I have tested the prototype with transformer and switching power supply...the distorsions are the same..
It is clear now that switching noise is getting into the analog stage..but i do not understand why..
It is clear now that switching noise is getting into the analog stage..but i do not understand why..
salut incearca sa pui in reteaua de reactie condensator polipropilena si rezistoare film metal 1%
hello try to place polypropylene capacitor and metal film rezistors in the feedback network
hello try to place polypropylene capacitor and metal film rezistors in the feedback network
Update(5)
At higher frequencies i can get lower power before heavy distorsion begins.
I get 8 amps at 50 hz and 4 amps at 20khz max..it sounds like positive feedback but i tried every filtering method i could think of in the singal path with no succes...😕
At higher frequencies i can get lower power before heavy distorsion begins.
I get 8 amps at 50 hz and 4 amps at 20khz max..it sounds like positive feedback but i tried every filtering method i could think of in the singal path with no succes...😕
Decreasing inductance increases the point(in Amps) where abnormal distorsion and noise begins.With 10uH that zone is near clipping at 18kz and 2 ohm load.With 1 ohm load and 15khz,18khz and beyond i still have problems and I do not want to decrease inductance any further.
distorsion begins here far beyond clipping level - 18khz 1 ohm load.
distorsion begins here far beyond clipping level - 15khz 1 ohm load.
distorsion begins here at clipping level for 18khz 2 ohm load
no distorsion at clipping level for 18khz 4 ohm load
So..as load impedance decreases and frequency increases heavy distorsion(white noise and lower frequency sounds) apear bofore clipping level..why?
distorsion begins here far beyond clipping level - 18khz 1 ohm load.
An externally hosted image should be here but it was not working when we last tested it.
distorsion begins here far beyond clipping level - 15khz 1 ohm load.
An externally hosted image should be here but it was not working when we last tested it.
distorsion begins here at clipping level for 18khz 2 ohm load
An externally hosted image should be here but it was not working when we last tested it.
no distorsion at clipping level for 18khz 4 ohm load
An externally hosted image should be here but it was not working when we last tested it.
So..as load impedance decreases and frequency increases heavy distorsion(white noise and lower frequency sounds) apear bofore clipping level..why?
Last edited:
Question:
Do you think changing lm311 with lt1016 will decrease distorsions?
It has 10ns response time.
Do you think changing lm311 with lt1016 will decrease distorsions?
It has 10ns response time.
The waveforms are exactly at the output..
Tomorow i will try to do a fft but i do not know if the noise will be visible on it.
Tomorow i will try to do a fft but i do not know if the noise will be visible on it.
Promised FFT:
No load:
8A load 1 ohm way before clipping level
No load:
An externally hosted image should be here but it was not working when we last tested it.
8A load 1 ohm way before clipping level
An externally hosted image should be here but it was not working when we last tested it.
So what is the problem here?
I can see only two things, one being, your switching freq. is low and other would be wrong size of filter for that freq.
Or your amp has oscillations, which won't go away if you won't do same big change
I can see only two things, one being, your switching freq. is low and other would be wrong size of filter for that freq.
Or your amp has oscillations, which won't go away if you won't do same big change
Switching frequency is 360khz..but there is big residual(1v rms) because i have only 10uH inductance.
About oscilations..what big changes do you suggest?
About oscilations..what big changes do you suggest?
anything from schematic level of change, maybe only some parts change... and can be the PCB... higher the freq. more important pcb is and components too, slow, 20years old components really don't work good with higher freq.
residual is not a problem, that is normal for this type of amp, but "residual" on residual is not something you would expect to be there, should be pure sine . what is your cutoff freq. of output filter?
residual is not a problem, that is normal for this type of amp, but "residual" on residual is not something you would expect to be there, should be pure sine . what is your cutoff freq. of output filter?
Residual is almost pure sine...the image has the 18khz signal on it..
I am sorry but i need more specific advice if you can give it to me..please read the last 3-10 pages for aditional information if you can.
I am gratefull for any ideea you may have.
I am sorry but i need more specific advice if you can give it to me..please read the last 3-10 pages for aditional information if you can.
I am gratefull for any ideea you may have.
lorylaci: +-6.5 V since i am using gain 10x..but the problem is the same with +-3 volts
luka: yes..those tests were with 18khz input signal..
Distorsion starting point is only influenced by output inductance.
-more inductance:distorsion starts at lower current.
-less inductance:distorsion starts at higher current.
luka: yes..those tests were with 18khz input signal..
Distorsion starting point is only influenced by output inductance.
-more inductance:distorsion starts at lower current.
-less inductance:distorsion starts at higher current.
Last edited:
You should not operate the LM311 over +-5 V, since 2N5904 can only take 5V at base safely. Higher voltage will make the LM311 operate slower, since higher voltage swing it will need to produce.
The output LC filter should be calculated for the load impedance. This is especially true for signal near to corner frequency. The UcD setup helps you getting down peaks and dips near corner frequency, but at the cost of phase responses etc... which will shrink your modulation limit.
Since this is a Class-D amplifier it cannot be controlled up to voltage rails, because modulation error starts at the 70-90% of the voltage rail. It depends on propagartion delay, whether you can take it up to 70 or 90%. If you change the LM311 to a much faster comparator, you will spare a lot of prop delay. Changing IR2110 ot IR2010, and making a faster level shift will decrease prop delay too. If you lower the switching frequency by changing feedback you can increase modulation limit too (but at the expense of higher switching residual).
Ok..thanks lorylaci for those details.
So it is normal for the original circuit to not work in 1 ohm load /high current and frequency.
I will try changing lm311 to lm319 or lt1016 and see what happens.
Ir2010 is only available here in smd package....
So it is normal for the original circuit to not work in 1 ohm load /high current and frequency.
I will try changing lm311 to lm319 or lt1016 and see what happens.
Ir2010 is only available here in smd package....
The original 30uH, 1,5uF is for 4 Ohm load, for 1 Ohm load the recalculated filter values are 7,5uH and 7,5uF (for 30kHz corner freq, a good compromise). So it is normal that for that load you need lower inductance value.
Change the switching frequency from 360kHz to about 250kHz. Then you can get out more power.
Attached you will find formulas for butterworth filter calculation.
