Hi Th3 uN1Qu3 ,
Ive check, Pin 6 on TL071 almost 0v, pin 3 LM311P is -8.5 vdc, pin7 LM311 i`ve got some DC there, i tried change LM311 and i have same result, i`ve 5 pcs with me LM311, try touching comparator there is nothing respond on speaker. only hear thumb and power on and off.
pin 3 LM311 goes to 1k resistor and connected to pin 6 TL071 and PIN2 LM311 to the feedback right?
Reagrds,
Azmi
Ive check, Pin 6 on TL071 almost 0v, pin 3 LM311P is -8.5 vdc, pin7 LM311 i`ve got some DC there, i tried change LM311 and i have same result, i`ve 5 pcs with me LM311, try touching comparator there is nothing respond on speaker. only hear thumb and power on and off.
pin 3 LM311 goes to 1k resistor and connected to pin 6 TL071 and PIN2 LM311 to the feedback right?
Reagrds,
Azmi
Hi Th3 uN1Qu3,
Thanks a lot for your info and to all who's helping me here, the problem is solved its my mistake it was my solder too much at pin 3 LM311 and its touch pin4 and thats the problem i`ve got -8.5v there, now i re-solder it back now everything is ok and amp is working now. Again Thanks a lot.
Best Regards,
Azmi.
Thanks a lot for your info and to all who's helping me here, the problem is solved its my mistake it was my solder too much at pin 3 LM311 and its touch pin4 and thats the problem i`ve got -8.5v there, now i re-solder it back now everything is ok and amp is working now. Again Thanks a lot.
Best Regards,
Azmi.
Hi Aucosticraft,
here is my conclusion about the noise, i`ve build ucd 25-1250 watts before, noise issue is depend on board design and important is very compact construction, but major i`ve found about noise in class D its because this amp is like pwm, working at high freq so its very sensitive. In my experience, when using linear supply the noise is very low but must meet the accuracy like correct inductance calc and right core, but when using smps its different story need many filter for it, like if your input source come from pc or any souce that using switching power supply then you will face many issue of noise.
For example, i`ve setup my class D with right core and inductance calc, and running with linear supply there is almost no noise at all, after that i`ve tried to connect with smps supply and noise is coming, same situation when using my pc sound card as sound source because pc power supply also is switching pwm, it may not happen if you`re using high quality pc supply because they have good dc filter on it and etc...
So this amp i`ve testing with linear supply and no big noise issue i`ve put 1nf at input souce to gnd to minimize it, almost no noise. from what i`ve tested so far and i`m not professional in class D and i`m still learning and thanks to DIY users here who helping me alot and share the experience together. About this amp supply +- 9V must be stable enough to drive if this voltage is not stable also you will get noise issue, LC filter and supply rail must be take care, and dont forget to make it compact design design as small as we can for class D its really lots diff between other class like AB or else.
I tried both T106-2 and ferrite core from pc supply as LC filter its good so far, no heat at coil and overheat at mosfet, i`ll take time to test at full volume with rail 28 0 28 vac at 6 amps each rails with linear supply lets see what happens, i do test at ucd 25-1200 before no problem it can handle it.
Regards,
Azmi
here is my conclusion about the noise, i`ve build ucd 25-1250 watts before, noise issue is depend on board design and important is very compact construction, but major i`ve found about noise in class D its because this amp is like pwm, working at high freq so its very sensitive. In my experience, when using linear supply the noise is very low but must meet the accuracy like correct inductance calc and right core, but when using smps its different story need many filter for it, like if your input source come from pc or any souce that using switching power supply then you will face many issue of noise.
For example, i`ve setup my class D with right core and inductance calc, and running with linear supply there is almost no noise at all, after that i`ve tried to connect with smps supply and noise is coming, same situation when using my pc sound card as sound source because pc power supply also is switching pwm, it may not happen if you`re using high quality pc supply because they have good dc filter on it and etc...
So this amp i`ve testing with linear supply and no big noise issue i`ve put 1nf at input souce to gnd to minimize it, almost no noise. from what i`ve tested so far and i`m not professional in class D and i`m still learning and thanks to DIY users here who helping me alot and share the experience together. About this amp supply +- 9V must be stable enough to drive if this voltage is not stable also you will get noise issue, LC filter and supply rail must be take care, and dont forget to make it compact design design as small as we can for class D its really lots diff between other class like AB or else.
I tried both T106-2 and ferrite core from pc supply as LC filter its good so far, no heat at coil and overheat at mosfet, i`ll take time to test at full volume with rail 28 0 28 vac at 6 amps each rails with linear supply lets see what happens, i do test at ucd 25-1200 before no problem it can handle it.
Regards,
Azmi
well about noise , Is it a part of EMI? inter-modulation of carrier frequency and input signals?
cause, i made a small note. when there is no input signal is connected the amp is super quiet. but the moment any signal is given at input. even 2Hz. this noise starts to appear.
I think snubber can lower noise. it will control over shoots at output. also i am thinking of multiple order filter. but taking feedback only after second order as we do now. this will lower the noise and keep basic ckt as it is. what do you say?
cause, i made a small note. when there is no input signal is connected the amp is super quiet. but the moment any signal is given at input. even 2Hz. this noise starts to appear.
I think snubber can lower noise. it will control over shoots at output. also i am thinking of multiple order filter. but taking feedback only after second order as we do now. this will lower the noise and keep basic ckt as it is. what do you say?
The input signal needs adequate low-pass filtering because any components above half the switching frequency will disturb the modulator. In poorly engineered circuits, radiated and conducted EMI will usually end up appearing at the input of the modulator, but only when a signal source is connected as you describe (unless the feedback path is picking up a lot of stuff too, then interference will be always there).
It's important to suppress any ringing because it involves producing EMI not just during the few dozen nanoseconds that a switching event takes to complete, but during a much longer time, which increases the chances of disturbing the modulator (or another channel) by introducing some error just when the comparator is determining that another switching event is required. Feedback can't correct that kind of noise pickup.
It's important to suppress any ringing because it involves producing EMI not just during the few dozen nanoseconds that a switching event takes to complete, but during a much longer time, which increases the chances of disturbing the modulator (or another channel) by introducing some error just when the comparator is determining that another switching event is required. Feedback can't correct that kind of noise pickup.
I agree Feedback can not correct noise. what I mean was , if we design 4th order filter instead of second order filter for speaker output. but the feedback will be given only from output of second order filter.
yes I will try this. might be the problem in my case.
Edit: Update: I connected typical RC filter and the noise was close to gone. that means it was pickup. now I have to try with audio in. but its late night here. can't pump my speakers. will test tomorrow.
Last edited:
I built a PCB for mine. The layout kinda sucks and i made some traces too thin, and forgot to change ExpressPCB's default clearance of 0,3mm, which means they've been a PITA to solder. But eventually i got it done.
Anyway. The linear regs i've bought SUCK. Another 7812 that starts only when it feels like it, first 7912 was dead, second 7912 decided to turn itself into a 7918. Great, isn't it? I'll be off to get a bunch more, maybe i find one that works. Btw i also think i've found the reason for the rather low 8v2 zeners on the mosfet gates - the 2k2 resistor doesn't pull the LM311 high enough. It needs 1k or lower as someone already mentioned before - but you will see no actual difference unless you use 12 volt zeners. I don't know if the difference in efficiency is worth it, there is one in simulation, in the real world it works about the same.
Anyway. The linear regs i've bought SUCK. Another 7812 that starts only when it feels like it, first 7912 was dead, second 7912 decided to turn itself into a 7918. Great, isn't it? I'll be off to get a bunch more, maybe i find one that works. Btw i also think i've found the reason for the rather low 8v2 zeners on the mosfet gates - the 2k2 resistor doesn't pull the LM311 high enough. It needs 1k or lower as someone already mentioned before - but you will see no actual difference unless you use 12 volt zeners. I don't know if the difference in efficiency is worth it, there is one in simulation, in the real world it works about the same.
It works good now. the input noise was the switching noise picked up by input wire.
I made this PCB for 7812/7912 regs, i'll keep it that way for now. I bought ST and Fairchild parts today, they work fine. The ones i had and worked poorly were from Taiwan Semiconductor - so avoid that maker.
I also discovered something that could explain the lack of performance some people have been getting from this circuit but it needs more thorough checking. I'll post about it later.
I also discovered something that could explain the lack of performance some people have been getting from this circuit but it needs more thorough checking. I'll post about it later.
Hi all,
Hi Th3 uN1Qu3, i do experiment myself with this baby amp, the experiment is about when we increase the volume then the spark distort will be hear from it, i`ve found the problem is unstable supply for +- 9v and voltage swing from main supply will produce distort when volume increase, its because i do regulate it from main supply, i use +-45 vdc mosfet is irf540 and 9540, so refer on m-tech schematic im using tl071 op amp and lm311 as comparator.
i do experiment on regulator section, i`ve build few model regulator supply using lm7809 and 7909, using zener 9.1v feed with 1.2k 3w resistor, esp model regulator using tip41c and tip42c and with high current regulator lm7809 with tips transistor... etc... lol
and for the result i do eliminate the distortion, it took me 1 week for it lol.. first i do regulate it from main supply i do test one by one regulator that i`ve build all comes with distortion result just the different it i can lower the distort. the best is 3 1k 3w resistor feed with 9.1v zener diode. second place is using tip 41 and 42. But still get spark distort can hear that lol. i`m almost no idea what i can do with it, so i do search for similar schematic and check it also experiment current draw by tl071, lm311 and 2n5551/2n5401. i`ve made few mod with the schematic, i put 1.2k on R4, adding 1nf caps after C1 to ground, R8 and R9 is 100 ohm, and i put 100nf at R5. After that i build another source supply from +-24Vdc with 500ma current each and regulate it with zener diode 9.1v to provide +-9vdc and the spark distort is totally gone, so i increase the volume still no distort, no over heat at mosfet P and N channel, i use T106-2 for the LC filter. Main supply for the mosfet is +-45 vdc at 5 amp. really loud lol.
thats so far for my test, i think other ppls may have another better solution for it, mine still looks ugly solution, but i`m happy with it at last i can eliminate it , test run it 6 hours with moderate volume its running smooth feed 12 inch mohawk speaker.
Regards,
Azmi.
Hi Th3 uN1Qu3, i do experiment myself with this baby amp, the experiment is about when we increase the volume then the spark distort will be hear from it, i`ve found the problem is unstable supply for +- 9v and voltage swing from main supply will produce distort when volume increase, its because i do regulate it from main supply, i use +-45 vdc mosfet is irf540 and 9540, so refer on m-tech schematic im using tl071 op amp and lm311 as comparator.
i do experiment on regulator section, i`ve build few model regulator supply using lm7809 and 7909, using zener 9.1v feed with 1.2k 3w resistor, esp model regulator using tip41c and tip42c and with high current regulator lm7809 with tips transistor... etc... lol
and for the result i do eliminate the distortion, it took me 1 week for it lol.. first i do regulate it from main supply i do test one by one regulator that i`ve build all comes with distortion result just the different it i can lower the distort. the best is 3 1k 3w resistor feed with 9.1v zener diode. second place is using tip 41 and 42. But still get spark distort can hear that lol. i`m almost no idea what i can do with it, so i do search for similar schematic and check it also experiment current draw by tl071, lm311 and 2n5551/2n5401. i`ve made few mod with the schematic, i put 1.2k on R4, adding 1nf caps after C1 to ground, R8 and R9 is 100 ohm, and i put 100nf at R5. After that i build another source supply from +-24Vdc with 500ma current each and regulate it with zener diode 9.1v to provide +-9vdc and the spark distort is totally gone, so i increase the volume still no distort, no over heat at mosfet P and N channel, i use T106-2 for the LC filter. Main supply for the mosfet is +-45 vdc at 5 amp. really loud lol.
thats so far for my test, i think other ppls may have another better solution for it, mine still looks ugly solution, but i`m happy with it at last i can eliminate it , test run it 6 hours with moderate volume its running smooth feed 12 inch mohawk speaker.
Regards,
Azmi.
Indeed it gets ugly distortion at high volume especially on bass, i thought a limiter was required but from what you say maybe not. I'll check the comparator and drivers supply.
My current setup has R8/R9 at 100ohm too but turn-off is too slow and there is a bit of cross conduction, i have to experiment. R4 is already at 1k in my circuit as far as i remember, and i'm still using 7812 and 7912 regs. My earlier mistake of putting the 1N4148s in backwards taught me a lesson or two about regulator output caps.
Also, what's your DC offset at the output? Mine sits at a few hundred millivolts all the time. I haven't connected the TL071's offset trim, can it be corrected with that or would a servo circuit be better? Ah, and i thought i would have a peaceful night. You made me whip out the soldering iron again.
My current setup has R8/R9 at 100ohm too but turn-off is too slow and there is a bit of cross conduction, i have to experiment. R4 is already at 1k in my circuit as far as i remember, and i'm still using 7812 and 7912 regs. My earlier mistake of putting the 1N4148s in backwards taught me a lesson or two about regulator output caps.
Also, what's your DC offset at the output? Mine sits at a few hundred millivolts all the time. I haven't connected the TL071's offset trim, can it be corrected with that or would a servo circuit be better? Ah, and i thought i would have a peaceful night. You made me whip out the soldering iron again.
hi Th3 uN1Qu3,
i`m sorry for bother you tonite lol, ah okay, i do experement too when what the diff between supply voltage at tl071,lm311 and 2n5551/2n5401, i`ve tried with +-8,+-9,+-10,+-12 vdc, when the supply is over 10vdc, the trans both 2n series will getting hot, as for mosfet too will get fast hot when play loud, then im lowering back to 9.0++ vdc it works great, no overheat. offset is about 200++ milivolts at idle.
Today i`ve tried play really loud with mohawk 12 inch and the speaker cone start getting hot, its not bad for the cheap amp i think, its suitable for the topic thread.
regards,
azmi.
i`m sorry for bother you tonite lol, ah okay, i do experement too when what the diff between supply voltage at tl071,lm311 and 2n5551/2n5401, i`ve tried with +-8,+-9,+-10,+-12 vdc, when the supply is over 10vdc, the trans both 2n series will getting hot, as for mosfet too will get fast hot when play loud, then im lowering back to 9.0++ vdc it works great, no overheat. offset is about 200++ milivolts at idle.
Today i`ve tried play really loud with mohawk 12 inch and the speaker cone start getting hot, its not bad for the cheap amp i think, its suitable for the topic thread.
regards,
azmi.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.