Hi everyone! I try to make a class-d amp , I using a TL084 do a triangle wave generator , and send to LT1016 (+) , LT1016(-) connect to the audio input
LT1016(7,8) ouput connect to the level shifting circuit ,
and then connect the level shifting output to the IC IR2011
at bootstrape circuit , i use a 100n to do the capacitor
IR2011's output connect to 23N20D MOSFET ,
when i turn on the power , the amp rise up to 10 amps
the LO side ouput mosfet get very hot
i do not use self-osc , so in my circuit i didn't use bootstrap startup circuit
i have double check my circuit , but i still can not check out!!
so , dose anyone get experence about it .... thanks for your helping
LT1016(7,8) ouput connect to the level shifting circuit ,
and then connect the level shifting output to the IC IR2011
at bootstrape circuit , i use a 100n to do the capacitor
IR2011's output connect to 23N20D MOSFET ,
when i turn on the power , the amp rise up to 10 amps
the LO side ouput mosfet get very hot
i do not use self-osc , so in my circuit i didn't use bootstrap startup circuit
i have double check my circuit , but i still can not check out!!
so , dose anyone get experence about it .... thanks for your helping
Hi there,
Glad to see your having a go at your own😀
Post a schematic and i'm sure we'll be able to sort you out and get it working. Have fun!
Mad.P
Glad to see your having a go at your own😀
Post a schematic and i'm sure we'll be able to sort you out and get it working. Have fun!
Mad.P
please , dose anyone can show me IR2011's Ho and Lo output waveform , because I don't what waveform is correct!!
so, please show me the waveform picture , thanks a lot !!
so, please show me the waveform picture , thanks a lot !!
Hmmm,
The drive to the ir2011 looks suspect, also check your boot strap supply. I'm not familliar with that chip but at a glance it looks wrong. Check your circuit against the ref circuit that should be in the data sheet for ir2011. Also whilst trying to get your amp to work for the first time you might find it easyier to remove the feedback since your design is carrier based.
Mad.P
The drive to the ir2011 looks suspect, also check your boot strap supply. I'm not familliar with that chip but at a glance it looks wrong. Check your circuit against the ref circuit that should be in the data sheet for ir2011. Also whilst trying to get your amp to work for the first time you might find it easyier to remove the feedback since your design is carrier based.
Mad.P
nitrate said:
Hi Nitrate
1.do you mean the boot strap supply voltage is wrong??
now i supply -Vdd+15v , 15v is too high ?? or too low ??
2.do you mean that remove the feedback circuit , and let the
circuit run fisrt ?? but if it do not have feedback circuit will not
effect the whole functional ??
3. i have another problem , in normal , the audio input is 0v
will the ir2011's output still have square wave ??
and the square wave's frequency is equare to triangle wave
frequency ??
Hi,
1. The bootstrap supply seems to be wired wrong. It should as i imagine ride on the bridge of the output fets. 15V sounds o.k to me but beware i HAVE NOT READ the ir2011 data sheet so i may be wrong on both accounts. I'm at work at the mo so i can't free up enough time to read through the data sheet. Maybe someone else on here who has used the chip befor might be better advising you on this point. It is a very poular chip on here.
2. Yes, run your circuit without the feedback. Your circuit is not self oscillating so you do not need feedback for the amp to run. The triangle wave will clock the amp along.
3. Absolutly correct yes. Your triangle will clock the ir2011 along at aprox 50/50% if you have a good triangle to compare the 0V input against.
P.S if you are using +/- 15V then yourdrive to the ir2011 will probably work.
Have Fun!
PP.S i created a sucsessful carrier based amp ( triangle job ). The schematiccs ect are located in my thread 'Breadboard classD'. Have a look it should give you some help.
Regards
Mad.P
1. The bootstrap supply seems to be wired wrong. It should as i imagine ride on the bridge of the output fets. 15V sounds o.k to me but beware i HAVE NOT READ the ir2011 data sheet so i may be wrong on both accounts. I'm at work at the mo so i can't free up enough time to read through the data sheet. Maybe someone else on here who has used the chip befor might be better advising you on this point. It is a very poular chip on here.
2. Yes, run your circuit without the feedback. Your circuit is not self oscillating so you do not need feedback for the amp to run. The triangle wave will clock the amp along.
3. Absolutly correct yes. Your triangle will clock the ir2011 along at aprox 50/50% if you have a good triangle to compare the 0V input against.
P.S if you are using +/- 15V then yourdrive to the ir2011 will probably work.
Have Fun!
PP.S i created a sucsessful carrier based amp ( triangle job ). The schematiccs ect are located in my thread 'Breadboard classD'. Have a look it should give you some help.
Regards
Mad.P
Your circuit looks about OK.
15V for the bootstrap voltage is oK, and the driver connection is also ok.
Some things to check. Let's go step by step.
1. First, disable the high voltage rails to the mosfets, leave only the +/-15V for the opamps:
2. Look at the triangle waveform, amplitude and frequency, and tell us how it is.
3. If there is a correct triangle waveform, there should be a square wave out of the comparator (with input tied to GND).
4. Disconnect feedback. For you to measure something useful, limit the gain of the input opamp by adding a resistor between pins 2 and 1 (- in and out), use something like 20 to 47Kohms. Adding a signal of enough amplitude should produce PWM at the output of the comparator.
5. If all these steps are correct, then connect the high voltage to the mosfets, but still without feedback. Current consumption should be ok, and you should have 15v between COM and VCC pins. (You will have bootstrap voltage only when the circuit is oscillating properly)
6. Check that, with 0V input, you have a square wave at the output (coil input), swinging from positive to negative rail, and with no significant overshoot nor ringing.
7. If so, then inject a signal and check that there is output coming out of the filter, and tell us how does it like (does it resemble to the input signal at all?)
If everything is ok, reconnecting feedback should work.
Tell us what step your circuit is failing at, and we will be able to put it to work properly.
15V for the bootstrap voltage is oK, and the driver connection is also ok.
Some things to check. Let's go step by step.
1. First, disable the high voltage rails to the mosfets, leave only the +/-15V for the opamps:
2. Look at the triangle waveform, amplitude and frequency, and tell us how it is.
3. If there is a correct triangle waveform, there should be a square wave out of the comparator (with input tied to GND).
4. Disconnect feedback. For you to measure something useful, limit the gain of the input opamp by adding a resistor between pins 2 and 1 (- in and out), use something like 20 to 47Kohms. Adding a signal of enough amplitude should produce PWM at the output of the comparator.
5. If all these steps are correct, then connect the high voltage to the mosfets, but still without feedback. Current consumption should be ok, and you should have 15v between COM and VCC pins. (You will have bootstrap voltage only when the circuit is oscillating properly)
6. Check that, with 0V input, you have a square wave at the output (coil input), swinging from positive to negative rail, and with no significant overshoot nor ringing.
7. If so, then inject a signal and check that there is output coming out of the filter, and tell us how does it like (does it resemble to the input signal at all?)
If everything is ok, reconnecting feedback should work.
Tell us what step your circuit is failing at, and we will be able to put it to work properly.
Pierre said:
Hi
1. The triangle waveform is right at 176KHz from +9 to -9 , but it
still have some dc offset , so it not symmetry .
2. i have use a 0.1u cap to to separate the DC , but it seems not
work , dose my circuit have problem ?? but in sometime , there
is no dc offset problem!!
3. the triangle wave out From tl084 , the peak to peak is 18v
when i just use a 10kohm to connect to lt1016(-) , the
triangle wave is cut by +5v and -5v , is that beacause
lt1016's power is +5v and -5v ?? so use another 10kohm
to divided the triangle wave, is that right??
4. Yes! when the input connect to GND , the LT1016's pin 7 and 8
have correct square wave out
5. i'sorry , i circuit have some problem, i draw some wrong with
LT1016 , i connect the triangle wave to LT1016's pin 3(-)
and connect the audio input to LT1016's pin 2(+)
is that will effect the whole circuit ability??
6. when i connect sine wave to lt1026(+) , i watch the lt1016's
pin3 (triangle wave in) , it seems get some sine wave signal
beacause the triangle wave's dc become not stable , it change
with sine wave input , is that right ??
7. i connect high rail to mosfet , but the current still higher than
5A , is that anything i lose ??
i have a question , the ir2011's HO and LO always output
square wave to open and close mosfet
is that will let the mosfet get high current ???
if no , in normal , how many current sumption is correct ??
thanks a lot !!
square wave to open and close mosfet
is that will let the mosfet get high current ???
if no , in normal , how many current sumption is correct ??
thanks a lot !!
I suggest that you start fixing it step by step and then continue:
- Yes, your input to the comparator should be something like +/-4V or so, so instead of forming a voltage divider, it is better to re-calculate the values of the triangle wave generator components. You can get any amplitude and frequency you want (up to a limit, and more with that slow opamp, of course).
- You shouldn't have offset in the triangle waveform. If it is not looking almost perfect perhaps you have some problem with the slew rate of the opamp. Don't you have a faster one (LF347 for example)?
- If you have connected the inputs to the comparator backwards, that won't work in closed-loop, as you will have positive feedback instead of negative feedback, and it almost for sure, is going to stick to one rail, delivering all the DC to the load. However, if you have followed our advice and connected it open-loop, that shouldn't produce any problem.
- About current consumption, when everything is ok, it shouldn't be more than 50mA per rail with no signal.
- Yes, your input to the comparator should be something like +/-4V or so, so instead of forming a voltage divider, it is better to re-calculate the values of the triangle wave generator components. You can get any amplitude and frequency you want (up to a limit, and more with that slow opamp, of course).
- You shouldn't have offset in the triangle waveform. If it is not looking almost perfect perhaps you have some problem with the slew rate of the opamp. Don't you have a faster one (LF347 for example)?
- If you have connected the inputs to the comparator backwards, that won't work in closed-loop, as you will have positive feedback instead of negative feedback, and it almost for sure, is going to stick to one rail, delivering all the DC to the load. However, if you have followed our advice and connected it open-loop, that shouldn't produce any problem.
- About current consumption, when everything is ok, it shouldn't be more than 50mA per rail with no signal.
First , thanks for your replying
1.now the triangle wave generator , I use tl084 to generate
is that fast enough for 200KHz ?? in the triangle wave circuit
i use the formula in http://www.play-hookey.com/analog/triangle_waveform_generator.html
f= (1/4RC)*(R2/R1) , i caculate the result is 716911Hz ..
but I only get 177KHz , is that beacause of TL084 ???
2. in your point 3 ...do you mean if i connect input to the pin2(+)
it will not work ?? so i will try to connect the audio input to
pin3(-) and connect the triangle wave to pin2(+) , and test!!
3. when the input still connect to the pin2(+) , i turn on the power
it always only low side mosfet or high side mosfet get high
temprature , is that because it's going to stick to one rail??
I will Try again , thanks again..!!
1.now the triangle wave generator , I use tl084 to generate
is that fast enough for 200KHz ?? in the triangle wave circuit
i use the formula in http://www.play-hookey.com/analog/triangle_waveform_generator.html
f= (1/4RC)*(R2/R1) , i caculate the result is 716911Hz ..
but I only get 177KHz , is that beacause of TL084 ???
2. in your point 3 ...do you mean if i connect input to the pin2(+)
it will not work ?? so i will try to connect the audio input to
pin3(-) and connect the triangle wave to pin2(+) , and test!!
3. when the input still connect to the pin2(+) , i turn on the power
it always only low side mosfet or high side mosfet get high
temprature , is that because it's going to stick to one rail??
I will Try again , thanks again..!!
Don't worry. I have successfully built very similar amps so I will help you step by step to fix it. But don't hurry too much!
- The TL084 has a slew rate of only 13v/us, that limits the triangle frequency to around 500KHz with a +/-6.5V amplitude, that's the absolute maximum. You are getting +/-9V, so the maximum would be around 350KHz. But my experiments show that this is impractical due to the very high bandwidth required for the "peaks". If you go higher than, say, 100KHz, use a faster opamp.
However, for debuggin, you can go to 100KHz if you get a reasonably accurate triangle. But do reduce amplitude to +/-4V max.
- By the moment, with NO feedback, no problem if the comparator input are connected backwards: that will only determine if the amplifier inverts or not. Try to fix it without feedback (don't forget the resistor between - input and out of the input opamp) and then we'll worry about that.
- If the amp sticks to the lower rail, then your low side mosfet will heat up (and your load, of course). BUT if it tends to stick to the upper rail, it won't produce any current, because there won't be oscillation and hence no bootstrap voltage to excite the high side mosfet, so you would get 0V output.
Let's fix the triangle/comparator first, then we'll start with the power stage. BTW: Don't make the error of disconnecting the gates of the mosfets, they will turn on simultaneously and short your supply, blowing instantaneously! Simply remove their supply.
- The TL084 has a slew rate of only 13v/us, that limits the triangle frequency to around 500KHz with a +/-6.5V amplitude, that's the absolute maximum. You are getting +/-9V, so the maximum would be around 350KHz. But my experiments show that this is impractical due to the very high bandwidth required for the "peaks". If you go higher than, say, 100KHz, use a faster opamp.
However, for debuggin, you can go to 100KHz if you get a reasonably accurate triangle. But do reduce amplitude to +/-4V max.
- By the moment, with NO feedback, no problem if the comparator input are connected backwards: that will only determine if the amplifier inverts or not. Try to fix it without feedback (don't forget the resistor between - input and out of the input opamp) and then we'll worry about that.
- If the amp sticks to the lower rail, then your low side mosfet will heat up (and your load, of course). BUT if it tends to stick to the upper rail, it won't produce any current, because there won't be oscillation and hence no bootstrap voltage to excite the high side mosfet, so you would get 0V output.
Let's fix the triangle/comparator first, then we'll start with the power stage. BTW: Don't make the error of disconnecting the gates of the mosfets, they will turn on simultaneously and short your supply, blowing instantaneously! Simply remove their supply.
1. OK ,,, i will recaculate the triangle wave generator value
2. I have already remove the feedback resistor (33k)
3. you said that (don't forget the resistor between - input and
out of the input opamp) , do you mean tow resistors ?
one in the audio input , in front of the tl082 (10K)??
one in the output of the tl082 (10k)??
4. you said if it tends to stick to the upper rail, it won't produce
any current, because there won't be oscillation!!
but the outpur of the ir2011 always has square wave out right ??
will the ir2011 stop the oscillation if bootstrap cap ??
5. you ever said remove the mosfet high rail pin , do you mean
only remove the rail pin or remove the mosfet out ??
now , i remove out the mosfet , not only the high rail pin .
will it effect out test ??
6. could you send me the circuit you has already made??
maybe i will find something out .... s0521089@cc.ncu.edu.tw
2. I have already remove the feedback resistor (33k)
3. you said that (don't forget the resistor between - input and
out of the input opamp) , do you mean tow resistors ?
one in the audio input , in front of the tl082 (10K)??
one in the output of the tl082 (10k)??
4. you said if it tends to stick to the upper rail, it won't produce
any current, because there won't be oscillation!!
but the outpur of the ir2011 always has square wave out right ??
will the ir2011 stop the oscillation if bootstrap cap ??
5. you ever said remove the mosfet high rail pin , do you mean
only remove the rail pin or remove the mosfet out ??
now , i remove out the mosfet , not only the high rail pin .
will it effect out test ??
6. could you send me the circuit you has already made??
maybe i will find something out .... s0521089@cc.ncu.edu.tw
Hello,
2. Ok, you have to remove the 33K resistor (R39).
3. No, you need a single resistor between pins 1 and 2 of U12A. The calculation of this resistor is as follows:
The overall gain of the amplifier is determined by the gain of the error opamp, along with your triangle amplitude and your supply rails. As you will have around 8Vpp of triangle, that means that with an error amplifier output of that same amplitude you will get maximum output excursion, for example 80Vpp if you use +/-40V rails. So your gain _after_ error opamp will be 80/8=10 V/V. For a more useful gain of around 40V/V, that means that you need 4V/V gain in your error opamp: so as you have R28=10k, place around 39K - 47K between pins 1 and 2 of U12A.
4. For driving the high side mosfet, you need to have around 12V between gate and source of that mosfet. If that mosfet was on, you would get a 40V output (assuming your supply rails are +/-40V, change to your value if you want), so you would need +52V in the gate. As there is no such voltage present in your circuit, a bootstrap supply is needed: that's what D1 and C25 are needed for: BUT it only works if there is some part of the time that the cap lower pin is connected to a lower voltage than the diode: that only happens when the lower mosfet is on and you get -40V at the output, so you have -40V at the source of the high side mosfet, and -40+15V at the diode and the capacitor can be charged for the next cycle.
If you don't have oscillation, the bootstrap capacitor won't ever be charged so the upper mosfet won't have drive voltage and it won't turn on. Do you follow me?
5. Yes, removing _both_ mosfets is the safer way to go until all the previous circuitry is fixed.
6. Please search for it until I find it, I published it under "mosfet reliability in class-d amplifiers" thread, if I remember ok. If you don't find it, I will re-post it here as soon as I find the file.
Best regards
2. Ok, you have to remove the 33K resistor (R39).
3. No, you need a single resistor between pins 1 and 2 of U12A. The calculation of this resistor is as follows:
The overall gain of the amplifier is determined by the gain of the error opamp, along with your triangle amplitude and your supply rails. As you will have around 8Vpp of triangle, that means that with an error amplifier output of that same amplitude you will get maximum output excursion, for example 80Vpp if you use +/-40V rails. So your gain _after_ error opamp will be 80/8=10 V/V. For a more useful gain of around 40V/V, that means that you need 4V/V gain in your error opamp: so as you have R28=10k, place around 39K - 47K between pins 1 and 2 of U12A.
4. For driving the high side mosfet, you need to have around 12V between gate and source of that mosfet. If that mosfet was on, you would get a 40V output (assuming your supply rails are +/-40V, change to your value if you want), so you would need +52V in the gate. As there is no such voltage present in your circuit, a bootstrap supply is needed: that's what D1 and C25 are needed for: BUT it only works if there is some part of the time that the cap lower pin is connected to a lower voltage than the diode: that only happens when the lower mosfet is on and you get -40V at the output, so you have -40V at the source of the high side mosfet, and -40+15V at the diode and the capacitor can be charged for the next cycle.
If you don't have oscillation, the bootstrap capacitor won't ever be charged so the upper mosfet won't have drive voltage and it won't turn on. Do you follow me?
5. Yes, removing _both_ mosfets is the safer way to go until all the previous circuitry is fixed.
6. Please search for it until I find it, I published it under "mosfet reliability in class-d amplifiers" thread, if I remember ok. If you don't find it, I will re-post it here as soon as I find the file.
Best regards
1. ok , now i use 80Vpp rail voltage , and 8Vpp triangle wave
so , the error amp's gain must be 80/8=10V/V right??
so i have to use 80k resistor into pin1 and pin2 of u12A ??
2. in your point 4 , thanks for you let me know how the bootstrap
work. so in my circuit's bootstrap , the value of the component
, is that all right ?? i use 1n4148 diode and 0.1u cap , is that
suitable for bootstrap ?
3. ok .. i will find out the "mosfet reliability in class-d amplifiers"
thread and read it !
Thanks a lot !!
so , the error amp's gain must be 80/8=10V/V right??
so i have to use 80k resistor into pin1 and pin2 of u12A ??
2. in your point 4 , thanks for you let me know how the bootstrap
work. so in my circuit's bootstrap , the value of the component
, is that all right ?? i use 1n4148 diode and 0.1u cap , is that
suitable for bootstrap ?
3. ok .. i will find out the "mosfet reliability in class-d amplifiers"
thread and read it !
Thanks a lot !!
Hi , Pierre
I have solved the problem already!!
I add a invertor circuit between the level shift and IR2011
and adjust the level shift Re resistor from 10k to 3.9k
because the level shift output high side voltage too low!
when i change Re to 3.9K , the high side voltage have 10V depend on -Vdd!
Thanks for your Help ,,,
Best Regard
I have solved the problem already!!
I add a invertor circuit between the level shift and IR2011
and adjust the level shift Re resistor from 10k to 3.9k
because the level shift output high side voltage too low!
when i change Re to 3.9K , the high side voltage have 10V depend on -Vdd!
Thanks for your Help ,,,
Best Regard
For the bootstrap cap, 0.1uF is too low, and about 1n4148, I am not sure if it will work.
Use 1uF instead and a schottky diode like MUR120
About the gain of the error opamp, it only depends on what final gain you need. You already have 10V/V gain due to the triangle and your rails.
If you need an overall gain of , say, 45, you need additional 4.5x in your error opamp, so you have to put a resistor between pins 1 and 2 of 4.5 times the input resistor.
Please tell us how it is behaving exactly, and what was your fix.
Use 1uF instead and a schottky diode like MUR120
About the gain of the error opamp, it only depends on what final gain you need. You already have 10V/V gain due to the triangle and your rails.
If you need an overall gain of , say, 45, you need additional 4.5x in your error opamp, so you have to put a resistor between pins 1 and 2 of 4.5 times the input resistor.
Please tell us how it is behaving exactly, and what was your fix.
1N4148 is rated too low for your design. It may be breaking down and discharge the small bootstrap capacitor, therefore the high side driver doesn't really work.
Test the circuit at below 70V pk=pk. It may work due to 1N4148 not breaking down.
Also remove the load during test but do put a resistor, say 100-ohm. That self-oscillating amp has highly peaking output impedance therefore it does need a load to damp the peaking.
Test the circuit at below 70V pk=pk. It may work due to 1N4148 not breaking down.
Also remove the load during test but do put a resistor, say 100-ohm. That self-oscillating amp has highly peaking output impedance therefore it does need a load to damp the peaking.
i will change 0.1u to 1u cap..and test !!
last time i only change the level shift resistor "Re" .
Change from 10k to 3.9k , then i measure then output
of the level shift Vc , its high side output voltage become
higher.
and i add a invertor "40106" between level shift and IR2011 to
shaping the square waveform. i think it's work!
The output waveform of the IR2011 is looking good!
But the power supply still isn't working stable!
i use a 15v Zener to reduce the 40v and then serise a 7815.
But the Zener always get hot , dose anyone have good ideal?
last time i only change the level shift resistor "Re" .
Change from 10k to 3.9k , then i measure then output
of the level shift Vc , its high side output voltage become
higher.
and i add a invertor "40106" between level shift and IR2011 to
shaping the square waveform. i think it's work!
The output waveform of the IR2011 is looking good!
But the power supply still isn't working stable!
i use a 15v Zener to reduce the 40v and then serise a 7815.
But the Zener always get hot , dose anyone have good ideal?
- Status
- Not open for further replies.