Sounds interesting - open a new thread on 'feedback vs power supply distortions' for this.
I'm all for it. But to overly complexify things by adding in a large number of subsidiary effects at once is an invitation to madness. One at a time. THEN add another effect. As the song says, Watch what happens.
Do less-than-voltage-source power rails have an effect? Of course they do. Can you mitigate those effects by better design? Yes.
Do nonlinearities make systems analysis more difficult? Certainly. And one could add And How! There is a book with a surpassingly trenchant contribution from the late great dynamical-systems mathematician Vladimir Arnold, which, early on, summarizes linear systems theory in two pages. If you can follow the notations, diff equations and complex-matrix-variable complex exponential functions and more, it's essentially the last word, despite the omitted details and how-to's, otherwise taking many volumes to explain in detail.
One at a time. And mind the sims. And don't forget measurements on real stuff.
I want to point out that I do appreciate your speculation on what feedback is and does. I am not going to attempt to 'shut anyone down' for their opinion. I can be pretty sure that there is a good deal of truth in a number of your inputs.
When I study or question something like negative feedback, it is from 50 years experience with it, coupled with opinions of those I have greatly respected over the decades. These people were always older than me when we first discussed negative feedback, and those who are still alive, still tell me about the problems with negative feedback even today. Elementary (typical college level) textbooks are not effective in answering my questions, there has to be more.
When I study or question something like negative feedback, it is from 50 years experience with it, coupled with opinions of those I have greatly respected over the decades. These people were always older than me when we first discussed negative feedback, and those who are still alive, still tell me about the problems with negative feedback even today. Elementary (typical college level) textbooks are not effective in answering my questions, there has to be more.
I think it's to do with circuit/device noise, embedded programme noise and interactions exaggerated by NF, and are not signal correlated.
This is aside from circuit generated HD and IMD which are signal correlated.
No proofs, but speculation based on subjective experience with BQP and another noise reduction method.
Dan.
I have spent 35 years packaging lasers. Some of them produce a thermal flux of 90 watts per mm^2. I contracted many years and I always got snickers from the big company guys. But my stuff typically 95% there on the first pass. The current company I am with has tyco, western digital, etc types. They snicker but have taken this company's development to a stand still. Hate to say it. But there are some esteemed here that are doing what Miles Davis fought to avoid. They hold on to expertise of the past. John curl, I don't care what anyone says, you continue to produce world class design, regardless of the tools your employer has at the moment. To me that is real engineering. I won't get into cooling damned semi conductors. I am sure I will be told they work just as good at 125c as 15c
Last edited:
Edit/repost.
This is aside from circuit generated HD and IMD which are tightly signal correlated.
When I use the term 'noise' in this context I mean 1/f noise, excess noise and white/thermal noise, and I am not convinced that perfectly flat white noise exists in real systems as opposed to mathematical models.
No proofs, but speculation based on subjective experience with BQP and other noise reduction/modification methods.
Dan.
I think it's to do with circuit/device noise, embedded programme noise and their interactions being exaggerated by NF in a non linear way/form which is not tightly signal correlated.
This is aside from circuit generated HD and IMD which are tightly signal correlated.
When I use the term 'noise' in this context I mean 1/f noise, excess noise and white/thermal noise, and I am not convinced that perfectly flat white noise exists in real systems as opposed to mathematical models.
No proofs, but speculation based on subjective experience with BQP and other noise reduction/modification methods.
Dan.
Yes. In fact, the hard clipping is because the feedback no longer works! It doesn't cause the hard clip, it prevents it. You have a forward amp with a gain of say 5000 and you drive it with a 1V input? Duh!!
It is only thanks to the feedback that there is any useful output from that amp. Clip it, which disables the feedback, and the full 5000 x gain slams the output into the rails.
That's how it works
Jan
Last edited:
Steven
The restrictions I have met are due to my knowledge limitations and not due to 2x4 MiniDSP hardware or software issues
For my first steps on DSP, I wouldn’t exchange usability and ease of connectivity for some better technical specifications.
The 2x4MiniDSP is ready to be connected to your analog source and your power amplifiers, so you can listen to what you are doing.
There are a lot of things to learn in the beginning and you will appreciate the ease of connectivity and freedom of compatibility issues the 2x4 offers to you.
With the mini-shark you need an extra board (AN-FP) for to input and output analog signals
The mini-Shark has many attractive technical options.
From all these I would only be interested on the increased time delay it offers for integrating subwoofers placed a long distance from the mains (0-300ms delay translating to 0-100m distance versus 0-7.5ms translating to 0-2.6m for the 2x4MiniDSP) and on the possibility for better room equalization (through re-phase).
I have no experience yet of the audible superiority of FIR implemented x-over filters over IIR filters but sofar, what I read haven’t agitated my curiosity enough to die for FIRing the filters, I may be wrong though.
George
Last edited:
George,
Thanks for the clarification on the need for more boards to use the mini-sharc. I just noticed or thought that it said you couldn't use FIR filtering with the mini-dsp, I thought that was what you would need to do the phase alignment or call it time alignment if you like. The mini-dsp is much cheaper to start with and perhaps I have just been listening to to many people being anal about what is needed.
What have you thought of the Rephase software, have you used it, it looks like there is a real learning curve to using that software.
Steven
Thanks for the clarification on the need for more boards to use the mini-sharc. I just noticed or thought that it said you couldn't use FIR filtering with the mini-dsp, I thought that was what you would need to do the phase alignment or call it time alignment if you like. The mini-dsp is much cheaper to start with and perhaps I have just been listening to to many people being anal about what is needed.
What have you thought of the Rephase software, have you used it, it looks like there is a real learning curve to using that software.
Steven
But there is no place for opinion or different versions of 'truth' in such matters. Feedback behaviour is a long time completely settled matter of verified fact. Deviation, by way of opinion or claims of fresh or new behaviour, can only be a matter of misunderstanding or misapplication, you can safely bet.
I did state 'subjectively'.
Remember that typical measurement methods incorporate 'smoothing' and so do not fully indicate subtleties that the ear picks up when listening to dynamic programme content, ie music be it live or recorded.
Dan.
The problem here is that once you have convinced yourself that feedback sounds bad (by peeking of course), while on the other hand there is no rational basis for that, you enter a state of Cognitive Dissonance (look it up).
It is VERY hard to get out of that - you either have to accept that your believe is not supported by reality, or slide down the slope that 'they're all against me', 'they are stupid', they are the 'thought-police', etcetera. Note that it is always 'they', never, 'me'.
You can see examples of Cognitive Dissonance here galore, and only very few examples of solving it successfully.
Jan
Cool.
If his "acausal" approach can help to dispel the pernicious notion of the "dog chasing its tail", it will be worthwhile for that alone.
https://archive.org/stream/bstj56-8-1337#page/n0/mode/2up
https://ia802703.us.archive.org/19/items/bstj56-8-1337/bstj56-8-1337.pdf
ChrisTemporary overdriving occurs very often on many installations and there lies the problem.
Monitor the input and feedback lines with an oscilloscope while listening.
Simulation is an eye opener.
George
You don't need FIR filtering to do time alignment; the MiniDSP can delay any channel up to 10 ms, which is enough for loudspeakers.
Rephase software is realy easy to use, but can only be used with a FIR capable DSP.
What I did to study the effect of linear vs. minimum phase is to have my speakers minimum phase, and use my PC, HolmImpulse, rephase and a convolver to turn the phase straight. Windows media player accepts a convolver plug in, so the total costs of this solution is about zero USD.
- Status
- Not open for further replies.