Hi subwo,
About the comparator I have understood.
HAHAHA- I was mistaken thinking that 2 zener better. My simulation is ok from 35v up to 62v for this circuit.
About the comparator I have understood.
HAHAHA- I was mistaken thinking that 2 zener better. My simulation is ok from 35v up to 62v for this circuit.
An externally hosted image should be here but it was not working when we last tested it.
Hi IVX,
I understand the circuit now.😉 I see quite a number of advantages with it. I still tend to find Figure B more attractive. How much slower is figure B than C? Take Care.
I understand the circuit now.😉 I see quite a number of advantages with it. I still tend to find Figure B more attractive. How much slower is figure B than C? Take Care.
Hi subwo,
<How much slower is figure B than C? - if fig B use very fast and mid power BJT (Vceo> 50v (> 65v for fig A 😡 ) Ptot> 2w) for lowering R1, R2 as possible, then slower but not so much. (just turn on Micro-cap
). Actually fig C working by capacitance feed (zener have 35pF) from comparator (BJT ` s eliminable too) to driver (logic IC) which zener DC offset. Anything maybe faster? Other question - whether such speed is necessary?
best regards.
<How much slower is figure B than C? - if fig B use very fast and mid power BJT (Vceo> 50v (> 65v for fig A 😡 ) Ptot> 2w) for lowering R1, R2 as possible, then slower but not so much. (just turn on Micro-cap
). Actually fig C working by capacitance feed (zener have 35pF) from comparator (BJT ` s eliminable too) to driver (logic IC) which zener DC offset. Anything maybe faster? Other question - whether such speed is necessary?best regards.
You can say that the faster your circuits, the more accurate can the timing be (I don't say that it would always be more accurate !!).
Errors in time AND errors in amplitude both contribute to errors in linearity, say distortion.
Regards
Charles
Errors in time AND errors in amplitude both contribute to errors in linearity, say distortion.
Regards
Charles
Hi IVX,
I was just thinking that speed was the reason for figure C. I don't think speed is as important for subwoofer amplifiers. 🙂
Maybe, in the future, I will work on a low to medium power class D amp for midrange and treble. Speed will be important for that because the switching frequency may be as high as a few megahertz.
I was just thinking that speed was the reason for figure C. I don't think speed is as important for subwoofer amplifiers. 🙂
Maybe, in the future, I will work on a low to medium power class D amp for midrange and treble. Speed will be important for that because the switching frequency may be as high as a few megahertz.
Proposal
Hello all.
What do you think about using a IR2184 driver, which as fixed dead-time and a single input, to simplify the design?
The output current (1700mA) seems enough (I limited it to about 1500mA with success in my IR2110 design). The thing is that the dead-time is about 450 ns. Is this practical for achieving low distortion using feedback for a switching frequency of about 200KHz? What do you think?
I was thinking in that using a small Maxim Class-D MAX4295 amplifier and this IR chip with 2 mosfets you can build a VERY small and simple but powerful amplifier. It worths a try, doesn't it?
Sergio
Hello all.
What do you think about using a IR2184 driver, which as fixed dead-time and a single input, to simplify the design?
The output current (1700mA) seems enough (I limited it to about 1500mA with success in my IR2110 design). The thing is that the dead-time is about 450 ns. Is this practical for achieving low distortion using feedback for a switching frequency of about 200KHz? What do you think?
I was thinking in that using a small Maxim Class-D MAX4295 amplifier and this IR chip with 2 mosfets you can build a VERY small and simple but powerful amplifier. It worths a try, doesn't it?
Sergio
The fact that ideally everything should follow the KISS principle makes this definitely worth a try. A dead-time that large is a bit much however.
Regarding fast control of an IR21xy's inputs:
Within reasonable limits it might be possible to use balanced line -drivers and -receivers like they are used for fast data interfaces. One has to take care of the offset voltage however. Maybe some PNP level shifters with parallel coupling Cs for speed would do the trick.
Regards
Charles
Sub Amp
Hello all,
I have completed the basic design for a sub amp. I have simplified the current source for the differential input pair down to a resistor from the previous posted diagram since the current mirror helps to maintain the bias point for minimal DC offset at the speaker.
The amp self oscillates at 100khz with no input signal and decreases to about 25khz at full output at 90% of the power supply rails. It is good not to let a self oscillating amp clip at the rails, but that disadvantage does not come close to negating the advantges of self-oscillation. 
C3 and C4 are internal to the input gate of the IR2113 and are included only for simulation. Decreasing the values R1 and R21 will let the amp oscillate at a higher frequency. I really want to use 2SA1123 transistors for the input pair, but I do not have the spice model for them and don't really know how they will affect the circuit operation, except I would expect the oscillation frequency to increase.
The node point "M" will be circuit ground since the digital part of the simulation of the 74C14 requires that its negative lead be at ground. That problem is due to my lack of talent at computer programming to change the spice files.
Increasing the output inductor, L1, also lowers the oscillation frequency. For now, I chose a low value because the air core inductor calculator program produced an inductor with nice physical size for that value.
I was thinking that it probably would be good if Q1 and Q2 of the current mirror were smaller and faster. But those worked well in the simulation.
Hello all,
I have completed the basic design for a sub amp. I have simplified the current source for the differential input pair down to a resistor from the previous posted diagram since the current mirror helps to maintain the bias point for minimal DC offset at the speaker.
The amp self oscillates at 100khz with no input signal and decreases to about 25khz at full output at 90% of the power supply rails. It is good not to let a self oscillating amp clip at the rails, but that disadvantage does not come close to negating the advantges of self-oscillation.
C3 and C4 are internal to the input gate of the IR2113 and are included only for simulation. Decreasing the values R1 and R21 will let the amp oscillate at a higher frequency. I really want to use 2SA1123 transistors for the input pair, but I do not have the spice model for them and don't really know how they will affect the circuit operation, except I would expect the oscillation frequency to increase.
The node point "M" will be circuit ground since the digital part of the simulation of the 74C14 requires that its negative lead be at ground. That problem is due to my lack of talent at computer programming to change the spice files.
Increasing the output inductor, L1, also lowers the oscillation frequency. For now, I chose a low value because the air core inductor calculator program produced an inductor with nice physical size for that value.
I was thinking that it probably would be good if Q1 and Q2 of the current mirror were smaller and faster. But those worked well in the simulation.
Attachments
Seems quite simple (what's very good).
Then the amplifier is to operate with symmetric suppy rails or only a positive voltage?
Let us know how it goes, ok?
Best regards
Then the amplifier is to operate with symmetric suppy rails or only a positive voltage?
Let us know how it goes, ok?
Best regards
Sorry! I dind't look at the voltage sources. It is now clear that the amplifier works with symmetric +/-80V rails. Good!
Sergio
Sergio
Some comments on your circuit, subwo1:
You seem to have used the same dead-time control that I did. However, I found by experimentation that that dead-time was not enough and had to add a 33pF capacitor in parallel with the input capacitance of the IR2113 (in my case I put a XOR gate before to make the signal square, altough I guess it is not necessary). I found the minimum dead-time to be about 120ns in order to minimize dissipation in the mosfets (that's for mine, perhaps that time is different for yours).
Another thing: Taking the pwm signal from one inverter and the inverted version from the output of other inverter cascaded with the first has a disadvantage: the second inverter introduces a delay. That causes more distortion. Perhaps with proper feedback and for subwoofer use that's not important, but at least have it mind.
Did you have in mind the voltage ratings of the input transistors? It is only a question, I have not checked it, only that we are talking about 160V supplies.
I am specially interested on the results obtained with a hand made inductor. Tell us how it goes with your prototype!
Best regards
You seem to have used the same dead-time control that I did. However, I found by experimentation that that dead-time was not enough and had to add a 33pF capacitor in parallel with the input capacitance of the IR2113 (in my case I put a XOR gate before to make the signal square, altough I guess it is not necessary). I found the minimum dead-time to be about 120ns in order to minimize dissipation in the mosfets (that's for mine, perhaps that time is different for yours).
Another thing: Taking the pwm signal from one inverter and the inverted version from the output of other inverter cascaded with the first has a disadvantage: the second inverter introduces a delay. That causes more distortion. Perhaps with proper feedback and for subwoofer use that's not important, but at least have it mind.
Did you have in mind the voltage ratings of the input transistors? It is only a question, I have not checked it, only that we are talking about 160V supplies.
I am specially interested on the results obtained with a hand made inductor. Tell us how it goes with your prototype!
Best regards
Or maybe use a cascode !
If this straigtforward and intelligent amp performs well, then there is nothing left to do than to congratulate !! 😎
I did some simulations on self oscillating PWM amps but I always had performance worse than using non-selfoscillating topologies. But I have to admit that 1.) simulations are simulations and 2.) I didn't study the subject with much depth.
The best ones among these were always the ones using hysteresis like yours. While the ones using phase-shift for oscillation should have some theoretical advantages, I never got them right.
Regards
Charles
The hysteresis to what Charles refers to is due to the schmidt-trigger inverter, 74C14, am I right?
By the way, subwo1, how are you going to implement the supply for the IR2113 (I wouldn't use less than 12V) and the 74C14?
I used a resistor in series with the input of a regulator in order to get 15V referenced to the negative rail from it without the need to use another power supply. However, with 80V you will need to use a BIG resistor.
What about a separate transformer, rectifying it and connecting the (-) of the bridge to the negative 80V rail as Crest LT does?
Another option is using a switching step-down regulator.
Best regards.
By the way, subwo1, how are you going to implement the supply for the IR2113 (I wouldn't use less than 12V) and the 74C14?
I used a resistor in series with the input of a regulator in order to get 15V referenced to the negative rail from it without the need to use another power supply. However, with 80V you will need to use a BIG resistor.
What about a separate transformer, rectifying it and connecting the (-) of the bridge to the negative 80V rail as Crest LT does?
Another option is using a switching step-down regulator.
Best regards.
Hello subwo,
this DT adjusting is dangerous in real life (IMHO), however D3,D4 have capacitance about 5pf too.
this DC adjusting have not good temperature stability. (you need twin BJT`s).
IR2184 best choice for sub application, i agree too!
this DT adjusting is dangerous in real life (IMHO), however D3,D4 have capacitance about 5pf too.
this DC adjusting have not good temperature stability. (you need twin BJT`s).
IR2184 best choice for sub application, i agree too!
- Status
- Not open for further replies.