Test number 2,
Inject an error current at the input of the output stage, sort of like there is some base current error due to driving a heavy load. Gee this current and all of its non-linearity appears in the feedback resistor on the so called current feedback op-amp while the input voltage error is hardly changed.
Inject an error current at the input of the output stage, sort of like there is some base current error due to driving a heavy load. Gee this current and all of its non-linearity appears in the feedback resistor on the so called current feedback op-amp while the input voltage error is hardly changed.
Certainly no problem with a JFET-input power amp, which is what I prefer anyway
.Cheers,
Bob
Just a gentle reminder that this Fourier etc sh*t relies on the system being Linear Time Invariant.
While amps are probably close enough to this most of the time to make FTs etc a really useful tool .. speakers are most certainly NOT even for some of the time.
It's just convenient for us to assume that they are ... as long as you keep the caveats always in mind
I think you're overstating here, the acoustic signal when captured by a mic becomes two dimensional time invariant data. The theory is still valid. The theory does not rely on that the conclusions gleened from the result do.
I though you were a big fan of Mr. Farina?
I am ... and also of Mr. Fourier. I've been using Jean Baptiste's stuff since late 70s and developed the theory for Angelo's method in the early 90s for production testing of speakers in the theoretically shortest possible time. Alas computing power and particularly good A/Ds & were too expensive for what we wanted in dem days.
But I'm quite happy to call it Angelo's method cos he published first and certainly popularized it.
Angelo's method makes it very easy to demonstrate non-Linear non-Time Invariancy in speakers
Just do response & THD curves using 1s & 10s sweeps at 2.83V. Especially with non Ferrofluid treble units.For sensible amps, not run into clipping, Angelo's 'weakly non-linear' criteria is certainly valid.
Pity about da practical bandwidth limitations with evil sampled versions of Mr. Fourier though.
But when we get EMSSACD (Even More Supa SACD), cheapo soundcards will do zillion bits at zillion MHz and we'll have reached the Millenium
_______________
So not questioning the validity of this Fourier sh*t .. just suggesting caution in interpreting its use especially with 'piecewise-linear-time-invariant, as the piece lengths go to zero', an oxymoron if only eye kud reed en rite
Last edited:
Perfectly valid to use oxymorons for humour. Shakespeare did. Not sure everyone got the joke though. Gotta be at least an FFT pseudo guru
Not all of about fft is useless, at least it could be used to determining wether the amp change the sound by non linearity or not (change means not always bad but sometimes also good changes). Its number is 0.01%THD @major tone and less (>0.01%) at least heared tone.Perfectly valid to use oxymorons for humour. Shakespeare did. Not sure everyone got the joke though. Gotta be at least an FFT pseudo guru
About IMD, TIM, and other time based/transient quality measurement, I have yet found its usefull. So, we should accept 0.01% THD for spectrum analyzer based test, and the other test it is depend on everyone religion.
um, how do you find the IMD products? - fft is great for multitone stimulus seeing IMD products offset at multiples of the multitone frequency difference
TIM is IMD - just with different phase relation than "AM" IMD so any TIM IMD product is only a part of the indicated IMD
you really should be familiar with a few of the pubs on "conventional audio measurements"
https://www.google.com/#q=cabot+audio+distortion+measurement
TIM is IMD - just with different phase relation than "AM" IMD so any TIM IMD product is only a part of the indicated IMD
you really should be familiar with a few of the pubs on "conventional audio measurements"
https://www.google.com/#q=cabot+audio+distortion+measurement
IMD19k+20k, did you find it is usefull?,
1st, I found that it is usually has numbers that lower than THD20k.
2nd, I couldn't make any true judgement of sound differences caused by IMD. Add an additional sinewaves of 18k and 21k to loudspeaker for eliminate IMD product seems not changing the sound quality. I can't hear, Idon't know if someone could.
CCIF 19+20kHz is generally my favorite measurement of distortion for power amplifiers. It is generally true that the numbers are a bit smaller than THD-20, so one needs to calibrate this out in one's expectations. It has the great advantage of showing nearly all of the nonlinearities out to quite high order. These are important. A typical THD analyzer may have a measurement bandwidth of only 80kHz, which obscures much of the important stuff. Also, one only needs to look on the low-side sidebands, which lie below 20kHz, so a computer-based spectrum analyzer can often be used.
19+20k CCIF is almost always measured by John Atkinson in his amplifier reviews, and that is always one of the first figures I look at.
Finally, it should be noted that THD-20 is sometimes nearly useless in measuring class D amplifiers because of their output filter.
Cheers,
Bob
I would like to add that (few) HD is not a problem: harmonics components are parts of most of all musical instruments sound. While IM adds a kind of modulated 'noise', non harmonic, not correlated, blurring the scene. IMs are the enemies.
But at lower frequency, the number also getting lower and more unhearable (for example 4k+5k). For amplifier with <0.01% THD, eliminating the additional distortion isn't change the audible sound.
Until now, I am still using 0.01% or lower for best THD figure, lower than that is useless. If an amplifier has 0.01% THD20, there is also non hearable IMD.
If someone could hear IMD effect on a 0.01%THD amps please let me know the measurement lay out (schematic), my common measurement is using headphone (could be attached to resistor series with dummyload). Don't worry about bandwidth, I could use HP audio analyzer at baristand (government research facility).
ClassD THD20 could be heard by adding 20k+21k signal and listen the 1k product, because 1k is non linearity product.
Yes I have. But it was to see if VAS is more correct than TIS. (Answer: VAS is more correct by the Wurcer criteria) Can't remember where I posted the results.
I'm quite happy to call this thread Amplifiers without LTP IPS that might be loosely called CFAs
As eye kunt spul or reed en rite, da names dun meen much to me. Only if I can make better & simpler amps.
Last edited:
No, it was to show that the current in the feedback R is what replaces the error in the forward transfer both linear and non-linear components (displacement current in C and current draw in the output stage). The experiment is easy to do with an AD844.