Sorry, but I can't find Re in your schematics. I suppose you are referring to R31 and R32, anyway. That's correct, in order to scale your 5V pulses from the comparator to around 12V, you need the same relationship between R33/R31 and R34/R32. 10K/3.9k=12/5 aprox. That's right. But in order to have better output to enter the IR2011, I recommend that you lower both resistors to, say, 1K and 2.2k.
Then you won't need the inverter you have added. (I have successfully built the amp without it).
About the diodes: in fact, you can't use 1N4148 with good results, specially for the bootstrap cap, and they have also too low current rating for the mosfet gates. Exchange all your diodes for MUR120 ones, for example.
About the zener used to reduce input to 7815, well, its dissipation depends on the current you are demanding by the driver. A normal value can be around 50-60mA, then you dissipate 0.06*15=0.75W, that can be too much for a small zener diode.
I recommend that you measure the current drawn (insert a multimeter in series) and then use a power (2W or so) resistor instead of the zener, calculated so its voltage drop is around 20V.
Is your triangle and comparator output OK now? What's the sw. frequency?
Then you won't need the inverter you have added. (I have successfully built the amp without it).
About the diodes: in fact, you can't use 1N4148 with good results, specially for the bootstrap cap, and they have also too low current rating for the mosfet gates. Exchange all your diodes for MUR120 ones, for example.
About the zener used to reduce input to 7815, well, its dissipation depends on the current you are demanding by the driver. A normal value can be around 50-60mA, then you dissipate 0.06*15=0.75W, that can be too much for a small zener diode.
I recommend that you measure the current drawn (insert a multimeter in series) and then use a power (2W or so) resistor instead of the zener, calculated so its voltage drop is around 20V.
Is your triangle and comparator output OK now? What's the sw. frequency?
1.do you mean R31 use 1k and r33 use 2.2k ??
2.i have bought the mur120 , i will try it !!
3. do you mean in front of the 7815 , use a resistor to serise with
7815?? so , if the input voltage is 40V , and assume the max
out current is 1A. so , i have use a (40-20)/1 = 20ohm
but it will need 20Watt , right ??
but 20Watt is too big, how cani do ???
4. the triangle wave, now , is amost good!
F = 200KHz , DCoffset is under 10mV
I think it's good enought , how do you think ??
My compartor LT1016 often breakdown , i don't why!!
I have Change 5 pcs , maybe is my power not stable.
2.i have bought the mur120 , i will try it !!
3. do you mean in front of the 7815 , use a resistor to serise with
7815?? so , if the input voltage is 40V , and assume the max
out current is 1A. so , i have use a (40-20)/1 = 20ohm
but it will need 20Watt , right ??
but 20Watt is too big, how cani do ???
4. the triangle wave, now , is amost good!
F = 200KHz , DCoffset is under 10mV
I think it's good enought , how do you think ??
My compartor LT1016 often breakdown , i don't why!!
I have Change 5 pcs , maybe is my power not stable.
Yes, I mean R31=R32=1k, R33=R34=2.2k.
I mean that, a resistor in series with the input of 7815. But it CAN'T have a current draw of 1A!!!! Measure it, and at most you must have in the order of 50-70mA. So R will be in the order of 20/0.07=285 ohm (270 ohm), and its dissipation will be around 0.07^2 * 285 = 1.4W max.
If the current consumption from the 7815 regulator is 1A, you have something really wrong!
The triangle must be very good. The amplitude can't exceed +/-4V or so. Your comparator shuldn't suffer at all if you supply it with 5V and don't input voltages exceeding +/-5V or so in any input (triangle or signal).
Where are you getting the power? It should be very stable, if not, your triangle won't be stable and hence your output.
I mean that, a resistor in series with the input of 7815. But it CAN'T have a current draw of 1A!!!! Measure it, and at most you must have in the order of 50-70mA. So R will be in the order of 20/0.07=285 ohm (270 ohm), and its dissipation will be around 0.07^2 * 285 = 1.4W max.
If the current consumption from the 7815 regulator is 1A, you have something really wrong!
The triangle must be very good. The amplitude can't exceed +/-4V or so. Your comparator shuldn't suffer at all if you supply it with 5V and don't input voltages exceeding +/-5V or so in any input (triangle or signal).
Where are you getting the power? It should be very stable, if not, your triangle won't be stable and hence your output.
i see the datasheet of the ir2011 , its max output current is 1A.
if we only supply 70mA , is it enought??
My PowerSupply is a SMPT. The amp that i used in the car !
so i have to use smpt system to get +/- rail voltage.
i say my power is not stable. Not mean the +/- 5v or +/- 15v.
because of i use smpt , and the voltage in car is not stable.
so my +/- rail voltage will get some ripple.
if we only supply 70mA , is it enought??
My PowerSupply is a SMPT. The amp that i used in the car !
so i have to use smpt system to get +/- rail voltage.
i say my power is not stable. Not mean the +/- 5v or +/- 15v.
because of i use smpt , and the voltage in car is not stable.
so my +/- rail voltage will get some ripple.
IR2011 peak current is +/-1A, but that's only during turn on/off of the mosfets. You only have to maintain the average current consumption, that depends on rise/fall time, peak current and frequency. For your frequency and mosfets, expect an average of around 60-70mA max. BUT it is important that you have bypass capacitors near the IR2011: 1uF in parallel with 100nF is enough: they provide the +/-1A peaks during the (short) transitions.
You can use only +/-5V for the modulator (opamps, comparator, triangle generator) instead of +/-15 and +/-5V. But if so, reduce your triangle to around +/-3V to improve linearity. This +/-5V supply should be regulated and well estabilized.
Best regards,
Pierre
You can use only +/-5V for the modulator (opamps, comparator, triangle generator) instead of +/-15 and +/-5V. But if so, reduce your triangle to around +/-3V to improve linearity. This +/-5V supply should be regulated and well estabilized.
Best regards,
Pierre
because then pre-amp must have +/- 15v , so i must reserv it!
now the +/- 15v and +/- 5v are serise a zener with 78xx .
ex. +5v ,input rail = +40v, i use two 15v-zener/1W serise with 7815.
But the zener get heating. the zener's Imax=70mA.
But its flow current seems not enought.
could i have to change bigger Watts Zener ??
Or what topology power regulator i can use!!
Thanks ,,,
now the +/- 15v and +/- 5v are serise a zener with 78xx .
ex. +5v ,input rail = +40v, i use two 15v-zener/1W serise with 7815.
But the zener get heating. the zener's Imax=70mA.
But its flow current seems not enought.
could i have to change bigger Watts Zener ??
Or what topology power regulator i can use!!
Thanks ,,,
Let's clarify some questions...
You start from +/-40V voltage, if it is from a car smps, it will be unregulated at all, so you need regulators for +/-5v and +/-15V, that is, 7805, 7905, 7815 and 7915. Ok so far?
As 40v is too much for a 78XX/79XX regulator, you need to drop the input voltage a little bit. Your +/-40V rails can drop quite a bit under load, so I wouldn't drop more than 15V for the +/-15V regulators (you have to give at least 3V margin, 18V input to the regulator, under worst case for it to work properly).
The easiest way is to measure the current that each rail needs (that will be fairly constant) and calculate a resistor value so that you have 15V voltage drop (R=15V/I). Minimum wattage needed will be I^2*R watts.
The same applies for the 15V regulator referred to -vss for the IR2011.
Until you don't have all this fixed, it doesn't worth the pain going on with the power stage and the rest.
You start from +/-40V voltage, if it is from a car smps, it will be unregulated at all, so you need regulators for +/-5v and +/-15V, that is, 7805, 7905, 7815 and 7915. Ok so far?
As 40v is too much for a 78XX/79XX regulator, you need to drop the input voltage a little bit. Your +/-40V rails can drop quite a bit under load, so I wouldn't drop more than 15V for the +/-15V regulators (you have to give at least 3V margin, 18V input to the regulator, under worst case for it to work properly).
The easiest way is to measure the current that each rail needs (that will be fairly constant) and calculate a resistor value so that you have 15V voltage drop (R=15V/I). Minimum wattage needed will be I^2*R watts.
The same applies for the 15V regulator referred to -vss for the IR2011.
Until you don't have all this fixed, it doesn't worth the pain going on with the power stage and the rest.
Titanchen,
have a look at this tutorial, you will find it very useful as you are starting with Class-D. It is a good introduction, I found it quite useful too.
http://sound.westhost.com/articles/pwm.htm
(BTW: I hope that _cool_ coldamp modules are available at last).
have a look at this tutorial, you will find it very useful as you are starting with Class-D. It is a good introduction, I found it quite useful too.
http://sound.westhost.com/articles/pwm.htm
(BTW: I hope that _cool_ coldamp modules are available at last).
Maybe it's because the output stage get powered before the control circuit gets powered and started working. Yesterday it just happened on mine.
Try to use seperate supply for control circuit and output stage; turn on the control circuit first.
Try to use seperate supply for control circuit and output stage; turn on the control circuit first.
Kenshin is right, it is always good to make the output stage work later than the control one.
With other driver chips it is very easily done by simply putting a RC network in the SD pin so the driver doesn't work until a second or two after power on.
With your chip, the only way to do that is to throw both inputs to GND with transistors or the like and add a delay.
If not, you will have to time the supplies properly.
With other driver chips it is very easily done by simply putting a RC network in the SD pin so the driver doesn't work until a second or two after power on.
With your chip, the only way to do that is to throw both inputs to GND with transistors or the like and add a delay.
If not, you will have to time the supplies properly.
I'm sorry , i don't really understand "throw both inputs to GND with transistors "
how will i have to do ??
i got another question , the bootstrape cap ,C25
how many working voltage i have to use ???
In my curcuit , only need 25v ??
how will i have to do ??
i got another question , the bootstrape cap ,C25
how many working voltage i have to use ???
In my curcuit , only need 25v ??
I was talking about a way of implementing the SD pin that IR2011 lacks. You have to put both inputs at logic 0 simultaneously. That can be done by passing them through AND (or other) gates controlled with a common SD signal, or by adding a NPN transistor in parallel (collector to signal, emitter to -VSS, both bases, with resistor in series, controlled by a common SD signal (referred to -VSS).
I was only thinking in loud.
The bootstrap cap rating of 25V is enough. (it needs to be driver voltage plus a security margin).
I was only thinking in loud.
The bootstrap cap rating of 25V is enough. (it needs to be driver voltage plus a security margin).
HI
I have rebuild a new PCB , and modify all component.
R31 and R32 become 1k , R33 and R34 become 2.2k.
D1 to D5 change to SF14.
But when i trun on , the current still got big current!!
I check the IR2011 pin5 and Pin6 , the waveform looks not good! The waveform's low side seems not at -Vdd. Looks like discharge not fast enought.
so i move out the IRfb23n20d output MOSFET and output inductor, connect the audio input to gnd , move out the feedback resistor R39. And change R31 and R32 to 510ohms , R33 and R34 to 1k. The pin8 of IR2011, Lo ouput, waveform looks right. so i reconnect the ouput MOSFET , but still got big current . So ,I add a invertor between output of Q1,Q2 and Ir2011 pin5 , pin6 again. And turn power on again , the current stop at 650mA, it\'s right? But the Lo-side output MOSFET get heating up. does anything got wrong ? and when i reconnect the output Indutor, the current rise up to 1.4A. Is that right?? I check the output waveform before thought the inductor. it have +Vcc and -Vdd square waveform.
so , i check the output waveform after LC filter, It got 10Vpp sine waveform at 200KHz.... It looks got something wrong!!
I have rebuild a new PCB , and modify all component.
R31 and R32 become 1k , R33 and R34 become 2.2k.
D1 to D5 change to SF14.
But when i trun on , the current still got big current!!
I check the IR2011 pin5 and Pin6 , the waveform looks not good! The waveform's low side seems not at -Vdd. Looks like discharge not fast enought.
so i move out the IRfb23n20d output MOSFET and output inductor, connect the audio input to gnd , move out the feedback resistor R39. And change R31 and R32 to 510ohms , R33 and R34 to 1k. The pin8 of IR2011, Lo ouput, waveform looks right. so i reconnect the ouput MOSFET , but still got big current . So ,I add a invertor between output of Q1,Q2 and Ir2011 pin5 , pin6 again. And turn power on again , the current stop at 650mA, it\'s right? But the Lo-side output MOSFET get heating up. does anything got wrong ? and when i reconnect the output Indutor, the current rise up to 1.4A. Is that right?? I check the output waveform before thought the inductor. it have +Vcc and -Vdd square waveform.
so , i check the output waveform after LC filter, It got 10Vpp sine waveform at 200KHz.... It looks got something wrong!!
I increase the output cap C26 became 0.44u and test the output wave form again .
The sine wave amplitude decreased.
But still got 3Vp-p.... the DC offset about 150mV.
I don't know what situation is right.
The sine wave amplitude decreased.
But still got 3Vp-p.... the DC offset about 150mV.
I don't know what situation is right.
650mA is not correct at all with no signal nor load!
Are you putting a load at the output? If not, your filter resonance may be messing things. You should put a 8r resistor as the load for testing.
Can you put the new schematics so we can have a look at it and figure out what's happening?
It looks as massive shoot-through due to no dead-time control but who knows...
About the output waveform of 10Vpp (changed to 3Vpp) at 200KHz, it may be ok. Please remember us the filter values and your supply rails voltage, as well as the test load you are using (if any), so I can check if that's logical.
Are you putting a load at the output? If not, your filter resonance may be messing things. You should put a 8r resistor as the load for testing.
Can you put the new schematics so we can have a look at it and figure out what's happening?
It looks as massive shoot-through due to no dead-time control but who knows...
About the output waveform of 10Vpp (changed to 3Vpp) at 200KHz, it may be ok. Please remember us the filter values and your supply rails voltage, as well as the test load you are using (if any), so I can check if that's logical.
OK, I have rechecked your schematics:
If you are running at 200KHz with a 10uH + 440nF filter, you will get too much ripple.
You should increase the coil inductance to around 20uH.
What core are you using? It is VERY important that you are using a suitable one, if not, you can get strange results like what you are getting.
If you are running at 200KHz with a 10uH + 440nF filter, you will get too much ripple.
You should increase the coil inductance to around 20uH.
What core are you using? It is VERY important that you are using a suitable one, if not, you can get strange results like what you are getting.
I'm sorry , i didn't caculate the smtp's power disspation
when i move out the output mosfet and output inductor(18u) , the idle current is 300mA for power system disspation.
when i reconnect the MOSFET , the idle current is 650mA.
So , the classd-amp only has 350mA disspation... is that OK???
And reconnect the output inductor , the power disspation is 1.4A.
The amp seemed getting 1.1A power disspation.
About the dead time control , does the IR2011 has dead time control by itself ??
Above all test environment were no LOAD and signal input was grounding.
I don't know what core i used , I just told the manufacture , and he made it for me.
I recheck my feedback circuit , i connect it before output inductor.
The waveform is square , doesn't i have to add a RC lo-pass filter to shape the square wave??
Or i should connect feedback throught Inductor ???
my new circuit :
when i move out the output mosfet and output inductor(18u) , the idle current is 300mA for power system disspation.
when i reconnect the MOSFET , the idle current is 650mA.
So , the classd-amp only has 350mA disspation... is that OK???
And reconnect the output inductor , the power disspation is 1.4A.
The amp seemed getting 1.1A power disspation.
About the dead time control , does the IR2011 has dead time control by itself ??
Above all test environment were no LOAD and signal input was grounding.
I don't know what core i used , I just told the manufacture , and he made it for me.
I recheck my feedback circuit , i connect it before output inductor.
The waveform is square , doesn't i have to add a RC lo-pass filter to shape the square wave??
Or i should connect feedback throught Inductor ???
my new circuit :
I am sorry, titanchen, but class D is not as simple as taking the first component you have in your drawer and putting it in your circuit. (no offence intended).
The output coil is CRITICAL. Even if the value is correct, from one core to another things can go perfect or burn just at start up (I say it from experience). Not all coils can work at 200KHz.
You should fix your output inductor before continuing. Some suggestions: drum cores from Wilco, 22uH is ok (ready made, they even offer free samples via their webpage) or type-2 toroid from Micrometals.
350mA is still too much just for the power stage. You should have 60-70mA per rail max, as I said.
And don't do tests with no load, please! Your filter can enter resonance and produce erroneous output, as well as bias your modulator circuit incorrectly when feedback is connected due to offsets in the output stage. Use a 4 to 8 ohm power resistor.
Don't worry about feedback until you have fixed the basics of your circuit, or you will not reach proper operation.
The output coil is CRITICAL. Even if the value is correct, from one core to another things can go perfect or burn just at start up (I say it from experience). Not all coils can work at 200KHz.
You should fix your output inductor before continuing. Some suggestions: drum cores from Wilco, 22uH is ok (ready made, they even offer free samples via their webpage) or type-2 toroid from Micrometals.
350mA is still too much just for the power stage. You should have 60-70mA per rail max, as I said.
And don't do tests with no load, please! Your filter can enter resonance and produce erroneous output, as well as bias your modulator circuit incorrectly when feedback is connected due to offsets in the output stage. Use a 4 to 8 ohm power resistor.
Don't worry about feedback until you have fixed the basics of your circuit, or you will not reach proper operation.
1. so , that mean the core must have 200kHz frequency response??
2."60-70mA per rail max",, so , total at +/v rail is 150 -200mA??
3. i want to make sure one , when all circuit work OK .
how many power disspation per rail is right?? 60-70mA??
2."60-70mA per rail max",, so , total at +/v rail is 150 -200mA??
3. i want to make sure one , when all circuit work OK .
how many power disspation per rail is right?? 60-70mA??
I mean you should draw around 60-70mA from one rail, 60-70mA from the other.
About the core, you have to select one with low losses at the frequency of operation.
For example, I once tried to substitute my hand-wound micrometals 20uH toroid by a J.W.Miller 20uH from Digikey. Size and aspect seemed identical.
But... as soon as I turned it on, the coil started to heat up very badly even with no load, and current consumption was much larger than usual.
As I said, another inductor that works quite well at that kind of frequency is Wilco drum core ferrite inductors. Select one with current rating of around 10-15A minimum, and 22uH is ok.
About the core, you have to select one with low losses at the frequency of operation.
For example, I once tried to substitute my hand-wound micrometals 20uH toroid by a J.W.Miller 20uH from Digikey. Size and aspect seemed identical.
But... as soon as I turned it on, the coil started to heat up very badly even with no load, and current consumption was much larger than usual.
As I said, another inductor that works quite well at that kind of frequency is Wilco drum core ferrite inductors. Select one with current rating of around 10-15A minimum, and 22uH is ok.
- Status
- Not open for further replies.