I learn from both those older and younger from me.
Right now, I am studying a number of books recommended by others, mostly older than me. Some I can barely understand, but I keep at it.
I REALLY want to know what is wrong with negative feedback. Why do we always lose something (in my opinion) when we use a lot of it? Is it the necessary 'open loop' bandwidth limitation? Is it the harmonic multiplication? TIM, PIM? My best efforts have been open loop, why?
Right now, I am studying a number of books recommended by others, mostly older than me. Some I can barely understand, but I keep at it.
I REALLY want to know what is wrong with negative feedback. Why do we always lose something (in my opinion) when we use a lot of it? Is it the necessary 'open loop' bandwidth limitation? Is it the harmonic multiplication? TIM, PIM? My best efforts have been open loop, why?
Brad,
I am reading what I can find online, I found a book that looks interesting without being nothing but math equations for dsp and am reading it, they just make it a pain to download as you have to do each chapter separately. It would have been simpler to just use an analog xo but the advantages of the dsp method just seem to make that a no brainer these days.
John,
Thanks for the compliment. I asked a question of you earlier today but I guess you didn't see it.
I am reading what I can find online, I found a book that looks interesting without being nothing but math equations for dsp and am reading it, they just make it a pain to download as you have to do each chapter separately. It would have been simpler to just use an analog xo but the advantages of the dsp method just seem to make that a no brainer these days.
John,
Thanks for the compliment. I asked a question of you earlier today but I guess you didn't see it.
I just did that. Was told ALL symmetrical designs were inferior / unreliable.
The "other " expert did say a Jfet symmetrical had pluses (blowtorch).
I stated the LIN design (you know the one) that said expert wrote a book on ,
had overload misbehavior. Then , he said ..."just don't clip it" .
.I get many more useful ideas from our younger members , how to overcome some
of these design limitations. While some of the older ones are "set in their ways" ,
discounting any idea that does not fall within their technical viewpoints.
OS
OTOH, you could be the hero of lots of people by explaining to the 'amateurs' in a respectful way how the issues really are.
AE was also a great teacher - that's what he did his last 40 years. Why not step in his footsteps here?
Jan
Someone asked me about the books I am reading. I can only say: The Collected works of Claude Shannon, and 3 books on Complexity.
Also, a book recommended by Richard Marsh:'Communication Acoustics'
I don't necessarily recommend any of them, yet. Best to get up to speed with Cordell, and other design manuals.
Also, a book recommended by Richard Marsh:'Communication Acoustics'
I don't necessarily recommend any of them, yet. Best to get up to speed with Cordell, and other design manuals.
Steven
Why don’t you start with a MiniDSP product and the accompanying software?
A 2x4 MiniDSP kit (~80 $ +10$ for the software) for minimum phase filters implementation.
MiniDSP 2x4 kit | miniDSP
It is not that difficult when you have it in front of you.
Start with that, there will be many months or years before you explore all the many capabilities it offers.
For around 300$ you buy their more advanced products with the AD Shark DSP chips inside but IMO better do the first steps with a simpler product
George
Chris, I went elsewhere with mine. I could have changed a lot of it with parts of better quality and added 40 years' worth of development, but didn't. Quite to the contrary, I intended to change only the parts which badly needed it after 40 years of work, namenly the caps. I wanted it to be returned to its original state as well as possible.
The reason is my own curiosity. I hadn't listened to one in well over 30 years, and my memory of it was that it sounded better than most everything I had heard for still sane money until then. So I wanted to check myself out, after so many years, just to see what I should think of it now, with a completely new and different system, in my view much better than I had before, but still with my trusty old AR94 speakers, which did the honours the first time round in 1986.
And I did learn two things. The first is that those AR94 speakers (with new rubber surrounds and much better parts in the XO) was in fact better sounding than I expected them to be, though hardly perfect, still very enojyable to listen to. The second is that the Marantz was in fact much better sounding than I rememberd it, as evidenced by what my 1041 speakers were able to reproduce when driven by the Marantz combo (3250b and 170DC). Lastly, when put in perspective by comparing it to the H/K PA 2400, which appeared in 1997 and thus embodied much of what was learnt in the meanwhile, it could still hold its ground, although in absolute terms, the H/K was better.
Lastly, I found that the Marantz of 1978 sounds better than any modern unit which I could borrow, and that the old saying that Marantz stopped being that good after it was sold off first to Sony and then to Philips. It just wasn't in the same league any more. Unfortunately.
Frank, I'm not sure exactly what you mean?
Are you opposed to using voltage regulators with their own NFB loop? It seems to me that well regulated voltage gain stages usually (but not by default!) produce a more transparent sound overall, since they never know what's going on with the high power current gain stages, as they remain isolated from them in every way except the overall NFB.
No, Dejan - just that NFB relies on being able to slew the instantaneous voltages within the circuitry at a very high rate, to compensate for the distortion at precisely the moment of worst non-linearity, which highly likely puts great demands on the PSs to maintain stability, not be modulated, at that precise moment. If they can't behave themselves, then the NFB also has to try and "cope" with the rails glitching - and it may often fail to do so.
I've run scenarios in Spice, putting some realistic behaviours into the mix - and it's easy to see the feedback failing to "fix the problem" - the mechanism can't deal with the multiple, simultaneous misbehaviours, and you get a glitch at the output.
I've run scenarios in Spice, putting some realistic behaviours into the mix - and it's easy to see the feedback failing to "fix the problem" - the mechanism can't deal with the multiple, simultaneous misbehaviours, and you get a glitch at the output.
Last edited:
As for teaching, it could be argued that it's done best by personal example. All one needs to do is to wonder why someone does something in a consistent way each and every time and why.
I was never a fan of FEts. But I did wonder why John consistently uses them across the board for his power amps, which I learnt by downloading as many service manuals of his gear as I could. So I bought some 2SK170, or what passes for it these days of Chinese manifacturers. Took some advice from Nelson Pass in matching them as best I could, which turned out to be better than I anticipated it to be. Made some rough models. I could find nothing I'd call conclusive evidence. Then I asked Damian Martin, who was kind enough to offer some advice and pointers, for which I thank him, and all of a sudden I did get some results. Then I wondered a bit more, looked over my notes made over the years, and some of my circuits suddenly had obvious measurebale and audible differences. It turns out that practically all the amps I consider to be better than the average, all to the last one, had FET inputs.
I have no idea how I could have missed this obvious fact for so long, but I reckon it's better ever than never. I certainly got a load of thought food and an interesting topic to investigate further.
And I won't even mention George, who is a natural born teacher.
I was never a fan of FEts. But I did wonder why John consistently uses them across the board for his power amps, which I learnt by downloading as many service manuals of his gear as I could. So I bought some 2SK170, or what passes for it these days of Chinese manifacturers. Took some advice from Nelson Pass in matching them as best I could, which turned out to be better than I anticipated it to be. Made some rough models. I could find nothing I'd call conclusive evidence. Then I asked Damian Martin, who was kind enough to offer some advice and pointers, for which I thank him, and all of a sudden I did get some results. Then I wondered a bit more, looked over my notes made over the years, and some of my circuits suddenly had obvious measurebale and audible differences. It turns out that practically all the amps I consider to be better than the average, all to the last one, had FET inputs.
I have no idea how I could have missed this obvious fact for so long, but I reckon it's better ever than never. I certainly got a load of thought food and an interesting topic to investigate further.
And I won't even mention George, who is a natural born teacher.
Last edited:
If there are participants strangling to integrate a woofer to a full range (or sub to a main speaker) here are some papers:
>Edit. All methods are based on adjusting time delay on one or two channels, so think of digital systems
http://www.excelsior-audio.com/Publications/Subwoofer_Alignment.pdf
http://www.doctorproaudio.com/doctor/cajondesastre/pdfs/Downloaded_from_Doctor_ProAudio_com-Phase_alignment-Joan_La_Roda-DAS_Audio.pdf
http://www.dv2.fr/article/delay_alignment_a_survey.pdf
For home use (small closed space), I have found the aligning method based on impulse response filtering as the easiest one for good results.
Next is wrapped phase matching around x-over freq by scaling down measurements, shifting x-over freq point higher.
Aligning using Group Delay differences is troublesome (again for small spaces).
I haven’t tried the ETC method
George
>Edit. All methods are based on adjusting time delay on one or two channels, so think of digital systems
http://www.excelsior-audio.com/Publications/Subwoofer_Alignment.pdf
http://www.doctorproaudio.com/doctor/cajondesastre/pdfs/Downloaded_from_Doctor_ProAudio_com-Phase_alignment-Joan_La_Roda-DAS_Audio.pdf
http://www.dv2.fr/article/delay_alignment_a_survey.pdf
For home use (small closed space), I have found the aligning method based on impulse response filtering as the easiest one for good results.
Next is wrapped phase matching around x-over freq by scaling down measurements, shifting x-over freq point higher.
Aligning using Group Delay differences is troublesome (again for small spaces).
I haven’t tried the ETC method
George
Last edited:
George, it is rumored that you are the man Google turns to when they can't find something. They must be right.
I am at a loss what Excelsior means by 'Envelope Time Curve'. Google's answer is recursive. Is it the envelope of the impuls response? If so, why not look at the impulse response directly to ascertain where the acoustic center of the drivers is. This is the only use for impulse response I can see in the context of driver alignment without further processing.
The second step to get the alignment right, after knowing the location of acoustic centers, consists of evaluating the phase plots (acoustic filter slopes and -6dB points).
If the acoustic centers don't align there are only three options. The first is to physically bring the drivers into alignment. This is not always possible. The second is to bring them into aligment by delaying one of the drivers. This can easily done with active filters, less so with analog ones (although Tannoy had one in its Little Red Monitor, just to give an example, it can be done at considerable cost). Third possibility is to live with the lobe where it falls.
I am at a loss what Excelsior means by 'Envelope Time Curve'. Google's answer is recursive. Is it the envelope of the impuls response? If so, why not look at the impulse response directly to ascertain where the acoustic center of the drivers is. This is the only use for impulse response I can see in the context of driver alignment without further processing.
The second step to get the alignment right, after knowing the location of acoustic centers, consists of evaluating the phase plots (acoustic filter slopes and -6dB points).
If the acoustic centers don't align there are only three options. The first is to physically bring the drivers into alignment. This is not always possible. The second is to bring them into aligment by delaying one of the drivers. This can easily done with active filters, less so with analog ones (although Tannoy had one in its Little Red Monitor, just to give an example, it can be done at considerable cost). Third possibility is to live with the lobe where it falls.
- Status
- Not open for further replies.