Hi All,
___________________________________________________
Andy C, I trust you are not indirectly infering that I am an 'Audio Mystic'.
Is it not the case that most Sim programmers suggect that users run 10 or more cycles when studying distortion ? That is steady sinewave.
If you have a query then ASK. Please don't make a blanket comment when I and everyone else do not know what you are on about !!! Though if you do ask please keep it to one question at a time.
One thing I don't like are challenges that are themselves technically unfounded, and I will not respond to those.
___________________________________________________
Hi Mikeks,
I did not write that NFB is a bad thing !
I have stated how it is possible to Sim a NFB loop delay generated problem and you tell me that this tells us nothing and the 'gyrations' are usually invisible.
Have you checked this or are you just assuming it, because assumptions can lead to flawed design ?
Nelson Pass warned against 'choke distortion' with his old A40.
I do not need to use a series output choke to keep out RF, (though some designs might) and series inductance 'isolates' an amplifier from capacitive loads by reactively introducing voltage distortion at the output terminal wrt genuine NFB loop controlled output. This is NOT Hi-Fi.
Remember the choke potential leads the resultant loudspeaker current, but loudspeaker current can be considerably shifted from NFB loop controlled voltage output, thus the choke error potential can be at considerable variance with signal waveform, and thus momentarily smear it in either polarity, especially on sibilance and transients. However you cannot observe this via steady sinewave investigation !
You quote me in Post 340 - this is what happens if you would care to Sim it for yourself.
Some software houses would say those results may not be valid, which would either put you off trying, or make you disbelieve the results acheived. However when you try this for many different amplifier circuits, whilst some show errors that you might disbelieve, others do not, and if the software is genuinely capable of first cycle generation/observation then the results themselves will not be wrong.
Sim 10kHz back EMF into a typical amplifier driven 6uH choke and you will be surprised, but don't expect me to do it for you !!!!!
You repeat a quote from Andy's posting,
WHY ?
What is it that you wish me to clarify ?
( I have family things that need doing so don't expect me to make instant replies.)
___________________________________________________
Hi Jan,
Chokes.
When a series output choke is being used, given that the loudspeaker is driven by NFB loop controlled voltage output plus any voltage developed by the output choke which is related to loudspeaker current flow alone, there is a need to ponder - would a NFB output voltage sensing error correction circuit actually correct loudspeaker drive voltage ?
Cheers .................... Graham.
___________________________________________________
Andy C, I trust you are not indirectly infering that I am an 'Audio Mystic'.
Is it not the case that most Sim programmers suggect that users run 10 or more cycles when studying distortion ? That is steady sinewave.
If you have a query then ASK. Please don't make a blanket comment when I and everyone else do not know what you are on about !!! Though if you do ask please keep it to one question at a time.
One thing I don't like are challenges that are themselves technically unfounded, and I will not respond to those.
___________________________________________________
Hi Mikeks,
I did not write that NFB is a bad thing !
I have stated how it is possible to Sim a NFB loop delay generated problem and you tell me that this tells us nothing and the 'gyrations' are usually invisible.
Have you checked this or are you just assuming it, because assumptions can lead to flawed design ?
Nelson Pass warned against 'choke distortion' with his old A40.
I do not need to use a series output choke to keep out RF, (though some designs might) and series inductance 'isolates' an amplifier from capacitive loads by reactively introducing voltage distortion at the output terminal wrt genuine NFB loop controlled output. This is NOT Hi-Fi.
Remember the choke potential leads the resultant loudspeaker current, but loudspeaker current can be considerably shifted from NFB loop controlled voltage output, thus the choke error potential can be at considerable variance with signal waveform, and thus momentarily smear it in either polarity, especially on sibilance and transients. However you cannot observe this via steady sinewave investigation !
You quote me in Post 340 - this is what happens if you would care to Sim it for yourself.
Some software houses would say those results may not be valid, which would either put you off trying, or make you disbelieve the results acheived. However when you try this for many different amplifier circuits, whilst some show errors that you might disbelieve, others do not, and if the software is genuinely capable of first cycle generation/observation then the results themselves will not be wrong.
Sim 10kHz back EMF into a typical amplifier driven 6uH choke and you will be surprised, but don't expect me to do it for you !!!!!
You repeat a quote from Andy's posting,
WHY ?
What is it that you wish me to clarify ?
( I have family things that need doing so don't expect me to make instant replies.)
___________________________________________________
Hi Jan,
Chokes.
When a series output choke is being used, given that the loudspeaker is driven by NFB loop controlled voltage output plus any voltage developed by the output choke which is related to loudspeaker current flow alone, there is a need to ponder - would a NFB output voltage sensing error correction circuit actually correct loudspeaker drive voltage ?
Cheers .................... Graham.
Graham Maynard said:
Are you claiming to be scientific then? Because if you are, there's the matter of "burden of proof" to consider. If someone engaging in science makes an assertion, the burden of proof lies with them to justify it. This is quite different from requiring the reader to engage in a game of "Mother May I" with the writer as you're doing above. The appropriate course of action for an unjustified assertion is to ignore it.
fab said:
It's all very well to make comments about the deleterious effects of output chokes and so on, but in the end we should keep in mind the frequency spectrum that we are dealing with. It seems quite obvious that an amplifier should have the highest possible bandwidth before feedback is applied and that it is quite undesirable to feed extraneous RF signals back into the signal, or feedback path of a wideband amplifier, that are wildly outside the frequency spectrum that we are dealing with. In the end the problems are not really with amplifier design as there are many good and adequate solutions there for audio, but in the signal sources and transducers. Perhaps some more attention should be given to what we are listening to and what with .
Graham Maynard said:
Hi Graham,
Let me see if I understand your point. If the amp is any good, the NFB will ensure that the amp output voltage faithfully follows the input signal.
Now, if you connect a speaker to this amp through a choke, your point is that, for some signal types, the speaker voltage/current is no longer strictly controlled by the NFB loop because of the influence of the choke. I think that is self-evident, and I cannot imagine anyone having any problems with it. The type of signals you mention can be seen as a combination of a sine wave and some kind of step function, and the step function having high order components excites the choke. So, yes, the speaker voltage/current at the initial step deviates from the one if there was no choke.
What I fail to see is how that would inpact on the amp and/or the NFB loop. I would think it would only make it easier for the amp to keep things under control, as the choke screens it from speaker/xover anomalies.
NB I just received the August EW issue. I shall have to put parts 1, 2 and 3 of your article together and see if I can make any sense of all this.
Jan Didden
Hi Jan,
Thank you for not trying to 'shoot the messenger'.
The present article runs to six parts before I can fully bring the 'concepts' together and show their distortion.
Yes the *loudspeaker* terminal voltage deviates, momentarily sometimes leading, sometimes lagging, wrt amplified sound waveform, and choke deviation is entirely dependent upon loudspeaker characteristics and current flow through it.
This cannot be simulated or bench observed via steady sinewaves and a 'dummy' resistor. As the choke near instantaneously develops its leading voltage, the error increases with frequency, i.e. first cycles of sibilants and transients, and on higher harmonics.
The series output choke voltage does not impact upon the amplifier that is enclosed within the global NFB loop - only its OUTPUT ! (which includes the choke induction potential)
___________________________________________________
Hi Fab,
Should we also not be getting rid of some of the capacitive loading, or at least ensure sufficient effective series resistance, plus a commensurate amplifier driving capability, like the 2 ohms suggested.
Or if we must use a stabilising output choke, then keep it as far below 1uH as possible.
Different non-choke internal feedback loops are possible. See;- Nelson, Cherry, JLH plus others.
Sibilance ? The choke will generate its leading potential according to the 'bulked' but individual crossover sections it feeds, and the error due to one loudspeaker driver circuit can induce a potential that will appear across the other; i.e. leading mid-bass reactance induction appears across the tweeter circuit and distorts its eventual output. Whether this would be sibilance or not I could not predict, though that potential exists.
Personally I regard the VAS Miller connected C.dom as being a more significant cause of reactive loudspeaker system induced sibilance.
____________________________________________________
Hi Ghost,
Of course I agree with you, but real-world loudspeakers DO give rise to amplifier induced amplifier-loudspeaker interface distortion, and this MUST be minimised, because it is far more significant for reproduction than the THD figures universally quoted.
Cheers for now ............. Graham.
Thank you for not trying to 'shoot the messenger'.
The present article runs to six parts before I can fully bring the 'concepts' together and show their distortion.
Yes the *loudspeaker* terminal voltage deviates, momentarily sometimes leading, sometimes lagging, wrt amplified sound waveform, and choke deviation is entirely dependent upon loudspeaker characteristics and current flow through it.
This cannot be simulated or bench observed via steady sinewaves and a 'dummy' resistor. As the choke near instantaneously develops its leading voltage, the error increases with frequency, i.e. first cycles of sibilants and transients, and on higher harmonics.
The series output choke voltage does not impact upon the amplifier that is enclosed within the global NFB loop - only its OUTPUT ! (which includes the choke induction potential)
___________________________________________________
Hi Fab,
Should we also not be getting rid of some of the capacitive loading, or at least ensure sufficient effective series resistance, plus a commensurate amplifier driving capability, like the 2 ohms suggested.
Or if we must use a stabilising output choke, then keep it as far below 1uH as possible.
Different non-choke internal feedback loops are possible. See;- Nelson, Cherry, JLH plus others.
Sibilance ? The choke will generate its leading potential according to the 'bulked' but individual crossover sections it feeds, and the error due to one loudspeaker driver circuit can induce a potential that will appear across the other; i.e. leading mid-bass reactance induction appears across the tweeter circuit and distorts its eventual output. Whether this would be sibilance or not I could not predict, though that potential exists.
Personally I regard the VAS Miller connected C.dom as being a more significant cause of reactive loudspeaker system induced sibilance.
____________________________________________________
Hi Ghost,
Of course I agree with you, but real-world loudspeakers DO give rise to amplifier induced amplifier-loudspeaker interface distortion, and this MUST be minimised, because it is far more significant for reproduction than the THD figures universally quoted.
Cheers for now ............. Graham.
janneman said:
Sorry about this, happens in monster threads: I overlooked something a few pages further.
So Jan, you are proposing to buffer the VAS (which is always a good idea) and then put a series impedance between buffer out and output stage in?
Wouldn't we still be faced with buffer output impedance nonlinearity?
This can be countered by using a larger series impedance, but I would be afraid to use anything beyond maybe 200 R because it will work as a low pass and hence reduce GBW.
Greetings,
Eric
Well Eric, actually you should also buffer the output stage input, because that is also non-linear. I did some sims with such a setup, using 1K for series resistance from the output of the Vas buffer to the input of the output stage input buffer (if you follow me...). The correction circuit had a gain of 1mA per Volt. It worked like a charm, almost a perfect gain of one power stage.
I'm working on a proto pcb for it to try it out irl.
Jan Didden
I'm working on a proto pcb for it to try it out irl.
Jan Didden
Eric,
This is the concept that takes also care of the secondary issues. Of course, the important question is how in practise you can cut away a lot of the pheripheral stuff while maintaining performance with a simple/small circuit.
That's where good designers make their money..
Jan Didden
This is the concept that takes also care of the secondary issues. Of course, the important question is how in practise you can cut away a lot of the pheripheral stuff while maintaining performance with a simple/small circuit.
That's where good designers make their money..
Jan Didden
Attachments
PMA said:
Well, I have reduced many Hawksford pages to just one drawing. Everything is there. This is the important and difficult part. The rest, as they say, are the boring details.
There must be 1000's of buffers and output stages floating around on this forum that can be used.
Hawksford, and others like Steven, have shown simple V->I converters of just one or two bjt's.
Of course, I intend to be original, but you will forgive me for not putting that up on the forum.
Jan Didden
Ouroboros said:
That's the spirit! Exactly the way I did it. I used a darlington complementary output stage with 'real' transistors, and ideal buffers.
IMHO, this is the way to verify proof-of-concept. You don't want to bother in this stage with non-linear opamps, oscillations etc. First you want to be sure your idea is not flawed. You want to see that the concept really takes care of the real output stage non-linearities. Then you start to put in 'real' components for the other parts, try to keep up the high performance, and even try some simplifications.
YMMV
Jan Didden
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.