Still, it's not that simple, with vision being a more dominant sense than hearing and occupying a bigger part of our brain in terms of activity. It's one reason why I don't need as good sound quality with movies as I want for my music. The visual experience takes much focus away from the sound for me and many other people, which is also probably why that nice TV is more important than speakers for the average movie watcher.
This is a moronic modification.
Go get an oscilloscope and discover what you have actually done to the output waveform by increasing the parasitic inductance of the output capacitor well over ten-fold. It will look awful with spikes and ringing. One of UcD main features is a clean sine-like output residual.
Distortion will also be greatly increased, as UcD modulator relies on a clean carrier residual, and these long-leaded wound-foil capacitors just can't provide it, they stop behaving as capacitors soon above 20khz, and UcD needs something behaving as a cap up to a few Mhz.
It won't last long if used at high volumes either, as erratic switching can arise.
Class D is RF-grade electronics, tweaking this is not for children!
In the class D stuff I do I'm routinely getting 3rd harmonic 80dB below fundamental up to insanely high power (far more than needed for critical listening) and other harmonics are not above that. That's 20dB better than any loudspeaker or the own air or vinyl (or our ears) can do. If that's not enough, nothing will be.
It's subject to further improvement too, but at the expense of degradation of other properties like efficiency, size, cost...
And for the moment, I want to keep the stuff I design affordable
Last edited:
Eva, could the parasitic inductance and resistance in a foil cap be thought to have possible positive audible effects though it deviates from the theoretically optimal?
Also, could these effects be controlled/illiminated by using a smaller faster cap in parallel, like when modding loudspeaker crossovers?
cheers,
Also, could these effects be controlled/illiminated by using a smaller faster cap in parallel, like when modding loudspeaker crossovers?
cheers,
This is a moronic modification.
Go get an oscilloscope and discover what you have actually done to the output waveform by increasing the parasitic inductance of the output capacitor well over ten-fold. It will look awful with spikes and ringing. One of UcD main features is a clean sine-like output residual.
Distortion will also be greatly increased, as UcD modulator relies on a clean carrier residual, and these long-leaded wound-foil capacitors just can't provide it, they stop behaving as capacitors soon above 20khz, and UcD needs something behaving as a cap up to a few Mhz.
It won't last long if used at high volumes either, as erratic switching can arise.
Class D is RF-grade electronics, tweaking this is not for children!
In the class D stuff I do I'm routinely getting 3rd harmonic 80dB below fundamental up to insanely high power (far more than needed for critical listening) and other harmonics are not above that. That's 20dB better than any loudspeaker or the own air or vinyl (or our ears) can do. If that's not enough, nothing will be.
It's subject to further improvement too, but at the expense of degradation of other properties like efficiency, size, cost...
And for the moment, I want to keep the stuff I design affordable
That's exactly my thought - the inductance of the leads of those caps will cause problems. I'm leaving mine alone. Not to mention I'd really rather listen to music than solder. There's a reason those parts are SMD!
Hi,
replace the filter capacitor with a better ... it's useless. As Eva says
To Sampleaccurate:
Fortunately I do not agree to "infinite samples without quantization.
first step (bit) is the comparator. what is the threshold of switching?
divide the amplitude of the carrier to obtain N.steps.
If selfoscillant, one of the problems (in an attempt to create an amp that sounds really good) is that the frequency shifts in relation to the modulation and consequently the amplitude of the carrier.
then the amplifier will not have a good linearity.
Another usual problem is the hysteresis, the audio signal and the resulting
MOSFET conduction (can not follow the modulation). this turns into thd that increases in proportion to the frequency of the audio signal.
In fact, any amp that I tested, even though the sound is different, have the exact same behavior.
thd increases in proportion to the modulation level and in relation to the frequency of the audio signal.
Apart from other defect in which.. we all know.
Regards
Roberto
replace the filter capacitor with a better ... it's useless. As Eva says
To Sampleaccurate:
Fortunately I do not agree to "infinite samples without quantization.
first step (bit) is the comparator. what is the threshold of switching?
divide the amplitude of the carrier to obtain N.steps.
If selfoscillant, one of the problems (in an attempt to create an amp that sounds really good) is that the frequency shifts in relation to the modulation and consequently the amplitude of the carrier.
then the amplifier will not have a good linearity.
Another usual problem is the hysteresis, the audio signal and the resulting
MOSFET conduction (can not follow the modulation). this turns into thd that increases in proportion to the frequency of the audio signal.
In fact, any amp that I tested, even though the sound is different, have the exact same behavior.
thd increases in proportion to the modulation level and in relation to the frequency of the audio signal.
Apart from other defect in which.. we all know.
Regards
Roberto
hi, interesting points of view!
I'm not the owner of a golden ears but judging from what I've heard. D class for me is playing something flat - without a dynamic and relatively dirty high frequency. It’s something like 5.1 multi channel receiver sound- I have a feeling that the power supply power is insufficient if there is small transformer, or it is too slow because of SMPS…. I have heard Rotel class D +SMPS and also 5.1 multi channel amps. They sound with some flatness for me…It may be that it is subjective – but for me the female vocals, classic and instrumental songs sounds best on class A iron pieces. I suspect it does not make sense for the holy war. One day, I also try to class D. but not yet. not now.
I'm not the owner of a golden ears but judging from what I've heard. D class for me is playing something flat - without a dynamic and relatively dirty high frequency. It’s something like 5.1 multi channel receiver sound- I have a feeling that the power supply power is insufficient if there is small transformer, or it is too slow because of SMPS…. I have heard Rotel class D +SMPS and also 5.1 multi channel amps. They sound with some flatness for me…It may be that it is subjective – but for me the female vocals, classic and instrumental songs sounds best on class A iron pieces. I suspect it does not make sense for the holy war. One day, I also try to class D. but not yet. not now.
osscar,
The top end aswell as the percieved dynamic contrast of my ICEPOWER amp do seem slightly flat too though it has a more than decent linear supply. Maybe the top end and the the reproduction of dynamic impulses of class D are the most difficult things to get right. On the other hand, the principle of class D do provide clear and undistorted rendition of music when well implemented, leaving most classic amp designs sound wolly and distorted in comparison...
My gut feeling is that output filter is a tremendous bottleneck, cutting ultrasonic harmonics and compressing the dynamics due to its energy storage.
cheers,
The top end aswell as the percieved dynamic contrast of my ICEPOWER amp do seem slightly flat too though it has a more than decent linear supply. Maybe the top end and the the reproduction of dynamic impulses of class D are the most difficult things to get right. On the other hand, the principle of class D do provide clear and undistorted rendition of music when well implemented, leaving most classic amp designs sound wolly and distorted in comparison...
My gut feeling is that output filter is a tremendous bottleneck, cutting ultrasonic harmonics and compressing the dynamics due to its energy storage.
cheers,
Oh, yes...
If it's not very heavy it's bad...
If it does not cost several thousand $$$$ it's bad...
If it's not big in size, it's bad...
If it is different from your audio fetish it must be bad...
Listening to music with headphones is bad...
MP3 is bad...
Anything except what you do is bad...
Because you never feel that you have enough self stem. You're always fighting to conquer external confirmation of your trends...
MP3 is like MPEG video. Image quality is far from perfect but this does not make any difference in the way you feel acting. In a movie, you know when a kiss or a caress is real and when it is fake, don't you?
Or are you too busy looking for pixel artifacts?
I don't need you at all to feel when what I do is good and when it isn't.
MP3 is very bad quality audio and i assume some uses it for there evaluation, understandable why they feel there is no difference between stupid hi-end stuff and low quality class-d ..
regards,
I've never hear the Sure Tripath 4x100 watt $50 board, but I have to admit that if you're on a budget that seems like an awesome price/performance ratio. I'm tempted to buy one just to hear it, although the 100watts is quoted at 10% distortion as I recall.
My only issue is that the frequency response specs aren't impressive. It appears to roll off the highs, but I can't figure out for sure (no pun intended).
I didn't say "infinite samples".
There are no discrete steps - period. You don't understand how a class D amp works. The threshold of switching occurs when the CONTINUOUS ANALOG program signal threshold exceeds the CONTINUOUS ANALOG output of the triangle wave generator. Go do some homework.
There is nothing discrete here.
Hi,
Right!, when it exceeds the threshold. when it is more or less, this is repeated with each cycle of the triangle at different levels of audio signal all the time.
then there's sampling, the linearity (triangular geometry and linearity). if you change the level (amplitude) of the triangle, then the exchange ratio (steps) this is the "gain ".
I do not think I need to study old things.
Last edited:
You studied on google, and when you got it right, then created a class D amplifier the same as already exist.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.