CC vs EI … I mean torroids are out of question
Related threads
http://www.diyaudio.com/forums/showthread.php?s=&threadid=45348&highlight=
http://www.diyaudio.com/forums/showthread.php?s=&threadid=51622&highlight=
On a number of occasions and threads references have been made to CC and EI cores for output trans as well as power trans. But I could not find any thread focussed on this matter …
I have a little experience so I can only report the ideas on CC I exchanged a while ago with a guy from Bartolucci, actually he is just selling these CC irons, so I just start the discussion to stimulate the sharing of your views/experiences.
CC Pros. The core material can be cut following the grains orientation so, once wound, the magnetic flux in the trans can travel along this preferred direction causing smaller losses and histeresis. This cannot happen in EI cores as the vertical leg of the E will present grains perpendicular to the flux.
The phenomenon allows CC cores to run at higher flux densities, that is power/voltage handling, than EI cores or, if you prefer, to have more headroom at the same power level and so working in a more linear way.
I believe many of you will have more to add.
Gianluca
Related threads
http://www.diyaudio.com/forums/showthread.php?s=&threadid=45348&highlight=
http://www.diyaudio.com/forums/showthread.php?s=&threadid=51622&highlight=
On a number of occasions and threads references have been made to CC and EI cores for output trans as well as power trans. But I could not find any thread focussed on this matter …
I have a little experience so I can only report the ideas on CC I exchanged a while ago with a guy from Bartolucci, actually he is just selling these CC irons, so I just start the discussion to stimulate the sharing of your views/experiences.
CC Pros. The core material can be cut following the grains orientation so, once wound, the magnetic flux in the trans can travel along this preferred direction causing smaller losses and histeresis. This cannot happen in EI cores as the vertical leg of the E will present grains perpendicular to the flux.
The phenomenon allows CC cores to run at higher flux densities, that is power/voltage handling, than EI cores or, if you prefer, to have more headroom at the same power level and so working in a more linear way.
I believe many of you will have more to add.
Gianluca
There are two schools of thought for where you set the peak flux density...
One is to set this according to the the max "core loss" you can tolerate... and like you mention the CC core has less losses...so you can drive the flux up a bit more...
This is usually the case when designing POWER transformers...
A Power transformers runs at a CONSTANT flux density,.ie...this is constant voltage @ constant frequency, since this is not load dependent....so you have a constant core loss...but variable copper loss dependent on load... SO your not interested in core distortion as much as you are core losses...
The other school of though is to design by seting the maximum "FLUX DENSITY" you are willing to tolerate based on core distortion, which is usually the case with AUDIO transformers....with this in mind...then CC-core or EI core are the same in this respect.... Assuming using the same core material, the CC-core has the advantage of having a higher "effective" permeablity over the EI core, thus giving a slight advantage in inductance..... BUT... I prefer the "sound" of EI cores because of the incremental INDUCTANCE curve is totally different.......
The CC-core incremental inductance curve is a gradual build up and PEAKS around full power output ...While the EI lamination curve will have a PEAK inductance at roughly half power output...
This means that the CC-core will have a thinner sound until you crank the amp up....while the EI core will have the "thump" going on at mid volume level..before the Flether-Munsson effect takes place.. SInce most Hi-Fi listeners don't crank thier gear full volume...the EI core give the most inductance at normal operating levels...
In many Push-Pull applications when designing by flux density..you end up with way more inductance then you need..so EI or CC-core would give roughly the same percievable low end performance...in that case I would use the CC-core....
BUT in SE amps where inductance is at a premium..the EI lamination is my choice for the reasons described above....
One of my favorite push-pull CC-core output transformers is the MacIntosh unity coupled outputs... This design always fasinated me because it was radically different... I use to wind these bifilar outputs years ago when I had more time...very interesting stuff..
Chris
One is to set this according to the the max "core loss" you can tolerate... and like you mention the CC core has less losses...so you can drive the flux up a bit more...
This is usually the case when designing POWER transformers...
A Power transformers runs at a CONSTANT flux density,.ie...this is constant voltage @ constant frequency, since this is not load dependent....so you have a constant core loss...but variable copper loss dependent on load... SO your not interested in core distortion as much as you are core losses...
The other school of though is to design by seting the maximum "FLUX DENSITY" you are willing to tolerate based on core distortion, which is usually the case with AUDIO transformers....with this in mind...then CC-core or EI core are the same in this respect.... Assuming using the same core material, the CC-core has the advantage of having a higher "effective" permeablity over the EI core, thus giving a slight advantage in inductance..... BUT... I prefer the "sound" of EI cores because of the incremental INDUCTANCE curve is totally different.......
The CC-core incremental inductance curve is a gradual build up and PEAKS around full power output ...While the EI lamination curve will have a PEAK inductance at roughly half power output...
This means that the CC-core will have a thinner sound until you crank the amp up....while the EI core will have the "thump" going on at mid volume level..before the Flether-Munsson effect takes place.. SInce most Hi-Fi listeners don't crank thier gear full volume...the EI core give the most inductance at normal operating levels...
In many Push-Pull applications when designing by flux density..you end up with way more inductance then you need..so EI or CC-core would give roughly the same percievable low end performance...in that case I would use the CC-core....
BUT in SE amps where inductance is at a premium..the EI lamination is my choice for the reasons described above....
One of my favorite push-pull CC-core output transformers is the MacIntosh unity coupled outputs... This design always fasinated me because it was radically different... I use to wind these bifilar outputs years ago when I had more time...very interesting stuff..
Chris
Concur with all, Cerrem - well stated.
Have also indulged in bi-filar stuff when I was younger and more wreckless; intruiging, especially with modern insulating materials on wire.
My personal preference in valve output stages (although I am actually a semiconductor amp designer!) is the Quad type: Distributed load with the [G2 - B+] section of the transformer on the cathode side (also used in Bogen).
But I would like to know whether anybody (I notice you will know) have used toroids for output transformers. There are a few Dutch designs around and the prospect looks attractive . . . but I would not know how to design the flipping thing, especially regarding leakage reactance. There are graphs for calculating layer-wound coils, but as far as I know it is impractical to go to some form of sectionalising with toroids. The ones I saw was simply wound "unlayered". Does one just go that way, knowing that the leakage reactance would probably be low enough, and then sort out feedback stability in situ with the final product? And what about tolerances - to what extent will the next ones measure the same? I would feel a little uncomfortable with the secondary lying somehwere in the middle in a rather undefined way - or am I just a little too set in yesterday's technology?
Have also indulged in bi-filar stuff when I was younger and more wreckless; intruiging, especially with modern insulating materials on wire.
My personal preference in valve output stages (although I am actually a semiconductor amp designer!) is the Quad type: Distributed load with the [G2 - B+] section of the transformer on the cathode side (also used in Bogen).
But I would like to know whether anybody (I notice you will know) have used toroids for output transformers. There are a few Dutch designs around and the prospect looks attractive . . . but I would not know how to design the flipping thing, especially regarding leakage reactance. There are graphs for calculating layer-wound coils, but as far as I know it is impractical to go to some form of sectionalising with toroids. The ones I saw was simply wound "unlayered". Does one just go that way, knowing that the leakage reactance would probably be low enough, and then sort out feedback stability in situ with the final product? And what about tolerances - to what extent will the next ones measure the same? I would feel a little uncomfortable with the secondary lying somehwere in the middle in a rather undefined way - or am I just a little too set in yesterday's technology?
Torroid calculations are not that much different....
I derived the equations from setting up some integrals and then solving diff equations..
The main assumption in simplifying the analysis is to slice the torroid at one location...(on paper) then to straighten the core into a long straight solid rectangular bar...
Now when you wind your wire layers...you would normaly wire L to right, then right to left..BUT NO...don't do that,,,,,
Instead you lay the wire layers down Left to Right, Left to Right, Left to Right... alway in the same dirrection....
Then you connect the layer from lower RIGHT to upper LEFT..lke a Zig Zag arrangement...
This will change the AC voltage gradient a bit.... when you solve the integral for winding capacitance on a normal winding arangement you have a Constant of Integration of 3 which is pulled out by 1/3 factor... But I leave it to you to solve this...
The aspect ratio of most torroid when viewed in this manner will show a wide winding width vs allowable winding height due to inner hole of torroid... This indicates a preference for lower leakage inductance BUT higher winding capacitance... which is applicable in many but not all types of output stages..
Chris
I derived the equations from setting up some integrals and then solving diff equations..
The main assumption in simplifying the analysis is to slice the torroid at one location...(on paper) then to straighten the core into a long straight solid rectangular bar...
Now when you wind your wire layers...you would normaly wire L to right, then right to left..BUT NO...don't do that,,,,,
Instead you lay the wire layers down Left to Right, Left to Right, Left to Right... alway in the same dirrection....
Then you connect the layer from lower RIGHT to upper LEFT..lke a Zig Zag arrangement...
This will change the AC voltage gradient a bit.... when you solve the integral for winding capacitance on a normal winding arangement you have a Constant of Integration of 3 which is pulled out by 1/3 factor... But I leave it to you to solve this...
The aspect ratio of most torroid when viewed in this manner will show a wide winding width vs allowable winding height due to inner hole of torroid... This indicates a preference for lower leakage inductance BUT higher winding capacitance... which is applicable in many but not all types of output stages..
Chris
cerrem said:
I am convinced that this peak in inductance is purely a function of the alternate stacking often used for EI cores creating two distinctly different airgaps, and the peak in perm at half power is simply the local saturation of the iron around the smaller airgap. When stacked with a butt gap and assuming a non-oriented core the EI and C cores should behave identically. Thowing the oriented materials into the mix does introduce the cross grain pattern of the EI's into the mix and i'm unclear as to what that actually does, but i cannot see it helping in any manner.
dave
Dave you are correct...
When alternate stacking of the EI laminations you introduce a small air-gap which is dominant path for the magnetic path length..due to the lower reluctance...
The flux "jumps" from E to E side-ways across this small oxide coated gap...you could pull all the I's out and it would not make a difference.. up to the point when the E's start to saturate and you get a PEAK inductance...then the flux jumps between E's to I's and then the airgap is bigger, butt gap, then the "effective" permeabilty drops and so does the inductance curve downwards..
I have modeled this mathamatically and have derived equations to deal with this..the acuracy of the equations is dead on...
I modified the equations to account for alternate stacking of groups of 2, 4, 5 ...n....
If you compare NON-oriented steels in CC-core vs EI laminations then it is a close call...why would you use NON-grain oriented C-cores...I don't even recall seeing them..whats the point??
There is a trick that gets the CC-core to behave better than EI..
You cut the gap in the CC-core on the greatest angle possible...
This way you increase the SURFACE AREA of the gap...this greatly reduce fringing on small gaped transformer designs and increases the "effective" permeability....
Chris
When alternate stacking of the EI laminations you introduce a small air-gap which is dominant path for the magnetic path length..due to the lower reluctance...
The flux "jumps" from E to E side-ways across this small oxide coated gap...you could pull all the I's out and it would not make a difference.. up to the point when the E's start to saturate and you get a PEAK inductance...then the flux jumps between E's to I's and then the airgap is bigger, butt gap, then the "effective" permeabilty drops and so does the inductance curve downwards..
I have modeled this mathamatically and have derived equations to deal with this..the acuracy of the equations is dead on...
I modified the equations to account for alternate stacking of groups of 2, 4, 5 ...n....
If you compare NON-oriented steels in CC-core vs EI laminations then it is a close call...why would you use NON-grain oriented C-cores...I don't even recall seeing them..whats the point??
There is a trick that gets the CC-core to behave better than EI..
You cut the gap in the CC-core on the greatest angle possible...
This way you increase the SURFACE AREA of the gap...this greatly reduce fringing on small gaped transformer designs and increases the "effective" permeability....
Chris
cerrem said:
nice explanation snipped.
the only benefit i can think of is lam thickness, you are pretty much stuck at .006 of an inch with stamped lams, and even they are quite delicate.
yes, I think mag metals calls that the "A" core. There are also the "uni-cores" which use a gap that is stepped (looks like a staircase) which has the same effect (as well as come other claimed benefits.)
dave
Thanks Chris (post #5).
I was thinking more of the practical implications of not having such a predictable situation as with neat layer winding. You would know that the physical distance between P and S sections is a factor in leakage inductance (I use the Crowhurst design graphs). This seems quite less predictable with a toroid where a definable flat layer does not exist because of overlapping turns on the inside - exact "gap" between P and S thus not well defined and will differ from one unit to the next, especially with wire of significantly different diameters.
But I have only seen toroids as power transformers, not output; not familiar with how the Dutch ones (or others) physically look. In any case, your remarks re C-cores also valuable. I make use of these (also question of what is readily obtainable here) and do not really have a good reason to switch to something else. (I do more semiconductor than tube amplifiers, but am often approached regarding the latter because I cut my electronic teeth in that era - now everybody knows my vintage!)
Johan
I was thinking more of the practical implications of not having such a predictable situation as with neat layer winding. You would know that the physical distance between P and S sections is a factor in leakage inductance (I use the Crowhurst design graphs). This seems quite less predictable with a toroid where a definable flat layer does not exist because of overlapping turns on the inside - exact "gap" between P and S thus not well defined and will differ from one unit to the next, especially with wire of significantly different diameters.
But I have only seen toroids as power transformers, not output; not familiar with how the Dutch ones (or others) physically look. In any case, your remarks re C-cores also valuable. I make use of these (also question of what is readily obtainable here) and do not really have a good reason to switch to something else. (I do more semiconductor than tube amplifiers, but am often approached regarding the latter because I cut my electronic teeth in that era - now everybody knows my vintage!)
Johan
mmm interesting post cerrem
do you have handy these inductance curves for EI and CC core?
But I have a doubt you can probably clarify...
Flux will be surely distorted, as it is obliged to turn and follow the iron shape, at the corners of the E's and there the grain orientation is also perpendicular to the flux worsening the scenario(lower perm). This will lead to a higher flux density in these points leading, possibly, to a "locally saturated core". In CC cores this wont happen due to the peculiar production/bending of the core.
Gianluca
do you have handy these inductance curves for EI and CC core?
But I have a doubt you can probably clarify...
Flux will be surely distorted, as it is obliged to turn and follow the iron shape, at the corners of the E's and there the grain orientation is also perpendicular to the flux worsening the scenario(lower perm). This will lead to a higher flux density in these points leading, possibly, to a "locally saturated core". In CC cores this wont happen due to the peculiar production/bending of the core.
Gianluca
As for inductance curves.... There really is no curves..
You start out with the Epstien ring curves....these are Ideal curves with no gap.... They provide you a "ideal" permeability for a given flux density.....Through use of equations, you then find the "effective" permeability...The effective permeabilty is based on many factors... The math is soo long to do, my hand falls off
SO I wrote computer programs with these equations to handle this...
The standard first order equations found in text books will get you in the ball-park but not really acurate....
As for the EI laminations, the flux density is fairly constant throughout the magnetic path..the corners around the holes don't become a real problem until really high frequencies...
The EI lamination is not perfect....
Chris
You start out with the Epstien ring curves....these are Ideal curves with no gap.... They provide you a "ideal" permeability for a given flux density.....Through use of equations, you then find the "effective" permeability...The effective permeabilty is based on many factors... The math is soo long to do, my hand falls off
SO I wrote computer programs with these equations to handle this...The standard first order equations found in text books will get you in the ball-park but not really acurate....
As for the EI laminations, the flux density is fairly constant throughout the magnetic path..the corners around the holes don't become a real problem until really high frequencies...
The EI lamination is not perfect....
Chris
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.