I personally never care about semantics, unless they have a fundamental essence in understanding things.
Or worse, that people take the wrong turn because they have some biased and emotional feelings with certain descriptions or words.
You can't amplify a digital signal, a true won't be a bigger true and a false won't be a bigger false.
You might want to improve or change the signal itself, but that doesn't change the property or message of the signal.
You can however, amplify analog signals, which also will change the value or message of that signal.
If you ever worked with real analog calculators and math blocks, you know what I mean.
So, what annoys me a bit about the digital vs analog debate for Class-D amplifiers, is more the subjective emotional feeling that some people bring to the table. With a clear feeling that everything that is digital is "bad" (not, really but I am exaggerating to make a point).
I personally don't care at all if that is what people feel, you do you, whatever makes you happy.
But in that case, I do think it's important to talk about semantics, because we are talking about an analog signal here, NOT a digital signal.
Or in other words, it's just simply an invalid argument (also known as a fallacy).
Btw, invalid is not the same as people not having these feelings and emotions for this type of amplifier.
Another reason is just pure educational.
Before you know just a small little change in definition will completely branch off in totally derailed myth stories.
Or just spreading plain misinformation and therefor disbelieves.
I just see people getting really sloppy these days to such an extend that we just might call ourselves just monkeys.
Because they are also just primates, so same thing right? .....
From an engineering point of view, there is a whole list of reasons why you really want to separate them.
You made a bell ring...
I read your entire response three times, but THIS... THIS... sinks it.
It seems so simple now...
Thank you!!!!!
Edited to add - By my own admission, I have to quit thinking about the "voltage levels" and the continuity of the actual electrical current, but the "signal" - the information being conveyed. WOW... seriously. Thank you!
Last edited:
No problem 🙂
Now you maybe also understand why the shape of the signal is a poor description.
It happens to be a on-off signal, that looks a lot like a square wave.
But you might as well define some kind of state with the shape of a horse. (just to name something silly)
Problem is that it's very hard to make that shape, and also to define when to switch to different states, not to talk about noise issues etc.
So as a carrier signal, we could even use sine waves if we want.
It doesn't make much sense to do so, since just a discrete signal is so much easier to make.
Especially when we also want speed!
The opposite is also true.
Just because something looks like a on-off signal, doesn't automatically makes it a digital signal.
It just happens to be super easy to convert to the digital domain.
^ Brilliant. That part of the discussion I feel like I had a reasonable grip on, and your explanation aligns with what I thought I knew. However, where (I think) I went off the rails is trying to reconcile in my wee brain my newly learned information about how the signal is 'created'//transported vs. the definition of the fundamental word "signal" => the information being conveyed.
I'm horrible at this, but if you'll allow me.
If you had asked me an hour ago if you could amplify a digital signal. I would have emphatically and with passion been WRONG and said, "Of course". What I'd have meant is that you could turn a 1Vpp electrical square wave into a 2Vpp electrical square wave. I'd have proudly exhibited my reasoning, and some of you would likely have had a good chuckle.
Where you made the bell ring is that I was focusing on my newly-learned knowledge of the electrical part, while completely ignoring what the word "signal" means. It is part and parcel tied to how the signal is used b/c that's what it means.
I'm never embarrassed to learn out loud.
I feel a weight has been lifted.
I'm horrible at this, but if you'll allow me.
If you had asked me an hour ago if you could amplify a digital signal. I would have emphatically and with passion been WRONG and said, "Of course". What I'd have meant is that you could turn a 1Vpp electrical square wave into a 2Vpp electrical square wave. I'd have proudly exhibited my reasoning, and some of you would likely have had a good chuckle.
Where you made the bell ring is that I was focusing on my newly-learned knowledge of the electrical part, while completely ignoring what the word "signal" means. It is part and parcel tied to how the signal is used b/c that's what it means.
I'm never embarrassed to learn out loud.
I feel a weight has been lifted.
Yeah.... It is pointless debating.
Many years back I indulged in an intense discussion (that eventually turned ugly) here on forums... I still believe what I believed then.... despite all the latest attempts and numerous "final conclusions" stated on these forums and in this thread.
It is good to see there's still a discussion happening.
And why all of this..? Because geogehifi invested in shares... and is trying to promote the product that uses them transistors. So, we know whom to blame... 🙂
My statement was maybe a bit short-spoken. But it is simple as that: Faster swiching makes using higher switching frequencies easier. And higher switching frequencies allow higher feedback factors for a given loop order.
Regards
Charles
None of the naysayers here has said anything about the digital pre-distortion circuit that's done within the dsp it's been talked about to me by a knowable tech that said he thinks the GaN-1 & Carina-GaN could very well have, that would keep the distortion and output impedance low, seeing it has no feedback. He said he may add to this discussion later.
Cheers George
Cheers George
Last edited:
Digital pre-distortion is a kind of feed-forward and would need to know the particular characteristics of the speaker being driven to have an impact on overall distortion and output impedance. Are speaker manufacturers providing such info?
It's not even a kind of feed forward, that is literally feed forward.
In this case it's about the distortion of the amplifier, not the loudspeakers
In this case it's about the distortion of the amplifier, not the loudspeakers
I was mostly talking about the distortion of the amplifier.
As for output impedance, that would only work if you know the impedance of the load and its (non linear) behavior.
There would of course be possibilities to measure the properties of the load before parametrising the pre-distortion algorithm. This isn't even a very new idea. And the properties of the output stage have also been taken into consideration within the digital modulator of the TacT amps a quarter of a century ago already
Regards
Charles
Regards
Charles
This idea is as old as the whole existence of control theory itself.
Usually a feedforward loop is used in conjunction with a feedback loop.
Because modelling any non-linear behavior (of any system) is practically very tricky or close to impossible.
Getting it somewhat/roughly close is much easier and than let the feedback loop take control over what's left over.
In this case it's only totally impractical, since you have to change the parameters on a case by case basis.
It makes very little sense to me to model and compensate a speaker in a feedforward loop.
Unless it's within a dedicated closed system whit very little other variables.
Still, it wouldn't make sense for an amplifier without a feedback loop.
Usually a feedforward loop is used in conjunction with a feedback loop.
Because modelling any non-linear behavior (of any system) is practically very tricky or close to impossible.
Getting it somewhat/roughly close is much easier and than let the feedback loop take control over what's left over.
In this case it's only totally impractical, since you have to change the parameters on a case by case basis.
It makes very little sense to me to model and compensate a speaker in a feedforward loop.
Unless it's within a dedicated closed system whit very little other variables.
Still, it wouldn't make sense for an amplifier without a feedback loop.
If I understood correctly it is not about correcting a speaker but an output stage and thereby taking the effects into account that the load has on the switching stage.
?
That is what I said 😉
But different speakers have all different loads = impedances.
And these impedances even change depending on the cone excursion as just one example.
Well, one could be interested in https://www.infineon.com/dgdl/Infin...N.pdf?fileId=5546d46264a8de7e0164b761f2f261e4
It takes 24bit I2S - but there is also a volume scaling in 0.25dB via the I2C control interface.
It seems interesting to me. Whether it is interesting enough to get a multi-channel USB-I2S and play is another matter at the moment.
But if you take an average of all "common" ones (esl's etc exculded), they "kinda" follow a similar to a degree impedance curve
Cheers George
Wouldn't it be nice if the amp had a "calculate speaker impedance parameters circuit within the DSP" setup button to do once only at first, so you have it all spot on impedance curve wise for that speaker that your using.
There's one for the designer team to work out.
Cheers George
There's one for the designer team to work out.
Cheers George
- Status
- Not open for further replies.