FR_and_Phase.png


Clearly the flat phase response of the DAC output as shown previously has been affected passing through this inexpensive Class D device. I've already measured Class AB devices where phase remains flat compared to this. It will be interesting to compare this with better Class D amps in the days ahead...

Another reason I wanted to measure this device is because this is a Class D amplifier. We are currently well into the ascent of Class D designs and I would not be surprised if by the end of the 2020's, almost all new devices are Class D by then for many good reasons! What we see in this device is a reminder of various limitations we should be aware of - potential low damping factor, phase variation, and ultrasonic noise.
Click to expand...
Hopefully soon we will have more phase measurements from more expensive class D and much better implemented.
 
Last edited:
You must log in or register to reply here.
Top Bottom