Looks exactly like the diagram on the data sheet, good stuff.
You mean the null adjustment I take it, just so we're clear and it's not the pot for adjusting frequency. What I find troubling is that you're only getting 0 to -5. It looks like it should operate from 5 to -5, I assume this is bothering you as well.
Check for a good 5 volts at pin 7 (VCC) if you have five volts there, I know you already tried adjusting the offset to no avail, why not try removing the pot, and run it, see what happens?
It isn't essential to operation anyway.
All I can think of maybe it is shorting your VCC to the offset pins or something.
While it's out you can double check the pot for proper operation as well, and make sure it's OK and is in/oriented properly.
Or see what kind of voltage drop you get across it as well.
You mean the null adjustment I take it, just so we're clear and it's not the pot for adjusting frequency. What I find troubling is that you're only getting 0 to -5. It looks like it should operate from 5 to -5, I assume this is bothering you as well.
Check for a good 5 volts at pin 7 (VCC) if you have five volts there, I know you already tried adjusting the offset to no avail, why not try removing the pot, and run it, see what happens?
It isn't essential to operation anyway.
All I can think of maybe it is shorting your VCC to the offset pins or something.
While it's out you can double check the pot for proper operation as well, and make sure it's OK and is in/oriented properly.
Or see what kind of voltage drop you get across it as well.
BTW, there's a good chance it's working fine and I just dont' understand why it's only operating from 0 to -5, but usually if there's symetrical rails there's a symetrical output right? That's all I base this assumption off.
Anyway, that bias adjustment is just to null the offset, and I doubt that amplifier would have much offset that would require nulling. You'd probably have to fine tune that with output stage disabled and input grounded, then zero it, just another guess though. I don't think you should see any obvious change from tuning it.
Anyway, that bias adjustment is just to null the offset, and I doubt that amplifier would have much offset that would require nulling. You'd probably have to fine tune that with output stage disabled and input grounded, then zero it, just another guess though. I don't think you should see any obvious change from tuning it.
yeah, it bugs me to not have +5 -5 swing at the output of the opamp, but at least, it works damn good, it fired up right away.
all the voltages to the op amp are fine. No matter how it is set, present or not, it dont change a thing. Yes, i may disable its input/output to see how it is at rest with only its Vcc...
I have raised the power from 20 to 30 volts rails, now i get right on 400khz, with dead on switching times. Perfect.
Now, hows it sounds...well, it sounds very good, its all i can say for now. I can bottom out a 12inch woofer rated at 100wrms.
I did made it play very loud, it gets barely hot. Only the inductor gets hot, even if it dont play, i think it is normal, isnt?
My plans is to build another channel like it, and bridge it. Running one channel out of phase will partialy cancel the pumping effect that i can cleverly see on the needle of my power supply. It pumps it goes from 30 to 35 volts rails at low freqs.
One thing that you may be aware of, its the turn on time controled by a timer (Q3 in my diagram, there is a mistake also there, C4 goes to ground, not to -vcc) so the output of the amp goes to -vcc close to 1 secs, a muting relay is totaly necessary. it is to charge the bootstrap cap, as stated into the document of IR.
I will build then an external protection board with dc detect.
I am happy, it works very good, with very few efforts. What to expect more...
all the voltages to the op amp are fine. No matter how it is set, present or not, it dont change a thing. Yes, i may disable its input/output to see how it is at rest with only its Vcc...
I have raised the power from 20 to 30 volts rails, now i get right on 400khz, with dead on switching times. Perfect.
Now, hows it sounds...well, it sounds very good, its all i can say for now. I can bottom out a 12inch woofer rated at 100wrms.
I did made it play very loud, it gets barely hot. Only the inductor gets hot, even if it dont play, i think it is normal, isnt?
My plans is to build another channel like it, and bridge it. Running one channel out of phase will partialy cancel the pumping effect that i can cleverly see on the needle of my power supply. It pumps it goes from 30 to 35 volts rails at low freqs.
One thing that you may be aware of, its the turn on time controled by a timer (Q3 in my diagram, there is a mistake also there, C4 goes to ground, not to -vcc) so the output of the amp goes to -vcc close to 1 secs, a muting relay is totaly necessary. it is to charge the bootstrap cap, as stated into the document of IR.
I will build then an external protection board with dc detect.
I am happy, it works very good, with very few efforts. What to expect more...
AHA!
So with R8 (on your schematic) having a DC grounded base, when the output of the op amp is high, the collector of Q3 is reversed bias, turning the transistor on and shunting your op amps high output. When the op amp swings low, below ground, BC then gets forward biased and it can no longer act as a transistor, turns off, and you get your negative swing.
Make sense? I think if you fix that error you spotted you'll have a symetrical output on the op amp. Or maybe I just need a drink I dunno you figure it out 🙂
Yeh the inductor will warm up, there's some idle current in it whenever it is switching because of the dead time, the more dead time the more idle current the more heat, I think.
Hey, nice work on this.
Yeah, but there is still R5, its a 1k, it should be high enough so the output of the opamp should be able to go thru its +vcc swing?
duno...it works anyway, but i will work around that, just to be sure.
😀
I managed to get 55 volts peak to peak without realy pushing hard with a 1khz sine wave, 4 ohm load, undistorted, 30 volts rails...
I will build a better power supply tomorrow, wich will give around 40 volts rails and plenty of current. I will push it to the death
we will see how good it is then.
duno...it works anyway, but i will work around that, just to be sure.
😀
I managed to get 55 volts peak to peak without realy pushing hard with a 1khz sine wave, 4 ohm load, undistorted, 30 volts rails...
I will build a better power supply tomorrow, wich will give around 40 volts rails and plenty of current. I will push it to the death
we will see how good it is then.
Yep being sure is good.
You mentioned pumping, bridging will totally cancel it, hopefully the circuit can handle the extra voltage/current??
Tuning for zero dc offset will be good with respect to pumping, 5 volts pumping seems like alot, but it depends on your caps too. Also if your op amp is indeed only working on the negative swing, and if it isn't supposed to, that could explain it as well. Try to use at least 10 000uF in the power supply caps and have the voltage rated at least 20% over the rails for a safety margin and pumping won't be a problem.
Anyway, consider the woofer in question. I have no idea what it is rated for, but since they used such high voltage with it, I have a feeling it might be like such high current a real amp will deliver. Let's see what smokes first 🙂
You mentioned pumping, bridging will totally cancel it, hopefully the circuit can handle the extra voltage/current??
Tuning for zero dc offset will be good with respect to pumping, 5 volts pumping seems like alot, but it depends on your caps too. Also if your op amp is indeed only working on the negative swing, and if it isn't supposed to, that could explain it as well. Try to use at least 10 000uF in the power supply caps and have the voltage rated at least 20% over the rails for a safety margin and pumping won't be a problem.
Anyway, consider the woofer in question. I have no idea what it is rated for, but since they used such high voltage with it, I have a feeling it might be like such high current a real amp will deliver. Let's see what smokes first 🙂
Ok, i did some real sound test with it.
The source is a Luxman tube cd player and the speaker is a Quest tower (2x 8 inches+1 tweeter, tweaked crossover)
It does sound very good, better than my Sony ES amplifier/receiver. It is very dynamic, and overall linear. My fear was a "chineese" sounds, like when the first cheap cd player came to birth many years ago. The sound is very pleasant. I am no more convinced that i will put this baby in a subwoofer, it plays too good !!!!😀
I didnt rework the op amp, it plays for now since 8 am, thats it.
The source is a Luxman tube cd player and the speaker is a Quest tower (2x 8 inches+1 tweeter, tweaked crossover)
It does sound very good, better than my Sony ES amplifier/receiver. It is very dynamic, and overall linear. My fear was a "chineese" sounds, like when the first cheap cd player came to birth many years ago. The sound is very pleasant. I am no more convinced that i will put this baby in a subwoofer, it plays too good !!!!😀
I didnt rework the op amp, it plays for now since 8 am, thats it.
I hope you still intend to fix it though!
It would probably sound even better.
Anyway, welcome to self oscillating class D🙂 You'll never want to mess with AB after that. Fun stuff or what? I kind of thought it would be too good for just a sub, glad you gave it a decent sound test.
It would probably sound even better.
Anyway, welcome to self oscillating class D🙂 You'll never want to mess with AB after that. Fun stuff or what? I kind of thought it would be too good for just a sub, glad you gave it a decent sound test.
There is still nothing that proves me that it is not ok...i cant believe no one ever try this amp.
I figured that i dont have +5 -5 at the op amp, but +4.67 -4.67, because i have misplaced my zeners with bad numbers, they are 5v now while they are suposed to be 5.6v...no big deal, it does play now 5th symphony to the hell and it still impress me a lot.
I did compare it with my pass labs repro, no way it sounds far better.
I figured that i dont have +5 -5 at the op amp, but +4.67 -4.67, because i have misplaced my zeners with bad numbers, they are 5v now while they are suposed to be 5.6v...no big deal, it does play now 5th symphony to the hell and it still impress me a lot.
I did compare it with my pass labs repro, no way it sounds far better.
some updates on this project.
Overall, there was no problem, just that it needs more than 25v to work per the specs. So now the op amp output swing +5 -5 triangle wave. Maybe there was not enough feedback to make it swing.
Some considerations about the -vcc+12v supply to the IR2011, it have to be strong and independant. This is why there is a DC-DC supply on the original design. Mine is "ok", but some hf stick to this vcc at high power output, wich gives some distortion at high power. I will work on that, i have done it fast and i didnt take car of it enough, but no big deal, i already know what to do.
Performance is still impressive, and sound quality is more than "oh, well, it sounds good".
i have experimented buss pump, it acts ok and is previsible, it just clip and then restart its oscillation. Adding big caps to the supply simply solve this issue, 2200uf starting point was already good enough to make it run 100% of its capacity. Final desing will include more than that for sure (8x 33 000uf)
I managed to run it with a 4ohm load at 25v rails. I did get a clean signal with close 25v peaks so it is very close to its 93% eff claimed. that gives then (25*.707)2 /4=78wrms. It gets barely hot, no need for a bigger heat sink.
The next step is to get proper voltage caps, and power it up with 38volts rails power supply, power supply that i have handy.
Note: it is designed to handle 50 volts rails, max 60.
50 volts @4ohm=312wrms
60 volts @4ohm=449wrms
(±400 @90% eff)
Overall, there was no problem, just that it needs more than 25v to work per the specs. So now the op amp output swing +5 -5 triangle wave. Maybe there was not enough feedback to make it swing.
Some considerations about the -vcc+12v supply to the IR2011, it have to be strong and independant. This is why there is a DC-DC supply on the original design. Mine is "ok", but some hf stick to this vcc at high power output, wich gives some distortion at high power. I will work on that, i have done it fast and i didnt take car of it enough, but no big deal, i already know what to do.
Performance is still impressive, and sound quality is more than "oh, well, it sounds good".
i have experimented buss pump, it acts ok and is previsible, it just clip and then restart its oscillation. Adding big caps to the supply simply solve this issue, 2200uf starting point was already good enough to make it run 100% of its capacity. Final desing will include more than that for sure (8x 33 000uf)
I managed to run it with a 4ohm load at 25v rails. I did get a clean signal with close 25v peaks so it is very close to its 93% eff claimed. that gives then (25*.707)2 /4=78wrms. It gets barely hot, no need for a bigger heat sink.
The next step is to get proper voltage caps, and power it up with 38volts rails power supply, power supply that i have handy.
Note: it is designed to handle 50 volts rails, max 60.
50 volts @4ohm=312wrms
60 volts @4ohm=449wrms
(±400 @90% eff)I have done a mistake while doing the supply, the base resistor that powers the zener should take its voltage before the collector resistor, directly at the vcc.
There is no cap for filtering the vcc at the collector as well, wich is something i simply forgot.
The output cap should be a 100uf while now its a 33uf.
I will try different scheme for this, i can cleverly see the hf with my scope, and with simple 0,1uf parralel to the zener, it droped nearly 80% of this hf. But there is still some fluctuations that is not hf, wich the biggers caps may solve.
BTW, i still dont have ideas on a simple preamp for this application. It seems that this amp is hungry at its input.
There is no cap for filtering the vcc at the collector as well, wich is something i simply forgot.
The output cap should be a 100uf while now its a 33uf.
I will try different scheme for this, i can cleverly see the hf with my scope, and with simple 0,1uf parralel to the zener, it droped nearly 80% of this hf. But there is still some fluctuations that is not hf, wich the biggers caps may solve.
BTW, i still dont have ideas on a simple preamp for this application. It seems that this amp is hungry at its input.
Ok, issue totaly solved. It turned to be simpler than i tough.
2x 0,1uf caps under the pcb close to 2 via and boom, problem gone.
It just clip the wave if it goes beyond its capacity, without any hf stuck to the signal.
Better, the square wave test now show a very clean square wave, where before it was very noisy.
And better better, the heat sink seems to be cooler at same power level.
I have a big variable resistor and i tough it was set at 4 ohm, it is in reality set at 3 ohm, wich gives close to 120wrms for now with a crapy power supply and only 2200 uf caps on the rails.
hummmmm....cant wait to see how it will go with a "real" power supply...
rock it !!!
2x 0,1uf caps under the pcb close to 2 via and boom, problem gone.
It just clip the wave if it goes beyond its capacity, without any hf stuck to the signal.
Better, the square wave test now show a very clean square wave, where before it was very noisy.
And better better, the heat sink seems to be cooler at same power level.
I have a big variable resistor and i tough it was set at 4 ohm, it is in reality set at 3 ohm, wich gives close to 120wrms for now with a crapy power supply and only 2200 uf caps on the rails.
hummmmm....cant wait to see how it will go with a "real" power supply...
rock it !!!
One more consideration into this specific design. 
the boostrap cap: dont be too hasty to put a "big" one. Initial design from IR showed a .33uf. There have to be a reason why they did not put a 100uf one, as some suggested.
I did try many sizes, from 100 uf to 0,1 uf, with very different results.
The bigger the cap, the more instable Vcc to the IR2011 is at high power. Looking at the original schematic, one may understand why. A bigger cap is more difficult to pump. It loads the 12v power supply too much for no reasons, as it is asked to be partialy discharged to the gate of the mosfet. I rapidly found this issue, at high power, there was all kind of harmonics and fluctuations present on the Vcc pin of the chip. tracing the signal with my scope, and with different sizing caps cleverly showed me that above 1uf, it is useless and makes the whole thing act like shi*. Not that the previsous modifications with the 0,1uf caps did solve it all, but for now everything seems purrrfect with the 1uf. There is very few modulation on the Vcc pin with the 1uf, and all the test (100-1k-10khz sine/square wave on a 4 ohm load) show good or better results.
the boostrap cap: dont be too hasty to put a "big" one. Initial design from IR showed a .33uf. There have to be a reason why they did not put a 100uf one, as some suggested.
I did try many sizes, from 100 uf to 0,1 uf, with very different results.
The bigger the cap, the more instable Vcc to the IR2011 is at high power. Looking at the original schematic, one may understand why. A bigger cap is more difficult to pump. It loads the 12v power supply too much for no reasons, as it is asked to be partialy discharged to the gate of the mosfet. I rapidly found this issue, at high power, there was all kind of harmonics and fluctuations present on the Vcc pin of the chip. tracing the signal with my scope, and with different sizing caps cleverly showed me that above 1uf, it is useless and makes the whole thing act like shi*. Not that the previsous modifications with the 0,1uf caps did solve it all, but for now everything seems purrrfect with the 1uf. There is very few modulation on the Vcc pin with the 1uf, and all the test (100-1k-10khz sine/square wave on a 4 ohm load) show good or better results.
I once thought of a diode and a resistor in parallel, conntected in series with the electrolytic, in order to charge the electrolytic slowly but having access to its charge quickly. I have never tried it out by myself though.
Regards
Charles
Regards
Charles
pat allen said:One more consideration into this specific design.
the boostrap cap: dont be too hasty to put a "big" one. Initial design from IR showed a .33uf. There have to be a reason why they did not put a 100uf one, as some suggested.
I did try many sizes, from 100 uf to 0,1 uf, with very different results.
The bigger the cap, the more instable Vcc to the IR2011 is at high power. Looking at the original schematic, one may understand why. A bigger cap is more difficult to pump. It loads the 12v power supply too much for no reasons, as it is asked to be partialy discharged to the gate of the mosfet. I rapidly found this issue, at high power, there was all kind of harmonics and fluctuations present on the Vcc pin of the chip. tracing the signal with my scope, and with different sizing caps cleverly showed me that above 1uf, it is useless and makes the whole thing act like shi*. Not that the previsous modifications with the 0,1uf caps did solve it all, but for now everything seems purrrfect with the 1uf. There is very few modulation on the Vcc pin with the 1uf, and all the test (100-1k-10khz sine/square wave on a 4 ohm load) show good or better results.
You mean as I suggested.
I also suggested the series resistors be removed! You can (and should) parallel a low ESR, small value cap with it, probably like 100nF.
If your 12V supply is that weak, add more "reserve".
😕
The reason why they didn't use a bigger cap is obvious, it's more expensive. When you do the math on these caps you take that as absolute min. value, then use X10 at least. The bigger it is the less droop you have on your gate voltage, and obviously the longer it can hold it during clipping.
FYI I use a 50V 100uf electrolytic parralled with a cheap ceramic 100nF and a 1n914 bootstrap diode with no problems. No series resistors.
It's OK I edited that already 🙂.... need my coffee ya know?
Anyway you could try beefing it up a little to make it work, but to be honest if it has protection against clipping (does it??) aaand there's little to no droop during switching, then I wouldn't bother with such a big cap, just use what works.
Anyway you could try beefing it up a little to make it work, but to be honest if it has protection against clipping (does it??) aaand there's little to no droop during switching, then I wouldn't bother with such a big cap, just use what works.
what works for me, its exactly what i am tryin to say. With this scpecific circuit, and surely because of the pcb design itslef, anything bigger than 1uf load the thing. It works good anyway with a 10uf for now, i am still working around this to remove as much as possible the riples.
Thanks for all the good sugestions.
Thanks for all the good sugestions.
Yep, cool.
10uF is much better than 330nF as well.
I also found a heavy demand on the supply for it, but using a discrete driver and messing with the values for that it also loaded down the driver alot so I can't draw a direct relation to how much it draws from the supply when it isn't loaded down.
Just using a 12V reg for it, I've had it run hot and perfectly cool, same cap arrangment.
I think the standard 200ms turn on time would be fine for a cap as big as 100uF even.
Have you tried without the series resistance at all?
BTW I personally have no ideas what you could use for a preamp, I just use a sound card for my experiments.
Oh yeah, I also have another 100uF right after my 12V reg.
10uF is much better than 330nF as well.
I also found a heavy demand on the supply for it, but using a discrete driver and messing with the values for that it also loaded down the driver alot so I can't draw a direct relation to how much it draws from the supply when it isn't loaded down.
Just using a 12V reg for it, I've had it run hot and perfectly cool, same cap arrangment.
I think the standard 200ms turn on time would be fine for a cap as big as 100uF even.
Have you tried without the series resistance at all?
BTW I personally have no ideas what you could use for a preamp, I just use a sound card for my experiments.
Oh yeah, I also have another 100uF right after my 12V reg.
- Status
- Not open for further replies.