Maybe I am reading between the lines too much, but you seem a little reluctant to assign a temperature to the core loss term in the measured winding resistance. Anyway, I am happy to agree that it will be about room temperature. So we have a resistance at a non-zero temperature. Will this generate thermal noise? If not, we seem to have a perfect noiseless resistor, yet at room temperature. If it does generate thermal noise, what is the ultimate source of that noise? The resistor does not exist as a physical item, unlike true ohmic winding resistance, so where does the noise come from?
Talk of current constriction is a separate issue, such as skin/proximity effect.
Yes.
Yes. They effectively cancel which I believe accounts for the effective reduction in permeability as frequency increases.
Ok.
So how does that end up as additional noise on the transformer's secondary?
I mean, the bucking flux is flux that's canceling some portion of the flux produced by the primary which means there's that much less flux coupling to the secondary.
So if we cancel the flux due to signal, we're ultimately left with the random motion of electrons in the core.
How would that be any different than if there were no signal and no signal-induced eddy currents?
se
Temperature? Do you mean an "equivalent" temperature? When you drive hf into the coil, the Rs is just higher. My first inclination would be to simply use the higher resistance in the model, with the same temperature. I would not be inclined to say the coil temp is increased, simply because it hasn't. Are you thinking about using a frequency dependent temperature instead of a frequency dependent resistance? Because a music signal has multiple frequencies, but the higher ones produce the dissipation.
A correct model has to use frequency dependent losses. Trying to use an equivalent temperature may be possible, but that still relies on frequency dependence..
The resistance does indeed exist as a physical entity, it is dissipation in the core as a result of eddy loss.
The noise of the eddy currents would certainly be dependent on the core temperature.
But without excitation from the coil, the core will not produce a magnetic field that can couple to the coil. It's a one way process.
My meter cannot make the distinction...but intellectually we agree there is a distinction..
Yup.
So how does that end up as additional noise on the transformer's secondary?
I mean, the bucking flux is flux that's canceling some portion of the flux produced by the primary which means there's that much less flux coupling to the secondary.[/quote]
Yes. And if the bucking flux is noisy, that will couple to the secondary.
Normal materials do not exclude flux 100%. It's not Meissner. So there is never full cancellation/exclusion.
Cheers, John
Last edited:
No Yawei, mine is older and was made in Japan. This machine was apparently built to order for someone in South America according to the "build" tag. It instead went on a tour of the machinery shows in North America to introduce the brand here. It used to be sold as the "Strippit" brand entry level toy. There are a number of variations all with the same or similar part numbers. Mine is hydraulic and has liners for the punch holders instead of just a turret. Takes Amada size tooling. I got a great deal on it when they decided to sell it.
Leather holes maybe, nibbling might be more of a problem. Thickness versus die clearance?
Most of the leather punching it seems uses a rubber back up for a sharp cutting hollow punch.
Going through the leather should be trivial, but jamming something is always really nasty.
I think you may be confusing a change in temperature with a change in resistance caused by frequency (or maybe you think I am confusing the two). I am simply trying to establish that every resistance, whether directly physical (e.g. a resistor) or reflected (e.g. via a transformer from a secondary or a core loss) or 'fictitious' (e.g. radiation resistance in an antenna) has an associated temperature. I am not trying to model a resistance increase by changing the temperature.
A resistance with a non-zero temperature will generate thermal noise, which you can always calculate by the normal Johnson/Nyquist formula. But where does the noise actually come from? In the case of a resistor it comes directly from charge in the resistor, so the temperature is the resistor temperature. In the case of an antenna the effective temperature is not the antenna physical temperature, but the mean temperature of the surroundings (and perhaps interferers too). What about core losses? Where does the noise come from? I assert that it is random fluctuations in the core. You deny this, so where does it come from? Are you saying no noise, or just inexplicable noise?
A resistance with a non-zero temperature will generate thermal noise, which you can always calculate by the normal Johnson/Nyquist formula. But where does the noise actually come from? In the case of a resistor it comes directly from charge in the resistor, so the temperature is the resistor temperature. In the case of an antenna the effective temperature is not the antenna physical temperature, but the mean temperature of the surroundings (and perhaps interferers too). What about core losses? Where does the noise come from? I assert that it is random fluctuations in the core. You deny this, so where does it come from? Are you saying no noise, or just inexplicable noise?
What does that mean?
We weren't talking about a diamagnetic material excluding flux. We were talking about the flux produced by eddy currents. That flux is what it is and because of its orientation, will cancel some small portion of the flux produced by the current flowing in the primary windings.
se
I know. We were talking about normal materials which will exclude part of the flux...superconductors will exclude all the flux because the eddy currents are not limited by resistivity. Sorry, I sometimes forget that others may not have superconductivy experience..especially you, you always seem to be on the game..
Yep...and if the cancellation flux is noisy, the net flux through either the primary or the secondary will have that noise component.
Cheers, John
Yes, but I simply didn't see what relevance it had to what we were discussing which as far as I'm aware were normal materials.
Mmmm. Ok. I think I'm on track now. If the flux from the eddy current that's taking a bite out of the primary flux is noisy, then the remainder of the primary flux will be noisy.
Got it.
Now if only someone could quantify it in a realworld transformer.
se
A step down on a practical eddy current issue
I would like to know how manufacturers of MC & Mic transformers deal will the problem of sharp edges of the stampings when they are using very thin laminations.
I have noticed with some small power transformers (0.5mm E-I thick laminations), that these sharp edges effectively short the adjacent electrically isolated laminations when measuring them with DC Ohm meter.
True, eddy currents do not behave like DC Current.
They (eddies) circulate locally, (that’s why we can not effectively “earth” eddy currents), but when laminations are of small length & width and of minute thickness, these edge shorts should act as shorts for quite an appreciable percentage of eddy currents too.
Regards
George
I would like to know how manufacturers of MC & Mic transformers deal will the problem of sharp edges of the stampings when they are using very thin laminations.
I have noticed with some small power transformers (0.5mm E-I thick laminations), that these sharp edges effectively short the adjacent electrically isolated laminations when measuring them with DC Ohm meter.
True, eddy currents do not behave like DC Current.
They (eddies) circulate locally, (that’s why we can not effectively “earth” eddy currents), but when laminations are of small length & width and of minute thickness, these edge shorts should act as shorts for quite an appreciable percentage of eddy currents too.
Regards
George
To quantify: measure resistance at your chosen frequency. Calculate Johnson noise for that resistance at room temperature. This sets a minimum. There may be extra sources of noise too. Noise measurements would show this. Has anyone actually done this?
Alternatively, measure power loss at your chosen frequency. Treat the transformer as a resistive attenuator with that loss, so it has that noise figure.
Alternatively, measure power loss at your chosen frequency. Treat the transformer as a resistive attenuator with that loss, so it has that noise figure.
Now we're getting somewhere, that paper says the opposite (my interpretation and if I'm wrong that's fine). In equilibrium in a radiative field a metalic conductor emits and absorbes flux, electrical and magnetic. They even treat superconductors at optical frequencies. As I said last week a resistive surface will induce noise in a plate held over it. This I have measured.
A step down on a practical eddy current issue
I would like to know how manufacturers of MC & Mic transformers deal will the problem of sharp edges of the stampings when they are using very thin laminations.
I have noticed with some small power transformers (0.5mm E-I thick laminations), that these sharp edges effectively short the adjacent electrically isolated laminations when measuring them with DC Ohm meter.
True, eddy currents do not behave like DC Current.
They (eddies) circulate locally, (that’s why we can not effectively “earth” eddy currents), but when laminations are of small length & width and of minute thickness, these edge shorts should act as shorts for quite an appreciable percentage of eddy currents too.
Regards
George
Good practice is to remove the sharp edges with a bath of vibrating abrasive material. It also should be dulled when the sides are "pickled" in an acid bath to reduce the surface conductivity to reduce the eddy currents.
In a proper stamping the punch actually only goes 1/2 way or less through the material before it pushes out. If you look at any punched metal piece on the side (magnification may be required) you should see part of the edge is smooth followed by roughness. The smooth is where the punch pushed. The rough is the breakaway!
When you have sharp edges remaining it can mean the die is worn and has too much clearance. If it is sheared too much of a gap between the knife and table will leave a burr.
Basically an edge means worn tooling or just poor quality.
Also the varnish used is not as an insulator but to reduce movement and mechanical noise.
Simon, thank you for your informative answer. Some more answers if you please?
This means that the metal strip has to be treated for electrical isolation application after stamping and abrasive bathing. I was under the impression that laminates were purchased with the insulation already applied.
Is this acid pickling the electrical isolation application I wrote above?
If yes, do you know what acid is used for silicon iron laminates?
Yes, I have looked with a X10 loop.
I thought so too.
When it comes to very small thicknesses, how easy it is to see these defects?
Do the small signal transformer manufacturers test for such electrical shorts between laminations? (Large transformers can be checked by monitoring the excitation current not exceeding a certain upper limit)
OK. But it comes handy as such (isolator) sometimes, No?
Regards
George
This means that the metal strip has to be treated for electrical isolation application after stamping and abrasive bathing. I was under the impression that laminates were purchased with the insulation already applied.
Is this acid pickling the electrical isolation application I wrote above?
If yes, do you know what acid is used for silicon iron laminates?
Yes, I have looked with a X10 loop.
I thought so too.
When it comes to very small thicknesses, how easy it is to see these defects?
Do the small signal transformer manufacturers test for such electrical shorts between laminations? (Large transformers can be checked by monitoring the excitation current not exceeding a certain upper limit)
OK. But it comes handy as such (isolator) sometimes, No?
Regards
George
Please, everyone. The most important thing is that laminations are important, and thin laminations improve the efficiency and lower the noise of the transformer like device that is built with them. I am fairly sure that most here have now come to realize this.
Now, beyond laminated cores, there are amorphous cores that have even better performance at mid to high frequencies. However, these cores tend to create 10 times or more low frequency distortion, so there is a trade-off. Which core to use and where? This is where we stand at the moment.
Now, beyond laminated cores, there are amorphous cores that have even better performance at mid to high frequencies. However, these cores tend to create 10 times or more low frequency distortion, so there is a trade-off. Which core to use and where? This is where we stand at the moment.
How on earth can you possibly say it's the most important thing when you haven't even quantified it!?
As children, the most important thing was to not leave our arms or legs hanging off the bed at night because we knew there was a monster under there that would grab any carelessly dangled limbs while we were sleeping and drag us under the bed and eat us alive.
For all you or anyone else here knows, the eddy current noise you so fear may be nothing more than a phantom under your bed.
I mean, if it's swamped by the noise due to the transformer's winding resistances (not to mention the various cable and contact resistances between the cartridge pins and the preamp's input), then what's the point of worrying about it, much less saying it's the most important thing?
Is it really nothing more than a numbers game for you, John? Just another 60's and 70's spec wars redux?
se
I will discuss something that many of you will have a hard time believing, but it has been measured and was published almost 50 years ago.
This is the measurement of a professional Ampex reproduce tape head of .5H.
Now, how does this compare to a standard transformer? First, it is composed of some sort of mumetal, it has 6 mil laminations, and it has a dc resistance of 100 ohms.
Now, this tape head was measured (as I did approximately 6 years later) by an engineer named Erling P. Skov. If you wish to see his patents and other papers, just 'Google' his name and follow up.
Now what is important is that this 'INDUCTOR' that is a normally used as a tape head, has a changing AC resistance that departs from the AC 'imaginary' impedance 2piFL that continually rises at 6dB/octave.
OK, effective series resistance starts at 100 ohms (the dc resistance), at 1KHz it is 350 ohms, at 10KHz it is 10,000 ohms and rising. This is called 'the real part of the impedance' and it virtually overwhelms the DC resistance above 500Hz.
This is shown on fig. 13, 'Noise Limitations in Tape Reproducers' JAES Oct. 1964
Now what does this mean? It means that finite laminations, even as thin as 6 mils, give eddy current losses that appear as a noisy resistor that can completely dominate the situation, and this can most probably be measured as high frequency noise in transformers, just like it is measured in tape recorder heads. This is one of the reasons that many transformer manufacturers have resorted to Amorphous cores, in order to keep the eddy current losses to a minimum.
This is the measurement of a professional Ampex reproduce tape head of .5H.
Now, how does this compare to a standard transformer? First, it is composed of some sort of mumetal, it has 6 mil laminations, and it has a dc resistance of 100 ohms.
Now, this tape head was measured (as I did approximately 6 years later) by an engineer named Erling P. Skov. If you wish to see his patents and other papers, just 'Google' his name and follow up.
Now what is important is that this 'INDUCTOR' that is a normally used as a tape head, has a changing AC resistance that departs from the AC 'imaginary' impedance 2piFL that continually rises at 6dB/octave.
OK, effective series resistance starts at 100 ohms (the dc resistance), at 1KHz it is 350 ohms, at 10KHz it is 10,000 ohms and rising. This is called 'the real part of the impedance' and it virtually overwhelms the DC resistance above 500Hz.
This is shown on fig. 13, 'Noise Limitations in Tape Reproducers' JAES Oct. 1964
Now what does this mean? It means that finite laminations, even as thin as 6 mils, give eddy current losses that appear as a noisy resistor that can completely dominate the situation, and this can most probably be measured as high frequency noise in transformers, just like it is measured in tape recorder heads. This is one of the reasons that many transformer manufacturers have resorted to Amorphous cores, in order to keep the eddy current losses to a minimum.
A small part of this resistance rise with frequency may come from skin/proximity effect. This is a rise in actual winding resistance. The noise temperature of this is the wire temperature.
Some will come from eddy currents. In series with the winding resistance there is an inductor. This is not a perfect inductor, so it adds further resistance. The noise temperature of this is the core temperature.
Finally, stray capacitance will make the whole thing look like a parallel tuned circuit so the resistance will rise as we approach resonance. This is just ordinary classical physics, but it means that you can't just measure DC winding resistance and expect to use just that in a thermal noise calculation.
Some will come from eddy currents. In series with the winding resistance there is an inductor. This is not a perfect inductor, so it adds further resistance. The noise temperature of this is the core temperature.
Finally, stray capacitance will make the whole thing look like a parallel tuned circuit so the resistance will rise as we approach resonance. This is just ordinary classical physics, but it means that you can't just measure DC winding resistance and expect to use just that in a thermal noise calculation.
Reminds me, did i spot a $2 LT1012 in the Indigo phono stage of the Frenchy branch of the Qualia animation biz ?
Once the punching force exceeds the shear strength, it will rip, no matter how far the punch went through.
The rip zone of any size ring can be seen without glasses by holding it against the sunlight.
There will always be an edge, how much also depends on tooling for non-turret punching.
In youth slavery days, i made 50cts an hour every 3 months, hand-sorting 1/4M stamped plates before they went through the bending die/mold, sharp edge side had to go on the inside.
(similar deal with 1/2 a Mil nylon sheet reinforcement bands for men's trousers, the sharp edge side turned hands into scratch battlefield within a couple of days. Wearing protective gloves dropped sorting speed by +50%. Ah, the things you do for love.
)
Last edited:
How is this possible? Eddy current formations themselves are independent of frequency and although eddy current losses are proportional to the square of the frequency due to inductive heating of the core, I don't see how the "bucking" current could cancel primary induction current to the degree we see losses of permeability at high frequencies. I believe the decrease in permeability is due more to domain switching issues.
John
I believe Scott's sideband test is the better and most sensitive method.
I suspect not. JC provided no actual mechanism to warrant a test, just observational suspicions and a bit of hand waving..Most sci based individuals tend to just ignore his pronouncements because no realistic cause was mentioned (nor obvious). In this particular case, ignoring his assertions may be disingenous. (sp)
Look how long it's taken for me to convince those involved here of the physics nature of eddy losses and the possibility of resulting noise. And this, three or four days after I provided the physics model, the expected results, and a methodology for testing the hypothesis.. sheesh.
Agreed. I've no idea how to derive overall noise given a frequency dependent component, I would leave that to those who know far more than I with respect to noise. Perhaps you and Scott can work that out..
I think the key point is in equilibrium in a radiative field. I haven't had a chance to read that paper, so haven't seen what they consider the radiative mechanism. Would you mind re-linking? thanks, this thread grows so fast.
I'd love details, as I don't know if it would be inductive, capacitive, radiative, or even casimir.
For small line level audio transformers, I suspect almost no contribution of skin/proximity even out to 100khz, so I'd reckon it's the wire temp and DC resistive value.
This is where I believe your model (what I believe is your model anyway) is inconsistent. As frequency increases, eddy dissipation rises while core temp does not. If the series resistance increases two or three orders of magnitude because of the core, the core temp is not increasing that way...
Unless of course, when you say "core temperature", you are speaking about the effective noise temperature of the core...the noise produced if the core were actually at that temp..
As I said, I leave the noise calcs and measurement to you technical types...me, I'm just a country doctor...
Agreed..
Cheers, John
Last edited:
- Status
- Not open for further replies.