Oki, good.
I have an output filter problem, no matter what amount of turns i have i get a ugly sinewave with almost 90v amplitude with the same freq as the oscillator if i have only a subwoofer connected or no load at all.
With a speaker it reduces to 19 volts, the coil also heats up no matter what.
I have tried 17, 15, 14 and 13 turns with the same result, also 24 turns but then the amp wouldent work at all.
According to the micrometals coil calc proggy around 15 turns is the right choice with the cores i got, but thats not what the amp thinks
Also the amp will not work with voltages over 2x17 volts, above it just makes nasty noises as well as heating up very fast.
It also suffers from dc offset drift, which none of the prototypes had, tho they ran on sigle supply and would blow up if u tried split supply because of a stupid wrong connection. When you cold start the amp it puts out over a volt, either positive or negative so i have to adjust it to a few millivolts before i connect the speakers directly as i have to use a cap during startup, but after just minutes its back at around 120-150 millivolts in either direction but i figure that is the coil heating up and then cooling off in cycles when the resistance changes from the heat.
I dunno if it is where i chose to take feedback that causes it to act like this and if it may do better if i move the feedback to before the op filter as Ledmania had in one of the schematics.
I have an output filter problem, no matter what amount of turns i have i get a ugly sinewave with almost 90v amplitude with the same freq as the oscillator if i have only a subwoofer connected or no load at all.
With a speaker it reduces to 19 volts, the coil also heats up no matter what.
I have tried 17, 15, 14 and 13 turns with the same result, also 24 turns but then the amp wouldent work at all.
According to the micrometals coil calc proggy around 15 turns is the right choice with the cores i got, but thats not what the amp thinks
Also the amp will not work with voltages over 2x17 volts, above it just makes nasty noises as well as heating up very fast.It also suffers from dc offset drift, which none of the prototypes had, tho they ran on sigle supply and would blow up if u tried split supply because of a stupid wrong connection. When you cold start the amp it puts out over a volt, either positive or negative so i have to adjust it to a few millivolts before i connect the speakers directly as i have to use a cap during startup, but after just minutes its back at around 120-150 millivolts in either direction but i figure that is the coil heating up and then cooling off in cycles when the resistance changes from the heat.
I dunno if it is where i chose to take feedback that causes it to act like this and if it may do better if i move the feedback to before the op filter as Ledmania had in one of the schematics.
Tekko...
Your material is not good if it heat...Try MicroMetal material -2 or -6 if you can. Nothing else of micrometal work. If you raise the value of your coil and amplifier stop to work, that's because your feedback loop is not designed correctly. Try to reduce closed loop gain, this should fix all your probleme of offset and DC unstability.
As I told you last time, remove the cap between feedback point and ground and add a capacitor over your feedback trimpot (need to find right value depend of the value of your pot).
A good way to find right value for your trim pot is to add 1uF to the load, with no capacitor in you feedback loop, then remove the 1uF and find the value to have stability without load. Dont forget your zobel circuit to avoid oscilation.
Fredos
www.d-amp.com
Your material is not good if it heat...Try MicroMetal material -2 or -6 if you can. Nothing else of micrometal work. If you raise the value of your coil and amplifier stop to work, that's because your feedback loop is not designed correctly. Try to reduce closed loop gain, this should fix all your probleme of offset and DC unstability.
As I told you last time, remove the cap between feedback point and ground and add a capacitor over your feedback trimpot (need to find right value depend of the value of your pot).
A good way to find right value for your trim pot is to add 1uF to the load, with no capacitor in you feedback loop, then remove the 1uF and find the value to have stability without load. Dont forget your zobel circuit to avoid oscilation.
Fredos
www.d-amp.com
Tekko:
You are wasting a lot of time and components due to your extreme lack of switching amplifier understanding.
Your output filter is resonating at a frequency too close to the switching frequency, that's why you get such a huge sinewave on the output.
You have to redesign the filter to make it resonate just above the highest audio frequency you want to play. Some RC damping may be useful to reduce Q. Also, you have to chose a switching frequency high enough in order to avoid exciting output filter resonance.
That's just switching mode power supply basics. Calculations and real inductance measurements are required in order to get the filter right, so forget about throwing components randomly at the circuit or using arbitrary turn counts.
You are wasting a lot of time and components due to your extreme lack of switching amplifier understanding.
Your output filter is resonating at a frequency too close to the switching frequency, that's why you get such a huge sinewave on the output.
You have to redesign the filter to make it resonate just above the highest audio frequency you want to play. Some RC damping may be useful to reduce Q. Also, you have to chose a switching frequency high enough in order to avoid exciting output filter resonance.
That's just switching mode power supply basics. Calculations and real inductance measurements are required in order to get the filter right, so forget about throwing components randomly at the circuit or using arbitrary turn counts.
The turn counts came from the micrometals coil proggy.
Even with 25kHz enterd in the proggy i still get the same turn count of 15 turns For cap i have a 0.47µF cap as C of the op filter.
the only cap in the feedback line is a 100pF cap after the pot (that should be after the 22k resistor between the pot wiper and the audio in on the pwm), between its wiper and ground, but if it has to go, its easy to remove. The feedback pot is a 10k trimpot. When i set the feedback pot so the wiper has less resistance to ground than to the output the gain skyrockets, and i have co compensate with the Vref pot, but it still drifts in dc offset.
None of the earlier prototypes drifted in dc offset, nor heated coils very mutch. Tho i never ran them @ split supply either.
I have a 1.2µF polyprop (or whatever it is) cap that i could connect across the load.
For a quiet load i use a busted tube that measures 2 ohms when cold, that has a filament voltage of 25 volts or so.
The amps switching freq is in the neighbourhood of 300kHz.
fredos, i want you on msn!!
Eva, u can only learn in one way, doing it the hard way (trial and error) I have came too far to stop now.
Even with 25kHz enterd in the proggy i still get the same turn count of 15 turns
For cap i have a 0.47µF cap as C of the op filter.the only cap in the feedback line is a 100pF cap after the pot (that should be after the 22k resistor between the pot wiper and the audio in on the pwm), between its wiper and ground, but if it has to go, its easy to remove. The feedback pot is a 10k trimpot. When i set the feedback pot so the wiper has less resistance to ground than to the output the gain skyrockets, and i have co compensate with the Vref pot, but it still drifts in dc offset.
None of the earlier prototypes drifted in dc offset, nor heated coils very mutch. Tho i never ran them @ split supply either.
I have a 1.2µF polyprop (or whatever it is) cap that i could connect across the load.
For a quiet load i use a busted tube that measures 2 ohms when cold, that has a filament voltage of 25 volts or so.
The amps switching freq is in the neighbourhood of 300kHz.
fredos, i want you on msn!!
Eva, u can only learn in one way, doing it the hard way (trial and error) I have came too far to stop now.
Tekko: I saw your designpage on hifi-musik.dk,which is the leading new forum in Denmark. At least you have a cool taste in music!
I think you will eventually get there, and get around to a good result. But it does take a lot of hard work, especially if you want to make a selling product.
If i can give you one advice, it would be to get a nice workbench, you will end up hurting your back, if you are working too much on that shelfpiece.
Good luck!
Lars
PS.. EVA: i have been following your posts for a while, you seem to becoming a significant factor in output filter expertise. Are you interested in some consultant work?
I think you will eventually get there, and get around to a good result. But it does take a lot of hard work, especially if you want to make a selling product.
If i can give you one advice, it would be to get a nice workbench, you will end up hurting your back, if you are working too much on that shelfpiece.
Good luck!
Lars
PS.. EVA: i have been following your posts for a while, you seem to becoming a significant factor in output filter expertise. Are you interested in some consultant work?
Eva: I can't because it seems to be blocked. Can you e-mail me on: lc@newclassd.com? Thanks
As i have been unable to find the right coil inductance, i gonna lift the RT resistor and install a pot wires as a variable resistor so i can adjust the oscillator freq instead and that way find the notch where the ugly sine reduces to a minimum or goes away completely.
Would the amp perform better with a subwoofer if i manage to murder the ugly sine ?
Would the amp perform better with a subwoofer if i manage to murder the ugly sine ?
Compensation and phase correction could only be reach if you remove the cap to ground in your feedback point and replce it with a cap over your trim pot. You should cloe the gain to X1 with out probleme if the compensation is done in the right way.
Fredos
www.d-amp.com
Fredos
www.d-amp.com
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.