Phase_accurate,
You're spot on, the current configuration with the transistors connected to gnd with the bases indeed greatly reduces the Miller effect and causes for a considerable speed gain.
Pierre, level shifting is indeed needed as the front-end, integrator and comparator, and the back-end, mosfet-driver and output stage, are running off of different power planes, hence interfacing them directly will cause for sparks to fly and parts to fry, not good.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
You're spot on, the current configuration with the transistors connected to gnd with the bases indeed greatly reduces the Miller effect and causes for a considerable speed gain.
Pierre, level shifting is indeed needed as the front-end, integrator and comparator, and the back-end, mosfet-driver and output stage, are running off of different power planes, hence interfacing them directly will cause for sparks to fly and parts to fry, not good.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Yes, I know, but what I was trying to suggest is that you can run the integrator and comparator with a single supply, then you can make their "gnd" be the negative rail. So you wouldn't need level shifting, the only matter is that you would need to AC couple the audio to avoid DC conflicts, and also the feedback.
I have done that in a synchronous design as a test, and it works, but I had some unstable DC offset at the output, but I am sure that can be solved easily.
I have done that in a synchronous design as a test, and it works, but I had some unstable DC offset at the output, but I am sure that can be solved easily.
Alright, on second thought there's no need for a resistor across the C at the integrator, because the integrator simply cannot drift. The capacitor voltage is caught between the comparator and integrator. When the switch level is reached the comparator will switch and so will the output stage and hence the voltage across the capacitor will drop untill the low switch level is reached and then the process repeats itself. That's the beauty of the SODFA concept, courtesy of Tilman (Tillg) and Stephan (Beobachter) and a few others at hifi-forum.de
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Pierre,
You're right about the solution you suggest, that would remove the need to level shift. However it will make the amplifier a lot worse in terms of performance, it might even not work properly at all anymore. Why?
If you add coupling Cs you're creating 1st order filters as well, that, despite their high Fres, will induce not linearities. Because you'll need a minimum of two Cs you're introducing a lot of non-linearities that will (greatly) reduce the performance of the amp in terms of harmonic distortion and overall linearity.
The voltage gain across the integrator will no longer be linear and that's at the basis of the concept, hence coupling Cs is not a good idea.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
You're right about the solution you suggest, that would remove the need to level shift. However it will make the amplifier a lot worse in terms of performance, it might even not work properly at all anymore. Why?
If you add coupling Cs you're creating 1st order filters as well, that, despite their high Fres, will induce not linearities. Because you'll need a minimum of two Cs you're introducing a lot of non-linearities that will (greatly) reduce the performance of the amp in terms of harmonic distortion and overall linearity.
The voltage gain across the integrator will no longer be linear and that's at the basis of the concept, hence coupling Cs is not a good idea.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Pierre,
Yes, the preamp will be DC free as that's the subwoofer filter I'll be using. That still needs to be calculated though, but my primary focus for now is the amplifier, as that's a challenge, a simple active subwoofer filter with variable phase, boost etc. is not that complex in comparison and is well documented throughout the net.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Yes, the preamp will be DC free as that's the subwoofer filter I'll be using. That still needs to be calculated though, but my primary focus for now is the amplifier, as that's a challenge, a simple active subwoofer filter with variable phase, boost etc. is not that complex in comparison and is well documented throughout the net.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Hi,
Sure the integrator is switched and reverses direction, but now add an offset error to the input of the integrating op amp and what will happen?
As it stands it will work fine indeally, such as in a simulator, but with world world op amps you might find a different story. For the cost of a single resistor... I believe for the same reason you'll want to pay careful attention to part selection in this area as well, like find an op amp with low input offset error.
You've seen the electronics texts showing an "ideal" text book active integrator Vs the "practical" integrator? The difference is a single high value resistor.
"SODFA".... yeah I know it stems from the German audio forum but I question its originality, it looks (and sounds) much more like it is strongly based on Bruno's "SODA", only the only notable difference is the integrated IC driver and dual N channels.
I know Tiki brought that up on their forum before and the web translator didn't allow me to decipher their response to that very well at all.
Best Regards,
Chris
Sure the integrator is switched and reverses direction, but now add an offset error to the input of the integrating op amp and what will happen?
As it stands it will work fine indeally, such as in a simulator, but with world world op amps you might find a different story. For the cost of a single resistor... I believe for the same reason you'll want to pay careful attention to part selection in this area as well, like find an op amp with low input offset error.
You've seen the electronics texts showing an "ideal" text book active integrator Vs the "practical" integrator? The difference is a single high value resistor.
"SODFA".... yeah I know it stems from the German audio forum but I question its originality, it looks (and sounds) much more like it is strongly based on Bruno's "SODA", only the only notable difference is the integrated IC driver and dual N channels.
I know Tiki brought that up on their forum before and the web translator didn't allow me to decipher their response to that very well at all.
Best Regards,
Chris
The thread starter actually mentioned the original patent:
http://v3.espacenet.com/origdoc?DB=EPODOC&IDX=DE19838765&F=0&QPN=DE19838765
It is in German only unfortunately. It is quite well written IMO although I don't believe all the statements about its superiority above carrier-based amps (apart fom a 6dB increase in NFB). Its biggest advantage is that its output-frequency remains almost constant and signal-independant. Basically it is a variant of the classic rectangular generator using an OP-AMP.
Regards
Charles
http://v3.espacenet.com/origdoc?DB=EPODOC&IDX=DE19838765&F=0&QPN=DE19838765
It is in German only unfortunately. It is quite well written IMO although I don't believe all the statements about its superiority above carrier-based amps (apart fom a 6dB increase in NFB). Its biggest advantage is that its output-frequency remains almost constant and signal-independant. Basically it is a variant of the classic rectangular generator using an OP-AMP.
Regards
Charles
This amp looks very similar to one of international rectifiers demo boards for a referents level class-d amp.
Pat Allen originaly built one for his sub woofer and is now using it as has full range amp because it sounded so good. I would look into it, you may get some good ideas at least. You can find more information at : http://www.diyaudio.com/forums/showthread.php?s=&threadid=49586 .
If you could get rid of the inverter chip they use that would be nice, because it’s only available in a “very” small SMT package.
Pat Allen originaly built one for his sub woofer and is now using it as has full range amp because it sounded so good. I would look into it, you may get some good ideas at least. You can find more information at : http://www.diyaudio.com/forums/showthread.php?s=&threadid=49586 .
If you could get rid of the inverter chip they use that would be nice, because it’s only available in a “very” small SMT package.
LBHadju,
I'll have to disagree with you there but it is NOT similar to the IRF reference design, that design suffers from a few basic errors that I obviously avoided. Furthermore using SMD devices is not a bad thing, but rather an asset in high-frequency switching circuitry such as a class-D amp.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
I'll have to disagree with you there but it is NOT similar to the IRF reference design, that design suffers from a few basic errors that I obviously avoided. Furthermore using SMD devices is not a bad thing, but rather an asset in high-frequency switching circuitry such as a class-D amp.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
I don’t understand much about class-d, that’s probably why I see them as similar.
I just see a comparator, inverter, and gate driver.
Do you think your amp will perform better then the IRF reference?
I certainly agree that short connections are better at these high frequencies. I’m just saying the SOIC package I think I can solder, However the SSOP (the inverter of the IFR amp) dose not look possible. That’s why at least your use of a different type of inverter looks like a big benefit to me.
Oddly enough I can’t get my hands on any triple inverters in the 8-SOIC package, but if your works better it may not be a problem.
Leve
I just see a comparator, inverter, and gate driver.
Do you think your amp will perform better then the IRF reference?
I certainly agree that short connections are better at these high frequencies. I’m just saying the SOIC package I think I can solder, However the SSOP (the inverter of the IFR amp) dose not look possible. That’s why at least your use of a different type of inverter looks like a big benefit to me.
Oddly enough I can’t get my hands on any triple inverters in the 8-SOIC package, but if your works better it may not be a problem.
Leve
LBHadju,
The IRF design is a classic hysteretic self oscillator with pre filter feedback, my design uses the SODFA concept, because I realize the difference isn't quite clear allow me to illustrate both concepts below and point out the differences:
Hysteric self oscillator:
As is evident from the above image the integrator gets its feedback before the low-pass filter of the amp. In essence this is a sigma-delta modulater, indentical to the concept used in sigma-delta D/A convertors.
SODFA self oscillator:
In the above image you can clearly see the difference, there's also feedback at the comparator input. Due to this feedback the amp is less prone to be affected by fluctuations in the supply lines, up to 20dB better than the hysteric self oscillator.
Obviously there's a lot more going on, but that's the gist of the matter. Overall the SODFA concept is a good one for a sub-amp as obviously the supply lines will be heavily taxed due to the high currents resulting from driving a large woofer at low frequencies.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
The IRF design is a classic hysteretic self oscillator with pre filter feedback, my design uses the SODFA concept, because I realize the difference isn't quite clear allow me to illustrate both concepts below and point out the differences:
Hysteric self oscillator:
As is evident from the above image the integrator gets its feedback before the low-pass filter of the amp. In essence this is a sigma-delta modulater, indentical to the concept used in sigma-delta D/A convertors.
SODFA self oscillator:
In the above image you can clearly see the difference, there's also feedback at the comparator input. Due to this feedback the amp is less prone to be affected by fluctuations in the supply lines, up to 20dB better than the hysteric self oscillator.
Obviously there's a lot more going on, but that's the gist of the matter. Overall the SODFA concept is a good one for a sub-amp as obviously the supply lines will be heavily taxed due to the high currents resulting from driving a large woofer at low frequencies.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Oh, and the invertor used in the IRF design is just a simple 7404, it is available in quite a number of different packages, including DIL14 and smd of course. If you ask me I think the use of digital logic ICs in their design is one of the design flaws, as is the way they do level shifting and there's a few other flaws as well. This does make the IRF reference design look more like a high-school project to me, rather than something you'd expect from a respectable manufacturer such as IR, but that's just my opinion really.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
@ Sander
Fairly you had add the source of your abowe images !
http://hifiakademie.de/.?id=0&si=MTExODQwNjY4Ny4xMTUzfDgwLjE0MC4xNzkuMTE4fCA
Fairly you had add the source of your abowe images !
http://hifiakademie.de/.?id=0&si=MTExODQwNjY4Ny4xMTUzfDgwLjE0MC4xNzkuMTE4fCA
Lenz,
I grabbed them from another forum, hifi-forum.de, as they contained no copyright notice I presumed they could be used freely. If they're yours then I apologize but cannot edit the post as it is closed after 30-mins.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
I grabbed them from another forum, hifi-forum.de, as they contained no copyright notice I presumed they could be used freely. If they're yours then I apologize but cannot edit the post as it is closed after 30-mins.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
http://rocky.digikey.com/WebLib/Linear Tech/Web Data/LT1016.pdf
See Page 2:
Are comparator pins 8 and 5 reversed in your drawing?
-=Randy
See Page 2:
Are comparator pins 8 and 5 reversed in your drawing?
-=Randy
Thanks Randy,
You're absolutely right. I've just checked the datasheet against the library for the LT1016 in my cad design software and it indeed has pins 5 and 8 swapped. I've made a new library that has the correct pinout. The below schematic is the correct one.
class_d_amp_005
http://hardwareanalysis.com/images/articles/large/11527.gif
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
You're absolutely right. I've just checked the datasheet against the library for the LT1016 in my cad design software and it indeed has pins 5 and 8 swapped. I've made a new library that has the correct pinout. The below schematic is the correct one.
class_d_amp_005
http://hardwareanalysis.com/images/articles/large/11527.gif
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Thank you for the explanation and drawings it makes it clear for us guys that are new to these concepts to keep up. I have some more questions if you don’t mind answering them.
Will your amp be usable full range?
Do you dislike the inverter in the IRF board, because it cause as time delay between the two connections going to the gate driver or is there another reason?
Would it be possible to use 2 optocouplers as the inverter/level shifter or are they to slow and / or require too much current? I can draw you an example of what I mean if you wish.
Will your amp be usable full range?
Do you dislike the inverter in the IRF board, because it cause as time delay between the two connections going to the gate driver or is there another reason?
Would it be possible to use 2 optocouplers as the inverter/level shifter or are they to slow and / or require too much current? I can draw you an example of what I mean if you wish.
LBHadju,
Opto couplers could work yes, but I'm trying to build something that'll match, or surpass, other class A/B and class-D amps and is also applicable full range. Optocouplers really don't belong in a design such as this, they're perfectly useable if this was a switch mode power supply, or a bucket convertor, or better yet; a motor driver. Besides the current solution with the two transistors is both elegant, simple and very fast, optocouplers would not offer a better solution in this case.
They might be if I was operating the driver stage at very high voltages and would like to keep the other electronics out of harms way in case of a shortcircuit of a failure in the output stage. In that case we quickly move into the realm of above mentioned SMPS or car-audio or PA use.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
Opto couplers could work yes, but I'm trying to build something that'll match, or surpass, other class A/B and class-D amps and is also applicable full range. Optocouplers really don't belong in a design such as this, they're perfectly useable if this was a switch mode power supply, or a bucket convertor, or better yet; a motor driver. Besides the current solution with the two transistors is both elegant, simple and very fast, optocouplers would not offer a better solution in this case.
They might be if I was operating the driver stage at very high voltages and would like to keep the other electronics out of harms way in case of a shortcircuit of a failure in the output stage. In that case we quickly move into the realm of above mentioned SMPS or car-audio or PA use.
Best regards,
Sander Sassen
http://www.hardwareanalysis.com
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.