Bonsai and Thorsten, please take it down a couple of notches. If you want to thrash things out do so off-line! Doing so in public paints neither of you in a flattering light. Any further comments on posts 13941 or 13960 will be deleted. 
Hi,
I don't remember anything about ferromagnetic cases or eddy current distortion in conductors adjacent to ferromagnetic materials or any of these topics, so don't ask.
I also don't remember anything about steel end caps and leads in resistors and steel leads in capacitors causing measurable levels of distortion.
Maybe John or Charles remember...
Meanwhile ferromagnetic cases are not used in the real High End Audio.
It is a requirement. You cannot do it any other way. You have to sign that you won't do it to be allowed to join the club and attend the secret meetings after the CES. If you do use ferromagnetic cases they revoke your licence. SO it must be really important.
Ciao T
PS, this post may be meant cynical or not. I cannot remember.
I don't remember anything about ferromagnetic cases or eddy current distortion in conductors adjacent to ferromagnetic materials or any of these topics, so don't ask.
I also don't remember anything about steel end caps and leads in resistors and steel leads in capacitors causing measurable levels of distortion.
Maybe John or Charles remember...
Meanwhile ferromagnetic cases are not used in the real High End Audio.
It is a requirement. You cannot do it any other way. You have to sign that you won't do it to be allowed to join the club and attend the secret meetings after the CES. If you do use ferromagnetic cases they revoke your licence. SO it must be really important.
Ciao T
PS, this post may be meant cynical or not. I cannot remember.
no, ferromagnetic cases can improve EMI resitance by having greater shielding effectiveness, particularly for low frequencies
one objection is that ferromagnetic materials are very nonlinear/hysteric so signal magnetic field coupling can cause distortion
residual magnetization could induce errors in signal loops from any relative motion
coaxial signal entry, low loop area wiring, air space, standoffs all make it possible for Scientific/Industrial Instrument manufacturer's to build 20 bit resolution systems in ferromagnetic boxes
another is that magnetostriction and forces from transformer's magnetic field can cause vibration - of course Eddy currents are also associated with forces in nonmagnetic conductive case materials too
compared to Al, Fe is very dense making a case of the same general dimensions, same panel stiffness or higher for the same mechanical resonance frequencies will be much heavier in Fe alloys
one objection is that ferromagnetic materials are very nonlinear/hysteric so signal magnetic field coupling can cause distortion
residual magnetization could induce errors in signal loops from any relative motion
coaxial signal entry, low loop area wiring, air space, standoffs all make it possible for Scientific/Industrial Instrument manufacturer's to build 20 bit resolution systems in ferromagnetic boxes
another is that magnetostriction and forces from transformer's magnetic field can cause vibration - of course Eddy currents are also associated with forces in nonmagnetic conductive case materials too
compared to Al, Fe is very dense making a case of the same general dimensions, same panel stiffness or higher for the same mechanical resonance frequencies will be much heavier in Fe alloys
Last edited:
Not a good day to ask a semi-retrospective question I presume Thorsten.
My curiosity was piqued when someone mentioned they would cause distortion, but may protect the components inside from the magnetic fields causing even more problems, than say a material that is almost transparent to the stray magnetic fields.
I will now erase these thoughts from my mind.
My curiosity was piqued when someone mentioned they would cause distortion, but may protect the components inside from the magnetic fields causing even more problems, than say a material that is almost transparent to the stray magnetic fields.
I will now erase these thoughts from my mind.If a steel case is bad, why is a steel chassis OK? Most circuits will have only small external magnetic fields so weak coupling to the case. The big exception is a valve power amp, with OPTs to generate significant leakage fields. Steel is often recommended for strength reasons.
I have a perverse theory. Good screening/filtering keeps out interference. Some people don't like this, because interference (if at a low level so not directly perceived as such) acts like dither so improves resolution or is misperceived as being 'air' (i.e. ambient noise) from the original recording. Getting rid of interference makes a unit sound 'dead'.
I have a perverse theory. Good screening/filtering keeps out interference. Some people don't like this, because interference (if at a low level so not directly perceived as such) acts like dither so improves resolution or is misperceived as being 'air' (i.e. ambient noise) from the original recording. Getting rid of interference makes a unit sound 'dead'.
I don't think that is perverse at all! and since the noise would more than likely be part of the noise floor it also lends itself to being ignored as far as measurements go other than that the noise floor is at X db
Perhaps the makeup of the noisefloor is something that needs to be considered 
Tony.
DF96, if you haven't used tungsten alloy, what you have there is a resonator. Attach a microphone to it, put it on a shake table along with two hundred golfballs, sweep it through the audio range at a +/-2cm displacement, then tell me it's a good chassis. That might be acceptable in mass market dog droppings, but there are those of us who appreciate the finer points of performance and cannot settle for metals with meager nuclei.
Don't be a "fool".
(while the response was indeed funny)I asked because I am building a superconducting magnet for a 1.25 megawatt electron and positron collider, the beam is 3 nanometers tall by 250 nanometers wide...The final quad is a 144 tesla/meter gradient, 3 meters long, cooled to 1.8 Kelvin to avoid helium flow turbulence.. If the magnet vibrates at a 4 to 5 nanometer level, two 3 nanometer beams will miss each other. The key word of course, being "collider".
We need to measure the vibration and location of the magnet, we need to provide active correction of the vibration. And it needs to be done by non contact measurement through a vacuum, from room temerature to 1.8 Kelvin, within a background field of 3 tesla (the detector solenoid field).
So it was not a glib question..
If you wish, you can PM me with the company name so I can persue this further.
Thank you
Cheers, John
Excellent responses.
Also, the magnetic permeability of the material changes as a result of the impinging magnetic field level and rate of change. A wall can end up with a frequency dependent permeability, and this can cause frequency product harmincs.
We also find that even really good non magnetic stainless will become magnetic at the surface if it is polished, scraped, in any way worked.
Cheers, John
I think the comment about scientific interments with 20bit resolution in ferromagnetic cases should guide our thinking for practical purposes here. In any event, getting -110db with an unscreened torroid is relatively easy. Do we really think the use of some ferromagnetic screening to take it down another 10-20db is going to screw the performance irretrievably? Paying attention to layout and loop areas will also make a significant difference in my view.
Hi,
Thank you, we do try and we are not all gold fuses and the like, we do know our basic engineering...
Now why would you doubt me? Surely my word is good enough, right?
Even if you do not take my word, as I mentioned it is a Chebychev (Type II to be precise) I really cannot see the reasons for you demurring...
Well, I designed and measured the filters and they are > 60dB/Octave for the first 30dB or so attenuation.
There are number of proponents of similar approaches, Neville Thiele has his Neville Thiele Methode, Jeff Joseph calls it infinite slope, Jason Cuadra had some nice articles. What I added to this is not a lot, but my adjusted methode does have advantages.
You can see a bit more on this page:
Abbingdon Music Research - Product - LS - 77 Features
Ciao T
Mr. Thorsten, I doubt everybody and every theory until proof comes along, so please don't take it personally. Thank you for pointing to the AMR site. As was already pointed out by DF96, achieving a 60 dB slope for some region does not equate to having a 10th order filter, but let's not go into semantics.
Looking at the LS77 and the filter slopes it becomes apparent which approach you have choosen. Putting an RLC notch behind a conventional filter stage may lead to impressive slopes at a low component count, but I consider it questionable loudspeaker engineering at best. This kind of Cauer filter causes the impedance of the loudspeaker system to vary widely as a function of F, with deep dips around xover. The reason being that these steep slopes are achieved by an almost short circuiting RLC at Fxover in parallel with the drivers.
The result is that a loudspeaker with these filters will only work properly for an amplifier with a pre-defined damping factor, an amplifier that furthermore is capable of driving very low ohmic, highly reactive loads.
regards,
Vac
Hi,
Well, if you are designing speakers like that, now that would be really stupid.
You can achieve notches not only by using parallel LC resonant circuits. Come on, this EE102, basic filter theory. Modern software allows us the ability to synthesise a wide range of filters, including monte carlo analysis against component tolerances, group delay and impulse response, including the drivers and we can nowadays measure acoustic systems quite well.
I really cannot see the point of using such questionoable engineering practices as those parallel LC shunts you seem to suggest as solution, so I don't.
The impedance of the LS-77 (excepting the LF tuning peaks) is actually quite flat and quite high. Minimum impedance is around 6 Ohm. Average impedance is close to 8 Ohm. While the speaker has a fairly low efficiency (around 87dB/W/m) and needs a lot of voltage to play loud, it is rather undemanding when it comes to current (basically a rather british design).
The real design challenges with such filters are to control the excess group delay and resultant impulse response distortions. Again, it is possible to synthesise filters that behave pretty close to complementary, but now we are into a discussion of speaker filter design for passive speakers that uses techniques and principles rarely even applied to active systems (but easily applicable to digital systems, where they in fact are found and inspired me).
If I was incapable of sensible design and had used the techniques you suggest, that would be indeed the case. You may however safely assume that I am a little "slicker than your average" (Craig David)...
Ciao T
Well, if you are designing speakers like that, now that would be really stupid.
You can achieve notches not only by using parallel LC resonant circuits. Come on, this EE102, basic filter theory. Modern software allows us the ability to synthesise a wide range of filters, including monte carlo analysis against component tolerances, group delay and impulse response, including the drivers and we can nowadays measure acoustic systems quite well.
I really cannot see the point of using such questionoable engineering practices as those parallel LC shunts you seem to suggest as solution, so I don't.
The impedance of the LS-77 (excepting the LF tuning peaks) is actually quite flat and quite high. Minimum impedance is around 6 Ohm. Average impedance is close to 8 Ohm. While the speaker has a fairly low efficiency (around 87dB/W/m) and needs a lot of voltage to play loud, it is rather undemanding when it comes to current (basically a rather british design).
The real design challenges with such filters are to control the excess group delay and resultant impulse response distortions. Again, it is possible to synthesise filters that behave pretty close to complementary, but now we are into a discussion of speaker filter design for passive speakers that uses techniques and principles rarely even applied to active systems (but easily applicable to digital systems, where they in fact are found and inspired me).
If I was incapable of sensible design and had used the techniques you suggest, that would be indeed the case. You may however safely assume that I am a little "slicker than your average" (Craig David)...
Ciao T
Vac,
The trick for a quasi 8th order LP does not require more than 2 coils, and 3 cap, plus a Zobel. The whole thing is amazingly simple. The network is almost identical to an electric fourth order network, but with an extra cap parallel to the second coil. Actually the impedance of this network is very well behaved.
How does one get there?
Load the woofers (in-box that is) .FRD and .ZMA files into LspCAD Pro, define starting values for the coils, caps en R's in the network, set your target curve, set a minimum impedance, define the optimzing frequency range and start the optimizer.
Before you have been able to grab a cup of coffee you wil be done.
All this fuss about higher order networks..
Eelco
The trick for a quasi 8th order LP does not require more than 2 coils, and 3 cap, plus a Zobel. The whole thing is amazingly simple. The network is almost identical to an electric fourth order network, but with an extra cap parallel to the second coil. Actually the impedance of this network is very well behaved.
How does one get there?
Load the woofers (in-box that is) .FRD and .ZMA files into LspCAD Pro, define starting values for the coils, caps en R's in the network, set your target curve, set a minimum impedance, define the optimzing frequency range and start the optimizer.
Before you have been able to grab a cup of coffee you wil be done.
All this fuss about higher order networks..
Eelco
Just as a point of interest, here is a distortion plot of a 1 meter length of RG-174 coaxial cable, which uses a copper-clad steel
The spike at 3kHz is the AP's residual distortion.
se
An externally hosted image should be here but it was not working when we last tested it.
center conductor.
An externally hosted image should be here but it was not working when we last tested it.
The spike at 3kHz is the AP's residual distortion.
se
- Status
- Not open for further replies.