depends
twin tones in band would be a most likely stimulus and out of band IMD products you actually want to be attenuated. the whole purpose of a LPF filter. ( that's one reason why I prefer iron laminations over ferrites > more core losses are a plus)
40-400Hz would be classic in band
twin tones in band would be a most likely stimulus and out of band IMD products you actually want to be attenuated. the whole purpose of a LPF filter. ( that's one reason why I prefer iron laminations over ferrites > more core losses are a plus)
40-400Hz would be classic in band
Last edited:
Ok. I think I follow. A similar test can be done with large AC machines, but mainly isn't now, in favour of easier tests.
It would.make.an.interesting test nevertheless. I wonder if it would be meaningful, and worth a shot.
But wouldn't the twin tones create IM beat freq of less than the stimulus? I.e. In band
And also of a higher (summed?) voltage?
I guess that makes a true reading of V and I difficult (increased crest factor). I also don't understand how the IMD products from an in band beat frequency are going to generate much out of band, as opposed to in band IMD.
But then, this is all new to me 😉
It would.make.an.interesting test nevertheless. I wonder if it would be meaningful, and worth a shot.
But wouldn't the twin tones create IM beat freq of less than the stimulus? I.e. In band
And also of a higher (summed?) voltage?
I guess that makes a true reading of V and I difficult (increased crest factor). I also don't understand how the IMD products from an in band beat frequency are going to generate much out of band, as opposed to in band IMD.
But then, this is all new to me 😉
Last edited:
that is pretty awesome inverter I wonder what the distortion figures are
I think I used one before to test some avionics gear. California Instruments or something
not sure it allowed inputs on the source
Last edited:
Its a programmable V, I, P calibrator supply, with phase angle adjustment. I recently used to evaluate a power analyser for accuracy at low power factors <0.1.
soft nonlinearities have a reasonable Taylor Series form - a power series in I or V
an x^n term creates n_th order harmonic with a single sine frequency
with 2 tone stimulus you get a*f1 +/- b*f2 where a, b are integers and add up to n, the order of the nonlinear term
if you step up your modeling to Volterra/Wiener you need multitones with n components to sweep n different frequencies vs level where n is the highest order nonlinear term you either suspect exists or is the highest in your band or above your noise threshold
it is in principle possible for a Volterra/Wiener model to have low single tone distortion and high multitone distortion - in practice its not often a big effect and you can see most nonlinearities with single tone harmonic distortion
but 2-tone IMD has long been used in audio because sometimes the difference products can be readily audible when harmonics and sum IMD are masked
an x^n term creates n_th order harmonic with a single sine frequency
with 2 tone stimulus you get a*f1 +/- b*f2 where a, b are integers and add up to n, the order of the nonlinear term
if you step up your modeling to Volterra/Wiener you need multitones with n components to sweep n different frequencies vs level where n is the highest order nonlinear term you either suspect exists or is the highest in your band or above your noise threshold
it is in principle possible for a Volterra/Wiener model to have low single tone distortion and high multitone distortion - in practice its not often a big effect and you can see most nonlinearities with single tone harmonic distortion
but 2-tone IMD has long been used in audio because sometimes the difference products can be readily audible when harmonics and sum IMD are masked
Last edited:
Well, I get a bit annoyed when people think that harmonic distortion measurements are the only measure of quality in town. They really only measure non-linearity.
As telecoms engineers we were often more interested in white-noise testing on a channel. Signal-to-noise ratio (S/N) is more fundamental, once you have maximised linearity IMO. The idea is you inject a sine wave tone to a channel. Then filter out the original test tone and look at the resulting white noise across the rest of the channel's spectrum. It's why soft domes, which measure well on distortion and frequency response, sound poor to me and might be suspect on S/N. They are a noisy channel in Information Theory terms.
Now Max Headroom supplied an extremely interesting noisy effect with cored coils. No-one was interested because you all think harmonic distortion. Because that is what oscilloscopes and spectrum analysers are good at.
https://en.wikipedia.org/wiki/Barkhausen_effect
Remember that in between the harmonics lies the Noise! And you're not seeing it.
A good speaker has a "quiet" sound that encourages you to turn the volume up. It's clear to me that air coils are going to work better. How much? Well that's your compromise on cost. Coils are measurably the worst components in a crossover, and show crosstalk and have microphonic effects more than anything else.
As telecoms engineers we were often more interested in white-noise testing on a channel. Signal-to-noise ratio (S/N) is more fundamental, once you have maximised linearity IMO. The idea is you inject a sine wave tone to a channel. Then filter out the original test tone and look at the resulting white noise across the rest of the channel's spectrum. It's why soft domes, which measure well on distortion and frequency response, sound poor to me and might be suspect on S/N. They are a noisy channel in Information Theory terms.
Now Max Headroom supplied an extremely interesting noisy effect with cored coils. No-one was interested because you all think harmonic distortion. Because that is what oscilloscopes and spectrum analysers are good at.
https://en.wikipedia.org/wiki/Barkhausen_effect
Remember that in between the harmonics lies the Noise! And you're not seeing it.
A good speaker has a "quiet" sound that encourages you to turn the volume up. It's clear to me that air coils are going to work better. How much? Well that's your compromise on cost. Coils are measurably the worst components in a crossover, and show crosstalk and have microphonic effects more than anything else.
Last edited:
telecom and system engineers know
noise testing is great for front ends where the key system gain is, but at the end of the chain (where the XO is) the S/N has already been established. In really low level pre-amps, sure microphonic issues and maybe the other magnetic curiosities might be worthy of some attentions, but come on, yer still pushing a red herring at this point. I think microphonics of ceramic caps around high gain stages (servos and such) is a real threat but microphonics of cored coils not too much. Besides phono cartridges they are rarely found in modern audio front ends anymore. I thought I addressed barkhausen Re dynamic range for XOs in a previous post. are you paying attention system7? any proper real system engineer worth his salt would be highly annoyed by your posting that low level noise introduced at the end of the chain having bearing on quality what so ever! I repeat doing a deep dive into noise levels where high level signals live is a red herring.
noise testing is great for front ends where the key system gain is, but at the end of the chain (where the XO is) the S/N has already been established. In really low level pre-amps, sure microphonic issues and maybe the other magnetic curiosities might be worthy of some attentions, but come on, yer still pushing a red herring at this point. I think microphonics of ceramic caps around high gain stages (servos and such) is a real threat but microphonics of cored coils not too much. Besides phono cartridges they are rarely found in modern audio front ends anymore. I thought I addressed barkhausen Re dynamic range for XOs in a previous post. are you paying attention system7? any proper real system engineer worth his salt would be highly annoyed by your posting that low level noise introduced at the end of the chain having bearing on quality what so ever! I repeat doing a deep dive into noise levels where high level signals live is a red herring.
Last edited:
Barkhausen effects is a small impact effect if you use a toroid it's not a one monolithic piece of marerial but the grain ferrite have a Gaussian distribution .
The noise have a large and uniform spectrum .
On the other hand the air inductor made a magnetic feedback with woofer (on small cabinet about 2 ways speaker) and not simple to know how it works....I think
this effect is not simple to know how it works
I'm from Missouri - Show Me
will someone please point to hard data, conditions, materials, units, coefficients, calcs, graphs of Barkhausen Noise in a real audio component?
will someone please point to hard data, conditions, materials, units, coefficients, calcs, graphs of Barkhausen Noise in a real audio component?
The level of the noise is important, and the spectrum of that noise more so.
Ferrite noise is audible and adds a characteristic signature to inductors and magnets.
Dan.
Last edited:
in crossovers says who?
there is something called dynamic range, maybe you don't understand the concept.
your head seems to be buried in noise too.
You seem to have cloth ears, or perhaps your head is up your ****.
Dan.
Note to mods - I am merely giving back as is given.
a troll calling me names your argument is done haha
do read anything in in this thread, or merely pop in periodically shouting "whoah dudes the sky is falling"
do read anything in in this thread, or merely pop in periodically shouting "whoah dudes the sky is falling"
Last edited:
Having never heard of Barkhausen noise, and reading the brief wiki explanation I would apply my usual attitude.
It would seem to me, given the definition, that it is current slew dependant. (rapid changes in magnetisation)
Due to the wooly definition the further question emerges:
What is rapid, as far as proven occurrence of Barkhausen noise? Or rather flux slew/field slew. This needs quantifying to understand if this noise can occur under conditions 'we' are likely to experience.
A further thought that follows: Is the deviation from ideal of great enough amplitude and period to be audible?
Without hard numbers, I have to assume that it either does not occur, or is buried in THD and not readily distinguishable. (it looks rather like very low amplitude slightly randomised quantisation error)
Id guess its more of a high flux slew, high frequency, high current slew effect. Like skin effect, a non issue for audio.
This is of course as no numbers, thresholds or other evidence were shown.
It would seem to me, given the definition, that it is current slew dependant. (rapid changes in magnetisation)
Due to the wooly definition the further question emerges:
What is rapid, as far as proven occurrence of Barkhausen noise? Or rather flux slew/field slew. This needs quantifying to understand if this noise can occur under conditions 'we' are likely to experience.
A further thought that follows: Is the deviation from ideal of great enough amplitude and period to be audible?
Without hard numbers, I have to assume that it either does not occur, or is buried in THD and not readily distinguishable. (it looks rather like very low amplitude slightly randomised quantisation error)
Id guess its more of a high flux slew, high frequency, high current slew effect. Like skin effect, a non issue for audio.
This is of course as no numbers, thresholds or other evidence were shown.
not trying to feed the troll, the effect might be an issue for a transducer with enough filtered gain using really good low noise floor preamp. certain materials have more of it and ferrites are not used in the demo not by chance
Last edited:
We have had many discussions of crossover components at diyaspergers.com. 🙂
The maths seems to say that 10W wirewound/ceramic resistors are hugely lower inductance than your average tweeter, and don't get hot, so ought to work well enough.
MKP capacitors seem to be about as good as it gets even in cheapish 250V thin foil implementations at 0.1% loss. MKT polyesters are measurably worse at around 1% loss. NP electrolytics add a small sort of white noise or sibilance to female vocal that I find audible in 1:1 comparisons.
Barkhausen_effect seems like the "smoking gun" measurable problem with cored coils. FWIW, most component tolerances manifest themselves most at the -3dB or -6db point. It doubtless applies to speaker magnets too, hence perhaps the AlNiCo camp who swear those old magnets sounded better. This is in addition to non-linear hysteresis effects.
For all that, I find 1mH ferrites work well enough unless I put tank notches across them, which multiply the Q and stored energy. Then they seem to break up and sound harsh. So, overall, I think we have a case for using aircoils sometimes.
The maths seems to say that 10W wirewound/ceramic resistors are hugely lower inductance than your average tweeter, and don't get hot, so ought to work well enough.
MKP capacitors seem to be about as good as it gets even in cheapish 250V thin foil implementations at 0.1% loss. MKT polyesters are measurably worse at around 1% loss. NP electrolytics add a small sort of white noise or sibilance to female vocal that I find audible in 1:1 comparisons.
Barkhausen_effect seems like the "smoking gun" measurable problem with cored coils. FWIW, most component tolerances manifest themselves most at the -3dB or -6db point. It doubtless applies to speaker magnets too, hence perhaps the AlNiCo camp who swear those old magnets sounded better. This is in addition to non-linear hysteresis effects.
For all that, I find 1mH ferrites work well enough unless I put tank notches across them, which multiply the Q and stored energy. Then they seem to break up and sound harsh. So, overall, I think we have a case for using aircoils sometimes.
Last edited:
so show the measurements already
or show the equations, units of the coefficients, calculations - I can't find any and I have plenty of ferrite manu catalogs, app notes, inductor, transformer design guides
I see Hysteresis Loop loss distinguished from Eddy current loss - and a vague handwaving "excess loss" lumped in with the Eddy current term - for ferrite Eddy current loss is really minor at audio
or show the equations, units of the coefficients, calculations - I can't find any and I have plenty of ferrite manu catalogs, app notes, inductor, transformer design guides
I see Hysteresis Loop loss distinguished from Eddy current loss - and a vague handwaving "excess loss" lumped in with the Eddy current term - for ferrite Eddy current loss is really minor at audio
Last edited:
jcx, your attitude amazes me! Why so hostile to considering a measurable effect? It gets a lot harder to figure out with boutique capacitors, that stuff about capacitor squeal around 15kHz.
MIT Physics Demo - Barkhausen Effect | MIT Video
THAT is what a cored coil sounds like. I don't care whether there was hum pickup or whatever due to high impedance load.
Now take it or leave it. I certainly don't want to set up a science laboratory to investigate the effect. Though I'm not averse to simple experiments on coil interaction myself. I satisfied myself that coils (here 1mH ferrite core) DO interact with each other audibly and with nearby earth planes. But I might think a couple of bucks extra spent on an aircoil worthwhile. Which is what the original poster was asking.
MIT Physics Demo - Barkhausen Effect | MIT Video
THAT is what a cored coil sounds like. I don't care whether there was hum pickup or whatever due to high impedance load.
Now take it or leave it. I certainly don't want to set up a science laboratory to investigate the effect. Though I'm not averse to simple experiments on coil interaction myself. I satisfied myself that coils (here 1mH ferrite core) DO interact with each other audibly and with nearby earth planes. But I might think a couple of bucks extra spent on an aircoil worthwhile. Which is what the original poster was asking.
Attachments
Last edited:
- Status
- Not open for further replies.