The AD1994 is out now. If I'm reading the datasheet correctly, the switching frequency would be 6MHz 
Are there any power stage design, that is capable of operating at such a high switching frequency?
Will IR2011 do? Or is it the time for discrete gate driver?
Are there any power stage design, that is capable of operating at such a high switching frequency?
Will IR2011 do? Or is it the time for discrete gate driver?
Hi,
You don't understand how this chip does what it does. Neither do I. However I can tell you it doesn't switch anywhere near that frequency. It would be silly to introduce a chip to the market that does, very few indeed would be able to make any use of it at all.
http://www.beis.de/Elektronik/DeltaSigma/DeltaSigma.html
I find that link to be an excellent read!!!!!!!!!
Really, it should almost be a sticky here. (hint)
Anyway from what I've gathered of the data sheet which is extremely complete, switching frequency varies somewhere around 475khz.
Cheers
You don't understand how this chip does what it does. Neither do I. However I can tell you it doesn't switch anywhere near that frequency. It would be silly to introduce a chip to the market that does, very few indeed would be able to make any use of it at all.
http://www.beis.de/Elektronik/DeltaSigma/DeltaSigma.html
I find that link to be an excellent read!!!!!!!!!
Really, it should almost be a sticky here. (hint)
Anyway from what I've gathered of the data sheet which is extremely complete, switching frequency varies somewhere around 475khz.
Cheers
That's more 12V powered boom-box stuff rated at 10% THD and with pre-filter feedback.
They don't tell anything about it, but they are using special inductors (maybe shielded air cored) to obtain such low THD figures. In practice, when using mundane output inductors and capacitors, only the hysteresis from the cores will already account for more than 1% THD at medium powers (but they measure at 1W output).
Also, the 12.288 Mhz clock is for the delta-sigma modulator, not for the carrier. In practice, carrier frequency varies wildly and may be anywhere above 48Khz (modulator sampling frequency).
They don't tell anything about it, but they are using special inductors (maybe shielded air cored) to obtain such low THD figures. In practice, when using mundane output inductors and capacitors, only the hysteresis from the cores will already account for more than 1% THD at medium powers (but they measure at 1W output).
Also, the 12.288 Mhz clock is for the delta-sigma modulator, not for the carrier. In practice, carrier frequency varies wildly and may be anywhere above 48Khz (modulator sampling frequency).
I think they simply measured THD before the output filter.
It is quite interesting that a product by one of the most capable IC manufacturers is delayed by more than a year BTW.
Edit: Maybe this chip by itself could be used as modulator/driver for a more powerful amp.
Regards
Charles
It is quite interesting that a product by one of the most capable IC manufacturers is delayed by more than a year BTW.
Edit: Maybe this chip by itself could be used as modulator/driver for a more powerful amp.
Regards
Charles
That was a great link! thanks 
So 6MHz is only the sampling frequency. And the output spectrum might be implicit due to the delta sigma scheme........
Any suggestion on external power stage to boost more power?
So 6MHz is only the sampling frequency. And the output spectrum might be implicit due to the delta sigma scheme........
Any suggestion on external power stage to boost more power?
phase_accurate said:
I've no doubt it could, very much like Zetex demonstrate in the data sheet for one of their modulators (I'm ignorant and won't check which).
Would it be worth doing though? I'd have to think performance would only get worse... I guess it depends alot on the drivers, both before and after.
Regards,
chris
I can also recommend a book on the same subject as the above link, it is the book where i got most of my inspiration for Class D amplifier construction.
It is IEEE no PC0274-1 Oversampling Sigma Delta Converters (Theory Design and Simulation) by James C Candy and Gabor C Temes.
It covers many of the same subjects as the link above, but maybe goes more indepth.
Sorry i have to say, this claim from EVA is simply untrue. If so it would be impossible to make any kind of Class D amplifier with less than 0.3% THD. And we all know it is. Sorry EVA
It is IEEE no PC0274-1 Oversampling Sigma Delta Converters (Theory Design and Simulation) by James C Candy and Gabor C Temes.
It covers many of the same subjects as the link above, but maybe goes more indepth.
Sorry i have to say, this claim from EVA is simply untrue. If so it would be impossible to make any kind of Class D amplifier with less than 0.3% THD. And we all know it is. Sorry EVA
Hi Lars,
Thanks for the book recommendation.
Regarding Eva's comment, keywords were "in practice", "mundane" and "medium powers", right along with the mention of them most likely using rather high end specialized inductors, just like you do to get good measurements.
It does make alot of sense that they'd go to great lengths in order to get optimal measurements for advertising, and such IC's often don't end up performing according to spec I understand.
It was also stated they don't measure at half power but at 1W, which further makes the measurements look better to the untrained eye, it really isn't telling of the true story (unless you know to look for all that they've omitted).
I think her comment was right on the money, and did a good job of demonstrating why it's just another junk box chip amp, destined for a 12$ radio.
Your quoted text took all that right out of context.
I find alot who don't begin to grasp class d yet and who are drawn in by the buzz going around often see these IC solutions as quick and cheap auto access to true high end audio. Personally I think that's a mistake, and it doesn't at all hurt to point out why it likely isn't as good as they try to make it seem.
That aside, Lars, what's your take on the IC solution, can ever be high end?
Regards,
Chris
Thanks for the book recommendation.
Regarding Eva's comment, keywords were "in practice", "mundane" and "medium powers", right along with the mention of them most likely using rather high end specialized inductors, just like you do to get good measurements.
It does make alot of sense that they'd go to great lengths in order to get optimal measurements for advertising, and such IC's often don't end up performing according to spec I understand.
It was also stated they don't measure at half power but at 1W, which further makes the measurements look better to the untrained eye, it really isn't telling of the true story (unless you know to look for all that they've omitted).
I think her comment was right on the money, and did a good job of demonstrating why it's just another junk box chip amp, destined for a 12$ radio.
Your quoted text took all that right out of context.
I find alot who don't begin to grasp class d yet and who are drawn in by the buzz going around often see these IC solutions as quick and cheap auto access to true high end audio. Personally I think that's a mistake, and it doesn't at all hurt to point out why it likely isn't as good as they try to make it seem.
That aside, Lars, what's your take on the IC solution, can ever be high end?
Regards,
Chris
Lars Clausen said:
Sorry i have to say, this claim from EVA is simply untrue. If so it would be impossible to make any kind of Class D amplifier with less than 0.3% THD. And we all know it is. Sorry EVA
Whilst I might disagree with EVA's 'statement' I still might pay attention for other reasons..... such as 'I wouldn't be so sure myself without looking closer'.
Of course you have done the sums before saying it is 'simply untrue'?
I think the point that was being made was that feedback was being taken prior to the output filter components.
I don't know much about this stuff myself other than the concept of a 1 bit oversampled A/D converter but I know as much as that a first order compensation (whoops it's a feedback loop) is naturally stable.
I would guess that references to second order loops really include a first order section at crossover.
What people care to do in the digital domain, after getting a digital answer, will be over my head.... and probably borrow terms from elsewhere, but it sounds like fiddling about.
Perhaps I should read more?
I think the basic point is this stuff needs pre-filter feedback because post filter feedback would make it go horribly wrong in a way that the digital johnnys would not have enough flexibility or the application engineers to deal with.
Fortunately, with pre-filter feedback they don't have to bother.
Problem solved!!(?????)
DNA
Hi Chris
My take on the AD1994 is that it's a shame it's fitted in a 9 by 9 mm lead frame chip carrier (with incredibly small DIY repelling solder pins), because if it was a little more accessible, it would be obious to make the Class D Gain Clone out of it 🙂
Even if it's not high end by any standard, it's still not all bad.
Is it a she?
All the best from
Lars 😉
My take on the AD1994 is that it's a shame it's fitted in a 9 by 9 mm lead frame chip carrier (with incredibly small DIY repelling solder pins), because if it was a little more accessible, it would be obious to make the Class D Gain Clone out of it 🙂
Even if it's not high end by any standard, it's still not all bad.
Is it a she?All the best from
Lars 😉
Lars:
I'm currently working with "mundane" gapped power ferrites, like: N27, N67, F44 and 3C90. Distortion due to hysteresis becomes already evident when looking at the inductor current ripple waveforms with all these materials, altough THD figures below 1% may be still attainable with pre-filter feedback because inductance remains more or less constant until saturation.
However, it's quite obvious that nobody uses gapped ferrites with these chips. My main concern was about the *iron-powder* inductors typically employed in real-world applications, whose inductance usually drifts by at least +-25% depending on instantaneous DC bias and whose hysteresis is far worse than in ferrites. Some (fair) datasheets and application notes reveal that devastating truth in form of 10% THD figures below clipping, while others just employ a high quality output filter (far outside the cost scope of the target applications) without telling anything about it.
Hooking the THD analyzer directly to the switching node through a very linear signal filter seems to be another common practice for THD measurement in the datasheets of ICs featuring pre-filter feedback, as it has been pointed out.
I'm currently working with "mundane" gapped power ferrites, like: N27, N67, F44 and 3C90. Distortion due to hysteresis becomes already evident when looking at the inductor current ripple waveforms with all these materials, altough THD figures below 1% may be still attainable with pre-filter feedback because inductance remains more or less constant until saturation.
However, it's quite obvious that nobody uses gapped ferrites with these chips. My main concern was about the *iron-powder* inductors typically employed in real-world applications, whose inductance usually drifts by at least +-25% depending on instantaneous DC bias and whose hysteresis is far worse than in ferrites. Some (fair) datasheets and application notes reveal that devastating truth in form of 10% THD figures below clipping, while others just employ a high quality output filter (far outside the cost scope of the target applications) without telling anything about it.
Hooking the THD analyzer directly to the switching node through a very linear signal filter seems to be another common practice for THD measurement in the datasheets of ICs featuring pre-filter feedback, as it has been pointed out.
Effects of hysteresis on a F44 ETD39 gapped inductor:
Effects of hysteresis on an inductor similar to the above but wound on a "26" material yellow/white iron powder toroid:
I have drawn straight lines for reference. Duty cycle is intentionally unbalanced to better show non-linearity. The iron powder core (40mm diameter) doesn't even seem to handle 50Khz well as temperature rise is higher than acceptable, while in the gapped ferrite it's negligible.
An externally hosted image should be here but it was not working when we last tested it.
Effects of hysteresis on an inductor similar to the above but wound on a "26" material yellow/white iron powder toroid:
An externally hosted image should be here but it was not working when we last tested it.
I have drawn straight lines for reference. Duty cycle is intentionally unbalanced to better show non-linearity. The iron powder core (40mm diameter) doesn't even seem to handle 50Khz well as temperature rise is higher than acceptable, while in the gapped ferrite it's negligible.
You are using 'hysteresis' but I think you are talking about the way the material saturates.....?... Your plots show it.
Esoteric mixes of worms with Nickel.
Looking at the Arnold data stuff shows some curves that say how much the permeability 'holds up' under DC bias.... (DC means, what something is sitting on, even if it is a bass drum)
They quote permeability figures for their cores, and 26 could be one of them.....
However if you design your inductor to be 20uH with zero amps through it it might become 10uH when someone hits a bass drum.
If the violin player is sitting on top of that then they might suffer a bit of 'transient intermodulation distortion'.
DNA
Esoteric mixes of worms with Nickel.
Looking at the Arnold data stuff shows some curves that say how much the permeability 'holds up' under DC bias.... (DC means, what something is sitting on, even if it is a bass drum)
They quote permeability figures for their cores, and 26 could be one of them.....
However if you design your inductor to be 20uH with zero amps through it it might become 10uH when someone hits a bass drum.
If the violin player is sitting on top of that then they might suffer a bit of 'transient intermodulation distortion'.
DNA
EVA: Thank You for your curves, i would have to agree with DNA that we are talking saturation here. The gapped core saturates at a much higher point than the ungapped core, and with a softer slope.
However i would have to disagree with DNA about the the violin and the drumbeat. A change in the inductance will only cause a change of inaudible residuals, and not cause any change to the audio signal. Anyway a properly dimensioned ferrite coil will not change it's value mentionably from 1 Amp to it's saturation point.
Let me show you an audio related curve, showing the output at 100W RMS 4 Ohms vs. freq. The signal shown actually passes through a ferrite choke, with no feedback after the choke.
I repeated the same test with an air coil, and it was not noticable better. So i am convinced that most of this (rather limited) distortion comes from other places in the amplifier, not the ferrite core.
However i would have to disagree with DNA about the the violin and the drumbeat. A change in the inductance will only cause a change of inaudible residuals, and not cause any change to the audio signal. Anyway a properly dimensioned ferrite coil will not change it's value mentionably from 1 Amp to it's saturation point.
Let me show you an audio related curve, showing the output at 100W RMS 4 Ohms vs. freq. The signal shown actually passes through a ferrite choke, with no feedback after the choke.
I repeated the same test with an air coil, and it was not noticable better. So i am convinced that most of this (rather limited) distortion comes from other places in the amplifier, not the ferrite core.
Attachments
Frankly there is not much of a difference. I have an aircoil with exactly the same value as my ferrite inductor. But i was never able to solve any 'problems' with the aircoil. So though in theory the aircoil should have a lower distortion, than the ferrite coil, i have not been able to confirm this with real measurements.
Another thing i would like to promote is, that every DIY'er should have a THD measurement system. The one i have is a really inexpensive PC based one, and many others alike are available. The only thing you have to make yourself is a breakout box with 4 / 8 Ohms load resistors, and two potentiometers for level equlization.
Nothing is better than tuning your amplifier to your own exact loudspeakers, and letting them perform at their best. It lifts the performance right up to a new level.
And also you can verify the technical specs some producer's have on their product sheet. They don't always match real life performance.
Another thing i would like to promote is, that every DIY'er should have a THD measurement system. The one i have is a really inexpensive PC based one, and many others alike are available. The only thing you have to make yourself is a breakout box with 4 / 8 Ohms load resistors, and two potentiometers for level equlization.
Nothing is better than tuning your amplifier to your own exact loudspeakers, and letting them perform at their best. It lifts the performance right up to a new level.
And also you can verify the technical specs some producer's have on their product sheet. They don't always match real life performance.
Lars Clausen said:Frankly there is not much of a difference. I have an aircoil with exactly the same value as my ferrite inductor. But i was never able to solve any 'problems' with the aircoil. So though in theory the aircoil should have a lower distortion, than the ferrite coil, i have not been able to confirm this with real measurements.
Another thing i would like to promote is, that every DIY'er should have a THD measurement system. The one i have is a really inexpensive PC based one, and many others alike are available. The only thing you have to make yourself is a breakout box with 4 / 8 Ohms load resistors, and two potentiometers for level equlization.
Nothing is better than tuning your amplifier to your own exact loudspeakers, and letting them perform at their best. It lifts the performance right up to a new level.
And also you can verify the technical specs some producer's have on their product sheet. They don't always match real life performance.
Tuning via equalizer or hacking into the speakers? One thing I find is moving my speakers an inch either way drastically changes the sound in this place, hardly worth the time for me anyway.
Wouldn't you want to buffer it as well?
Seems worthy of a new thread 🙂
- Status
- Not open for further replies.