Thanks for the input about the transformer interactions with casings.
Just to clarify - I claim I read was that an EI line level transformer sounded better without any screening at all.
i.e.
Present: the core, the bobbin & the coils, wooden mounting box.
Absent: coil covers, mu metal casing, any metalic casing.
Judging by the way a mains EI interacts with a steel box - even as it comes close even without touching - It's clear that there is considerable leakage flux.
Given that jn says a difference due to the proximity to metal casing can measured I would be surprised if it could not also be heard.
But I can't help thinking that instead of fiddling around with leaky EI transformers would be much more benefitial to switch to a C core device if we're chasing the best quality - if they can be found.
This whole topic, more generally, of how casings impact on final performance of audio electronics is very interesting for me.
Most manufacturers choose metal casing while some like denis moorecroft insist, after extensive listening that non resosnant plastic is the way to go.
I'm planning to do some testing of my own later this year to see whether screened with eddy currents sound better than unscreened but eddy free . . . or some kind of mix.
mike
Just to clarify - I claim I read was that an EI line level transformer sounded better without any screening at all.
i.e.
Present: the core, the bobbin & the coils, wooden mounting box.
Absent: coil covers, mu metal casing, any metalic casing.
Judging by the way a mains EI interacts with a steel box - even as it comes close even without touching - It's clear that there is considerable leakage flux.
Given that jn says a difference due to the proximity to metal casing can measured I would be surprised if it could not also be heard.
But I can't help thinking that instead of fiddling around with leaky EI transformers would be much more benefitial to switch to a C core device if we're chasing the best quality - if they can be found.
This whole topic, more generally, of how casings impact on final performance of audio electronics is very interesting for me.
Most manufacturers choose metal casing while some like denis moorecroft insist, after extensive listening that non resosnant plastic is the way to go.
I'm planning to do some testing of my own later this year to see whether screened with eddy currents sound better than unscreened but eddy free . . . or some kind of mix.
mike
mike
There is a specialty in industrial engineering dealing with Non Destructive Testing methods.
One of these methods utilises eddy currents to test components made of electrical conductive materials.
In simple terms, a frequency generator feeds a coil which generates a magnetic field.
When a metal is placed within this field, eddy currents are induced into it, which in turn generate a counter magnetic field. This affects the generating coil.
The method monitors the coil’s voltage and current which is displayed in a plain with axes of real and imaginary part of coil’s impedance.
This impedance is affected -independently and in different ways- by:
Excitation frequency
Amplitude of excitation signal
Coil construction
Relative position of coil to the test piece
Geometrical attributes of test piece
Electrical conductivity of test piece
Magnetic permeability of test piece
Homogeneity of test material.
The sensitivity and discrimination of retrievable information is amazing.
Now, as you can understand, all these bear a direct resemblance to the case of a transformer/shield/box interaction.
Monitoring primary/secondary/both coil impedance is feasible and results will be more informative than you may want.
The difficult part is the correlation of that data with observed acoustic differences.
George
Apart for using specialized expensive equipment which will require some modifications, one can take the following route to perform the technical part of the test:
Use a soundcard, build a simple interface and perform impedance measurements just like testing speaker impedance.
Export the data to a spreadsheet. There, work out the data scaling and presentation, as well as normalize the data sets in respect to one or many other data sets, so any parameter can be studied in selective isolation.
George
Use a soundcard, build a simple interface and perform impedance measurements just like testing speaker impedance.
Export the data to a spreadsheet. There, work out the data scaling and presentation, as well as normalize the data sets in respect to one or many other data sets, so any parameter can be studied in selective isolation.
George
But if channel A's power transistors are mounted on a heatsink at one end of an enclosure and channel B's power transistors are mounted let's say 16" away on their heatsink at the other end - on the face of it a reasonable spacing . . . can they not still talk to each other and create interference etc virtually unimpeded via eddy currents through the virtually zero impedance of the casing ?
Via eddy currents? Is it not the eddy currents in a chassis (or dedicated shielding) that actually provide the shielding when it comes to magnetic field interference.
To my understanding, shielding works two ways. Reflection loss and absorption loss.
Reflection loss occurs when the impedance of the material is significantly different from that of the interfering wave's impedance and therefore the interference is reflected off the shielding.
Absorption loss works by way of eddy currents which convert the interfering energy into heat.
se
Please correct me if I'm mistaken . . .
change in current in transistor A attempts to induce an equal and opposite current in any adjacent conductor - in this case the adjacent conductor is the heat sink & casing which offers effectively no resistance, current flowing around heat sink and casings looks for and adjacent conductor and finds transistor B . . . .
change in current in transistor A attempts to induce an equal and opposite current in any adjacent conductor - in this case the adjacent conductor is the heat sink & casing which offers effectively no resistance, current flowing around heat sink and casings looks for and adjacent conductor and finds transistor B . . . .
One effect of eddy currents is to try to reduce the field stimulus which causes them (Lenz's Law IIRC). Less external field mean less coupling.
To get magnetic coupling via a screen you can cut a slot in it, thus reducing the local eddy currents (and so not reducing the fields) but forcing what currents are induced to travel up the other end of the slot and induce interference there. Slots cut in conductors can thus both prevent coupling and create coupling - details matter. Lack of slots probably means lack of coupling. To get a definitive answer would require solving the field equations, perhaps by using an EM simulator.
To get magnetic coupling via a screen you can cut a slot in it, thus reducing the local eddy currents (and so not reducing the fields) but forcing what currents are induced to travel up the other end of the slot and induce interference there. Slots cut in conductors can thus both prevent coupling and create coupling - details matter. Lack of slots probably means lack of coupling. To get a definitive answer would require solving the field equations, perhaps by using an EM simulator.
The vendor was incorrect. When we specify the iron qualites for magnet laminations, one spec is coercivity of the iron.. After being pushed at or near a tesla, the remnant field will be about two orders of magnitude lower. Running a power core over a tesla at 60 hz is begging for heat problems. But setting the core such that it'll blow fuses as they stated...no.I sincerely doubt that.
I concur with you. I wonder if the vendor used a zero crossing switch during tests.
I only presented what occurs magnetically. If measurement I mentioned found nothing, audibility will be zero as well.
Well said. I use an HP4284A precision LCR meter. It automatically sorts out the real and imaginary parts, and presents it as R and L. What it cannot do is sort out the coil resistance from the eddy current dissipation. This is because the IR drop of the coil and the eddy equivalent resistance are both 90 degrees from the inductance.
The ITER project has the exact same problem. They need to be able to sort out the inductive and resistive component of the main solenoid. Since it's 40 feet tall, about 20 feet across, and runs at 45kA, if the coil itself goes even slightly resistive, they need to detect it and shut it down. Otherwise, the stored energy will be released at the bad spot. 30 gigajoules with no place to go is in general, not a very good thing. Their worst nightmare is during ramp and plasma initiation, because the entire stainless structure will eddy, and when the plasma initiates, it will pull heavily from the field energy.. They have been told how to do it right, but what they will end up doing is unknown.
The difficult part is a current viewing resistor which has very low inductance and current rate of change pickup.
Actually, I would be more concerned with the star grounding configuration. The input reference and moreso the feedback divider ground reference cannot loop trap any magnetic fields internal to the chassis. If the supply rails are not routed to minimize inductance and therefore flux generation, they may couple to the reference wires. Also, keep the input pair E-B loops away from stray fields..it's full gain there.
Accurate.
jn
Last edited:
Distance
So a split power supply on two chassis with the transformers and diodes and some filtration on one and the final filters ETC on the circuit chassis would work if properly done. What would be the major areas of causes of problem doing it this way. Cost not a major one being it is DIY .
But of course that series R can be measured independently. Odd that the 4284A doesn't have that feature (?).
The issue is much more complex than that.
When I desire to know the Ls and Rs of an entity which has magnetic storage, the meter will present to me how much energy the entity stores as a result of current (L), and it will present to me the voltage that the device presents at it's terminals as a result of the current (subtracting of course the voltage portion of the storage mechanism), expressed as a ratio (V/I).
When there are no external loss mechanisms, the meter will report the resistance of the wire. When there are loss mechanisms, the meter will report those as a part of the first loss mechanism, wire resistance.
If the wire proximities as a result of current path modulation caused by the rate of change of energy storage, that additional loss will be added to the first mechanism of IR. My meter of course, integrates that proximity loss to a simple number even though it is actually a second harmonic term of Rs.
Externally generated losses are another component which the meter interprets as series resistance.
When the meter is in LsRs mode, it lumps:
The inductance of the primary field plus the inductance of fields generated by external objects (opposite sign).
The resistance of the wire comprising the coil, plus the added resistance caused by proximity effect (integrated of course), plus the added resistance caused by dissipation of external objects which alter the field.
jn
It can be easily shown that a typical audio chassis will not prevent hum pickup from external power supplies IF they are not extensively shielded or separated from the audio circuitry. This is especially true in hi gain circuits like phono stages.
The biggest 'downside' to dual cases, one for the audio circuits and one for the power supply is the extra cost of two cases, and some way of connecting them together.
The biggest 'downside' to dual cases, one for the audio circuits and one for the power supply is the extra cost of two cases, and some way of connecting them together.
Interesting. But are you in essence saying that the d.c.r. is a measurement of no value? I'm presuming that the meter is given plenty of time to settle.
No. DCR may be of value, but not at any frequency. DCR does not present any information regarding what the coil or magnet or transformer will do with time varying signal.
It is the value a coil will approach asymptotically as frequency goes down. But it doesn't say what the system will do as frequency goes up, it doesn't say what the external eddy losses will be.
As a measure of the influence an external magnetic or conductive box would have on a component, DCR is useless.
jn
- Status
- Not open for further replies.