I read a mention that "Audio transformers ideally should have close to 1:1 core aspect ratio because then the magnetic flux density is evenly distributed across the AF band"
But I can't find a single article on this, nor any book.
Does anyone know of any articles on this or books that explain this?
But I can't find a single article on this, nor any book.
Does anyone know of any articles on this or books that explain this?
You can't find it because it's not true.
The distribution of magnetic flux density does not depend on the turn ratio otherwise audio output transformers would be all bad...... If your target is to use the core in a uniform fashion, the quantity you want to look at is the magnetic magnetomotive force (m.m.f.) which is of course closely related to the magnetic flux density. The reason is that the distribution of m.m.f. is direct consequence of winding geometry and volume usage. In other words, it's consequence of partition between primary and secondary and coupling.
The distribution of magnetic flux density does not depend on the turn ratio otherwise audio output transformers would be all bad...... If your target is to use the core in a uniform fashion, the quantity you want to look at is the magnetic magnetomotive force (m.m.f.) which is of course closely related to the magnetic flux density. The reason is that the distribution of m.m.f. is direct consequence of winding geometry and volume usage. In other words, it's consequence of partition between primary and secondary and coupling.
Last edited:
Lots of stuff written all over the Jet.
Some make sense, some not that much.
Fwiw I wind transformers commercially and if at all possible, try to use "square" cores, a.k.a. 1:1 ratio, mainly because it's more efficient.
We are talking geometry/shape here, not turns ratio.
It gives you the shortest average turn length, so it lowers d.c.r, minimizes copper weight, etc.
Makes no sense to me.
Not surprising.
Best case, it was some kind of statement taken out of context.
Ah ok, in that case it gets worse in the high frequency domain because the leakage flux increases. The core length does not change because the lamination is still the same! The turn length changes, unless we are talking of C cores of different sizes. However, the increase in leakage inductance is modest, unless there is a big disproportion between the 2 sides. If the ratio between the sides is 1.5:1 or less, then one can expect an increase in leakage inductance in the 5-10% for the worst case and depending on the geometry.
What I said before still stands and it's not neglectable. A transformer, say, with 40x50 mm core with the ideal winding geometry can be much better than one with 40x40 mm with a less ideal one.
Well then why did old transformers use a very low aspect ratio instead of a square core? In the 25Hz days transformers were very long.
It only makes sense if they somehow "knew" that 60Hz would be the standard in the future, so they made a standard lamination size which would result in a more of less square core for 60Hz, even tho they were using 25Hz.
That makes less sense to me than the aspect ratio influencing frequency response.
Also, the windings would also influence the frequency response. Because if they're too thick, they will have high eddy currents at higher frequencies.
That's the reason almost no one uses foil windings, because above ~6khz the eddy currents skyrocket.
It only makes sense if they somehow "knew" that 60Hz would be the standard in the future, so they made a standard lamination size which would result in a more of less square core for 60Hz, even tho they were using 25Hz.
That makes less sense to me than the aspect ratio influencing frequency response.
Also, the windings would also influence the frequency response. Because if they're too thick, they will have high eddy currents at higher frequencies.
That's the reason almost no one uses foil windings, because above ~6khz the eddy currents skyrocket.
In the 50ies a frequency response 1 dB down at 15 KHz was considered good. Today is pretty poor. Different standards.
Low frequency has nothing to do with core cross-section shape. 60Hz is just the line frequency and often it is used as a standard for output transformers mainly because it makes them look bigger (more power handling). It's a trick that often works if someone does not know that power rating should be specified at the lowest frequency of interest. Back in the old days max power at 60Hz was good spec. Again, today it is not.
Square cross-section or not only the effective area matters at low frequency. If you have larger area with the same magnetic path length you will get more inductance (for the same number of turns, of course) but because turns will be longer it will be less efficient (more copper loss). For inductance it's the ratio between effective core area and magnetic path length that matters. If you increase the area and the length at the same time, for example using a different lamination, you might end up with the same ratio and the only advantage would be more space for the windings (can put more turns or use larger wire for better efficiency). A good transformer design always results in a good balance of all things that might have opposite requirements.
Also a larger area with the same laminations means cost saving because one only needs to buy 1 type of laminations for different transformers and applications. That is the main advantage of EI laminations over C cores and toroidals.
In terms of copper there might also be some saving because a smaller wire will be used (on a smaller lamination with non-square cross-section) and the transformer's weight will be less. Shipping cost saving for larger quantities, too. The copper is sold by weight too in most cases.
Low frequency has nothing to do with core cross-section shape. 60Hz is just the line frequency and often it is used as a standard for output transformers mainly because it makes them look bigger (more power handling). It's a trick that often works if someone does not know that power rating should be specified at the lowest frequency of interest. Back in the old days max power at 60Hz was good spec. Again, today it is not.
Square cross-section or not only the effective area matters at low frequency. If you have larger area with the same magnetic path length you will get more inductance (for the same number of turns, of course) but because turns will be longer it will be less efficient (more copper loss). For inductance it's the ratio between effective core area and magnetic path length that matters. If you increase the area and the length at the same time, for example using a different lamination, you might end up with the same ratio and the only advantage would be more space for the windings (can put more turns or use larger wire for better efficiency). A good transformer design always results in a good balance of all things that might have opposite requirements.
Also a larger area with the same laminations means cost saving because one only needs to buy 1 type of laminations for different transformers and applications. That is the main advantage of EI laminations over C cores and toroidals.
In terms of copper there might also be some saving because a smaller wire will be used (on a smaller lamination with non-square cross-section) and the transformer's weight will be less. Shipping cost saving for larger quantities, too. The copper is sold by weight too in most cases.
Last edited:
In my experience, a ratio up to 1.6:1 can be used for relatively small transformers without troubles. For example, the EI84 (European metric) with core area of 28x45 can be very good. Going up in size the max ratio should be reduced a little bit to 1.4-1.5, because increasing the lamination size winding surfaces increase and so unwanted capacitances. As said, the core ratio will have an influence on leakage inductance but if both leakage inductance and stray capacitances increase the transformer resonance will shift to lower frequency faster. That is not desirable. You might compensate the larger winding area of the larger core using more insulation to keep capacitances in check but there is a limit. Balance, that's the keyword. Always.
So you're saying there's absolutely 0 reason that intelligent engineers in the 40s decided to use disadvantageous core geometry?
And then when 60Hz became the standard, they for some reason all of a sudden decide to use different core geometry.
And then when 60Hz became the standard, they for some reason all of a sudden decide to use different core geometry.
I did not say that was no reason. There is at least one reason (commercial) for which a disadvantageous core shape can be a great advangate (more profit).
60Hz has never been the standard for output transformers and it's not to be confused with frequency response. Power response involves distortion too.
As I said, some manufacturers use it as reference for THEIR CATALOGUE because it makes their transformers look better/bigger and they pick the core on the same basis of PSU transformers without doing assessments or calculations.
I suggest you read the entire article about the Olson 6F6 DYI amplifier to undestand the requirements from a good amplifier in the 50ies (like frenquecy response 50-15000 Hz @-1dB etc...) and compare it to modern standards.
Finally, I do not see all these disadvantagous core gemetries used in vintage stuff.
60Hz has never been the standard for output transformers and it's not to be confused with frequency response. Power response involves distortion too.
As I said, some manufacturers use it as reference for THEIR CATALOGUE because it makes their transformers look better/bigger and they pick the core on the same basis of PSU transformers without doing assessments or calculations.
I suggest you read the entire article about the Olson 6F6 DYI amplifier to undestand the requirements from a good amplifier in the 50ies (like frenquecy response 50-15000 Hz @-1dB etc...) and compare it to modern standards.
Finally, I do not see all these disadvantagous core gemetries used in vintage stuff.
Last edited: