Pre-distorsion is a concept of injecting the anti-phase of a functions inherent distorsion to reduce/null said inherent distorsion on its output. You seem to describe it as an icing of a cake which is it not.
//
This is not a new concept and is used atm in some commercial offers in loudspeakers.
The idea is neat and smart the real world application of it not as easy as it seems: you have to try to solve a non dynamic behavior for it to work and as we know not all distortions are linear or have a 'predictable' behavior...
That said it is used by Presonus in it's Sceptre range of coax loudspeakers ( with horn, Altec style rather than Tannoy style) to counteract the issues related to the horn diffraction ( the original algorythm come from Fulcrum Acoustic own research on coax and is licenced to Presonus). It works nice but the issue's behavior is more or less fixed and somewhat predictable ( at least it was managable to measure it's effects and then make a 'profile' from them) .
Temporal Equalization—TQ™ | Fulcrum Acoustic
If it had been used to compensate for breakups of a membrane the outcome would have been much more hazardous...in broader term to compensate for any non linear behavior is a no-no ( well atm, nothing tell us that increasing dsp power and 'better' measurements techniques will not be availlable in (near) future for non linear artefacts).
Here without measurements of the 'good' parameters there is no ways the principle will work. And it works so measurements and parameters are valid, theorically and in real world.
For those believing measurements doesn't have sense, they have to understand that in loudspeakers the latest breakthrough in the field comes from measurements and models of them: iow from simulations. If anyone doubt they just have to download VituixCAD and play a bit with it...
Last edited:
Andrea, I am afraid that you will need to pay for this information
Good night myself
I'm sorry but I'm not interested in paying for this, it was just a curiosity.
My business is IT and not audio and the project was closed with the expected results.
We have a big advantage here, we design for ourselves with virtually no budget limit.
Half-right, is not right, and you do not have the power to make it right outside your home and your business
Not true what you say.
Impedance analyzer (which is a specific type of VNA+current sources, e.g. not Z0 referenced) is the right tool for loop measurements. Else, the trivial VNA will report S-parameters to the Z0. Now, all loops have Z0 at your measuring physical points and all frequencies? Never !
Yes, you can still re-scale the S-parameters obtained from a standard VNA, to give you the reality of the loop, but then you need postprocessing code&equations as I said.
Your "dislilke" of current sources based tools (modern ones), just because of your emotional impact with 70s era, is totally injust, technically incorrect, and in the end is your problem
To measure true loop gain, the VNA is the one and only proper instrument.
You can measure it open loop and let the VNA apply the Randall-Hock equation.
In Rhea's book, this is on page 9 and 10 of 450, so you might consider to learn
some beginner's stuff.
Gerhard
< Discrete Oscillator Design: Linear, Nonlinear, Transient, and Noise Domains (English Edition) eBook: Rhea, Randall W.: Amazon.de: Kindle-Shop >
Attachments
Last edited:
Hi Demian,
there is no reason to measure the crystal unless you suspect it was damaged.
When you order a crystal you have to provide the desired specs to the manufacturer according to the oscillator circuit you want to build.
Then the manufacturer provides all the measured specs.
For example in the Driscoll oscillator where the crystal is placed in the emitter circuit, the lowest impedance point, high Q and high ESR are strongly desired, because they mean the higher loaded Q.
So SC-Cut type and higher overtone as possible.
But the most important spec cannot be measured: the surface finish.
This heavily affects the random walk of the crystal which dominates the close in phase noise of the oscillator.
There is no way to measure this spec and to predict the result.
We have found crystals with higher Q but at the end the oscillator measures worse than another using a crystal with lower Q.
The only way is measuring the oscillator with a proper phase noise analyzer tool.
Andrea
Yes it may. Or not.
I will play devil's advocate and use your case as an example: you claim about a 0.3% gain but at which output level? 10W, 5W 1W?
And what happen along the power output range? If we consider a 20db dynamic range signal (20 db over rms level is what we usually use as headroom for non compressed ( dynamic) signals) and a 64/50w amp this will make rms level in the vicinity of 1w so we should have to characterise distortion from 1 to 64w and then from 1 to 0.01w ans see if the behavior is constant or not ( situation described by Anatech some pages ago) to be sure that this basic approach is really a benifit for the whole range of operqtion of your amp or not...
Given it is somewhat circuit's topology related we can already deduce things but still it needs measurements.
I am in no way telling your approach is wrong Indra and encourage you to follow your own instinct and experience but this have to be objectively evaluated afterwards to help make links between theorical approach and real world gain.
And for this we need measurements.
@ scottjoplin
Probably balanced by reduction of driver induced imd.
Of course I measure, otherwise I won't be able to make any sense at all. But it would be kitchen table grade of measurement, no fancy logo on my plot.
And the approach is yet far from theoretical, it is still in a logically suspicious stage. At least you told me of an implementation that works.
As you know, the implementation is for personal use. So I work on it with my own pace, no need for me to hurry.
Probably balanced by reduction of driver induced imd.
~1W
Of course I measure, otherwise I won't be able to make any sense at all. But it would be kitchen table grade of measurement, no fancy logo on my plot.
And the approach is yet far from theoretical, it is still in a logically suspicious stage. At least you told me of an implementation that works.
As you know, the implementation is for personal use. So I work on it with my own pace, no need for me to hurry.
Last edited:
As stated repeatedly by our experts, both THD and IMD are two different ways of observing the same aspect of nonlinearity, unless they are mistaken. Anyway, my hands will be tied up the next few weeks, I'll be able to work some more on the cancellation stuff afterwards. But in any event, measurements I made won't be able to directly benefit anybody else due to difference of gears being used.
Hi bimo,
Interesting. I greatly prefer an accurate amplifier for listening. Possibly the difference is in the interface between the amplifier and the speaker? I'm talking about how the amplifiers he has tried interact with the load and phase shift of his speakers.
That's an entirely new topic, how an amplifier reacts to a reactive load. We do test this with capacitance added across the load resistor in the lab. There isn't really a standard for this. I tend to follow the Marantz ways of testing this from the 1970's. That seems to be effective.
Don't get dragged down the stupid path of using extreme levels of impedance. That never ends and really only explores idiotic speakers like the Kappa 9. I always wonder why a competent speaker designer would design a load with extreme impedance shifts. To me, that shows complete incompetence. No matter what amplifier you are considering, they don't perform well when they are "unhappy". So why would anyone design a speaker that makes amplifiers "unhappy" and expect good sound?
Why would anyone buy such a speaker?
-Chris
Interesting. I greatly prefer an accurate amplifier for listening. Possibly the difference is in the interface between the amplifier and the speaker? I'm talking about how the amplifiers he has tried interact with the load and phase shift of his speakers.
That's an entirely new topic, how an amplifier reacts to a reactive load. We do test this with capacitance added across the load resistor in the lab. There isn't really a standard for this. I tend to follow the Marantz ways of testing this from the 1970's. That seems to be effective.
Don't get dragged down the stupid path of using extreme levels of impedance. That never ends and really only explores idiotic speakers like the Kappa 9. I always wonder why a competent speaker designer would design a load with extreme impedance shifts. To me, that shows complete incompetence. No matter what amplifier you are considering, they don't perform well when they are "unhappy". So why would anyone design a speaker that makes amplifiers "unhappy" and expect good sound?
Why would anyone buy such a speaker?
-Chris
Speakers are different to amplifiers, mainly because of the damping, anyway, you might find this thread interesting. Geddes on Distortion perception
- Status
- Not open for further replies.