The minimum sample frequency that allows reconstruction of the original
signal, that is 2M samples per unit distance, is known as the Nyquist frequency .
eva :
Nyquist just tell you "The minimum sample frequency"
never tell u "the best sample frequency"
if u can , why not use a 44k pwm signal to make a full rang class-d amp?
rg
fumac
signal, that is 2M samples per unit distance, is known as the Nyquist frequency .
eva :
Nyquist just tell you "The minimum sample frequency"
never tell u "the best sample frequency"
if u can , why not use a 44k pwm signal to make a full rang class-d amp?
rg
fumac
The only two advantages are a higher bandwidth that nobody needs because we are completely unable to hear stuff above 30Khz (not to mention at 200Khz) and a very low carrier residual that forces you to get part of your feedback signal before the output filter for proper oscillation,
----------------------
no body can heard about 30k signal, yes , right!
but u can heard about the phase shift of treble
in hifi audio , phase shift is important same like frequency respond.
so we just can rise up the filter to higher ,
then keep the phase shift lower and lower.
u are in spain, so u can heard about the mcd in spain,
and u can carry a ucd amp to comparat the diffrent with MCD.
i 'v never have a ucd on hand , just heard about it once, good sound .i never tell others ucd have a poor sound.it is a great class-d
we just give users another choise, so need to told other the diffrent,
some body like ucd , then they can buy ucd, yes they r buying.
if some body like to try a more higher speed and more width frequency respon, less phase shift, then they can choose us .
xmax
rg
fumac
----------------------
no body can heard about 30k signal, yes , right!
but u can heard about the phase shift of treble
in hifi audio , phase shift is important same like frequency respond.
so we just can rise up the filter to higher ,
then keep the phase shift lower and lower.
u are in spain, so u can heard about the mcd in spain,
and u can carry a ucd amp to comparat the diffrent with MCD.
i 'v never have a ucd on hand , just heard about it once, good sound .i never tell others ucd have a poor sound.it is a great class-d
we just give users another choise, so need to told other the diffrent,
some body like ucd , then they can buy ucd, yes they r buying.
if some body like to try a more higher speed and more width frequency respon, less phase shift, then they can choose us .
xmax
rg
fumac
There's one good paper of Sovadk's PhD paper http://www.student.dtu.dk/~s042302/diy/low_dist_amp_master.pdf about what happens with higher switching frequency. It must be divide into 2 kind, hysterisis (before LC) or phase shift (after LC). In case of phase shift, the lower residual voltage, automaticly will rise the switching frequency and loop gain. Increasing loop gain means lower THD and lower output impedance, naturally. Sovadk made a phase shift type selfoscilating, put LPF in the middle, and the switching frequency can rise to 790khz without any hysterisis feedback.
fumac said:
You don't seem to be used to mathematical definitions. The minimum sample frequency is the one that allows perfect representation of the waveform.
The DAC inside every standard CD player does this. It converts a PWM signal with 44Khz carrier back into a nice and good sounding analog waveform.
A class D amplifier with an ideal modulator and a very high order output filter would produce perfect 20Khz output operating at just 44Khz. These are the real world limiting factors. In practice we use a higher Fsw just to ease the task of removing frequencies above Fsw/2 from the input and the output with practical filters.
Eva said:
hi eva
why not make a class-d amp running at 44.1k for hifi audio ?
low pwm frequency, can be very high effect,
also will be very low THD at very long dealtime(depend on ur thinking )
just as u say , "It converts a PWM signal with 44Khz carrier back into a nice and good sounding analog waveform."
why not use 44k to make it ?
tell me the reason.
pls
fumac
fumac said:
Sorry, we can only hear differential phase between channels (phase differences between sound reaching left and right ears). It is just a matter of component tolerance to get this part of the amplifier right.
It has been widely demonstrated that we can't hear phase shift between frequencies reaching the same ear, particularly for trebble and midrgange (bass still gives rise to some controversy). You can use a phase shifter to flip your highs 180 degrees or you can even delay them one whole milisecond with DSP and you will never notice it as long as the group delay transition is smooth (like in an analog filter). Sorry again.
fumac said:
hi eva
why not make a class-d amp running at 44.1k for hifi audio ?
low pwm frequency, can be very high effect,
also will be very low THD at very long dealtime(depend on ur thinking )
just as u say , "It converts a PWM signal with 44Khz carrier back into a nice and good sounding analog waveform."
why not use 44k to make it ?
tell me the reason.
pls
fumac
Eva said:
The reason is that very sharp input and output filters are required for 44Khz operation, while 300Khz switching allows to use plain 2nd order filters. I'm serious. What other exotic reason were you expecting? uh?
Eva said:
evily: sure thing , u tell me the truth, thanks eva:
"because u cant find a very high order output filter " for class-d amp(high power high order, l)
so , u need to push the pwm frequency higher and higher
all about 1mhz class-d is depend on the filter.
we cant! get the very very very good filter to make a very good class-d amp at low frequency.
now , u know why we go to 1mhz or 2.56mhz
because the largest problem is the output filter.
the 22k filter in true live will course the treble respond slower and the phase shift large.
so we goto the 1mhz and just use a simple filter to output a
very good performance music.
u true know this .
u said that u cant heard the phase shift ,
ok , but this is also meaning: at the 1mhz the phase shift is lower than the 300k ones
thanks ,eva
do we true know our body ?--pls ask Physiologists, perhaps they dont know the answer too.
and also u can ask the speaker maker : tannoy, kef,
they have many Coaxial speakers, ask them for why use Coaxial speakers?
what's the diffrent with Coaxial speakers and nomal speakers
xmax
fumac
The phase and amplitude response of the output filter is completely linearized once it gets enclosed in the proper feedback loop. The self-oscillating arrangements that I use don't exhibit major phase shift at 20Khz, even with switching frequencies as low as 80Khz and output filters resonating at 10Khz. Maybe you were just unable to implement the proper feedback scheme and you had to resort to a brute force approach like increasing switching frequency?
No, phase shift is not the problem with filters.
No, phase shift is not the problem with filters.
eva
sohappy the talk with u
i 'm thinking a fare game
any body like 300k, then goto 300k
anybody like 1mhz , then goto 1mhz
also they can own both.
this is quite fare for anyone here
we just product an amp
that we do like , that my agents do like
and my users buy it ,perhaps like it ,
they use it to listen to music, make them happy .
this is the important thing.
in the world, there r many many amps,
class-a , class-ab, class-d, class-h
many many designers r working hard to make them quite better
for what ? just for happy.
i think they can run together without any bad words
in this topic, ihave learn something, thanks all
too busy this days , back to work , bye
rg
fumac
sohappy the talk with u
i 'm thinking a fare game
any body like 300k, then goto 300k
anybody like 1mhz , then goto 1mhz
also they can own both.
this is quite fare for anyone here
we just product an amp
that we do like , that my agents do like
and my users buy it ,perhaps like it ,
they use it to listen to music, make them happy .
this is the important thing.
in the world, there r many many amps,
class-a , class-ab, class-d, class-h
many many designers r working hard to make them quite better
for what ? just for happy.
i think they can run together without any bad words
in this topic, ihave learn something, thanks all
too busy this days , back to work , bye
rg
fumac
"Maybe you were just unable to implement the proper feedback scheme and you had to resort to a brute force approach like increasing switching frequency?"
i post a mcd-500 squear pwm signal.
working at 135v , 935k , 4ohm load,
"some body cant do it " not meaning "no one can do it "
best rg
fumac
i post a mcd-500 squear pwm signal.
working at 135v , 935k , 4ohm load,
"some body cant do it " not meaning "no one can do it "
best rg
fumac
Attachments
I think the task of building a PWM amp that does work well at this switching frequency is not an easy task. And fumac did it well IMO.
But I agree that the motivation for doing so is not the right one. To me it looks a little like the "Boss Hoss" motorcycle.
OTOH I don't fully agree with the following statement:
This is true but group-delay distortion decreases the ability of the ear to exactly measure the inter-aural time delay.
So there is a (old fashioned !!) way to achieve the phase-linearity that fumac is calling for and the low switching- and cutoff- frequency that Eva is calling for. It is called phase-linearisation and it simply trades increased total time-delay against improved phase-linearity.
With this principle it is theoretically possible to build an amp with a cutoff-frequency of 10 kHz that has a flat group-delay up to 100 kHz. A less extreme example of such a phase-linearisation can be seen here:
http://www.diyaudio.com/forums/showthread.php?postid=1310284#post1310284
Regards
Charles
But I agree that the motivation for doing so is not the right one. To me it looks a little like the "Boss Hoss" motorcycle.
OTOH I don't fully agree with the following statement:
This is true but group-delay distortion decreases the ability of the ear to exactly measure the inter-aural time delay.
So there is a (old fashioned !!) way to achieve the phase-linearity that fumac is calling for and the low switching- and cutoff- frequency that Eva is calling for. It is called phase-linearisation and it simply trades increased total time-delay against improved phase-linearity.
With this principle it is theoretically possible to build an amp with a cutoff-frequency of 10 kHz that has a flat group-delay up to 100 kHz. A less extreme example of such a phase-linearisation can be seen here:
http://www.diyaudio.com/forums/showthread.php?postid=1310284#post1310284
Regards
Charles
phase_accurate said:
hi charles
nice to meet u
thanks for ur deep thinking reply
my english is not good for telling other my thinking at technolegy,
but ur words is good to me
"phase-linearisation "
and ur reply at that topic is helfull.
speaker box designers more care about the phase .
about 1mhz , we had slow down the frequency to 600k at same board, but after we do hearing test,
we click out the 600k design.
xmas
fumac
There are so many factors influencing fidelity that one factor cannot be taken as a hard criteria. But in my own experience in listening with speakers of around 87db sensitivity, I prefer good clean frequency response up to 50KHz. Currently the speaker is the most limiting factor masking the audibility of most electronics performance.Eva said:
I do ear clearly the effects of absolute polarity in music. If the speakers are not good enough, especially the spectral decay performance in the first 0.4ms, then these effects are not that audible. Spectral decay is like noise in a system.Eva said:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.