Oh, designing the gate-driverstage is the most complicated thing with ClassD...
I think, it's very important to have the correct choice of devices,
these transistors must be able to deliver up to 1 ampere for a short
time, so hfe-curve should not drop to low above ~500ma.
I am not familiar with the 2SC8050...
Cascading more EFs does not really help if the last stage is not able
to deliver the current needed.
Mike
I think, it's very important to have the correct choice of devices,
these transistors must be able to deliver up to 1 ampere for a short
time, so hfe-curve should not drop to low above ~500ma.
I am not familiar with the 2SC8050...
Cascading more EFs does not really help if the last stage is not able
to deliver the current needed.
Mike
Yesterday I resonanced a output filter at 30KHz and got 150V peak output.
I wonder if the swords of JEDI warrior in <<Star Wars>> is done this way.
I wonder if the swords of JEDI warrior in <<Star Wars>> is done this way.
MikeB said:
JohnW: Your work looks really nice!!!!
The 2.2$ was that including the 2 regenerator coils, and bypassing caps? Because in our ZAPpulse we use quite a bit more than 2$ alone for the bypassing caps. 😀
Will the amp get overheated if loaded with 4 Ohms (200W) and will it shut down, or just blow up?
Again i think it's beautiful work! 🙂 Really compact!
Best regards
Lars
The 2.2$ was that including the 2 regenerator coils, and bypassing caps? Because in our ZAPpulse we use quite a bit more than 2$ alone for the bypassing caps. 😀
Will the amp get overheated if loaded with 4 Ohms (200W) and will it shut down, or just blow up?
Again i think it's beautiful work! 🙂 Really compact!
Best regards
Lars
Fast enough, yes ! Powerful enough ? No !
It does nearly the job, the 2n5401 works fine, but the 2n5551
enters some strange breakdown visible with scope.
I wanted to try 2n2222a or bc327, i already bought them, but
had no time yet for this project.
Mike
It does nearly the job, the 2n5401 works fine, but the 2n5551
enters some strange breakdown visible with scope.
I wanted to try 2n2222a or bc327, i already bought them, but
had no time yet for this project.
Mike
I don't know, the 2n5401 didn't show these symptoms, even with
more gatecharge to drive... I don't understand !
Mike
more gatecharge to drive... I don't understand !
Mike
Sorry to come late to this thread.
I'm looking at the first post schematic, but from skimming through the extensive developments since it appears that there has been no explicit schematic revision to reflect all of the changes---or did I miss something? If not, maybe it's time for a new one (hint hint)?
The one thing that astonished me so far is the scope photo of the gentleman's 2nS risetime P-channel output, showing essentially no ringing! I haven't seen a waveform that clean since I designed some clock drivers for a spectrometer using a self-scanned photodiode array back in ~1978. As it was still the dawn of good power MOSFETs I used RF bipolars in quite an elaborate config., finally steering currents in and out of 100 ohm R's with fast diodes.
I don't know whether (again) I may have missed it, but in my experience it is vital to use a probe with a ground ref springy thing adjacent to the tip, or even better a superfast differential probe like the old Tek product. Otherwise you are looking mostly at the ringing in the scope ground lead.
Anyway, good stuff. As a newbie I am enjoying the forum.
I'm looking at the first post schematic, but from skimming through the extensive developments since it appears that there has been no explicit schematic revision to reflect all of the changes---or did I miss something? If not, maybe it's time for a new one (hint hint)?
The one thing that astonished me so far is the scope photo of the gentleman's 2nS risetime P-channel output, showing essentially no ringing! I haven't seen a waveform that clean since I designed some clock drivers for a spectrometer using a self-scanned photodiode array back in ~1978. As it was still the dawn of good power MOSFETs I used RF bipolars in quite an elaborate config., finally steering currents in and out of 100 ohm R's with fast diodes.
I don't know whether (again) I may have missed it, but in my experience it is vital to use a probe with a ground ref springy thing adjacent to the tip, or even better a superfast differential probe like the old Tek product. Otherwise you are looking mostly at the ringing in the scope ground lead.
Anyway, good stuff. As a newbie I am enjoying the forum.
This seems to be the only thread where someone has actually built a working module, not just simulated.
How does it sound, for the brief amount of time you can run it?
Do you think it will become a functional DIY project?
Leve
How does it sound, for the brief amount of time you can run it?
Do you think it will become a functional DIY project?
Leve
Hi !
It does sound quite good, the problem is, the hotter the better...
I think, if solve these "small" problems it will be a functional
DIY-project. I will continue this project later !
Mike
It does sound quite good, the problem is, the hotter the better...
I think, if solve these "small" problems it will be a functional
DIY-project. I will continue this project later !
Mike
Driver Stages
Sorry if I missed anything. You were talking about driver stages. have you considered using a MOSFET driver for the output stage or are we designing completly with BJT's here? I was working on a Class-D amplifier as my honors project for my BE degree last year where I used a IR2113 and I successfully got 60-80Wrms out and I could actually listen to decent audio.
The problem I got was the ringing on the edges (because they were so fast!) was getting back into my small signal stages. This was the sole factor stopping me getting more power outputs.
I would like to spend more time developing it, but now I'm tied up with my masters project designing a Hi-Fi CD pickup. Check out my site for the basics of the design (CB800D).
http://carlslectronics.webhop.net
Sorry if I missed anything. You were talking about driver stages. have you considered using a MOSFET driver for the output stage or are we designing completly with BJT's here? I was working on a Class-D amplifier as my honors project for my BE degree last year where I used a IR2113 and I successfully got 60-80Wrms out and I could actually listen to decent audio.
The problem I got was the ringing on the edges (because they were so fast!) was getting back into my small signal stages. This was the sole factor stopping me getting more power outputs.
I would like to spend more time developing it, but now I'm tied up with my masters project designing a Hi-Fi CD pickup. Check out my site for the basics of the design (CB800D).
http://carlslectronics.webhop.net
Re: Driver Stages
My own reasons for designing fully discrete circuits are cost, flexibility towards optimisation and (believe it or not)... simplicity. In a half-bridge circuit on a split supply, the extra circuitry needed to interface a comparator living around GND and a FET driver living on the negative rail is about as complicated as and more finicky than both functions executed in 3-legged, 2-junction parts.
Umm the thread is called "my fully discrete Class D" 🙂 Presumably it's at least partly a matter of sports.CarlBenton said:
My own reasons for designing fully discrete circuits are cost, flexibility towards optimisation and (believe it or not)... simplicity. In a half-bridge circuit on a split supply, the extra circuitry needed to interface a comparator living around GND and a FET driver living on the negative rail is about as complicated as and more finicky than both functions executed in 3-legged, 2-junction parts.
Oops, Sorry bruno. My bad!😱 and yes you are right, is is difficult when dealing with the driver chip referenced to the minus rail! One of the problems I was having.
Well, lets keep it discret from now on then. lol
I am interested in seeing this design. Like "bcarso" said, I too would like a schematic.
Well, lets keep it discret from now on then. lol
I am interested in seeing this design. Like "bcarso" said, I too would like a schematic.
Hi !
I did not post latest schematic because it's not fully functional and i
can't recommend for building... I'll have to redesign the driverstage
completely. I will keep it fully discrete, just for simplicity and to keep
it easy to build.
Mike
I did not post latest schematic because it's not fully functional and i
can't recommend for building... I'll have to redesign the driverstage
completely. I will keep it fully discrete, just for simplicity and to keep
it easy to build.
Mike
It looks to me like you're suffering from a awful lot of shoot throught in your design.
The schematic:
The simulation
The uppler simulation shows the gate drive voltages. The lower simulation shows the output voltage and the current through the output transistors.
The problem is that the n-mos turns on before the p-mos has turned off. This gives a lot of shoot through when the output svings from high to low. You'll probably get a lot of distortion because of this.
The deadtime when the output swings from low to high, is very long. You should try to minimize this.
I've attached the orcad files.
The schematic:
An externally hosted image should be here but it was not working when we last tested it.
The simulation
The uppler simulation shows the gate drive voltages. The lower simulation shows the output voltage and the current through the output transistors.
The problem is that the n-mos turns on before the p-mos has turned off. This gives a lot of shoot through when the output svings from high to low. You'll probably get a lot of distortion because of this.
The deadtime when the output swings from low to high, is very long. You should try to minimize this.
An externally hosted image should be here but it was not working when we last tested it.
I've attached the orcad files.
I've worked on the power stage. This (see the schematic) seams to work well. Theres nearly no shoot through, however the switching time might be too long. You can try to minimize R34, R35 together with R21 to target this.
The basic idea behind this gate driver configration, is to connect the two mosfet gates through capasitors, so that they are always in sync. This also decreeses deadtime.
Shoot through is avioded by taking advantage of the gate threshold voltages. You can't turn one mosfet on before the other one is off.
Notice that the mosfet currents have been multiplyed with a factor of 10 to make them visible.
I've attached the project if any of you would like to work further.
Regards
Kaspar Sinding Meyer
The basic idea behind this gate driver configration, is to connect the two mosfet gates through capasitors, so that they are always in sync. This also decreeses deadtime.
Shoot through is avioded by taking advantage of the gate threshold voltages. You can't turn one mosfet on before the other one is off.
An externally hosted image should be here but it was not working when we last tested it.
Notice that the mosfet currents have been multiplyed with a factor of 10 to make them visible.
An externally hosted image should be here but it was not working when we last tested it.
I've attached the project if any of you would like to work further.
Regards
Kaspar Sinding Meyer
- Status
- Not open for further replies.