Nonlinear system use feedback to correct for nonlinear effects. Was used in woofer systems decades ago. Phillips developed a woofer with "Motional Feedback" in the 70's. Don't know if any once has applied it to the mids in an active speaker. I think the pain may not be worth the gain.
You assume too much. Who says I was talking asbout drivers or passive crossovers.
Yes, it was the X10.
There is a vid of John Meyer talking about his company's historical milestones, where he brings up the X10's servo control. I think it might have been at an AES awards ceremony for him...can't find the vid or I'd link it.
Interestingly he said the servo didn't work as anticipated, because it reacted to every motion of the 15", such as the driver moving from opening and closing doors into the studio, etc.
The servo would act to counter those room induced motions of the driver, making for some unnatural and unwanted sound.
Devils in the details, huh !!
There are modern methods for distortion digital correction using DSP, but for some reason they are not widely used.
Schurer et al. Comparison of Three Methods for Linearization of Electrodynamic Transducers
Schurer's phd thesis
Nope, I was only talking about removing the fixed phase rotation caused by a crossover as that was the subject of this thread.
You drag a lot of variables into this discussion, I didn't address any of those. As I stated. One can still use FIR to remove phase rotation caused by the crossover (for instance, an active crossover before the amp, or another DSP unit with these options, but not having specific linear phase crossovers) and you'd still be able to remove the group delay.
I'm not talking about correcting the errors you mentioned. As I'm not trying to drag stuff into this thread that doesn't belong here. That was my point. Why drag all that other stuff into a discussion about FIR filters. We were discussing two different methods of applying filters. And a side discussion about using other methods to arrive at something similar.
Anyway, I see you got what you wanted, the discussion is now about all of the other variables. Good luck discussing that, I'm out.
Who cares what anyone claims, just use them for what they can do. Most uses go wrong due to wrong expectations, that part is true.
Don't blame the tool, blame the user.
Last edited:
wesayso said:
As long as the system is linear (and satisfies superposition principle) you could apply your filters in any order you want. However, if your driver(s) enter the non-linear region (due to excursion, heat or power compression), then the system ceases to be non-linear and the order in which filters are applied starts to matter. So, whenever one says that you could EQ either way, linearity is still assumed (no offence again).
Note that linearity is assumed since your filter type (FIR / IIR / passive) is also linear and time-invariant (LTI), as already mentioned in post #499. The following is an article on linearity, but if you find it boring, skip to the section "Classification summary", at the end of the article.
https://www.tokyodawn.net/meaning-of-system-linearity-in-audio-production/
wesayso said:
Please note that I'm not part of that discussion as I understand it to be futile, you simply can't adjust for all those. Personally, I would prefer linear phase only upto the crossover stage, without including the drivers.
wesayso said:
As far as I understand, most people are tricked / fooled into buying equipment by the salesmen/showroom. Now, as to why they fall for it depends on many things including sales pitch, budget, fancy features etc. but the most important point would be knowledge. It's hard to trick some someone when they know enough on the topic and also know what they want and what would get them what they want.
Unfortunately, a lot of people do not understand any signal processing but still end up buying DSPs in hopes of getting great sound.
Last edited:
Sorry man, i've tried to stay quiet, but this post is nonsense.......
No one who's been participating this thread, is going to make the mistake of thinking those issues you list are correctable by FIR.....or by any dang filter.
I'm personally thankful for the very high level of expertise, sharing contributors have provided, knowing such irrelevant issues won't be introduced.
In fact, this thread has been more about questioning what FIR can do just at line-level electrical,.... with only slight swerves into small signal global acoustic corrections.
To bring in large-signal issues, or driver/box/horn/ room issues, or ******* air anomalies /space time distortion ....is plain bonkers imo....
Must say, it seems like you have some real biases against DSP in general, and FIR in particular.
Ps i hit like 'Like' by accident to your post #510.. I was trying to mutli-quote/reply to it, not with favor, but made that mistake instead.
All for the better I guess, I've said enough already....
Last edited:
That's the wrongest judgement of me in years, I just made myself quite a large processor (3 DSPs) from scratch (below).
https://www.diyaudio.com/community/...ents-preferences-opinions-ideas.381423/page-3
That part was intended for the case where the loudspeaker is included within the correction loop. I'm actually surprised because audio people usually seem to know that SPL/phase (frequency domain response) is a dynamic thing that cannot be corrected using static filters.
Replace with "linear".
I guess no one is doubting the effects of thermal compression and dynamic non-lineraities here. But this thread is about correcting errors in the linear operation range of a system. The thermal compression and other level-dependant effects take place whether a systems frequency response is corrected by FIR (or whatever) or not. But with the solutions discussed here it is at least possible to improve a given system imn terms of phase and frequency.
BTW: Huge efforts are being made to improve drivers regarding their dynamic and high-power behaviour. I guess this is still easier than a digital correction of such errors which would also have to take ageing into consideration.
Regards
Charles
BTW: Huge efforts are being made to improve drivers regarding their dynamic and high-power behaviour. I guess this is still easier than a digital correction of such errors which would also have to take ageing into consideration.
Regards
Charles
Newvirus, i don't get the point.
Nobody is telling you those issues doesn't exhist (and certainly not Mark which folowed courses about technology used for live events and is in my view as credible as any serious pro in the field about all this...) just that no one ever thoughts about compensating for them using FIR.
When that happens the speaker is being used in excess of it's realistic limits. This does happen of course if someone buys a small bookshelf and turns it up to 11. I do understand your point that when a graph is shown of a complete inversion it could give the impression that all irregularities have been fixed when that is not the case in reality.
A well designed speaker should have no real significant change in frequency and phase response due to level over it's intended range. Erin's audio corner has compression tests on various speakers and most stand up really well, varying less than a dB or so across most of the range. Some are not cut out for going loud and it shows clearly.
In most cases it is your own ears that introduce the greatest non linearity to frequency response with level and this applies to all speakers. They will not sound the same at 60dB, 80dB and 100dB. Without some form of dynamic EQ or rebalancing any tuning is only good for a limited range of SPL levels.
fluid said:
Dear fluid, I understand non-linearity to be an instantaneous thing, rather than an accumulated quantity like temperature. The primary reason why I call loudspeakers non-linear and amplifiers linear is because amplifiers use feedback loops while most loudspeakers do not.
Now, there's motional feedback and I'm not against it. However, it is necessary to feed back the acceleration / force directly as opposed to the voice-coil current, as these quantities are proportional only if the Bl product is constant, which is not the case, as it depends on excursion.