Hello guys, im a bit new to this forum!
So i am currenty developing a feedbackless Hi Fi class-D amplifier! Here is the schematic: http://ftp.torespont.hu/solarian/classd/kapcs-uj.JPG My problem is that, that the LM319 comparator picks up really lot of noise(oscillating or sy)! Ive disconnected the output of the comparator from the 4070, totally grounded the audio_in pin, and the noise is still there! The Triangle wave is almost perfect, no dc offset, but the square wave's duty cycle at the output of the lm319 is not 50%!
Here is the PCB: http://ftp.torespont.hu/solarian/classd/uj.jpg
So i am currenty developing a feedbackless Hi Fi class-D amplifier! Here is the schematic: http://ftp.torespont.hu/solarian/classd/kapcs-uj.JPG My problem is that, that the LM319 comparator picks up really lot of noise(oscillating or sy)! Ive disconnected the output of the comparator from the 4070, totally grounded the audio_in pin, and the noise is still there! The Triangle wave is almost perfect, no dc offset, but the square wave's duty cycle at the output of the lm319 is not 50%!
Here is the PCB: http://ftp.torespont.hu/solarian/classd/uj.jpg
Hello Solarian
Maybe it's because the LM319 has about 50 nS shorter delay for a positive transition, compared to the negative transition.
It's hard to read your switching frequency, when we don't know the value of R4 ..
Maybe it's because the LM319 has about 50 nS shorter delay for a positive transition, compared to the negative transition.
It's hard to read your switching frequency, when we don't know the value of R4 ..
The IR drivers are noisy things. You should put them at low distance to the fets with some distance to the comparator.
And use a RC-lowpass for the logic & comparator supply (or the IR supply, but don't tie them together like you do it now).
digi
And use a RC-lowpass for the logic & comparator supply (or the IR supply, but don't tie them together like you do it now).
digi
Hi,
Good luck! Sounds like an ambitious project, it may not be all that high fi but it will be cool and you'll learn alot.
You haven't described this "noise" in any great detail.
What Lars Clausen describes makes alot of sense, coupled along with real world inputs and parasitics you're liable to have some DC offset as well which will contribute to an error in duty cycle, and Workhorse's idea should help with that in order to give the input a good dc operating point.
What I think you are likely to be a victim of is a slowly transitioning signal(s) on the input of a fairly fast comparator causing multiple/spurrious switching because it's having a hard time deciding which way to go while the inputs are close to one another.
You might try adding some positive feedback (hysteresis) around your 319 to help with this, it's a schmitt trigger is is somewhat useful in class d.
Looks like the perfect circuit to start with. Keep posted.
Regards,
Chris
Good luck! Sounds like an ambitious project, it may not be all that high fi but it will be cool and you'll learn alot.
You haven't described this "noise" in any great detail.
What Lars Clausen describes makes alot of sense, coupled along with real world inputs and parasitics you're liable to have some DC offset as well which will contribute to an error in duty cycle, and Workhorse's idea should help with that in order to give the input a good dc operating point.
What I think you are likely to be a victim of is a slowly transitioning signal(s) on the input of a fairly fast comparator causing multiple/spurrious switching because it's having a hard time deciding which way to go while the inputs are close to one another.
You might try adding some positive feedback (hysteresis) around your 319 to help with this, it's a schmitt trigger is is somewhat useful in class d.
Looks like the perfect circuit to start with. Keep posted.
Regards,
Chris
Hello classd4sure!
Thank you all for these infos, ill try the positive feedback today! The noise is like when you tune an old radio tuner! So sometimes the noise is there (not whitenoise) and sometimes not!
Thank you all for these infos, ill try the positive feedback today! The noise is like when you tune an old radio tuner! So sometimes the noise is there (not whitenoise) and sometimes not!
You might use an input lowpass RC filter with pole at less than 100kHz to avoid any 'parasitic' modulation in a comparator.
- Status
- Not open for further replies.