| wimms |
While trying to get a better idea what kind of distortions occur in a simulated amp I found an interesting way to present the signals.
Basically, you take the output of the amp, reduce its amplitude by gain to match that of input signal, subtract. Now feed the the result to the Y-axis of "scope" and the input signal to the X-axis of scope.
Ideal linear amp with zero latency must show a straight line. Real amp has propagation delay, that is ideally constant wrt input signal, and this must with sinusoidal input manifest as a perfect ellipse on the scope (wrapped sinewave). By stepping through few input amplitudes, and superimposing outputs of such graphs after normalizing, we expose the deviations from the ideal *shape* of the signal. It is visually very easy to see, and imo even gives very useful hints as to what is going on in the amp.
See the graphs below. In this case, graph is made for outputs of 50uW (THD 0.000006%), 55W (THD 0.0014%) and 178W (THD 0.0012%). THD of first 9 harmonics. Notice that THD calculation shows slightly lower THD for 178W output wrt to 55W output.
Look at the graphs. Blue is reference, 50uW output with lowest THD. Green is 55W output and red is 178W output. All graphs are zoomed suitably to approximate circle shape. Absolute values of graphs are meaningless, its only shape that matters here. It is immediately evident that 55W and 178W output graphs deviate very seriously from ideal circle, despite that total THD was 0.001% for both.
How to read the graphs. X-axis shows input voltage instead of time axis. Timing is hidden. As input sinewave develops in time, output signals circulate on this graph, rotating counterclockwise. There are 9 full circles of sinewaves, and they are there behind each other.
When input signal goes from zero to positive slope, watch output response in south-east sector. When input signal wraps from its positive peak and goes down towards zero, watch north-east sector, etc. Because output lags input, difference is initially negative (we start saving after 1st period), later when input goes down, difference becomes positive. Time is shown as signal on graph, just to help visualize how input signal is changing wrt time. Its not any kind of timing reference there.
Nothing quantitative can be drawn from such graphs, but only relative *shapes* of the signals can be compared. We know for sure that for sinusoidal input, ideal shape would be exactly circle or ellipse. Any deviation from that is distortion.
In this particular NFB amp simulation, it can be easily seen that positive going half of the 55W waveform oscillates around reference "ideal" (NFB in action?), and that the amp is asymmetric - negative slope is pretty nasty.
178W output shows more severe overcompensation, probably due to NFB lag and arising late compensation. In both high power cases negative slope on first glance seems to lead input, but I suspect it is late "catching up" due to NFB action (exact timing is lost on this graph).
Also note, that THD calculation gave 178W output lower THD than for 55W output. Its not because the distortion was lower, but only because nature of it was slightly different (less higher order harmonics due to evident slew rate limiting).
Because we are graphing *difference* between input and output, our sensitivity is pretty large, canceling out absolute output levels.
In conclusion, it seems to me that this kind of graph is exposing some details very hard to notice by any other means. I'd like to think that this helps to understand the nature of distortion, and perhaps even shed some light on how non-NFB amps differ from NFB amps.
Comments? Anyone used such graphing before and how its actually called? |
|
|
| Graham Maynard |
Hi Wimms.
Your graphical representation clearly shows that the amplifier under test is not internally balanced, but are these measurements taken using a resistor load that ideally matches zero current crossover activity with the negligible power demand of zero voltage crossover ?
If yes you are not invoking the loudspeaker back-emf induced distortions that alter the sonic character of reproduction, and which will generate different diagonally opposite phase shifted wiggles.
Also can I suggest you study at 10kHz, because any half decent amplifier is okay at 1kHz.
Would it not be possible to zero null your input reference to match output before subtraction so that you end up with more of a straight line with +/- deviations ?
Unfortunately this would apply only for that given amplitude and frequency, for the phase shift changes with frequency, and the propagation delay with amplitude, the latter being impossible to observe by monitoring resistor/sine thd figures in output isolation with continuous sine wave drive !!!!!
This is why I have simulated use of my suggested X-Y monitor circuit, where when you use two amplifiers, both have the same nominal propagation delay to whatever input is applied..
Then any ovality will represent time and voltage shifted error arising at the loudspeaker terminals wrt that arising across a perfect resistor; ie; the display will show unwanted zero phase amplitude based as well as back-emf induced non-linearities.
Ovality = amplifier impedance. Shows back-emf induced time fuzzing of loudspeaker voltage !
Straight line = perfectly resistive.
Voltage errors = amplitude non-linearity, crossover distortion, slew rate limit induced error, voltage/current clipping etc.
This display could be observed in real time with sine or music input, just watch that scope is isolated.
I will also attach approximate virtual equivalent of the well known 'Ariel' loudspeaker that is nicely awkward to drive at 10kHz but sounds excellent with a good amplifier; usually tube types. Sine display gives time expansion at centre of horizontal axis. The first four half cycle sinewave currents with this loudspeaker are quite different at 10kHz !!!!!
And finish off with four two 10kHz cycle X-Y / Ariel simulations with +/- 10V peak at the output terminal. (Which will always be more idealistic than real life)
A = D Self 'Blameless' like circuit showing effect of resistor damped series output inductance. (Loss of image accuracy)
B = same amplifier without choke. (C.dom causing amplifier impedance + degrading crossover suppression )
C = ditto but without 'C.dom' as well. (Phase shifted/back emf induced class-AB crossover still not gone with approx 80dB nfb)
D = my 25W class-A. (Not perfect, but excellent to listen to)
Cheers ............. Graham. |
|
|
| Graham Maynard |
| Useful virtual loudspeaker load. |
|
|
| Graham Maynard |
| Virtual scope traces. |
|
|
| wimms |
hi Graham,
thanks for your interesting reply.| quote: | Originally posted by Graham Maynard
Your graphical representation clearly shows that the amplifier under test is not internally balanced, but are these measurements taken using a resistor load that ideally matches zero current crossover activity with the negligible power demand of zero voltage crossover?
If yes you are not invoking the loudspeaker back-emf induced distortions that alter the sonic character of reproduction, and which will generate different diagonally opposite phase shifted wiggles.
Also can I suggest you study at 10kHz, because any half decent amplifier is okay at 1kHz. |
Yes, that was resistive load. My surprise was actually that the method so well exposes intrinsical nonlinearity of the amp into resistive load and at 1khz where it should be easy job for the amp. Remember, highest harmonics was 100db(!) down the 177W test signal. I didn't really expect to be able to see the nature of the distortion so clearly. THD plot is pretty useless in comparison. And even normal timebased subtracted error voltage is less striking.
| quote: | | Would it not be possible to zero null your input reference to match output before subtraction so that you end up with more of a straight line with +/- deviations ? |
You mean to imitate ideal amp with matched prop delay of the DUT? It is possible, though difficult. But the result is much harder to interpret. See attach. Note that reference blue is basically straight line, its just so much zoomed into (1nV/div) that it appears as oval. There, nonlinearity errors of higher power signals exceed in magnitude of normal prop delay induced voltage, and it becomes hard to correlate with the signal. Imo slight suitable phase shift before subtraction actually helps to understand the result. Its about balance, too much, and you'll see nice ovals only.
| quote: | | Unfortunately this would apply only for that given amplitude and frequency, for the phase shift changes with frequency, and the propagation delay with amplitude, the latter being impossible to observe by monitoring resistor/sine thd figures in output isolation with continuous sine wave drive !!!!! |
Agree on the first part, but I'm not sure I follow you on the second part.
In my view it is rather impossible to perform a null test with reactive load. Perfectly normal phase shifts stick out too much, and it is rather difficult to account for what is normal and what is not.
The same delay dependence on amplitude with reactive load is in my view pretty normal thing to happen.
| quote: | | This is why I have simulated use of my suggested X-Y monitor circuit, where when you use two amplifiers, both have the same nominal propagation delay to whatever input is applied.. |
No doubt this is very interesting approach, and perhaps even easier to conduct in real life. But I think this is cancelling out several amp nonlinearities (if they are perfect match), and exposes only effects that becomes visible thanks to reactance of speaker.
I understand that you are much more interested in amp-speaker interaction, thats what your method seems to try to expose. but wouldn't it be more correct to compare ideal amp into speakerload with DUT into speakerload? At least in simulations. Ideal amp would have to have roughly similar phase response as DUT though.
| quote: | Then any ovality will represent time and voltage shifted error arising at the loudspeaker terminals wrt that arising across a perfect resistor; ie; the display will show unwanted zero phase amplitude based as well as back-emf induced non-linearities.
Ovality = amplifier impedance. Shows back-emf induced time fuzzing of loudspeaker voltage !
Straight line = perfectly resistive.
Voltage errors = amplitude non-linearity, crossover distortion, slew rate limit induced error, voltage/current clipping etc. |
Interesting. I may be wrong, but I interpret differently. Ovality that appears when you compare resistive load with reactive load is in my view reactance of the speaker, time shift between current and voltage. I don't see this as impedance of amp, although nonzero impedance of amp allows us to see that. Back-emf is not necessarily real there. Time fuzzying shouldn't happen for sinusoidal signals. This hints that you used discontinuous test signal, like positive half of sine starting right from zero. This is like impulse that causes some oscillation, and that is what manifests as time fuzzying. It occurs in sim too. To make "clean" test, you'd want to use AM modulated sinewave that rises smoothly (and cosinusoidally) from zero. Imo that kind of time fuzzying disappears then.
That straight line in your test. This is what my sim above zooms into. Its not so straight at all.
| quote: | 'Ariel' loudspeaker..
Sine display gives time expansion at centre of horizontal axis. The first four half cycle sinewave currents with this loudspeaker are quite different at 10kHz !!!!! |
If you simulate with sinewaves starting from 0 volts, then this is wrong. Not only first 4 cycles are different, but theoretically all of them. Sim assumes all currents and voltages to be zero at time0. It is unrealistic increase of amplitude. You get impulse response of the bw-limited amp-speaker system superimposed with the test signal.
I've tried speakerloads with bandwidth limited squarewave (composed from 20 harmonic sinewave generators) with simulated amps before (though not with this method), and it very well shows how stored back-emf energy forces output stage to crossover currents upto several times after voltage crosses zero without voltage errors actually developing. Very interesting to observe - you realize that during these moments NFB is completely detached and amp output impedance is doing crazy dancing on its own.
In my case the speakerload is not wanted. It actually makes it harder to observe amp nonlinearities. The phase shift is unavoidable, especially if we use most convenient pure sinewave test signal. To use speakerload, I'd have to use the raised cosine windowed test signal, and it produces spirals instead of ovals. Becomes pretty cluttered there. I'll attach an example graph. It isn't really much different from what I learned with resistive load.
| quote: | A = D Self 'Blameless' like circuit showing effect of resistor damped series output inductance. (Loss of image accuracy)
B = same amplifier without choke. (C.dom causing amplifier impedance + degrading crossover suppression )
C = ditto but without 'C.dom' as well. (Phase shifted/back emf induced class-AB crossover still not gone with approx 80dB nfb)
D = my 25W class-A. (Not perfect, but excellent to listen to)
|
Very interesting graphs. It is evident though that you indeed use sinewave starting with rectangular windowing. I'd distrust the fuzzying part and also compare only general shapes of these graphs. Of course, the crossover distortion spikes are pretty striking. Though, when you look at the scope divisions, then despite awful shape in graph C, it must be admitted that this amp is showing *least* of error voltage. This only underlines that phase shift is causing "static" error that hides the details and should be ideally canceled to a reasonable degree. |
|
|
| wimms |
| The raised cosine test signal into speakerload graph. |
|
|
| Graham Maynard |
Hi Wimms.
Nice to be able to chat with you about this, even though in public.
You wrote agreed first - not second. Amplifiers quite literally slow down as output stage current demands increase, the nfb loop then causes their internal impedance, and the delay caused by that impedance, as seen by the load, to become momentarily increased.
The X-Y method shows the error due to resistor load wrt loudspeaker, plus error due to loudspeaker load wrt resistor. As the amplifier's internal impedance reacts differently to each load I am not so sure that there is any 'cancellation' of non-linearities. My thoughts are that the amplifier will actually look worse, but then there is at least an opportunity to deal with all errors resulting from passive and dynamic loading.
Ovality is due to amplifier impedance allowing the reactive load to drive its output terminal wrt amplified input voltage; if you simulate with 'ideal' amplifiers the result is a straight line.
You appear to concerned that I am using a suddenly starting signal.
Don't forget that audio waves start suddenly to the bandwidth limit of the system itself.
It is pointless to filter say a 1kHz signal before applying it to an amplifier so that it cannot suddenly start with a change equivalent to real-world sources !
Strike a triangle and it does not slowly build up its first cycle, there is a singularly rich harmonic leading edge, some of which is both inaudible and not picked up by microphones !
Currently CDs are circa 20kHz, DVD-A/SACD 40kHz+, so these should set our filtering bases.
Besides if you run the X-Y examinations for many cycles they do not change much beyond say the third cycle, so the fuzzying (time based amplifier-loudspeaker interface energy storage and release exchange) is not merely a leading edge problem. Also the back emfs from different composite loudspeaker elements all arise at different time periods after initial music energisation, which is why the early half cycles are so asymmetrical.
So many folks seem to think that I do not understand this suddenly starting aspect, but equally I think that those who express such comments are the very ones who wrongly test amplifiers with 'non-musical' waveforms.
By rigidly applying their theoretical pre-filtering they are denying themselves the opportunity to see and make their designs capable of coping with what really can happen as a result of sibilant and transient loudspeaker energisation.
Re your harmonic square wave test and cosine test illustrations - very interesting. So many designers once believed that nfb 'protected' their amplifiers, and that increased levels made for endless improvement - not so.
You say that the speaker load is not wanted because it makes it harder to observe amplifier linearities. But maybe it is actually the amplifier's nfb reaction to the phase shifted loudspeaker generated back emf that is the significant problem, and, not the Nth degree of nfb loop generated amplitude linearity. Tube amps can be as 'bad' as 1% in the thd stakes, yet still drive real world loudspeakers better than 0.001% solid staters - they do not react as badly.
Yes figure 'C' has the lowest amplitude error. The phase shifted loudspeaker current is causing the complementary output stage to reverse commutate through a portion of its fixed common bias before the nfb loop regains control, and thus the waveform appearing at the tweeter will have an entire new and non coherent waveform across its terminals. This cannot happen with class-A, but phase shift can still introduce transient induced offsets where the amplifier has insufficient bandwidth/speed.
'D' has much less damping, but the error shows little in the way of generating a separately identifiable product.
Cheers ............. Graham. |
|
|
| Mark A. Gulbrandsen |
Guys,
While I have not had time to digest what you've written those traces look to have been made with a virtual "Etch-A-Sketch".....
Mark |
|
|
| wimms |
hi Graham,| quote: | Originally posted by Graham Maynard
You wrote agreed first - not second. Amplifiers quite literally slow down as output stage current demands increase, the nfb loop then causes their internal impedance, and the delay caused by that impedance, as seen by the load, to become momentarily increased. |
Shouldn't amp slow down happen equally with resistive load? I don't understand well how you relate variance of delay of amp with reactive loading; you said its impossible to observe with R load. How you measure that delay when current and voltage are shifted wrt time as normal part of reactance? When I tried, looking at zero crossing instants, then variance of prop delay was so unbelievable that I had to stop thinking that voltage on amp outputs is right thing to look at. In fact, I started to suspect that its even not right to use voltage as a feedback signal. I wonder even if perfect voltage source is the ideal amp to drive reactive speakers..
What are your thoughts on this?
| quote: | | The X-Y method shows the error due to resistor load wrt loudspeaker, plus error due to loudspeaker load wrt resistor. As the amplifier's internal impedance reacts differently to each load I am not so sure that there is any 'cancellation' of non-linearities. My thoughts are that the amplifier will actually look worse, but then there is at least an opportunity to deal with all errors resulting from passive and dynamic loading. |
Y-axis shows difference between outputs. That imo cancels errors that are common with both loadings, cancellation not necessarily complete. Thats what I meant. Reactance would cause the errors to manifest shifted in time, and exposed, but imo they are then mixed with speaker late reaction to new signal and that makes it difficult to tell what is exactly what, or when. Very drastic things stick out, like the crossover spikes, but fine details stay hidden.
| quote: | You appear to concerned that I am using a suddenly starting signal.
Don't forget that audio waves start suddenly to the bandwidth limit of the system itself.
It is pointless to filter say a 1kHz signal before applying it to an amplifier so that it cannot suddenly start with a change equivalent to real-world sources !
Strike a triangle and it does not slowly build up its first cycle, there is a singularly rich harmonic leading edge, some of which is both inaudible and not picked up by microphones ! |
Please don't take this as arguing, but just as clarification of my view.
When I was exposed to fourier analysis, I realised that there exist no instantaneous changes in nature. BW limiting is not limiting only maximum bandwidth, but also maximum rate of change of the spectrum. Thats nature of filtering.
Perfect full amplitude 10khz sine signal appearing from silence is not realistic signal due to the first cycle. It would require huge bw. It never happens in reality either. When you strike a triangle, first there appears cosinusoidal increase of amplitudes of large number of harmonics, including the lowest one. Its like a preringing of digital filters. The signal develops for finite amount of time. Check it: http://www.pcabx.com/technical/reference/triangle.wav
I understand that you try to simulate 'attack' of a musical instrument, but I've found that suddenly starting sine is overloading the system's expected bw usage more than is realistic. It is not a problem normally in sims, as it is wrappable/extendable, but with reactive loads this isn't resolved right due to missing past history. It overloads the reactive load and produces ripples that aren't real. That imo only clutters the graphs and confuses.
I've found that I'm able to better analyze results when signal spectrum is better controlled. I've opted to use toneburst signals instead of more severe pulses or sudden sines. For eg. see this http://www.diyaudio.com/forums/show...8152#post528152 as example of most severe toneburst.
But afterall, it doesn't matter what signal you use if you know what you're doing and looking for.| quote: | | Besides if you run the X-Y examinations for many cycles they do not change much beyond say the third cycle, so the fuzzying (time based amplifier-loudspeaker interface energy storage and release exchange) is not merely a leading edge problem. Also the back emfs from different composite loudspeaker elements all arise at different time periods after initial music energisation, which is why the early half cycles are so asymmetrical. |
Can you be sure only first few cycles are affected? That depends on the load reactance, signal, source impedance. Check the attach. Thats speakerload I use, energized from perfect amp, just static 1mOhm impedance added. What can we say about that perfect amp? Wildly different shapes, fuzzy, and rings for long time. If I'd see that on amp output, I'd not know - to worry or not. But this one is perfectly normal behaviour, there is no first cycle distortion there, no changing output impedance, no NFB.
Have you checked the step response of Ariel load with your method? It does not settle in 100ms. IIR. That fuzzying is purely due to impulse energy from the first cycle imo. I can't extract any amp impact there.
| quote: | So many folks seem to think that I do not understand this suddenly starting aspect, but equally I think that those who express such comments are the very ones who wrongly test amplifiers with 'non-musical' waveforms.
By rigidly applying their theoretical pre-filtering they are denying themselves the opportunity to see and make their designs capable of coping with what really can happen as a result of sibilant and transient loudspeaker energisation. |
Yes, but isn't impulse or step response better method for testing such things? It seems easier to interpret. Or the harmonic synthesized squarewave, it has benefit of being spectrally clean and not overload the system.| quote: | | You say that the speaker load is not wanted because it makes it harder to observe amplifier linearities. But maybe it is actually the amplifier's nfb reaction to the phase shifted loudspeaker generated back emf that is the significant problem, and, not the Nth degree of nfb loop generated amplitude linearity. Tube amps can be as 'bad' as 1% in the thd stakes, yet still drive real world loudspeakers better than 0.001% solid staters - they do not react as badly. |
When I said not wanted, I meant my initial post approach. There, maximal zooming effect occurs when phase shift is minimal and controllable. It shows magnified nonlinearity of amp with closed loop. Of course you may be right and we should focus on amp-load interaction instead of amp linearity, but my first post focused on amp nonlinearity only.
Adding speakerload changes the game alot. Phase shift is huge, partly due to finite output impedance. That makes the X-Y ovals very large, hiding fine details. Size of the oval is depending most of all on the time shift. It could be impedance, delay, NFB induced error. When the oval is large, its resolution is low. The variations around the ideal line (oval) are exposed only when you can get the timeshift small.
For eg. in your virtual scope traces, it is not easy to compare them, because they are in different scales. Fig. A is so far apart from the rest, that it is impossible to say if it is actually better or worse than say C. Series inductor increases output impedance, but *tubes* have huge Z too, that can't be bad for the sound and imaging? I'd try to bring them to common scale, by adjusting reference amp's delay and impedance. If there are similar problems in all of them, they should appear in sorta fair comparison.
I thought you might want to compare DUT with ideal amp into the same load. If ideal amp's phase response is matched to that of DUT, then time shift induced errors are minimized, and Y-axis will show more detail about DUT issues.
Btw, have you tried connecting the load resistor instead of to ground to another signal generator? It is effectively controlled reactive load. You avoid that way resonances and oscillations of passive speakerload. |
|
|
| AudioWizard |
| A spectrum analysis (both magnitude & phase) seems more accurate and simpler to interpret. |
|
|
| Nelson Pass |
I'm going to suggest that simulations will not accurately
predict distortion level at -100 db levels. I would not
even trust them at 1% (-40 dB). |
|
|
| Mark A. Gulbrandsen |
| quote: | | A spectrum analysis (both magnitude & phase) seems more accurate and simpler to interpret |
I completely agree, I've been reading this stuff and a spectrum analusus of the distortion is all to easy to deciper. Some of these charts look like the burn graph for a rocket engine!!
Mark |
|
|
| Graham Maynard |
Hi Wimms.
I am hurrying here so I hope I am not making any errors.
Yes amplifiers slow down with resistor loading too, but this tends occur when the rate of change in voltage/current output demand is least - around natural amplitude clipping.
I know you understand what is going on, but music driven moments, or for that matter specific sinewave frequencies, where loudspeaker impedance is purely resistive and either 0, 360, 720 degrees etc. are very rare, and thus effective X-Y cancellations would be too.
Regarding sudden starting at 10kHz.
I have NEVER stated that this is a realistic waveform ... acceleration during the first 90 degrees is much greater than required, but it can be used during simulation to observe aspects of amplifier operation, load induced instability etc. Its only a method of looking for problems that should not arise; sure, some folks use square-waves or a step response - which are equally unrealistic - but useful.
Amplifiers that do not cope with square waves do not sound good, similarly, any amplifier that can cope with a suddenly starting simulated sinewave is not going to be limited with audio. The JLH and my class-A are * not upset * by it, and we should not forget that our audio chains remain capable of passing transient amplitude level changes that are more extreme than maximum linear 20kHz sinewave.
Also as you say - adding a speaker load changes the game a lot.
Actually doing this modifies amplifier output so much that you can hear it on headphones in another room. Different LS different sound! Thus I would love to see you making loudspeaker load (or RCL equivalent) investigations with your distortion microscope, and especially with your cosinusoidal input increase.
Your post#9 circuit illustration proves that loudspeaker back emf 'reflects' at the zero impedance of a perfect amplifier. The more nfb any amplifier has, the lower the output impedance, the more the ringing and the greater the potential for a modifying effect upon loudspeaker terminal waveform any series amplifier/cable impedance will have.
Like you I wonder if a perfect voltage source is the ideal way to drive reactive loudspeakers !?! I often add a question mark when discussing this.
JLH's solution was a good amplifier with a 220 milliohm resistor in series with the output terminal. The resistor reduced loudspeaker circuit ringing and reduced loudspeaker/amplifier interface distortion because it was in series with the amplifier's much lower nfb loop generated output impedance.
The different voltage scales on my illustrations show how badly a 6uH series output choke at the output of a high nfb amplifier in 'A', allows the loudspeaker to modify its own loudspeaker terminal voltage. Driver motor current flow is similarly affected, and bass/mid driver/crossover circuit back emf can affect tweeter terminal voltage.
This is separate from the loudspeaker generated crossover spikes illustrated in 'B' and 'C', also the waveform distortion due to a choke is insignificant if checked using a resistor load alone !
The error shown in 'C' is still there on 'A' but is choke masked and would need axis expansion to illustrate it.
Tube amps have higher series output R, and much less feedback (if used), so loudspeaker driven interaction with internal amplifier impedance is much less, as also is amplifier-loudspeaker resonance damping.
Yes I now always simulate by reverse driving an amplifier via its load resistance. There is an excellent correlation between this, the X-Y testing, probably your distortion microscope too, and, fundamental nulling with a virtual reactive load, though accurate nulling with the latter is much less easy to finalise.
Anyone who measures thd with a plain resistor misses out on the opportunity to visualise so many other waveform induced amplifier problems.
Hi AudioWizard.
Yes spectrum analysis can reveal the overall outcome of distortion products, but unlike observations discussed above you cannot similarly tie down the waveform related instant of distortion causation in order to deal with it.
Equally spectrum analysis is not going to be useful if you do not use a representative loudspeaker load because nfb loop induced current flows are quite different and induce quite different internal amplifier responses to when the load is a 'dummy' resistor !
Hi Nelson and Mark,
But would you ignore the results, especially when headphone monitoring of amplifier output confirms the effects being illustrated.
Of course our ears must be the final arbiters !
Cheers ............. Graham. |
|
|
| wimms |
hi Graham,
I tried nulling the output of amp to input with speakerload. It is quite daunting and confusing, especially with toneburst. At the precision level I wanted, even minute FR phase nonlinearities translate into notable amplitude errors that aren't caused by amp. I had to measure precise output impedance, amp lag and its gain into testload, and copy that all to ideal amp. That works for single amplitude, frequency, lag. I couldn't get it perfect, but I think the null is at reasonable level.
As I suspected, changes due to reactive load are such that it is quite unhelpful to graph them as ovals. So instead I found that linear-time graph is better in this case.
See attach. There you can see simultaneous graphs of output voltage, amp output current I(Rout), and error voltages for 180W peak load. Input signal is 3 periods of toneburst at 10Khz (10Khz AM-modulated by 10Khz) starting from 100us. Speakerload is 3way speaker model instead of Ariel. Tested amp is what I posted in your AABB thread - JLH topology. Its measured Zout is 2m ohms.
V(out-inp) is amp output loaded with speaker minus output of delayed linear amp into no load.
V(ref-out) is delayed ideal amp loaded with speaker minus output of tested amp with exact same speakerload.
V(ref-res) is delayed ideal amp loaded with speaker minus output of delayed ideal amp loaded with resistive impedance of speaker.
I'm not sure if exact shapes mean anything, but this is what I see from such measurements.
V(ref-res) serves as reference - linear amp loaded with speaker. It shows expected voltage errors due to reactive loading on finite output impedance.
Out-inp graph should match ref-res. Deviations there are due to nonlinearities, lag variance and Zout changes.
We can see that error voltage varies around reference, trying to track voltage, NFB at work. Initial excess negative error is likely result of forward prop delay, openloop gain generated, until NFB catches up. V(ref-out) shows difference between expected voltage on reactive load and real amp - supposedly exposing all amp nonlinearities arising due to reactive loading. Positive ref-out error voltage means amp output lags behind reference, though it also includes all amp nonlinearities. Difference is mainly lag (mostly positive error). Negative error dropouts on ref-out line seems to be due to back-emf.
Compared to ref-res, ideal amp with same load, we can see that output of real amp sort of lags ideal ref-res line.
Interesting event occurs near 60us. Seems like due to initial lag amp is compensating to reduce error, but due to backemf is flipping over to other extreme, overdoing, and is stuck tracking delayed NFB, being chronically late. I think this is where we can see error generated due to NFB lag.
I'd like to note that output current needs to lead the voltage. Initial current transient is very close to voltage in time, but as the waveform develops, the timing difference between current and voltage increases. This suggests that load is capacitive dominantly, and to develop voltage NFB expects and targets, current must ideally lead input transient at least for amount of amp lag. In other words, I'm having impression that for voltage NFB to work right, it must start compensating before input signal arrives. As this is impossible, at all transients error voltage is developed, and the "catching up" rush begins. That rush causes several problems by trying to correct voltage error that may have been safely ignored. For one, apart from nonlinearities of amp, I think this is causing phase modulation on leading edges of transients.
I'd compare this to jitter. In fact, while trying to achieve null, I've found that it is nearly impossible to distinguish errors arising from nonlinearities vs timing, and basically all nonlinearities could be reduced to signal correlated jitter. Remembering that incredibly low amounts of jitter is audible, that makes you wonder.
I'm getting general impression that voltage NFB should be taken before output stage where minimal reactance is, and output stage should be unity buffer outside global NFB. I can't see how NFB would otherwise work flawlessly. And by looking at different output impedances, I don't see that as very important thing byitself. I'm thinking that if output stage buffer has series feedback with the load, that is pretty much sufficient for damping.
Do you agree with my reasoning? |
|
|
| amplifierguru |
Agree with Mark.
Wimms your first FFT looks 'unnatural' ( what generated that?) and all the others are less than clear.
I'll have to have less beer I can see. |
|
|
| Graham Maynard |
Hi Wimms,
My - my, this takes some thinking about, hence the delay.
Before I go further. I choose to stick with the virtual 'Ariel' because it is a challengingly awkward load for amplifier testing.
When a crossover circuit has plain 6dB, or greater than 12dB per octave networks I have noted that it becomes much easier for an amplifier to drive, but then sound waveform recomposition via driver motor output could become much more difficult to optimise.
Another point. You are here displaying an amplifier circuit that cannot exhibit back-emf driven fractional bias voltage reverse commution as do the majority of conventional class-AB 'hi-fi' amplifiers ! Thus the waveforms here will be quite unlike and possibly less discontinuous that with other 'hi-fi' amplifier designs.
I agree that linear time display is very useful, but I would like to be able to see either the input voltage or the voltage at the output of the unloaded ideal amplifier divided by an appropriate factor (say 100 ~ 1000) to give a visual reference for zero level waveform voltage crossovers.
Without an original voltage waveform reference I am struggling to interpret what is happening.
At 60uS has the load induced output terminal current already reversed before the waveform voltage has passed through zero, and thus the nfb loop sensed current correction is no longer in phase with normal input voltage driving current ? This is where amplifier propagation delay generates entirely new parasitic waveforms that 'shimmy' or in the case of class-AB 'zig-zag' about normal. I would agree that it is similar to jitter, though I prefer to leave that term to its more logical meaning within CD technology, though both problems are related to differential image focussing accuracy and reproduction clarity.
Voltage nfb taken before the output stage will remove nfb loop induced interaction but still not be ideal either, especially with any output stage that is not perfect class-A. Series impedance will still allow output terminal voltage shift (wrt original waveform) to be generated by loudspeaker induced back emf such that a crossover or one driver back emf can still affect another driver, say modifying tweeter harmonics.
JLH chose to produce a good class-AB output and then use a 220 milliohm resistor between amp and loudspeaker, this would protect each from the worst reactive effects induced by the other.
The other approach is of course to construct an amplifier that is so fast (very low propagation delay) and powerful (high current capability without device protection) as to produce a perfectly amplified voltage no matter what loudspeaker loading effects might subsequently arise.
But as to which any design will sound best still cannot be determined theoretically, only empirically. Maybe Susan already has hit the mark with her 'no global feedback' circuits.
Cheers ............. Graham. |
|
|
| wimms |
Graham, I'm in a hurry, so just to clarify a little.
I chose another speakermodel instead of Ariel because I've found it to be more difficult load in sims. I'll post its circuit later. I realise that I should have used Ariel for more common ground, but I didn't want to repeat all the nulling again. I'll look into that later.
I played with this JLH/AABB combo just to look if and how does reactive load impact class-A amps as well. I realize that in AB amps it would have been worse.
Re input voltage reference. It is there as out*4u. This amp follows input so well that graphing input voltage separately isn't showing any visible differences. The error voltages are -100db below output levels. The voltage lag between input and output is around 10ns. This isn't visible. So you can safely assume that V(out) is the input waveform.
btw, I made an error in previous post, the wave starts at 10us not 100us.
I'm not sure if perfect voltage source no matter what loading is actually right goal to pursue. More on that later. |
|
|
| Graham Maynard |
Hi Wimms.
My method for an initial setting up the fundamental null.
Run say three full cycles. Measure the plus and minus peak amplitudes for the second cycle and use their average for the null source amplitude.
Expand the horizontal axis for the end of second cycle zero crossover, and measure the microseconds of delay; eg. 1uS.
Using this figure as a percentage of the cyclic time period calculate the angle of delay wrt 360 degrees; eg. 3.6 degrees @10kHz.
Subtract this figure from 360 degrees, and use the result as the starting angle for the nulling source; eg. 356.4 degrees.
This can speed things up a bit.
__________________________________________
I am still at a loss to understand your traces in post#14.
I would normally show the input or output terminal voltage scaled down so that an energising voltage reference is clear. If I saw an outline sketch of the test arrangement showing the meter positions it might help.
I also found that because any amplifier is differently influenced by its load/signal, a fundamental null must be separately established for each specific amplifier/load/sinusoid.
Take the green trace.
This looks like the sort of trace I have had, though here there may be a need to adjust the amplitude by about 50uV.
While the distortion amplitude is low it reveals output stage imbalance which I have concluded as being a cause of audible distortion when you get any JLH class-A like circuit going in real life and momentarily overloaded by music waveforms.
The green trace shows an error relating to inherent inability for the positive pulling output device to fully match nfb loop control output slew due to its limited pure current positive drive and the lack of a proper turn off for the lower device; ie. no pull-down.
Moving on to 60uS, the negative swing becomes accurately nfb loop control driven through a negative slew because the negative pulling device has its current drive via the phase splitter emitter while the phase splitter collector simultaneously pulls down the positive pulling output device. Charge storage within the upper device provides an additional source of current for the phase splitter to energise the lower output device.
Cheers ............. Graham. |
|
|
| wimms |
Hi Graham,
See attached a sketch of my test setup. It took me long to figure out how to let the spice do most of the nulling. I'm running this setup several times to approach null. First simple run, and observe the Zout and Gain in meas results. These are amp specs into given load. Copy these into params, rerun the test. Observe newgain2, if it differs from gain2, put new value into gain2. Observe Lag_error. It should be kept towards zero. Add the difference to lag of wlet2 input. Rerun the sim. I'm done when Zout, Gain2 doesn't change, and lag_error is below picosec.
| quote: | | The green trace shows an error relating to inherent inability for the positive pulling output device to fully match nfb loop control output slew due to its limited pure current positive drive and the lack of a proper turn off for the lower device; ie. no pull-down. |
I find that the trouble is not in the output stage, but the splitter. Especially without pulldown resistor. In that case all driver current simply must split perfectly. Trouble comes from splitter base current, and from gm changes of splitter with voltage swing.
I don't see purpose for pulldown resistors in driving 100% class-a stage. The current never reverses from bases, thus pulldown never occurs. It simply loads the splitter. CCS seems more fit for that.
| quote: | | Moving on to 60uS, the negative swing becomes accurately nfb loop control driven through a negative slew because the negative pulling device has its current drive via the phase splitter emitter while the phase splitter collector simultaneously pulls down the positive pulling output device. Charge storage within the upper device provides an additional source of current for the phase splitter to energise the lower output device. |
But, upper CCS is never exhausted, there is always enough current for positive device, like there is no other current source for lower device. For upper device to provide its base storage, there must occur base current reversal. How can that happen if bootstrap isn't exhausted? That would cause output switching. I don't think upper device base charge is contributing any splitter current. It only causes lag in current modulation of output devices, and upper device base charge is sucked out through crossconducance or load.
At 60us, I'm not sure the nfb loop has any control there. The lower device pulling is to a degree canceled by upper device excess base charge induced current. The sudden drop of error signal is imho simply because of inductive reactance kicking its backemf.
I'm wondering that backemf can't help consuming base charge of lower device, and they stay excessively conducting for longer than upper device, giving impression that its the upper devices that are slow pulling up. |
|
|
| scott wurcer |
| quote: | Originally posted by Nelson Pass
I'm going to suggest that simulations will not accurately
predict distortion level at -100 db levels. I would not
even trust them at 1% (-40 dB). |
Nelson, I simply suggest that you are wrong based on my experience with our internal simulators. That which is available to the masses just does not match up with the state of the art. |
|
|
| andy_c |
Scott,
I noticed a problem when trying to extract the parameters of the MJL3281a from the characteristic curves of the data sheet. Estimating the Early Voltage from the slope of the characteristic curves gave one number that was relatively low, while estimating it from the beta vs. Ic and Vce (for which beta varied only a small amount) gave a much higher number. The beta curves were taken at lower voltages and currents than those on the characteristic curves. So it appears that the devices don't fit the Gummel-Poon model of Early Voltage behavior particularly well.
In RF and microwave circles, the MEXTRAM model is becoming more popular now http://www.semiconductors.philips.c...ipolar/mextram/.
I noticed it models a bias-dependent Early effect. Have you ever used MEXTRAM, and if so, does it give more accurate distortion predictions than Gummel-Poon? I suppose this question might be more than you're willing to disclose, so I won't feel bad if you ignore it :). |
|
|
| scott wurcer |
| quote: | Originally posted by andy_c
Scott,
I noticed a problem when trying to extract the parameters of the MJL3281a from the characteristic curves of the data sheet. Estimating the Early Voltage from the slope of the characteristic curves gave one number that was relatively low, while estimating it from the beta vs. Ic and Vce (for which beta varied only a small amount) gave a much higher number. The beta curves were taken at lower voltages and currents than those on the characteristic curves. So it appears that the devices don't fit the Gummel-Poon model of Early Voltage behavior particularly well.
In RF and microwave circles, the MEXTRAM model is becoming more popular now http://www.semiconductors.philips.c...ipolar/mextram/.
I noticed it models a bias-dependent Early effect. Have you ever used MEXTRAM, and if so, does it give more accurate distortion predictions than Gummel-Poon? I suppose this question might be more than you're willing to disclose, so I won't feel bad if you ignore it :). |
We use our own extended Gummel-Poon model for Bi-polars. A simply linear model of the Early effect will generally not give you much distortion. I would say these days simulators start falling apart at the 70-80dB level. One must be careful when discussing distortion from subtle device related effects or when it is inherent in the circuit topolgy and bias levels. |
|
|
| Terry_Demol |
| quote: | Originally posted by scott wurcer
We use our own extended Gummel-Poon model for Bi-polars. A simply linear model of the Early effect will generally not give you much distortion. I would say these days simulators start falling apart at the 70-80dB level. One must be careful when discussing distortion from subtle device related effects or when it is inherent in the circuit topolgy and bias levels. |
Yes, so true. I get 0 feedback (global/interstage) circuits of
various function to go well below 0.001%.
IME the linearity is foremost based on your 2nd point,
topology and bias levels; however it is impossible to go
further than a certain point without addressing the first
point; subtle device related effects.
T |
|
|
| Graham Maynard |
Hi Wimms,
Thank you for showing your outline circuit.
Due to other committments it has taken me some time to get my mind around your investigations.
I note that the reference virtual amp does not include output C or L, which both change with amplitude, frequency and load in a real propagation delayed NFB amplifier due to load current variation affecting gain bandwidth product throughout each waveform cycle.
If the input trace is V(out)*4u, then this itself is asymmetrical, and therefore time shifted errors will not have a smooth symmetrical basis upon which to base conclusions.
Note how I(Rout) does not cease immediately. Whether this error arises at the start or finish of a toneburst is irrelevent, it is a propagation delay/ NFB loop/ loudspeaker impedance artifact, here having independent tweeter energising (waveform distorting) capability, as illustrated by V(out-inp) and V(ref-res).
It would be interesting to see V(ref-out) with the virtual amp having internal C and L, in Farads and micro-ohms, for this should properly isolate propagation delay/ NFB loop induced error for the circuit under test.
Also, although you are concerned about sudden start-up, the error trace due to continuation after a sudden start-up via a good amplifier is unlikely to be much different to your own delayed modulation trace after the same 10uS, with the benefit that any introduced errors may be viewed against a symmetrically energising sinewave input.
Re your impression that the NFB 'needs to compensate before the input arrives'.
This is of course is impossible. A series output inductor can help the amplifier here, but it also increases amplifier-loudspeaker interface distortion because it is outside the NFB loop and driven by extremely low impedance.
Quad very cleverly use a series output inductor as part of the NFB output sensing current dumping bridge, but I found this less good sounding than simple class-A.
I'm all ears to hear any suggestions that voltage source drive is not best, but I believe the main differences causing amplifier 'sound' relate to series output impedance, as opposed to output resistance, and maybe a standardised 220mR, as used by JLH so long ago would prove useful.
Re my JLH/class-AABB. The splitter currents are imbalanced and become increasingly so due to NFB with approaching load or input induced overload. That is why I eventually, but reluctantly, gave up on it.
It is not the bootstrap that becomes exhausted, but that the lower device receives additional imbalancing currents through the driver transistor from the base of the upper device, plus, driver base current passing straight through to the lower output from the input stage. This is all simulatable, as well as audible!
Pulldown resistors can be seen as being necessary during high frequency audio output to minimise base carrier/charge build up so that an output device quickly drops out of high load current conduction as its V.ce is suddenly increased by a push-pull partner. This applies more to non-complementary output stages than to centrally biased NPN/PNP output stages.
Hi Nelson.
I quite agree.
We can only gain knowledge of the mechanisms involved, not read the simulations as if factual.
Cheers ........... Graham Maynard. |
|
|
| wimms |
Hi Graham,| quote: | Originally posted by Graham Maynard
I note that the reference virtual amp does not include output C or L, which both change with amplitude, frequency and load in a real propagation delayed NFB amplifier due to load current variation affecting gain bandwidth product throughout each waveform cycle.
..
It would be interesting to see V(ref-out) with the virtual amp having internal C and L, in Farads and micro-ohms, for this should properly isolate propagation delay/ NFB loop induced error for the circuit under test. |
I'm not sure I follow you. In what way could I have added output C and L to the virtual amp? Please advise how to do that meaningfully. The DUT amp had only 1 compensation (splitter) cap in it, no L. Still, imo the exercise was to graph the difference between ideal and real amp. The difference between DUT into Rload and DUT into speaker load you already illustrated.
| quote: | | If the input trace is V(out)*4u, then this itself is asymmetrical, and therefore time shifted errors will not have a smooth symmetrical basis upon which to base conclusions. |
My goal was to reduce harmonic pollution to minimum, to avoid (still excessive) impact of phase response to different frequencies. I don't think asymmetry is much of trouble here.
| quote: | | Also, although you are concerned about sudden start-up, the error trace due to continuation after a sudden start-up via a good amplifier is unlikely to be much different to your own delayed modulation trace after the same 10uS, with the benefit that any introduced errors may be viewed against a symmetrically energising sinewave input. |
Sudden startup created spikes, oscillation at both ends, and unknown time shifts that I can't make sense of. Later pure sinewave shows no transient details. I picked the waveform that has severe yet harmonic transient that both amp and speaker are still expected to follow. Imo this rids us from worries that we have overdriven the system with signal to which the system can only answer with spurious response.
But for completeness I include run of same amp with sudden starting sine. It is indeed pretty similar. The main difference is output current that in sudden sine case wobbles for considerable time on time axis like spring that has been hit instead of compressed.
| quote: | Re your impression that the NFB 'needs to compensate before the input arrives'.
This is of course is impossible. A series output inductor can help the amplifier here, but it also increases amplifier-loudspeaker interface distortion because it is outside the NFB loop and driven by extremely low impedance. |
Do you agree that my impression is valid? Because I'm not 100% confident myself. Though I can see that current-drive amps and non-NFB amps are free from that issue, as neither is stuck on tracking speaker terminal voltage.
| quote: | | Note how I(Rout) does not cease immediately. Whether this error arises at the start or finish of a toneburst is irrelevent, it is a propagation delay/ NFB loop/ loudspeaker impedance artifact, here having independent tweeter energising (waveform distorting) capability, as illustrated by V(out-inp) and V(ref-res). |
This is the main point of wondering what is the right method of speaker driving. This non-ceasing of current is present with voltage drive. With current drive, its the voltage that does not cease.
| quote: | | I'm all ears to hear any suggestions that voltage source drive is not best, but I believe the main differences causing amplifier 'sound' relate to series output impedance, as opposed to output resistance, and maybe a standardised 220mR, as used by JLH so long ago would prove useful. |
The crux of the matter is: which is more harming - shorting the stored speaker energy, creating delayed currents in it, or letting it float. In current drive, the current ceases with input signal, but due to high output impedance, terminal voltage is doing dances. Which is worse?
Re voltage drive, it really starts here - cone motion force is proportional to *current* through it, voltage on crossover input is pretty far from controlling linearity of that one.
| quote: | | Pulldown resistors can be seen as being necessary during high frequency audio output to minimise base carrier/charge build up so that an output device quickly drops out of high load current conduction as its V.ce is suddenly increased by a push-pull partner. This applies more to non-complementary output stages than to centrally biased NPN/PNP output stages. |
I need some help here. I still don't see how pulldown resistor can help when splitter is driving BJT that never exits class-A, isn't floating but tied to PS, and base current can't reverse until it or splitter drops out of class-A. Can you help me with this? I see this only required in class-AB to minimize crossconductance. In your class-AABB setup, there is in addition fact that high-current B and AB devices have their base pulldowns. |
|
|
| wimms |
| quote: | | I'm all ears to hear any suggestions that voltage source drive is not best, but I believe the main differences causing amplifier 'sound' relate to series output impedance, as opposed to output resistance, and maybe a standardised 220mR, as used by JLH so long ago would prove useful. |
I don't know if this is all familiar, but I'll repost here verbatim what Thomas Dunker wrote on the JoeList. This was also posted here http://www.diyaudio.com/forums/show...10&pagenumber=6 Hopefully this doesn't come as boring.
---------------------------------------------------------
Fri, 20 Sep 2002
Subject: [JN] Field-Coils and missing considerations
My assorted ramblings follow:
If you had read up on the issue of flux modulation and resulting harmonic and IM distortion in electrodynamic speaker units, perhaps you'd realize that the only magnet structures likely to rival field coils in that regard would be neodymium ones. Next would be alnico, and finally, the most inferior of the lot, ferrite types. It's all in the BH curves.
The extreme opposite of an old, high efficiency field coil driver, would be a modern, low efficiency ferrite PM driver. It has been assumed that one could make up for a lousy magnet system by calling for greater flux variations from the voice coil, which only increases flux modulation. If you have "something" with a so-so linearity to begin with, and feedback is impractical, the only way to keep distortion down is to make the signal component use as small as possible a fraction of that nonlinear characteristic.
This applies everywhere in an analog audio reproduction chain, speakers being but one example.
Modern speakers have evolved AWAY from literally ALL the wise and scientifically sound rules of thumb that the speaker pioneers of the 1920s-1930s adhered to.
Why does one often find that one is wasting one's time trying to explain that certain "obsolete" technologies may in fact have technological advantages over the contemporary "state of the art"? The main reason that field coils went out of fashion was that permanent magnets eventually became ECONOMICALLY more viable than field coils and because they're not as LABOR-INTENSIVE as coils, and finally that PMs are more CONVENIENT in that they don't require a steady supply of DC. Concerns such as these, having less to do with objective performance, and more to do with profit margins, also pushed horn speakers, triodes, (and eventually pentodes), hardwired circuitry, vinyl, analog tape and various other advanced technologies off the consumer audio scene.
Since the ridiculously shallow "tube vs transistor" discussion sort of came up again ("hey, I once/never heard a good transistor amp"), I can't help but comment on that as well.
If one puts ANY amplifier inside a feedback loop designed to make the amp a "perfect" voltage amplifier - one ALWAYS measures voltage distortion - you have made an amplifier ideally suited to drive a theoretical speaker that basically does not exist - a speaker which has all the same electrical properties as a noninductive power resistor. If we could enjoy music directly from resistive dummy loads, these amps would be all we could ever ask for. Or if we could make a truly voltage controlled SPEAKER with no complex impedance components...
The minimum distortion possible from ANY electrodynamic speaker is a matter of how much distortion there is on the CURRENT driving it, not the voltage, as the industry appears to believe. Rather, if you have the feedback trying to - and more or less succeeding at - maintaining undistorted output voltage while driving a speaker that is very far from a purely resistive load, the feedback not only FAILS TO REDUCE current distortion (it was never asked to) but in fact INCREASES it.
This, however, is no point to make to discredit the IDEA of feedback. Instead we need to look at why, where and how it's being used. For a number of reasons, any form of feedback works better with predictable, stable loads. Should we still want to use feedback in power amps driving speakers, it would seem a lot more sensible to use it to maintain a linear relationship between the INPUT VOLTAGE SIGNAL and the OUTPUT CURRENT SIGNAL.
But since modern speakers are every bit as stupid as modern amps, we'd have to redesign both. Modern speakers depend on low output impedance from the amp for bass damping, and the drivers and crossovers are optimised to work "optimally" in the "undistorted voltage" situation. Here the logic is getting so absurdly twisted that it's "understandable" how the industry pretends nothing is wrong.
I haven't looked into digital amps a great deal. Maybe they avoid some of the problems with conventional amps, but they're still asked to produce "undistorted voltage amplification" - to suit conventional speakers with inadequate self-damping. And speakers still aren't resistive loads, so there can be no proportionality between voltage and current, hence the digital amps don't solve the main problem: There is NO WILL within the industry to radically redesign the whole amp-speaker system , as the known problems seem either unknown or ignored, so we'll be stuck with all the same old market-optimized set of compromises.
The point to make is: An amp that is driving a speaker is part of a system made up of the amp and the speaker, with all sorts of interdependencies, interactions and general MESS. If the speaker is left out of the discussion or analysis of amplifier technology, we can't expect to learn anything useful about the amps. Audiophiles have been puzzled at the uselessness of commonly made measurements on amps and the way some amps are more sensitive to the choice of speakers than others etc. etc.
Seeing no obvious explanations for these discrepancies (which are easily explained), the reaction was to dismiss all attempts at objectively assessing the quality of audio reproduction. It quickly became very unhip to get too scientific about audio. The mantra was "trust only your ears".
Triodes, pentodes, BJTs, FETs etc. are ALL devices which allow us to control a current, and should all be interesting for amplification once we decide to focus on the distortion of a power amp's output CURRENT rather than the voltage.
An SE triode is an obvious example of a sensible amplifier to use with the speakers they made in the 1930s. A stage or two of pure voltage amplification, to the grid of the output triode, producing a corresponding current swing in the transformer primary, and producing a proportional output current on the secondary. If the tubes are all really linear and the speaker can get away with a few hundred milliwatts of power, this is as simple and direct as amplification gets. Given that they didn't rely as heavily on feedback back then, they developed the most linear amplification devices they could manage. Some of the most linear triodes, like 211, RE604 and AD1 are still some of the most linear amplifiying devices EVER made. This was before the idea caught hold that everything can be corrected after the fact, which we're still stuck with.
If speakers hadn't "evolved" the way they did, amplifier technology wouldn't have "evolved" the way it has either.
If an amp today has to have 48 transistors, several feedback loops, a kilowatt's worth of power supply and 50 pounds of heat sinks just to produce the 300 watts required to drive the 0.1% efficient speaker to decent SPLs but still no convincing dynamics, and it still doesn't deal with any of the major known distortion problems, how can anyone talk about progress?
One of my sources for inspiration in audio has been Ragnar Lian, co-founder of Scan-Speak back in the 70s and a living loudspeaker legend here in Scandinavia. He wrote dozens of articles with an inclination similar to mine, explaining known problems and suggesting solutions, and pointing to the decline of audio technology during the second half of the 20th century.
I'm under the impression that he eventually felt that the only people who listened to him were people OUTSIDE the industry, and that within the confines of commercial audio (yes, even "high-end") there is only so much you CAN do because the LAST thing any conservative institution will do is to admit to its past mistakes. Ragnar's vision was to start over from scratch and build integrated amp/speaker SYSTEMS addressing all the problems that the industry has failed to solve since the fateful split between amp manufacturers and speaker manufacturers.
He grew tired of the stubbornness and conservativity of the Scandinavian speaker industry, I guess. If he hasn't retired yet, he's still working as a magnetics engineer on heavy industrial magnetic systems.
Most ironic of all is that now, two decades after CD came out on the market, and with emerging new media that are theoretically capable of insane dynamic range far beyond that of CD, we're STILL struggling to build systems that can competently deal with the best dynamic range available from the old analog media.
Aw, time to step down from the soap box.
Thomas
---------------------------------------------------------
Another interesting post by Thomas is
Theory of the Amplifier / Speaker Interface
http://melhuish.org/audio/article5.html |
|
|
| andy_c |
| quote: | Originally posted by scott wurcer
We use our own extended Gummel-Poon model for Bi-polars. A simply linear model of the Early effect will generally not give you much distortion. I would say these days simulators start falling apart at the 70-80dB level. One must be careful when discussing distortion from subtle device related effects or when it is inherent in the circuit topolgy and bias levels. |
Hi Scott,
The reason I mentioned the Early effect is a previous discussion about the JLH amp and some simulations that were done of its output stage. The thread is here. The JLH output stage is very sensitive to beta, such that the sharing of current between the upper and lower output devices ends up being quite a strong function of the Early voltage. If the Early voltage of the output device model is made very large, the current sharing is almost perfect between the upper and lower output devices. If the Early voltage is made smaller, the current sharing becomes quite asymmetrical, as the sim in jcx's first post in that thread indicates. This makes the simulated output power somewhat less than it should be. In trying to figure out which of the computed Early voltages to use (based on beta vs Ic with Vce as parameter, or from the slope of the output characteristic curves), it became clear that a bias-dependent Early voltage would be required to really model what's going on.
In the majority of cases I agree that the effect won't be terribly pronounced. But in that thread, it became clear that the JLH design is exceptionally sensitive to beta variations of the output devices.
Anyway, that's the history behind my earlier comments and question.
Andy |
|
|
| Graham Maynard |
Hi Thomas,
I guess we try to satisfy our own requirements rather than trust the commercial designs from those who don't 'start afresh' because they would not profit from doing so.
In my class-AABB there are output 'pulldowns', but there is also a fixed minimum bias from the previous stage which is maintained until very high power demand. Also the drivers do not turn off, and these maintain continuous low level output load contribution.
___________________________
I saw you had given your ideal amplifier an 'output resistance'.
This is part way between an ideal amplifier, and an ideal amplifier plus output impedance that matches the amplifier under test.
An ideal amplifier with passive output impedance is not the same as fundamental nulling because loudspeaker system induced back-emf can generate a fractional voltage across that output resistance.
An ideal amplifier with series output R+L+C that matches the nominal R+L+C values of the amplifier under test will develop the same nominal fractional loudspeaker induced voltage, such that any difference between their outputs will be due to propagation delay/NFB variation in the amplifier under test alone.
Your test arrangement could do this if you add series L+C to the 'output resistance' component already shown connected to your ideal amplier output. I though this was your aim.
To obtain approximate values, run a voltage amplitude plus phase plot for your DUT with the loudspeaker load alone; say 1Hz to 100KhZ Then simulate with (m)R+(u)L+C(F) components in place of DUT and adjust their values until the plots match.
___________________________
It does not matter whether we go for pure voltage or pure current drive, neither alone sounds right via loudspeakers that have reactive characteristics within the AF range.
For quality reproduction should we go for low distortion voltage drive and add a series resistor to minimise increased power draw through dynamically induced impedance dips, followed by a parallel load resistor to minimise instability due to a dynamically induced impedance rise; say 0.22 and 10 ohms, for 8 ohm nominal ?
In the past I have noted that some class-AB ampliers sounded better with a small series resistor or a parallel output resistor. Both options increased the thd at the loudspeaker terminal, and the latter became hot as it increased amplifier loading, yet the audio WAS better.
Maybe there would be merit in trying such components on non-class-A NFB amplifiers, for neither option has been worthwhile on a JLH based class-A.
Cheers .......... Graham. |
|
|
| Graham Maynard |
Hi Thomas,
Just realised that I did not mention to reverse drive the amplifier to obtain the output impedance characteristics.
You can do this via the loudspeaker, or via a nominal load such as 8 ohms.
Cheers ........ Graham. |
|
|
| sam9 |
Nelson Pass wrote:
| quote: | I'm going to suggest that simulations will not accurately
predict distortion level at -100 db levels. I would not
even trust them at 1% (-40 dB).
|
It ocurrs to me that whether they will or not, one doesn't really know until an equivalent measurement is made on a real life unit. Based on the scale and resolution appearing on the some of the graphs posted achieving that would a heroic act deserving three cheers and drinks on the house. |
|
|
| wimms |
Hi Graham,
Do not confuse me with Thomas Dunker, I only reposted one of his old postings. He himself isn't afaik visiting this forum.
| quote: | Originally posted by Graham Maynard
I saw you had given your ideal amplifier an 'output resistance'.
This is part way between an ideal amplifier, and an ideal amplifier plus output impedance that matches the amplifier under test.
An ideal amplifier with passive output impedance is not the same as fundamental nulling because loudspeaker system induced back-emf can generate a fractional voltage across that output resistance. |
Hmm, let me explain that decision. Perfect voltage source, as in zero output impedance, is clearly impossible physically and thus imo unreasonable goal or reference. We can't get rid of wire resistence. It makes no sense to simulate speakerload attached to perfect voltage source. In ideal Amp with finite output impedance we try to achieve linearity. Simulating with perfect voltage source and finite Zout I'm trying to simulate that. If I add C+L, then I can't help but think of simulating distortion of the amp under test, but my goal was to expose that distortion, nonlinear or reactive.
Even though finite Zout changes waveforms due to backemf, it must be of same kind as with realistic amp if latter was strictly linear and had static output impedance. Comparing VS+Zout with DUT imo cancels out expected and unavoidable effects of reactive load, and exposes best what nonlinearities DUT leaves, at merely better zoom factor. Thats why DUT isn't really expected to produce null output after fundamental nulling, but similar waveform as VS+Zout is showing. Thats why I tried simulating such waveform, to compare the actual output of DUT to it.
But I'll try your suggestion sometime to see if it makes a difference. I'm just worried that I'll loose solid ground of reference due to handpicking.| quote: | It does not matter whether we go for pure voltage or pure current drive, neither alone sounds right via loudspeakers that have reactive characteristics within the AF range.
For quality reproduction should we go for low distortion voltage drive and add a series resistor to minimise increased power draw through dynamically induced impedance dips, followed by a parallel load resistor to minimise instability due to a dynamically induced impedance rise; say 0.22 and 10 ohms, for 8 ohm nominal? |
I hope you gave the post of Thomas a good thought and also checked the link I gave. What Thomas said gathers together alot of background. Dynamic speaker driver is fundamentally current-driven device. He's not alone in touting this. It is true though that neither is best, pure current vs pure voltage drive. Speaker and amp fundamentally needs to be matched to each other. When one realises that, it becomes clear that building ideal amp that would fit any kind of standalone speaker with passive XO is futile - after we build an ideal voltage amp, we hit the issues that perfect voltage isn't ideal signal to drive the passive speakers.
Imo we really need to look into effects of perfect voltage drive to the sound, without the freedom to ignore the physics of the transducers. And the physics indeed hints that we should be more concerned about linearity of *current* through transducers, not so much about linearity of voltage on amp terminals.
I find it much more meaningful to attribute audible differences between tubes or SS-no-NFB and SS NFB to that of output drive mode rather than any of the THD. Higher output impedance seems to suit better many, especially efficient drivers. There are physical reasons why that reduces IMD and compression.
http://www.firstwatt.com/current_source_amps_1.htm Nelson certainly can say a word or two on that matter.
http://sound.westhost.com/project56.htm - there is a nice example how to play with controlled output impedance of the amp without wasting too much power on series resistors.
Re sticking with voltage drive.
I believe that reactance of load must be isolated from NFB loop. Quite likely simple series resistance achieves that to a degree, but imo isn't a complete solution. Or NFB must be extremely fast, on the order of (sub)nanosec lag. Current-fedback amps like JLH are approaching that, and perhaps that is the reason why they don't benefit as much from small series resistence. Yet we still need to look into currents in transducers and corresponding acoustic output. I've tried to find a good speakerload model that could reasonably simulate acoustic output, but haven't found one. It would be nice to observe acoustic effects of different approaches.
Does anybody have any good model of speaker that reasonably simulates expected acoustic output, or could help developing one? |
|
|
| noname |
"This method takes advantage of the characteristics of a low-distortion instrumentation amplifier."
from
http://www.elecdesign.com/Articles/...&ArticleID=8188
"The delay causes the biggest error in the fundamental, which is not important. The harmonics exhibit error also, but as they are so far down in amplitude (with respect to the fundamental), the errors are second order and have little effect on the total distortion calculation." |
|
|
| d3imlay |
| I seem to recall that Hafler used to suggest that the ouput of an amplifier channel be inverted and summed with the original signal and fed into the second channel. The result in the second channel is the non linearities. You listen to the second channel and get a relative idea of amplifier quality. |
|
|
| Graham Maynard |
Hi Wimms,
Please forgive my occasional confusions, these are due to time shortages and cerebral stress from unresolved head injury, also the nature of this inkless forum. I am okay sitting at a table with paper and written replies. Went through another bad patch recently, so am only now catching up since my last reply.
Thanks for the links - good to see their content published.
Of course all of this was previously realised here, but there has still not been any expression of possibility for back emf induced amplifier non-linearity due to amplifier NFB loop/propagation delay.
Unfortunately you modelled my 'relatively linear' AABB circuit, whereas I feel that either your 'Distortion Microscope' development or my 'X-Y' monitor would show up significant error products when other more conventional amplifier circuits are virtual loudspeaker loaded.
These 'interface' distortions have nothing to do with the unavoidable voltage/current loudspeaker drive arguments that tend to be simple in nature, but are much more complex and can be orders of magnitude greater than thd figures suggest in non-class-A designs. They also remain largely unrecognised - unaccepted - undiscussed.
This was why I suggested your reference amplifier could have output L and C as well as R, so that back emf induced loudspeaker error would not mask amplifier induced error. Your last test circuit could be better than mine at imaging interface induced distortion in isolation. Try it with another circuit, not one where I have attempted to already minimise the errors you are looking for.
Cheers ......... Graham. |
|
|
|
