Class D amplifiers have very little phase shift


Too many diagrams to post here, so I just blogged it:

https://inatinear.blogspot.com/2020/02/class-d-amplifiers-do-not-have-horrible.html

The 20 degrees at 2 kHz argument is bunk.
erik_squires
the dynamic load of the drivers and crossover will cause massive changes in the impedance of the speaker system. The drivers shift all over the map, thousands of times per second.

As high as a few hundred ohms, and as low as into almost negative numbers.

This is reacting against the Class D output filter.

This means the filter and the driver combination, is all over the map in their effectiveness of blocking or dealing with the amplifier’s output.

The trick is that the drivers are not static in their impedance value.

Which in turn, causes the effectiveness and the nature of the output filter to be all over the map.

If one measures with low values of signal and/or a sine wave, none of this real world stuff will be visible in the measurements.

I’m not taking umbrage with any side of the debate on sound quality, in this. I’m just noting the reality about the class D output filter shifting all over the map, thousands of times per second, in it’s basic effectiveness, in any real world scenario. almost as if the filter was designed in a static formula/static load scenario. Like a single frequency and wave shape of output and single unchanging load.

The real world is all over the map and dynamic as hell.

Thus, class D output filters as they stand today = Wrong. Incomplete. Not correct. I don't have a solution, but I did see the problem, in the very first few seconds where I saw a class D amplifier for the first time. I do have some ideas, though.
Hmm all my Class D amps sound fabulous. Best ever ( I have 8 Class D amps that I know of in my house).
I guess this is one reason for that.

Can you guess I am a Class D fan?  
In a self oscillating class D amp, the feedback itself is used to correct phase shift. Its pretty effective in this regard.

But I've been listening to a class D amp with zero feedback, switching at a fairly low frequency (400KHz), with a rather conventional filter, and it does not sound like phase shift is a problem :)


The article at the link is correct. Understanding filter theory is really helpful in conversations like this. If only one party understands the theory though, its very difficult to make any progress.
the dynamic load of the drivers and crossover will cause massive changes in the impedance of the speaker system. The drivers shift all over the map, thousands of times per second.


Not really relevant and also, what does this even mean?
True enough, Ralph, I get it, but that is horse chasing after barn doors are open, and there is always a better way.
but that is horse chasing after barn doors are open, and there is always a better way.
Sorry, I can't make out what you're trying to say here.
Really, I posted on a very very narrow piece of data regarding Class D and already the anti-Class D people are here.
When you throw the bait out in public you don’t get to decide which varmints will decide to bite.
@mapman

Ain't that the truth?
@erik_squires
Would you please Xsim with a 1mH inductor in series to your speaker and also with a 10uF capacitor parallel to your speaker and show us the result?
@atmasphere Thanks for the vote of confidence!!

@imhififan Is that typical for a Class D stage??
1 mH sounds very very high, isn't it?
@imhififan

I wrote this:

The actual values are irrelevant for the purposes of this discussion.

I meant to say that I just picked whatever XSim uses for the defaults.

Best,

E
I'm not worry about the small inductor, but the capacitor parallel to the speaker.
It is important to note that output impedance chart varies from 5 milliohms to 50 milliohms, so the effect at worst case of the magnitude and angle of the impedance on the speaker are tiny, so the lower chart with the actual output should give a very good idea of how the amp would perform.
There's no way I would put a 10uF cap in parallel with a speaker!! :)
Shown values don't make any sense.  10uF would represent about 0.8 ohm capacitive reactance at 20kHz, while series resonance of  1mH and 10uF would be at 1.5kHz

@kijanki 


The values were picked at random. They aren't real. :) As I wrote, the point was not the values, but where the parts were. :)

E
@erik_squires

Thank you. As for the phase shift, datasheet for the modules in my previous class D amp (Icepower 200ASC) show about 18-30deg phase shift at 20kHz dependent on the impedance of the speaker. Every amplifier has some phase shift at 20kHz because of the limited bandwidth. My amp had bandwidth limit of 65kHz/-3dB but many amplifiers with even lower bandwidth, hence larger phase shift, are praised for wonderful and extended highs. Lamm ML3 that has only about 40kHz/-3dB limit was praised by Stereophile as "airy and extended on top". Yes, I understand that phase shift can suppress highs by improper summing of harmonics, but dynamic speakers at these frequencies have usually inductive character, meaning that they accelerate phase (that would also negatively affect treble). Perhaps such phase shift is not that bad or effect of it is so negligible (who can hear 20kHz) that we don’t even hear it.
The first graph is the phase response of the impedance the second graph is of the frequency.
The first graph is the phase response of the impedance the second graph is of the frequency.

To be more precise: The second is the graph, both frequency and phase, of the entire amplifier.

The first plot measures impedance, the second magnitude of the voltage. It is the difference between measuring a passive component and an active one. :)

Ok, thanks I have a hard time with these graphs of specs. 
It may help to think of where the test probes are. For the first chart, the probes are:

Amplifier --> (+) Output Impedance (-) --> Speaker

For the second:

Amplifier --> Output Impedance  --> (+) Speaker (-)




Yes, yes, I am a new here and only ever read these forums usually for a laugh, and have posted this time because I feel the need to.

 I'll post this once just for the ignorance of those that believe all Class-D have the next to no phase shift from the upper/mid right up to the highs. It has been quoted to look at graphs of x Class-D as it "only" has x phase shift and they used this to compare it to other Class-D modules. Some even being cheap subwoofer modules used.

It's false,  and those here who “consider themselves” to be EE’s should know better, you know who you are and should lay low to giving your tick of approval to this garbage.

 Because some of these comparison modules are measured with/without their output filters insitu, and the manufacturer has used Audio Precision "very high order" external “test filtering” instead for the graphs, and/or the frequency knee of the output filter they use if used, is totally different to the one it's being compared to by the one doing the comparing, which will directly effect the amount of phase shift down lower in the audio band as those of us all know.

This thread is just plain bunkum.
That helps,  it's where I thought they were measuring and that's  why the first graph was misleading they put the probes on the PCB. Not sure what jesskerry is on about I don't  consider myself to be anything other than trying to understand. 
@djones51

I don’t think the graph itself is misleading. It’s that is’ so very very rare. I have seen output impedance plots, rarely, but never one with phase also.

It’s truly something for the geeks.

Best,

E
Not sure what jesskerry is on about I don’t consider myself to be anything other than trying to understand.
Yet you were one of the first to to pat the OP (who has no idea) on the back in the other now deleted thread, without knowing as you say yourself " consider myself to be anything other than trying to understand" try to understand what I just posted above.
Good one! it’s a no wonder I don’t post.
 Because some of these comparison modules are measured with/without their output filters insitu, and the manufacturer has used Audio Precision "very high order" external “test filtering” instead for the graphs, and/or the frequency knee of the output filter they use if used, is totally different to the one it's being compared to by the one doing the comparing, which will directly effect the amount of phase shift down lower in the audio band as those of us all know.

@jesskerry

I only know of one person who could put together such a long sentence that in the end is so convoluted that it cannot be proven or disproven.  I would try to answer it, but it is the opposite of building an argument from parts that logically form a whole.

Hahahaha.
What thinking human could care when the insensitivety to phase argument stands? If you want to bow down to measurements please design us the perfect amp. That doesn't waste power 
like incandescent light bulbs.
And doesn't sound like dog waste.
Well jesskerry the best advice I can give is do as you say and don't  post. 
I meant ,misleading the way someone was trying to use the graph by saying ice 1200 modules had a huge frequency phase drop at 2khz.
I meant ,misleading the way someone was trying to use the graph by saying ice 1200 modules had a frequency phase drop at 2khz.

@djones51

You are right.

I just meant to say that the geek who put it together never intended it to say what is being claimed. I make a lot of plots, so I feel sympathy for having your data ripped off to mean something you never intended. :)
Somewhere out there is a person who is so concerned with this thread his digestion is in serious danger of shutting down. 



into almost negative numbers
Not almost, negative. When a voice coil is flying backward (relative to the now-changed drive signal), it is a generator of back emf.

We can argue about how long it is out of phase, but the fact remains.
Just agreeing and reinforcing what you said. You were too timid :-)

Now, the practical question is "how well does the amp control the voice coil (aka damping)?" If its very, very good, this overshoot and negative emf will be less. Not zero, but less. No i have not measured - its very hard to do so with non-repetitive signals which is where the problems would likely lie.

BTW i found the blog interesting but need to go backa nd think about what the (other party’s) original chart really meant. Unloaded? Hard to say what it is. Amps frankly are pretty easy if it weren’t for the damned loads :-)
I've seen that graph used on DIY and ASR to claim the same thing. Most of the stuff posted there is way over my head, I never post on those sites just read them so I was trying to figure out what the deal was with this graph and why is didn't  mean what the guy was saying. 
Yet you were one of the first to to pat the OP (who has no idea) on the back in the other now deleted thread


George, you are one of the only people who even know that thread was deleted.  I'm happy to say the moderators brought it back, but this one has better pictures:

https://forum.audiogon.com/discussions/no-class-d-amps-don-t-have-horrible-phase-shift


Here's the problem, I have two charts. Both from the same source.  Both clearly labelled.

Several people have used the first chart to claim it is the amplifier output phase chart. In fact, the actual amp phase is the second. Then George#3 (or 5?) rolls in and says that charts can't be trusted because (gobbledygook).  So... if charts can't be trusted, why is he relying on the first one to make his point????

Here we go, I added similar charts from a speaker simulator.

See the bottom. You have both output and impedance for the same speaker charted. Both have phase angles. They are not the same chart.

https://inatinear.blogspot.com/2020/02/class-d-amplifiers-do-not-have-horrible.html


Also, look at the phase angle for the output? I make damn good speakers. :) Others do too, but I am so proud of how well those two drivers integrated.
Above all, everyone, I want to point out that I’ve made 1 claim.

The impedance chart is being misused.


There are several great ways to argue that Class D has a lot of phase shift. Show your work. Show the chart. Post specifics. Find a review, or data sheet.

Instead we get personal attacks, conflagration of a number of possible Class D issues, but the people (person) who relies on that chart to begin with is completely unable to actually refute my claim.

I welcome any and all sources of credible data (public review, manufacturer data sheet, etc.) which expand our knowledge here.
TI has a nice Class D calculator tool.
http://www.ti.com/tool/LCFILTER-CALC-TOOL

I have 3 tube amps, 1 bipolar and 1 Class D.
Each has its strengths and weaknesses.

In over ½ c. no amp has ever had it all.
Ya do have to wonder about the obsession about phase and then throw it all away with 2nd order or worse or wire the midrange out of phase speakers... 



it’s a system people...
Oh, so to be clear, the phase angle of the impedance refers to the relationship of the voltage to current.

The phase angle of the output refers to timing.

The only thing they share is that impedance and output are complex numbers. That is, they have a magnitude and phase. None of those four numbers are equivalent.
...time for a station break....;)

https://www.youtube.com/watch?v=n_Cfsktanus

....and the following song is....appropo...? *L*
I am on my 3rd Pioneer Class D3 AV Receiver and LOVE the Sound as well as the efficiency and accuracy!  My present AV Receiver is a Pioneer SC-LX701
And music, the waveform is “ really complex numbers “

I enjoyed the video station break.
@asvjerry  thank you very much--wasn't aware he was putting forth an album. . and the comment on the following video also entertaining :)
@wyoboy....& @robert53....Y'all are welcome. *S*

I wasn't aware of him at all....but now I am educated in what I avoided so ways ago....;)   As for the album, hope it meets expectations.

T. Petty's passing left a void that stands to be filled and fulfilled....

...and as for the segway vid....dedicated to all of our fav trolls....*L*

...and I don't object to inclusion in that category... ;)....although honestly making the attempts at minimum to 'pointed entertainment'....however obscure....

*mock sigh*  Bad habits & diehards....*G*
...ran into a pair of these @ a 'vintage' MCM/modern furnishings store here....

https://www.ebay.com/itm/JBL-Aquarius-IV-S109-Space-Age-Mid-Century-Modern-Speakers-White-Possible-toShip-/303205269560

"For sale?"  No....only audiophiles seem to like them....
"*L* You're talking to one....but I don't blame you.  The white ones' are on the rare side.  Just curious, and I didn't think you'd want to part with them."

Not to everyones' taste, but a 'pleasant sounding' JBL.   Kind of an era omni in its' way.  These were an 8~8.5/10, the white finish needed a touch-up cleaning and wax..

Nice to see an appreciated 'oldie'...*S*