I looked through the forum and didn't see any discussion of this.
With the Slagle interstage transformers I got it was recommended to experiment with different shims in the gap. With my AVC it was recommended to experiment with restacking the lams. I see others are reporting results from restacking their MC step ups.
I know they both change the amount of inductance but why choose one over the other.
what is the difference in adj air gap and restacking lams??
let the games begin
Up to this point I have not played much with restacking and gapping as I have been busy just getting my amps to the final (ha ha) design but I must say I am very suprised how much difference it makes.
So far I have experimented with my AVC and mc stepups and the results have been positive. I also have 2 sets of interstage transformers in my phono stage and plate chokes in my amp so I have a ways to go before I try them all but I'll get to it eventually.
So far I have experimented with my AVC and mc stepups and the results have been positive. I also have 2 sets of interstage transformers in my phono stage and plate chokes in my amp so I have a ways to go before I try them all but I'll get to it eventually.
Bruce Bosler

 Posts: 96
 Joined: Sun Jan 23, 2005 10:58 pm
Bruce, that's a good question and it probably should have been explained here before.
When you try different lam interleaves you really are trying different gaps. The reason is that a simple butt gap (no spacer) is effectively a small gap. To make the gap even smaller you need to interleave the lams. A 10x10 interleave will effectively be smaller than a butt gap, but larger gap than a 5x5 interleave. 1x1 is as close to ungapped as you can get, but even that will have some effective gap.
Some texts say a butt gap is equivalent to .0005 inches. My guess is that it probably depends on a number of factors including how well the lams are cut and how carefully the stack is assembled. I suppose someone could deduce the effective size of a butt gap and that of a number of different interleaves by measuring the inductance of them all along with a series of different size spacers (under identical conditions) then working the problem backwards.
 Dave
When you try different lam interleaves you really are trying different gaps. The reason is that a simple butt gap (no spacer) is effectively a small gap. To make the gap even smaller you need to interleave the lams. A 10x10 interleave will effectively be smaller than a butt gap, but larger gap than a 5x5 interleave. 1x1 is as close to ungapped as you can get, but even that will have some effective gap.
Some texts say a butt gap is equivalent to .0005 inches. My guess is that it probably depends on a number of factors including how well the lams are cut and how carefully the stack is assembled. I suppose someone could deduce the effective size of a butt gap and that of a number of different interleaves by measuring the inductance of them all along with a series of different size spacers (under identical conditions) then working the problem backwards.
 Dave
Thanks Dave. That makes sense but it still amazes me that making these seemingly insignificant changes can affect the sound so much. Some days it drives me insane that there are so many variables and others I just enjoy playing with it. At least I'm to the point that if I never made another change I have a system i can enjoy.
Bruce Bosler

 Posts: 2099
 Joined: Sat Jan 22, 2005 3:54 am
 Location: NYC
 Contact:
Hey Bruce,
EI and other stamped lams offer some interesting options for the selection of gap size. At one end of the spectrum you can place all of the E's on one side and all of the I's on the other and get what is often called a "butt" gap.
The Butt gap is the equivalent of the smallest gap available from a "standard" ccore. In both cases the smoothness and uniformity of the pole surfaces play the major roll in the actual gap size.
From the butt gap, increasing the gap size is simply a matter of inserting some sort of spacer to increase the distance between the poles, going in the other direction and attempting to reduce the size of the gap is where things get interesting.
Essentially there are two ways to reduce the gap size, decrease the distance between the pole faces or increase the surface area of the pole faces. With Ccores, you are limited by polishing the surfaces, and with EI's your limit is by the quality of the stamping and the ability to align pairings of E's and I's. Essentially you can only get the mating surfaces so perfect and that dictates the smallest gap possible.
Beyond the physical limitations of the "butt gap" the gap can be made smaller by increasing the surface area of the gap. This can be taken seen in it's extreme form but looking at the tape wound toroid. The gap in a toroid is the entire surface area of the tape which nets a huge surface area and thus a very small effective gap. Moving back to the Ccore, if you make the cut an an angle you also increase the surface area which allows a smaller effective gap than a perpendicular cut (assuming the mating of the surfaces ideal) Magnetic Metals actually offers cores like this. Another option I have seen is called the unicore where they step the pole faces which doubles the surface area of the gap.
Moving on to EI's, we have the option of alternating the e's to form the core. Imagine a core stacked solely with scrapless E laminations and try to envision the nature of the gap. The gap becomes the surface area between the touching "tines" of the e's and in order for the flux to complete its loop, it must traverse two gaps (e1e2 and then back from e2e1). This gap is much smaller that the butt gap scenario for two reasons, first the surface area is much larger, and second, the physical spacing between lams is much smaller than across the traditional butt gap pole face.
So back to out 1X1 stacking of e's. If we measure the inductance with just the E's in place and get 100hy's of inductance and then insert the 1's we will only see a small increase in inductance. I still haven't completely gotten my head around this since there are a few things at work here. Both flux and inductance are based on core area and if you double the core area you cut the flux in half and double the inductance. SO a EE core stacked without the I's has effectively 1/2 the core area which should double the flux and half the inductance, but that does not happen. So apparently the gap is made so much smaller that it nearly doubles the inductance. There is a lam shape that fully takes advantage of the "NoI" configuration that has the upright of the E at double the width of a standard EI. When stacked in the EE configuration, the core area remains constant since the double width of the upright makes up for the lack of an I. These lams are often seen in small modem transformers and in an ideal world, would be my choice if a minimally gapped core were desired.
My main problem (conceptually) with alternate stacking is that you create two distinctly different airgaps. It is almost as if you have tow different cores. With 1/2 of the core having a very small gap and the other half having a butt gap. With any applied DC (intentional or not) the very small gapped parts of the core will locally saturate at about 1/2 the Bsat of the core leaving the remainder of the core to handle the duties above that level. I envision this causing a "Kink" in the transfer function (is that the right word) of the device. I am not saying that this would be a bad thing, and actually if properly done, that "kink" could offset another nonlinearity in the core netting you a more linear device.
A few weeks back I was trying to better figure out the nature of the alternately stacked gap so I tried a number of various different stackings starting with a butt gap and ending with a 1X1 pattern, after measuring each stacking I simply inverted the two outer lams and remeasured. Assuming 36 lams, this would convert a 2 section 18X10 stacking to a 4 section 1X17X17X1 pattern. What I quickly noticed was that the number of sections and not the lams per section was the dominant factor in determining the overall inductance. (plots attached below)
The First thing you will notice about the plots is they are not nice clean curved lines that you see published in books. They are what they are, a quick down and dirty test of three samples. I would expect if I repeated this test 10 times and averaged the results the line would smooth but at this point of the game I am content visualizing the apparent pattern. The main reason for the jumpiness has to do with the fact that it is really hard to set a precision gap due to the various physical parameters involved. As a sample of this, you can take a 10 inch strip of lams and stack 19 identical bobbins as neatly as possible with a butt gap and still see a 20% variation of inductance. I think it is fair to assume that the perm of the material remains constant throughout the stack so the only variable is the consistency of the gap. The butt gap is actually the most problematic and as you go in either direction your the range of your sample will close. By the time your spacer becomes a few thousandths of an inch or you get to 1X1 your consistency can be closer to 12% and consideration of the physical variables involved this actually seems to all fit together.
When you increase the gap beyond the butt gap with a physical spacer, suddenly the small variations between the E and I pole faces are swamped out by the increasingly larger physical spacer. Going in the other direction, the larger physical surface area of the EE gap and the more consistent spacing between them also explains why the gaps become more repeatable as the number of sections increases.
So where does this leave us? My current thinking is that the number of alternate sections dictates the inductance and it would be very interesting to compare a few different stacking possibilities with the same number of sections.
dave
EI and other stamped lams offer some interesting options for the selection of gap size. At one end of the spectrum you can place all of the E's on one side and all of the I's on the other and get what is often called a "butt" gap.
The Butt gap is the equivalent of the smallest gap available from a "standard" ccore. In both cases the smoothness and uniformity of the pole surfaces play the major roll in the actual gap size.
From the butt gap, increasing the gap size is simply a matter of inserting some sort of spacer to increase the distance between the poles, going in the other direction and attempting to reduce the size of the gap is where things get interesting.
Essentially there are two ways to reduce the gap size, decrease the distance between the pole faces or increase the surface area of the pole faces. With Ccores, you are limited by polishing the surfaces, and with EI's your limit is by the quality of the stamping and the ability to align pairings of E's and I's. Essentially you can only get the mating surfaces so perfect and that dictates the smallest gap possible.
Beyond the physical limitations of the "butt gap" the gap can be made smaller by increasing the surface area of the gap. This can be taken seen in it's extreme form but looking at the tape wound toroid. The gap in a toroid is the entire surface area of the tape which nets a huge surface area and thus a very small effective gap. Moving back to the Ccore, if you make the cut an an angle you also increase the surface area which allows a smaller effective gap than a perpendicular cut (assuming the mating of the surfaces ideal) Magnetic Metals actually offers cores like this. Another option I have seen is called the unicore where they step the pole faces which doubles the surface area of the gap.
Moving on to EI's, we have the option of alternating the e's to form the core. Imagine a core stacked solely with scrapless E laminations and try to envision the nature of the gap. The gap becomes the surface area between the touching "tines" of the e's and in order for the flux to complete its loop, it must traverse two gaps (e1e2 and then back from e2e1). This gap is much smaller that the butt gap scenario for two reasons, first the surface area is much larger, and second, the physical spacing between lams is much smaller than across the traditional butt gap pole face.
So back to out 1X1 stacking of e's. If we measure the inductance with just the E's in place and get 100hy's of inductance and then insert the 1's we will only see a small increase in inductance. I still haven't completely gotten my head around this since there are a few things at work here. Both flux and inductance are based on core area and if you double the core area you cut the flux in half and double the inductance. SO a EE core stacked without the I's has effectively 1/2 the core area which should double the flux and half the inductance, but that does not happen. So apparently the gap is made so much smaller that it nearly doubles the inductance. There is a lam shape that fully takes advantage of the "NoI" configuration that has the upright of the E at double the width of a standard EI. When stacked in the EE configuration, the core area remains constant since the double width of the upright makes up for the lack of an I. These lams are often seen in small modem transformers and in an ideal world, would be my choice if a minimally gapped core were desired.
My main problem (conceptually) with alternate stacking is that you create two distinctly different airgaps. It is almost as if you have tow different cores. With 1/2 of the core having a very small gap and the other half having a butt gap. With any applied DC (intentional or not) the very small gapped parts of the core will locally saturate at about 1/2 the Bsat of the core leaving the remainder of the core to handle the duties above that level. I envision this causing a "Kink" in the transfer function (is that the right word) of the device. I am not saying that this would be a bad thing, and actually if properly done, that "kink" could offset another nonlinearity in the core netting you a more linear device.
A few weeks back I was trying to better figure out the nature of the alternately stacked gap so I tried a number of various different stackings starting with a butt gap and ending with a 1X1 pattern, after measuring each stacking I simply inverted the two outer lams and remeasured. Assuming 36 lams, this would convert a 2 section 18X10 stacking to a 4 section 1X17X17X1 pattern. What I quickly noticed was that the number of sections and not the lams per section was the dominant factor in determining the overall inductance. (plots attached below)
The First thing you will notice about the plots is they are not nice clean curved lines that you see published in books. They are what they are, a quick down and dirty test of three samples. I would expect if I repeated this test 10 times and averaged the results the line would smooth but at this point of the game I am content visualizing the apparent pattern. The main reason for the jumpiness has to do with the fact that it is really hard to set a precision gap due to the various physical parameters involved. As a sample of this, you can take a 10 inch strip of lams and stack 19 identical bobbins as neatly as possible with a butt gap and still see a 20% variation of inductance. I think it is fair to assume that the perm of the material remains constant throughout the stack so the only variable is the consistency of the gap. The butt gap is actually the most problematic and as you go in either direction your the range of your sample will close. By the time your spacer becomes a few thousandths of an inch or you get to 1X1 your consistency can be closer to 12% and consideration of the physical variables involved this actually seems to all fit together.
When you increase the gap beyond the butt gap with a physical spacer, suddenly the small variations between the E and I pole faces are swamped out by the increasingly larger physical spacer. Going in the other direction, the larger physical surface area of the EE gap and the more consistent spacing between them also explains why the gaps become more repeatable as the number of sections increases.
So where does this leave us? My current thinking is that the number of alternate sections dictates the inductance and it would be very interesting to compare a few different stacking possibilities with the same number of sections.
dave
 Attachments

 The blue and pink represent the same core size and the green is just some old data from a smaller core. It's interesting to note the pattern remains similar.
 image001.gif (9.49 KiB) Viewed 12320 times