I use IR2110 to build a class-d amp.
But It always has Crackle sound , when output amplitude get clipping.
I don't have any ideal why it happen and don't know how to improve it ...
The class-d is different with class-ab , i know.
But why class-d amp get clip will occure the crackle ???
They were getting clip , but class-ab will not get crackle....
somebody said the crackle is normal ,, perhaps....
But can not we solve it or improve it ???
Or it is the disadvantage with class-d amp???
I use HIP4080 to build a class-d amp before and try to clip the output amplitude.
It get very stable ...
The amplitude can rise up to +RAIL and clip.
So , i think it has other way can do it .
But It always has Crackle sound , when output amplitude get clipping.
I don't have any ideal why it happen and don't know how to improve it ...
The class-d is different with class-ab , i know.
But why class-d amp get clip will occure the crackle ???
They were getting clip , but class-ab will not get crackle....
somebody said the crackle is normal ,, perhaps....
But can not we solve it or improve it ???
Or it is the disadvantage with class-d amp???
I use HIP4080 to build a class-d amp before and try to clip the output amplitude.
It get very stable ...
The amplitude can rise up to +RAIL and clip.
So , i think it has other way can do it .
There are several reasons for "crackle sound" to be introduced when clipping:
- IR211x boostracp capacitor being drained too quickly and under-voltage lockout protection kicking in.
- The control loop can no longer control the output filter during clipping, so the filter will resonate at its natural frequency until normal operation is resumed. RC damping may improve that a bit, altough filter resonance should be chosen above the audio band to start with.
- Control loop instability. In practice, phase margin of systems using global feedback is reduced when the output comes very close to the rails, so careful adjustment of compensation networks is required. Also, for oscillator based designs, sawtooth edge precision is critical there.
There are too many ways to get it wrong.
- IR211x boostracp capacitor being drained too quickly and under-voltage lockout protection kicking in.
- The control loop can no longer control the output filter during clipping, so the filter will resonate at its natural frequency until normal operation is resumed. RC damping may improve that a bit, altough filter resonance should be chosen above the audio band to start with.
- Control loop instability. In practice, phase margin of systems using global feedback is reduced when the output comes very close to the rails, so careful adjustment of compensation networks is required. Also, for oscillator based designs, sawtooth edge precision is critical there.
There are too many ways to get it wrong.
Eve put you on the good way. First check your bootstrap capacitor if it can handle long pulse at clipping. In my design, I use a 47uF and 2X 0.1uF Tantal dip. Look to at the recovery time of your bootstrap diode...Check the output wave with a scope if their any wave over the signal when your near the clipping, this will indicate that your core is saturing. Try to reduce the overall gain too and drive your amp with more voltage. Reducing too the amplitude of the triangle wave could improve thing. Maybe another source, try to add more decoupling on the main power supply, between power supply and ground, and a place that nobody try, between the aux power for the Ir2110 and the -15V of your drive circuit. This will canmcle lot of noise between low voltage power supply and hight power power supply..
Fred
www.d-amp.com
Fred
www.d-amp.com
It's not so easy to get a floating power supply because the "midpoint between MOSFETs" is being switched back and forth at speeds grater than 1V/ns. Anything connected to that node will be a nice antenna, and the required common-mode filter to solve that problem may be bigger than the output inductor itself!
The bootstrapped resistor/capacitor or diode/capacitor supply is far more practical and quiet. Also, IR211x ICs draw less than 0.5mA supply current on each output cell as long as output state is not changed, so a 4.7uF capacitor already allows for more than 10ms of clipping while discharging by just 1V. Figure out that I'm using 2.2uF without any starvation problems (with IGBTs and feeding 18V to the gates).
The crackle is probably being caused by more complex phenomena. Your oscilloscope will tell you...
The bootstrapped resistor/capacitor or diode/capacitor supply is far more practical and quiet. Also, IR211x ICs draw less than 0.5mA supply current on each output cell as long as output state is not changed, so a 4.7uF capacitor already allows for more than 10ms of clipping while discharging by just 1V. Figure out that I'm using 2.2uF without any starvation problems (with IGBTs and feeding 18V to the gates).
The crackle is probably being caused by more complex phenomena. Your oscilloscope will tell you...
Eva,
I didn't mean it is easy to solve practically, only that it is conceptually simple (and works in the simulations as well ;-)
I agree in that the bootstrap cap. approach is much better. In fact it has worked very well for me even at clipping. I really doubt your crackle problem is due to that.
I didn't mean it is easy to solve practically, only that it is conceptually simple (and works in the simulations as well ;-)
I agree in that the bootstrap cap. approach is much better. In fact it has worked very well for me even at clipping. I really doubt your crackle problem is due to that.
If it "crackles", something is wrong. I've the same thing happening with my very prototype IR2110 + 2 x TC4421 + 2 x IRFB31N20D halfbridge setup. The "crackle" is potentially damaging to the speaker(s).
To easily rule out problems with the bootstrapping: remove bootstrap diode completely, connect a small external 12V..15V battery with correct polarity(!) to the bootstrap capacitor on the IR2110 high side, and don't connect this battery anywhere else. Run your bridge. If there's no crackle, you know where the problem is...
- Jan
To easily rule out problems with the bootstrapping: remove bootstrap diode completely, connect a small external 12V..15V battery with correct polarity(!) to the bootstrap capacitor on the IR2110 high side, and don't connect this battery anywhere else. Run your bridge. If there's no crackle, you know where the problem is...
- Jan
And by the way, if you get clipping and you are using direct digital PCM to PWM conversion which never should clip, then seeing clipping also indicates a problem with the bootstrap supply. Most likely the IR2110 goes into undervoltage lockout and shuts down the upper gate drive, which you'd see as a big gap or clip in the audio output.
But if you are using "natural" analog to PWM conversion, the of course clipping is "normal" like in class AB etc whatever.
But if you are using "natural" analog to PWM conversion, the of course clipping is "normal" like in class AB etc whatever.
Eva said:
EVA,
Simply, Imagine a Sinewave of 20Hz 50mS clipped heavily and IR2110 driving the mosfets with Total Gate Charge Qg easily exceeding 200nC.....with Switching frequency of 250KHz...resulting in 50mA of minimum gate current demand .....what would happen then.....
Either use a floating supply
Or use a Chargepump based on 555 Timer
OR use enough big capacitor for bootraping 100mFd likewise....
Nothing would happen. The bootstrap capacitor becomes fully charged on each switching cycle, and on the other hand, there are no switching events during clipping. Also, current to charge the gates is drawn from the capacitor only once after each charging event, so clipping has no influence there (it can be merely considered as a switching period much longer than usual).
Furthermore, shall the bootstrap capacitor had something to do with the clipping problems, such problems should only appear on positive clipping.
Furthermore, shall the bootstrap capacitor had something to do with the clipping problems, such problems should only appear on positive clipping.
Eva said:
I was also talking about positive high side switch during Clipping only....
You were incorrect in your saying that nothing would happen, In case of IRFP260N with Qg = 234nC with switching frequency at 250KHz the high side during clipping would show more switching losses...but if you use some means of charging the bootstrap capacitor or with a charge pump, it will not shown much switching losses and the mosfet , the high side switch wouldnot destroyed eventually...during clipping...
It's probably much less critical with a driver IC than with a "standard" discrete driver circuit, but during positive half cycle while the body diode conducts Miller would serve to load Vgs which would draw from the bootstrap capacitor.
This could be even more detrimental if you'd opt to select a more "optimal" gate source voltage for high speed switching, like 8 volts for instance.
I think Charles had a real neat Idea to solve this issue, easily and cheaply, and were I using such a driver IC I believe I'd give it a serious look.
This could be even more detrimental if you'd opt to select a more "optimal" gate source voltage for high speed switching, like 8 volts for instance.
I think Charles had a real neat Idea to solve this issue, easily and cheaply, and were I using such a driver IC I believe I'd give it a serious look.
Workhorse!
Eva is almost right, there is no switching event during hard clipping! 100% duty cycle! But in the border, at 98...99 % duty cycle there is switching on the gate-driver, but the IGBT or FET can't switch off, so bootstrap cap can't be charged is this cycle. This is a very rear state of operation, so a 2,2 uF can have enough charge to supply if it charged to 18 V.
But there is a different explanation of crackle sound. Control loop can be overcharged during clip. If there is an integrator in the loop, its voltage should be limit to shomewhat more then it is neccessary for full drive.
The output waveform must be observed. If there is breakdown on positive side -> bootstrap problem, if output signal stuck on supply line for 20-100 us after the end of input overload -> control loop problem. Observation of the integrator is advisable too.
Eva is almost right, there is no switching event during hard clipping! 100% duty cycle! But in the border, at 98...99 % duty cycle there is switching on the gate-driver, but the IGBT or FET can't switch off, so bootstrap cap can't be charged is this cycle. This is a very rear state of operation, so a 2,2 uF can have enough charge to supply if it charged to 18 V.
But there is a different explanation of crackle sound. Control loop can be overcharged during clip. If there is an integrator in the loop, its voltage should be limit to shomewhat more then it is neccessary for full drive.
The output waveform must be observed. If there is breakdown on positive side -> bootstrap problem, if output signal stuck on supply line for 20-100 us after the end of input overload -> control loop problem. Observation of the integrator is advisable too.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.