Hi guys, i`m new here only few weeks ago and i saw many people doing class D projects, saw some intresting projects arround here. Cheers to all class D fans 
I`m not new in smps/pwm domain, in the last 3-4 months i have learn things about smps fenomens that a year aqo where quite strange to me.
Let me tell you what this project is all about, i want to build a class D amplifier with a rated power of 1000W even more, probably in the future i want to achive 2-2.5 kw. I know it's a lot of power, but there is no need to be sceptic around here, so i want to start this project, i have over 2 years already in pcb design so i think it won't be a problem.
Let me get to the point of this project.
What i want to build is a 1kw @8 ohms in full bridge, non self oscilating triangle, but this will be somewhere at 250-300khz
I want to use LM361 for the main comparator, some 6N137 optocuplors for the optical isolation betwen the Hard switching drives and mosfets.
The output chokes for the Low pass filter wil be some iron powder cores.
I'm not home right now, but tonight i will post a principle schematic of what i wanna achive.
What i wanna ask you guys if you know some documentation regarding the design of feedback uppon this kind of amplifiers, any sugestions are wellcomed.
I wanna say in the end that i want to use this amplifier for 8ohm load and 4ohm load, reading some aplications note from IRF, i saw that the LPF filter is design specific for a single load (ex 8ohm,4ohms,2ohms) i want to do something universal to say so.
What do you think guys?
I`m not new in smps/pwm domain, in the last 3-4 months i have learn things about smps fenomens that a year aqo where quite strange to me.
Let me tell you what this project is all about, i want to build a class D amplifier with a rated power of 1000W even more, probably in the future i want to achive 2-2.5 kw. I know it's a lot of power, but there is no need to be sceptic around here, so i want to start this project, i have over 2 years already in pcb design so i think it won't be a problem.
Let me get to the point of this project.
What i want to build is a 1kw @8 ohms in full bridge, non self oscilating triangle, but this will be somewhere at 250-300khz
I want to use LM361 for the main comparator, some 6N137 optocuplors for the optical isolation betwen the Hard switching drives and mosfets.
The output chokes for the Low pass filter wil be some iron powder cores.
I'm not home right now, but tonight i will post a principle schematic of what i wanna achive.
What i wanna ask you guys if you know some documentation regarding the design of feedback uppon this kind of amplifiers, any sugestions are wellcomed.
I wanna say in the end that i want to use this amplifier for 8ohm load and 4ohm load, reading some aplications note from IRF, i saw that the LPF filter is design specific for a single load (ex 8ohm,4ohms,2ohms) i want to do something universal to say so.
What do you think guys?
Here is how the finalstage looks http://www.webpix.ro/imagini/final.gif
I used TC429 for drives, they are strong enough to handle 6A peaks. I use them so in the near future probably i will want to handle more power on output.
Anyway now i have a little time for this, i am stil working on designing the amp.
By the way nikula, your schematic looks nice. How did you design the feedback circuit? That's what intrest me more, i saw many problems regarding feedback control. In my opinion a bad feedback causes nasty smoke...
Regarding the choper transistors, i too was thinking to use IRFB4227, they are probably the best solution for high power class D amps.
I will come back latter, now i gotta run to work
Cheers
I used TC429 for drives, they are strong enough to handle 6A peaks. I use them so in the near future probably i will want to handle more power on output.
Anyway now i have a little time for this, i am stil working on designing the amp.
By the way nikula, your schematic looks nice
. How did you design the feedback circuit? That's what intrest me more, i saw many problems regarding feedback control. In my opinion a bad feedback causes nasty smoke...Regarding the choper transistors, i too was thinking to use IRFB4227, they are probably the best solution for high power class D amps.
I will come back latter, now i gotta run to work
Cheers
Moro696!
With post-filter feedback: PI controller, and D in the feedback. The PWM modulator can be replaced by a gain of Vdd/V_triangle in the simulation, this way you can adjust the controller's parameters.
Be careful! If you use post-filter feedback, then you must pay many attention for the filter's parasitic effects!
With post-filter feedback: PI controller, and D in the feedback. The PWM modulator can be replaced by a gain of Vdd/V_triangle in the simulation, this way you can adjust the controller's parameters.
Be careful! If you use post-filter feedback, then you must pay many attention for the filter's parasitic effects!
Upper side driver's VCC is 5V?
No dead time in front of gate driver and high value of gate turn-on resistor?
Very low input impedance?
Care to comment?
Here is how the finalstage looks http://www.webpix.ro/imagini/final.gif
I used TC429 for drives, they are strong enough to handle 6A peaks. I use them so in the near future probably i will want to handle more power on output.
Anyway now i have a little time for this, i am stil working on designing the amp.
By the way nikula, your schematic looks nice
Regarding the choper transistors, i too was thinking to use IRFB4227, they are probably the best solution for high power class D amps.
I will come back latter, now i gotta run to work
Cheers
This looks nice, I like the idea of buffering the gate driver, several comments:
- you must oversize the bootstrap capacitors to handle extra supply current from the buffer
- you should experiment with schottky turn-off, this may be tricky to control dV/dt with such a high gate charge current and parralleled mosfets
- output filter should also have capacitors to ground to kill common mode ripple (you filter only differential mode here). Don't let speaker cable to become an antenna
- why do you want an optocoupler?
I think there is error here, since lower one has 12v, I assume other will have it too
http://www.webpix.ro/imagini/full2.gif
what do you think about this fullbridge, especialy the feedback circuit?
what do you think about this fullbridge, especialy the feedback circuit?
This control loop is likely to oscillate, or will be inefficient. You have to place a series resistor to C8 (this makes the proportional part of the controller), and parallel capacitors to R11, R12 (deriving).
I agree with darkfenriz, two capacitors from the outputs to the GND would be nice.
I agree with darkfenriz, two capacitors from the outputs to the GND would be nice.
Last edited:
Principally, for a clocked designed as it is the feedback loop looks workable if you cut the feedback bandwidth properly (you need to experiment here a bit).
The schematic looks a bit dirty in terms of voltage levels, level shifting etc., the IR2110's should be supplied with around +15 related to negative rail, same as optocouplers' secondaries. Hope you already know that.
One possible pitfall is the high/low side PWM shapes. You don't have any dead time in front of optocouplers, I don't know the turn on/off times for them and the transmission delay spread. This may lead to complementary PWM overlaps and cross-conduction problems, maybe it would be better to opto-couple just single ended PWM and symmetrize it afterwards, possibly with dead time adjustment.
The schematic looks a bit dirty in terms of voltage levels, level shifting etc., the IR2110's should be supplied with around +15 related to negative rail, same as optocouplers' secondaries. Hope you already know that.
One possible pitfall is the high/low side PWM shapes. You don't have any dead time in front of optocouplers, I don't know the turn on/off times for them and the transmission delay spread. This may lead to complementary PWM overlaps and cross-conduction problems, maybe it would be better to opto-couple just single ended PWM and symmetrize it afterwards, possibly with dead time adjustment.
To achieve stable feedback in this topology the cutoff freq must be limited under 10 kHz. Do you think it's good for anything? ;-)
Why do you think 12V is not enough?
I'm almost sure he already know this.
HCPL2630 can't be used with this high supply voltage.
It's not a problem, dead time of IR2011 + gate diode can be enough, but if not, the pull-up resistor of the optocouplers with the input cap of IR2011 provides additional dead time (and capacitance can be increased any time).
_______________________________________________________________________________
TC429 is inverting! Use non-inverting gate-driver!
Last edited:
Probably not, but depends a bit on switching frequency and output filter, can be tailored to be "workable"
12V is OK.
I mean they should be related to negative rail, they are mistakenly related to ground at schematic.
Do you like such solutions? I would prefer single optocoupler and glue logics.
Good catch !!
At corner freq an LC filter has 90 degree phase-shift. Integrator has an other 90 degree. Feedback turns to positive, so if its gain is more then 0 dB, then it will oscillate. What is even worse: without load the gain of the filter goes up (theoretically infinite), so it can oscillate with a very weak feedback also. It's practically unusable.
I think some of the earth signs actually means the nagative rail. Otherwise the optos wouldn't be neccessary.
Yes, I do. It's simple but still precise. (As long as logic treshold of IR2011 stays the same. I don't know this IC from practice, I only used IR2110 with HCPL2630, and it has different input).
Sorry, I was wrong! In connection with dead-time I like it indeed, but otherwise it has a nasty behaviour: if you don't drive any of the LED's, then both optos put out high, wich makes all FETs turn on. Actually this is the main reason why I don't use it this way for years (sorry, I forgot it!). With tied power supply turn-on and off sequence it can be avoided, but this makes the circuit too complicated.
Maybe moro696 thinked that the additional inverting at gate-driver solves this problem, but it causes an other: when undervoltage happens, IR2011 will give low level, the gate driver translates it to high, so MOSFET will conduct when it shouldn't.
Pafi, what's your experirnce with feedback in clocked vs. self-oscillating designs?
I agree that both should ideally have a phase-lead network in feedback network to cancel one of filter poles, but to my intuition self-oscillating always allows for more feedback to keep loop stable, don't you agree?
I agree that both should ideally have a phase-lead network in feedback network to cancel one of filter poles, but to my intuition self-oscillating always allows for more feedback to keep loop stable, don't you agree?
First of all sorry for the mess with the ground planes, on my schematic i forgot to note diferent thouse two type of grounds, meaning the signal ground, and the power ground from the final stage.
Now regarding HCPL2630 and IR2011 voltages, the HCPL2630 maximum power suply range from datasheet i know it's 5Volts, and i also know that IR2011 alows TTL 3.3-5 volts signal at imput.
A normal mosfet (except igbts) saturates at Vgs 10Volts, asuming a 12Volts supply voltage for IR2011, i don't thik on thouse drain-source jonctions there will be a fall greater than 2Volts.
Now regarding the feedback, i`m not so experienced in feedback design, but i found out a document that in my opinion shares some light uppon my situation
here is this document: http://users.ece.gatech.edu/~mleach/ece4435/f01/ClassD2.pdf
For the feedback design, i was thinking takeing a lower frequency than the pwm (300khz), i was thinking of 30-50khz for the feedback.
Btw is there any solution to put a optical barier in the feedback circuit? i dont want to use the common ground for both of the stages.
Oh and another thing, i don't want to use inverting drivers for the mosfets (TC429), i want to use TC4420 wich is a noninverting driver
Now regarding HCPL2630 and IR2011 voltages, the HCPL2630 maximum power suply range from datasheet i know it's 5Volts, and i also know that IR2011 alows TTL 3.3-5 volts signal at imput.
A normal mosfet (except igbts) saturates at Vgs 10Volts, asuming a 12Volts supply voltage for IR2011, i don't thik on thouse drain-source jonctions there will be a fall greater than 2Volts.
Now regarding the feedback, i`m not so experienced in feedback design, but i found out a document that in my opinion shares some light uppon my situation
here is this document: http://users.ece.gatech.edu/~mleach/ece4435/f01/ClassD2.pdf
For the feedback design, i was thinking takeing a lower frequency than the pwm (300khz), i was thinking of 30-50khz for the feedback.
Btw is there any solution to put a optical barier in the feedback circuit? i dont want to use the common ground for both of the stages.
Oh and another thing, i don't want to use inverting drivers for the mosfets (TC429), i want to use TC4420 wich is a noninverting driver
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.