Graham Maynard said:
Graham, I my argument isn’t concerned only with THD and I wasn't singling out any single aspect of amplifier performance as having any particular precedence over all others. I am simply making the point that enhancing an amplifiers performance in one area inevitably compromises the performance in another.
My amplifier measures excessively well in the other relevant parameters (damping factor/slew rate/bandwidth etc) and "lightenig up" the nfb will not make the amplifier sound one jot better. Your postualtion that such may be so is interesting, but I have never built an amplifier that behaved as you suggest.
I build my amplifiers to be transparent - they are not supposed to have a "sound", and none of my solid state designs have ever sounded any different form the others (and I have listened to them a lot), except for in the instances that a fault had developed.
Cheers,
Glen
Originally posted by Graham Maynard:
"Think of an amplifier driving a first audio cycle into a LS.
The amplifier voltage always starts out leading current"
Always ?
"but by the end of that cycle the current can be leading"
I think there's a misconception here. How can current lead at the end of the cycle ?
Are you trying to say that the current near the end of the cycle is not zero ? I think that's not the same with leading.
thus the amplifier's NFB response can become modified and there can be waveform distortion within that first cycle
to some extent yes, more than predicted by the resistive load case.
because of finite open loop gain,bandwidth, output impedance and open loop linearity.
"LS do not generate leading current until after they have been energised, and initially they conduct normally at the resonant frequency until resonance has been energised"
I'm lost here
"Think of an amplifier driving a first audio cycle into a LS.
The amplifier voltage always starts out leading current"
Always ?
"but by the end of that cycle the current can be leading"
I think there's a misconception here. How can current lead at the end of the cycle ?
Are you trying to say that the current near the end of the cycle is not zero ? I think that's not the same with leading.
thus the amplifier's NFB response can become modified and there can be waveform distortion within that first cycle
to some extent yes, more than predicted by the resistive load case.
because of finite open loop gain,bandwidth, output impedance and open loop linearity.
"LS do not generate leading current until after they have been energised, and initially they conduct normally at the resonant frequency until resonance has been energised"
I'm lost here
Hi Rodolfo,
What I am trying to do is get people to think about unavoidable *fundamentals* when they apply NFB around a real circuit which does not work as instantly as the method requires for coherent error correction, so that they might themselves understand.
(Unfortunately the NFB aspect is now being discussed in the output coil thread.)
I myself can see what you are asking for in the traces, so I do not know what more to say.
For example the choke/resistor group delay variations are measured by two separate voltmeters as shown and all traces are for separate circuits simultaneously driven in the same timeframe; nothing is differential there.
I saw your equivalent loudspeaker back in 2005. I thought that the loudspeaker would work less well than it might do due to lack of individual tweeter Zobel and impedance compensation at the crossover frequency. Also I did not like 8R in series with a reactive tweeter network, and was/am not sure that the 'box' and driver resonance representations can be correct - a box is in parallel with a driver but unless restrictively small has its own separate tuning characteristic. Also if the mid-bass driver were to be checked alone it develops an incredibly well damped and low frequency resonance peak.
Yes transduction group delays do have a significant effect upon the loudspeaker reproduction we hear, and indeed some aspects which cause them can be reflected back into the composite impedance characteristic, but these do not relate directly as to how an amplifier 'sounds' when we so clearly recognise that its output is not 'original' when auditioned through many different loudspeakers.
Hi Glen,
Don't we all know that changing one aspect of any circuit affects performance in another?
Folks who design amplifiers tend to aim for a particular set of characteristics, just as you state you measure, and thus it is most likely their(your) finished products will drive a reactive loudspeaker system similarly. Do your amplifiers drive and sound the same as those of other designers though ?
Those who build with different priorities ? I doubt it.
Does that make those others designers wrong for what they achieve ?
If you have designed an acceptably transparent amplifier, then that is good, but I have no way of knowing. All I have have read from you here is your challenge about compromises, yet no clues as to what compromises you might be thinking about.
Do you accept that other folks amplifiers can sounds subtly different to yours, and yet their's too can bench-test the same as your's at human hearing frequencies ?!
I have been trying my best to show how these reactive loudspeaker loading interface generated amplifier 'characteristics' can arise.
Do you have any explanations to add in relation to your amplifier's stated transparency ?
Would you care to show your circuit here, or provide a link to it ?
Cheers ......... Graham.
What I am trying to do is get people to think about unavoidable *fundamentals* when they apply NFB around a real circuit which does not work as instantly as the method requires for coherent error correction, so that they might themselves understand.
(Unfortunately the NFB aspect is now being discussed in the output coil thread.)
I myself can see what you are asking for in the traces, so I do not know what more to say.
For example the choke/resistor group delay variations are measured by two separate voltmeters as shown and all traces are for separate circuits simultaneously driven in the same timeframe; nothing is differential there.
I saw your equivalent loudspeaker back in 2005. I thought that the loudspeaker would work less well than it might do due to lack of individual tweeter Zobel and impedance compensation at the crossover frequency. Also I did not like 8R in series with a reactive tweeter network, and was/am not sure that the 'box' and driver resonance representations can be correct - a box is in parallel with a driver but unless restrictively small has its own separate tuning characteristic. Also if the mid-bass driver were to be checked alone it develops an incredibly well damped and low frequency resonance peak.
Yes transduction group delays do have a significant effect upon the loudspeaker reproduction we hear, and indeed some aspects which cause them can be reflected back into the composite impedance characteristic, but these do not relate directly as to how an amplifier 'sounds' when we so clearly recognise that its output is not 'original' when auditioned through many different loudspeakers.
Hi Glen,
Don't we all know that changing one aspect of any circuit affects performance in another?
Folks who design amplifiers tend to aim for a particular set of characteristics, just as you state you measure, and thus it is most likely their(your) finished products will drive a reactive loudspeaker system similarly. Do your amplifiers drive and sound the same as those of other designers though ?
Those who build with different priorities ? I doubt it.
Does that make those others designers wrong for what they achieve ?
If you have designed an acceptably transparent amplifier, then that is good, but I have no way of knowing. All I have have read from you here is your challenge about compromises, yet no clues as to what compromises you might be thinking about.
Do you accept that other folks amplifiers can sounds subtly different to yours, and yet their's too can bench-test the same as your's at human hearing frequencies ?!
I have been trying my best to show how these reactive loudspeaker loading interface generated amplifier 'characteristics' can arise.
Do you have any explanations to add in relation to your amplifier's stated transparency ?
Would you care to show your circuit here, or provide a link to it ?
Cheers ......... Graham.
Hartono,
>>Originally posted by Graham Maynard:
"Think of an amplifier driving a first audio cycle into a LS.
The amplifier voltage always starts out leading current"
Always ? <<
Please refer me to a real world loudspeaker with stationary transducer where there is not sufficient series inductance+resistance for this to be the case.
----------------------------------------------------------------------------------
"LS do not generate leading current until after they have been energised, and initially they conduct normally at the resonant frequency until resonance has been energised"
I am sorry to say that I have posted this several times, and I cannot sit here endlessly typing.
Look at loudspeaker impedance plots and check out the electrical phase relationships through their unavoidable reactivities.
Cheers .......... Graham.
>>Originally posted by Graham Maynard:
"Think of an amplifier driving a first audio cycle into a LS.
The amplifier voltage always starts out leading current"
Always ? <<
Please refer me to a real world loudspeaker with stationary transducer where there is not sufficient series inductance+resistance for this to be the case.
----------------------------------------------------------------------------------
"LS do not generate leading current until after they have been energised, and initially they conduct normally at the resonant frequency until resonance has been energised"
I am sorry to say that I have posted this several times, and I cannot sit here endlessly typing.
Look at loudspeaker impedance plots and check out the electrical phase relationships through their unavoidable reactivities.
Cheers .......... Graham.
Graham Maynard said:
Graham, this is getting ridiculous. Due to your continual objections, I have repeated them over and over and over again. If you haven’t got a clue as to what compromises I have been talking about with respect to the utility of an output inductor, then you simply haven’t been reading my posts. I suggest that to go back to the one at which Bob Cordell interjected.
My patience has limits. I am not going to reiterate what I have already said 5 times or more.
Also, by your own admission, you won’t even look at an amplifier design that doesn’t comply with your largely unsubstantiated negative feedback ideas, so why should I even bother?
Glen,
Please be so kind as to link the objections you refer to.
Remember I was addressing your comments to me, not those addressed to someone else before I joined the thread !!!
Bothering is what this is all about, in order to make progress.
Thank you ....... Graham.
Please be so kind as to link the objections you refer to.
Remember I was addressing your comments to me, not those addressed to someone else before I joined the thread !!!
Bothering is what this is all about, in order to make progress.
Thank you ....... Graham.
Graham Maynard said:
In a response to Bob you put forth your objection to my reply to you, in a conversation we had already begun.
http://www.diyaudio.com/forums/showthread.php?postid=1204482#post1204482
Glen,
Bob agreed with you, I responded to Bob !
Thereby I suggested I could have a different opinion to what you - and Bob - concur; this being an unqualified blanket statement covering inductorless circuitry.
I have already clarified one difference of opinion you mentioned - an aspect you appeared to suggest might be a compromise re lightening up on NFB, which I think can positively affect reproduction quality. I cannot say more because no other circuit/design pointers have been raised.
Cheers ........ Graham.
Bob agreed with you, I responded to Bob !
Thereby I suggested I could have a different opinion to what you - and Bob - concur; this being an unqualified blanket statement covering inductorless circuitry.
I have already clarified one difference of opinion you mentioned - an aspect you appeared to suggest might be a compromise re lightening up on NFB, which I think can positively affect reproduction quality. I cannot say more because no other circuit/design pointers have been raised.
Cheers ........ Graham.
Graham, I don't know why Glen thinks that he knows more about amp design than we do. Perhaps he thinks we are trying to make them 'sound' a certain way. However, I don't know how to do that. I try to make ALL of my designs free from anything that I think that can be audible that is a departure from the original signal.
Hi John,
My letter to Doug Self was when his original Blameless had an output choke - Jun 94. Doug's 'normal type of reply' appeared in Aug 94; including -
>> "back EMFs from reactive loads do not cause detectable inter-modulation and even if they did, a higher feedback factor would surely reduce rather than increase the effect ” <<
Of relevence to this thread is that this remains a widely held view, in spite of NFB being unable to do anything about choke potential generated by phase shifted loudspeaker system current flow.
Hi Hartono,
I see I did not answer your -
>> Are you trying to say that the current near the end of the cycle is not zero ? <<
Yes. Dynamic loudspeakers have a zero phase response at just a couple of spot frequencies.
Initially the drive starts with zero phase, but it is the crossover reactivity and loudspeaker generated back-EMFs which change the phase of current wrt voltage, as seen in loudspeaker impedance charts, and this is what an audio amplifier has to cope with. The current phase varies constantly with music signal waveform energisation.
Textbooks do not make it clear that these are steady sine based characteristics only, such that amplifier currents relating to (music) voltage drive are much more variable, and can momentarily be much higher at any resonant frequency than the sine impedance suggests.
Cheers ......... Graham.
My letter to Doug Self was when his original Blameless had an output choke - Jun 94. Doug's 'normal type of reply' appeared in Aug 94; including -
>> "back EMFs from reactive loads do not cause detectable inter-modulation and even if they did, a higher feedback factor would surely reduce rather than increase the effect ” <<
Of relevence to this thread is that this remains a widely held view, in spite of NFB being unable to do anything about choke potential generated by phase shifted loudspeaker system current flow.
Hi Hartono,
I see I did not answer your -
>> Are you trying to say that the current near the end of the cycle is not zero ? <<
Yes. Dynamic loudspeakers have a zero phase response at just a couple of spot frequencies.
Initially the drive starts with zero phase, but it is the crossover reactivity and loudspeaker generated back-EMFs which change the phase of current wrt voltage, as seen in loudspeaker impedance charts, and this is what an audio amplifier has to cope with. The current phase varies constantly with music signal waveform energisation.
Textbooks do not make it clear that these are steady sine based characteristics only, such that amplifier currents relating to (music) voltage drive are much more variable, and can momentarily be much higher at any resonant frequency than the sine impedance suggests.
Cheers ......... Graham.
Hi John Curl
I don't know whether you have already answered my questions elsewhere (in other threads etc)... but you haven't said what is audible about the output inductors.
Here's what could apply for starters ...
frequency limit: the inductor simply slows the amplifier down
problem with capacitor loading
trips protection circuits due to high in-rush currents of the type Graham M has been discussing
In itself, a 3 uH inductor gives a -3dB point of 385 kHZ with 8 ohms loads. It is difficult to understand how this is audible. However, if the inductor were one of several time-constants, then perhaps they could all add up. For example, if an input filter also had 385kHz cut-off, the two add up and may begin to affect audio response. Add a few time constants from other stages and pre-amps and it is possible that the overall losses impair audio performance (transient/frequency response).
If you load an amp. with a 2 uF capacitor - and who would do this except for those using electrostatic speakers ?- then some designs might be unstable and need an inductor to help. The transient response of an amp. causes ringing, and generally if you add (say) a 3 uH inductor the ringing gets to be a lower frequency, but this is still over normal audio limits (typ. 50 kHz for 2uF-3 uH).
Perhaps it is possible that this creates intermodulation distortion in amplifiers with generally low performance (i.e. higher distortions where IMD is more likely), and this is what is audible.
What might also be audible is that the presence of the inductor resonates with a capacitor load to generate a low dynamic load - the "first cycle" Graham talks about- where no out-of-phase currents have started. Currents up to 10 A or more are possible with resonant loads. Perhaps these trip the protection circuits, or cause some transistors to saturate, or gain-limit, and this is what makes the inductor "audible"?
In which case(s) it is not the inductor which is audible per se, but it is leading to situations the amplifier can't handle, and this is what makes it audible.
Or,maybe if you have poorly wound inductors high currents physically cause the wires to vibrate and literally become audible.
or, ... still waiting ...
cheers
John
I don't know whether you have already answered my questions elsewhere (in other threads etc)... but you haven't said what is audible about the output inductors.
Here's what could apply for starters ...
frequency limit: the inductor simply slows the amplifier down
problem with capacitor loading
trips protection circuits due to high in-rush currents of the type Graham M has been discussing
In itself, a 3 uH inductor gives a -3dB point of 385 kHZ with 8 ohms loads. It is difficult to understand how this is audible. However, if the inductor were one of several time-constants, then perhaps they could all add up. For example, if an input filter also had 385kHz cut-off, the two add up and may begin to affect audio response. Add a few time constants from other stages and pre-amps and it is possible that the overall losses impair audio performance (transient/frequency response).
If you load an amp. with a 2 uF capacitor - and who would do this except for those using electrostatic speakers ?- then some designs might be unstable and need an inductor to help. The transient response of an amp. causes ringing, and generally if you add (say) a 3 uH inductor the ringing gets to be a lower frequency, but this is still over normal audio limits (typ. 50 kHz for 2uF-3 uH).
Perhaps it is possible that this creates intermodulation distortion in amplifiers with generally low performance (i.e. higher distortions where IMD is more likely), and this is what is audible.
What might also be audible is that the presence of the inductor resonates with a capacitor load to generate a low dynamic load - the "first cycle" Graham talks about- where no out-of-phase currents have started. Currents up to 10 A or more are possible with resonant loads. Perhaps these trip the protection circuits, or cause some transistors to saturate, or gain-limit, and this is what makes the inductor "audible"?
In which case(s) it is not the inductor which is audible per se, but it is leading to situations the amplifier can't handle, and this is what makes it audible.
Or,maybe if you have poorly wound inductors high currents physically cause the wires to vibrate and literally become audible.
or, ... still waiting ...
cheers
John
Hi John Ellis,
The amplifier current will be the same whether there is an output choke or not, it is the external leading voltage wrt current flow it develops that affects audio. One moment subtractive of amplifier voltage output, another additive to, but not directly related to the audio signal input, and with the peak potential being proportional to both current and frequency.
I have given my findings of how it was audible (though subtle), and I would wonder if John Curl could add his own description, for mine alone is hardly acceptable.
The only thing similar to the typical choke which was once used would be about 20ft/6m of LS cable. So if anyone has heard a difference between using 20ft of cable and just 3ft, then they can hardly say that chokes are not audible.
However if anyone is already using LS cables more than a few feet, then the effect of a series choke will be masked by those cables limiting discrimination.
Cheers ........ Graham.
The amplifier current will be the same whether there is an output choke or not, it is the external leading voltage wrt current flow it develops that affects audio. One moment subtractive of amplifier voltage output, another additive to, but not directly related to the audio signal input, and with the peak potential being proportional to both current and frequency.
I have given my findings of how it was audible (though subtle), and I would wonder if John Curl could add his own description, for mine alone is hardly acceptable.
The only thing similar to the typical choke which was once used would be about 20ft/6m of LS cable. So if anyone has heard a difference between using 20ft of cable and just 3ft, then they can hardly say that chokes are not audible.
However if anyone is already using LS cables more than a few feet, then the effect of a series choke will be masked by those cables limiting discrimination.
Cheers ........ Graham.
John Ellis, you are at square one in understanding inductors placed on the output of power amps. First, what happens WHEN you add a serious capacitance to the load of an amp? Have you ever heard of RINGING?
However, it is more complex than that, and I suspect it will be a long road for you to catch up with the discussion here..
Start from the beginning, and in the end, you might be able to understand more than you do now.
However, it is more complex than that, and I suspect it will be a long road for you to catch up with the discussion here..
Start from the beginning, and in the end, you might be able to understand more than you do now.
Graham Maynard said:
AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAARRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRRGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHHH!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
I'm now going to retreat from this thread and spend my spare time poking my nipples with kebab skewers instead.
Graham Maynard said:
This argument cannot be repeated over an over without proper qualification.
For the sake of putting some numbers, I checked a simulation from one of my designs sporting about 800 KHz 3dB open loop bandwidth. Other designs will vary but what matters is the ballpark range where we are working.
Group delay within the passband goes all the way down to 100 nS. Let's asume nonetheless as a very conservative criteria, that actual delay from amplifier input back through feedback to be 1 uS.
One very extreme test signal for evaluating GNF impact could be for example a step, which should transverse the forward chain unmodified at least for 1 uS before feedback steps in to limit its amplitude according to the set gain. This means we should have very roughly speaking a 1 uS spurious pulse. Now the frequency spectrum for this pulse comprises a very small DC component, plus harmonics at 1 MHz and multiples where most of the energy resides.
Note this is an extreme case taking very conservative assumptions, real world situations are far friendlier.
The only way to evaluate impct as for most other forms of amplifier induced perturbations, is actual comparison of input and output with real musical program, and to measure the residulals both in amplitude and correlation with music, and only then to draw conclusions.
Your post here shows Delay ch-R, Delay ch-LS and BACK EMF voltage probes, the plot has no label whatsoever, thereby my insistence on having the point clarified. Yet I guess whatever it corresponds to, the point is not relevant in the light of what has been commented before.
I confess I find difficult to deal with your reasoning when on the one hand you complain the particular model I choose may not be the best one, formerly stated
in this post that "even the cheapest loudspeakers" should pass the amplifier "inner activities" and above you support you can recognize this activities through many different loudspeakers.
I have tried fruitlesly to have auditioning inclined people to flesh out for those of us not endowed with particuarly sensitive or trained ears, the features that betray otherwise unmeasurable distortions.
This situation is the kind that leads others to quit in despair, though it is sometimes interpreted as acceptance, something that obviously is not.
Much of these situations stem from not sharing a common signal processing fundamentals knowledge. With due respect I highly recommend this free online book by Steven Smith, which with fairly accesible math covers in several chapters a solid understanding on the subject.
Rodolfo
ingrast said:
Excellent link. Those convinced that the low-pass pole of an amplifiers load-isolating network of several hundred kHz will hinder transient response and cause audible ringing, really should read the parts on post-DAC filtering.
Cheers,
Glen
Excellent link Ingrast.
I'm a bit lost on this thread now. Can someone list the things the output coil does to destroy the sound of an amp?
fuzzy treble?
loose bass?
soft top end?
Oscillatory tendencies?
Sound stage collapses? (Careful, can be quite heavy)
Loss of 'air' (hell, maybe that's the problem - we're suffocating)
Loss of 'first cycle' (swallowed by coil, or maybe even speaker cable inductance?)
😀
I'm a bit lost on this thread now. Can someone list the things the output coil does to destroy the sound of an amp?
fuzzy treble?
loose bass?
soft top end?
Oscillatory tendencies?
Sound stage collapses? (Careful, can be quite heavy)
Loss of 'air' (hell, maybe that's the problem - we're suffocating)
Loss of 'first cycle' (swallowed by coil, or maybe even speaker cable inductance?)
😀
Have you guys arrived on some specific Conclusion yet, if not, then do tell me also when something Concrete is found....😀 😀
You all are painstaking the "Audibility of Coil", but have you figured about the passive crossover coils/caps, capacitance/inductance in connecting wires, complex speaker impedances which have far greater effect on sound then this mere 5 to 10 turn air core inductor....has...
You all are painstaking the "Audibility of Coil", but have you figured about the passive crossover coils/caps, capacitance/inductance in connecting wires, complex speaker impedances which have far greater effect on sound then this mere 5 to 10 turn air core inductor....has...
Bonsai said:
I pick fuzzy treble. You see, output inductors limit the frequency response into capacitive loads to maybe a few hundred kilohertz and cause ringing on very fast transients.
Just ignore the fact that very fast transients of sufficient magnitude to cause such an effect don't actually exist in real life audio sources, such as any recording which is digitally mastered, whose bandwidth is brutally chopped off above 20kHz as part of the normal D-A conversion process.
Don't take time to ponder how anyone could hear a little ringing at one or two hundred kilohertz either.
Cheers,
Glen
- Status
- Not open for further replies.