if you crank up the frequency to 20k, you can pull the lams and the transformer will still work,
Roll your own Output Transformers
- Thread starter All Ears
- Start date
Help Support GroupDIY Audio Forum:
Well that's true!
But what I find is that with my EE24/25 bobbins, I can easily fit a stack of 19 lams in, or squeeze 20 in there. With 19 I get more inductance at 1KHz, whereas with 20 I get less at 1 KHz, but more at 100 Hz.
And it is repeatable with many transformers and a couple of different meters.
Really we don't worry about 1KHz too much because the roll-off is not anywhere close to it, but i find it interesting nonetheless. It could also affect efficiency if it is leakage.
But what I find is that with my EE24/25 bobbins, I can easily fit a stack of 19 lams in, or squeeze 20 in there. With 19 I get more inductance at 1KHz, whereas with 20 I get less at 1 KHz, but more at 100 Hz.
And it is repeatable with many transformers and a couple of different meters.
Really we don't worry about 1KHz too much because the roll-off is not anywhere close to it, but i find it interesting nonetheless. It could also affect efficiency if it is leakage.
Not leakage, losses.zebra50 said:Well that's true!
But what I find is that with my EE24/25 bobbins, I can easily fit a stack of 19 lams in, or squeeze 20 in there. With 19 I get more inductance at 1KHz, whereas with 20 I get less at 1 KHz, but more at 100 Hz.
And it is repeatable with many transformers and a couple of different meters.
Really we don't worry about 1KHz too much because the roll-off is not anywhere close to it, but i find it interesting nonetheless. It could also affect efficiency if it is leakage.
Leakage in an iron-core xfmr is about 0.1%, so changing it by as much as -50/+100% would have no significant influence on mid-band efficiency.
Iron losses increase when pressing lams because the contact resistance decreases.
When squeezing more lams, the pressure increases so the losses increase, which creates a reduction in impedance that the L-meter translates as less inductance; at the same time more iron-> more LF inductance.
contact resistance, hmmmm,
Between lams.CJ said:contact resistance, hmmmm,
abbey road d enfer said:Between lams.CJ said:contact resistance, hmmmm,
no, freq is too low for that effect,
stacking factor K goes up when lams squeezed,
flux goes down as B=1/k,
perm goes down with flux if below the BH knee,
inductance goes down with perm,
leakage inductance ? no, me bad,
meant coupling factor, exact same thing only different,
coupling factor is covered in RDH4-Ch 10, Inductance, but for air core coils only, however, the effect does exist in steel core transformers, but probably to a much lesser extent, this is why some people like paper wound coils over bobbin wound, as you can get the wire closer to the core for more inductance,
other factors? -Lenz type effect from close lams influencing mag field from lam to lam, as there will be an electrical field associated with the mag field in each lam, this field being of the same polarity will tend to repel the adjacent field in the lam which will buck the mag field that is trying to be formed in the neighboring lam, this may or may not be a factor, as the frequency usually has to be high for this effect as seen in conductors being close together,
also, maybe Q factor is changing due to core getting away from optimum geometry of stack which is around 0.6 length to width,
to verify insert insl between lams while squeezing and watch L meter,
another possible effect for 80 Ni would be lam expansion and contraction would be hindered by clamping, this is why most hi perm cores are not clamped, as the lams need to "breathe", otherwise perm will be slightly reduced, magneto-restrictive force.
Eddy currents appear at any frequency, but get worse as frequency increase. If they didn't happen at 50/60Hz, power xfmr cores would be solid.CJ said:abbey road d enfer said:Between lams.CJ said:contact resistance, hmmmm,
no, freq is too low for that effect,
So inductance goes down at low level, goes up at higher level, then goes down again; nothing new.
Overall, the inductance at LF will increase compared to loose lams. But at HF, eddy currents increase because the electrical short is more significant, so apparent inductance decreases (in fact Rloss decreases). Exactly what zebra 50 describes.
As I wrote earlier, what's the difference in midrange performance (<2kHz) bet k=0.9995 and k=0.9998 (in other words Ll=0.005Lnom and Ll=0.0002Lnom).leakage inductance ? no, me bad,
meant coupling factor, exact same thing only different,
yes, eddy currents and hystsrisis occur at 50/60 cycles, remember the old shaded coil record player motors with the copper wire shorting rings?
but also remember that many cheap power transformers have uninsulated hardware running thru the core and no, this is not a shorted turn as many people believe,
we need to do some experiments in order to get a better understanding of this lam pressure thing,
we have left out some important details in the discussion, such as
1) are we talking about steel or nickel lams? nickel lams have no insulation, Microsil lams like the ones used in power transformers have an oxide coating, there is a burr that can short the lams at the edges, but this does not create a block of steel as the eddy currents are still broken up on 99 percent of the lam surface,
2) what effect does frequency and voltage level have on inductance changing when the lams are pressed tighter?
3) what percentage of inductance loss are talking about? 1 percent? 5 percent? 50 percent?
4) does the amount of turns affect the lam pressure vs inductance effect?
5) how much pressure are we talking about and is there a point where squeezing the lams harder no longer changes the inductance?
so lets wind a 100 turn and a 1000 turn coil and lace it with Si and Ni lams and get out the signal generator to see what happens, oh boy, can't wait, always new things learned,
here is a link to some eddy current stuff>
http://www.ndt-ed.org/EducationResources/CommunityCollege/EddyCurrents/Physics/depthcurrentdensity.htm
but also remember that many cheap power transformers have uninsulated hardware running thru the core and no, this is not a shorted turn as many people believe,
we need to do some experiments in order to get a better understanding of this lam pressure thing,
we have left out some important details in the discussion, such as
1) are we talking about steel or nickel lams? nickel lams have no insulation, Microsil lams like the ones used in power transformers have an oxide coating, there is a burr that can short the lams at the edges, but this does not create a block of steel as the eddy currents are still broken up on 99 percent of the lam surface,
2) what effect does frequency and voltage level have on inductance changing when the lams are pressed tighter?
3) what percentage of inductance loss are talking about? 1 percent? 5 percent? 50 percent?
4) does the amount of turns affect the lam pressure vs inductance effect?
5) how much pressure are we talking about and is there a point where squeezing the lams harder no longer changes the inductance?
so lets wind a 100 turn and a 1000 turn coil and lace it with Si and Ni lams and get out the signal generator to see what happens, oh boy, can't wait, always new things learned,
here is a link to some eddy current stuff>
http://www.ndt-ed.org/EducationResources/CommunityCollege/EddyCurrents/Physics/depthcurrentdensity.htm
Attachments
CJ, Abbey... Great discussion, I been sitting back and learning.
So, for my experiments I can't answer all of CJ's questions, but
Well, I see it on both primary and secondary of a 1:34 ribbon transformer, and the ratio seems well behaved.
In my case, the difference between 19 lams stack is that occasionally a lam will drop out so quite loose really - I would say no pressure there. But squeezing that last lam in can be quite a challenge.
(I can add that the 80/20 lams are slightly thinner than the 50/50, so you can get 20 of the 80/20 lams in without squeezing).
I guess one could stack 19 and squeeze with a micrometer whilst measuring L. I'm not sure how I would measure the actual force / pressure created, but it would at least give us a scale.
Interesting stuff!
Stewart
So, for my experiments I can't answer all of CJ's questions, but
In this case I use a mixture of 50/50 Nickel-Steel (Sold as "Permanorm"), and 80/20 Nickel-Steel (Mumetal).
Yes, I believe that see this effect too. If the lams are stacked as they arrive, with the same sides up and all the burrs in the same direction, then the transformer specs are more reproducible than if you drop them on the floor and mix them up! The difference is a % or two.
5 to 10%. So, measurable with some degree of confidence.
Well, I see it on both primary and secondary of a 1:34 ribbon transformer, and the ratio seems well behaved.
In my case, the difference between 19 lams stack is that occasionally a lam will drop out so quite loose really - I would say no pressure there. But squeezing that last lam in can be quite a challenge.
(I can add that the 80/20 lams are slightly thinner than the 50/50, so you can get 20 of the 80/20 lams in without squeezing).
I guess one could stack 19 and squeeze with a micrometer whilst measuring L. I'm not sure how I would measure the actual force / pressure created, but it would at least give us a scale.
Interesting stuff!
Stewart
ok first experiment was done last night, V72 choke coil stuffed with just 10 lams of hi perm nickel, this was so the lams could be spread out inside the bobbin with large air gaps in between,
at freqs from 20 hz to 1000 hz and at 20 volts AC the inductance went way up when the lams were squeezed together, and went back down as the lams were spaced out in the bobbin,
so inductance will not always go down when the lams are squeezed, sometimes it will go up,
depends on how many we start with i guess,
so a small stack of lams will tend to concentrate the flux better if they are next to each other rather than spread apart by a large air gap
i will try this with steel lams next,
at freqs from 20 hz to 1000 hz and at 20 volts AC the inductance went way up when the lams were squeezed together, and went back down as the lams were spaced out in the bobbin,
so inductance will not always go down when the lams are squeezed, sometimes it will go up,
depends on how many we start with i guess,
so a small stack of lams will tend to concentrate the flux better if they are next to each other rather than spread apart by a large air gap
i will try this with steel lams next,
ok with steel Si grain lams there was always an increase in inductance with the lams pressed tighter, this happens with 15 lams and 48 lams at 20 volts, at all frequencies,
going back to supermalloy lams we noticed the same thing but at the point of max Q, where input current was lowest at about 800 hz, there was a slight drop in inductance,
so either we are shifting the frequency of max Q for the core by changing the air gap between the lams, or we are stressing the crystals in the lams by bending the lams slightly due to the clamping points being at places which cause deformation,
so we are narrowing down this inductance loss to a specific frequency,
time to try a lower turns coil to see if the same thing happens, lowering the turns will get us up higher on the BH curve, ie, more flux.
the increase in inductance for steel lams by squeezing happened to a lesser extent as we approached the max Q freq, and increased again at freqs higher than the max Q point,
all this is the same thing that Zebra sees with 19 vs 20 lams,
most LC meters measure L at 1 k hz, which is close to max Q freq of the core, this may explain why you see the L drop with hand held meters,
going back to supermalloy lams we noticed the same thing but at the point of max Q, where input current was lowest at about 800 hz, there was a slight drop in inductance,
so either we are shifting the frequency of max Q for the core by changing the air gap between the lams, or we are stressing the crystals in the lams by bending the lams slightly due to the clamping points being at places which cause deformation,
so we are narrowing down this inductance loss to a specific frequency,
time to try a lower turns coil to see if the same thing happens, lowering the turns will get us up higher on the BH curve, ie, more flux.
the increase in inductance for steel lams by squeezing happened to a lesser extent as we approached the max Q freq, and increased again at freqs higher than the max Q point,
all this is the same thing that Zebra sees with 19 vs 20 lams,
most LC meters measure L at 1 k hz, which is close to max Q freq of the core, this may explain why you see the L drop with hand held meters,
Cool beans.... My meter does 100Hz, 1KHz and 10 KHz, but I also have an impedance bridge which I can run at any frequency. Would an impedance plot against frequency cast any more light on this, with one side of the transformer terminated? I think it might.
Definitely. However, letting the xfmr unterminated will make deviations more measurable.zebra50 said:
buildafriend
Well-known member
no thread jacking intended
how can you measure the inductance of a transformer? if I remember right inductance was the increase in voltage and decrease in current from wrapping a wire around a magnet. perhaps it can be obtained by finding the difference from voltage gain and current loss?
( sorry for being a stupid head )
how can you measure the inductance of a transformer? if I remember right inductance was the increase in voltage and decrease in current from wrapping a wire around a magnet. perhaps it can be obtained by finding the difference from voltage gain and current loss?
( sorry for being a stupid head
)
if you do not have a L meter you can simply use a signal generator and a current meter,
think of the coil and core as an AC resistor,
measure the voltage across the coil while writing down the AC current,
use Ohms Law to get the resistance,
then plug that into the Reactance formula for inductors which is
XL= 6.28 x freq x Inductance,
rearranging the formula to isolate Inductance is left to the student as an exercise,
just kidding,
Inductance = ohms/6.28f
using the generator method allows you to measure inductance at different freqs and voltages,
and remember, eddy currents can be a good thing, this is idiot cj on the Merced River using an eddy current to stop for a scout of an innocent looking rapid that actually has a series of underwater tunnels in the boulders on the left, so after you get flipped in that hole on the left and eject, you are washed into the underwater tunnels which is not a good thing,
and yes, that is a Telefunken decal,
Nightmare Island Class 4
think of the coil and core as an AC resistor,
measure the voltage across the coil while writing down the AC current,
use Ohms Law to get the resistance,
then plug that into the Reactance formula for inductors which is
XL= 6.28 x freq x Inductance,
rearranging the formula to isolate Inductance is left to the student as an exercise,
just kidding,
Inductance = ohms/6.28f
using the generator method allows you to measure inductance at different freqs and voltages,
and remember, eddy currents can be a good thing, this is idiot cj on the Merced River using an eddy current to stop for a scout of an innocent looking rapid that actually has a series of underwater tunnels in the boulders on the left, so after you get flipped in that hole on the left and eject, you are washed into the underwater tunnels which is not a good thing,
and yes, that is a Telefunken decal,
Nightmare Island Class 4
Attachments
this thing is getting more complicated by the minute, there may be many reasons for what we are seeing,
warping of the lams is a definite factor,
eddy currents due to decreased contact resistance as Mr Abbey pointed out can not be ruled out,
flux level also plays a role, as well as the number of turns on the coil,
and there may even be some capacitive effects going on at 10 k hz and above,
thought a few experiments would solve the problem but now we need to start over and write everything down for core type, coil turns, voltage level, frequency and pressure in order to quantify what happens where, ???
with 15 Ni lams inside a 1000 turn coil, inductance can be increased by moving the lams together, and then decreased by bending the lams, it is easy to bend the lams when the stack is very small so the effects can be seen easily, also, this effect is obviously going to be more pronounced on the hi perm cores as the crystals are more fragile,
a smaller turns coil raises the max Q freq, so squeezing the core will cause a decrease in inductance due to increased eddy currents rather than crystal deformation as eddy currents increase as the square of the frequency, and the core is contributing less to the inductamce as the freq goes up,so not much eddy current effects at 20 hz, but at 10 k they could become more of a factor,
so high turns coils will have a lower Q freq which means that crystal deformation will take over as the dominant factor in the decreased inductance,
flux levels will change with turns count, so the core loss factor will change as the turns change, on and on we go...
warping of the lams is a definite factor,
eddy currents due to decreased contact resistance as Mr Abbey pointed out can not be ruled out,
flux level also plays a role, as well as the number of turns on the coil,
and there may even be some capacitive effects going on at 10 k hz and above,
thought a few experiments would solve the problem but now we need to start over and write everything down for core type, coil turns, voltage level, frequency and pressure in order to quantify what happens where, ???
with 15 Ni lams inside a 1000 turn coil, inductance can be increased by moving the lams together, and then decreased by bending the lams, it is easy to bend the lams when the stack is very small so the effects can be seen easily, also, this effect is obviously going to be more pronounced on the hi perm cores as the crystals are more fragile,
a smaller turns coil raises the max Q freq, so squeezing the core will cause a decrease in inductance due to increased eddy currents rather than crystal deformation as eddy currents increase as the square of the frequency, and the core is contributing less to the inductamce as the freq goes up,so not much eddy current effects at 20 hz, but at 10 k they could become more of a factor,
so high turns coils will have a lower Q freq which means that crystal deformation will take over as the dominant factor in the decreased inductance,
flux levels will change with turns count, so the core loss factor will change as the turns change, on and on we go...
I think one reason for this difference between effects that you are seeing when compressing the lams has to do with the fact that two different things are changing.
1) The potential for eddy currents increases as the lams become more electrically linked
2) The flux path (not sure of terms) or magnetic path is increasing as the (for example) I and E pieces are magnetically linked.
This #2 magnetic linking is seen (and was discussed by you CJ in another article) in the ferrite material cores, as you tighten the screw or clamp and align the pieces and clean the faces you get more inductance. ( note: If you tighten too much the faces deform and separate and your inductance reduces. It is amazing that ferrite can be flexed at all)
So I think to some extent it depends upon how you are tightening and the geometry and condition of the core.
If it is an IE core, and you clamp the I's closer to the E's or even clamp the lams closer together allowing I's to magnetically link to their cheek facing (stack adjacent) E's you get a better flux path and more inductance (the flux travels from E to I through part of the adjacent E).
If the E's are clamped tighter to the matching (same lam layer) I, the flux increases. If you get to the point where you effectively have a solid piece, and the flux path between the same layer IE cannot be improved (as if there was one solid one piece lam that looked like EI together (like an 8 or a B)) then clamping lams closer together doesn't link the I's and E's because they are fully linked, and all you get is more Eddy current and the inductance goes down. This could also happen with unwarped Lams with polished faces in perfect condition and alignment, where the flux path cannot be significantly improved by clamping.
Just a theory, but I think it matches the facts.
1) The potential for eddy currents increases as the lams become more electrically linked
2) The flux path (not sure of terms) or magnetic path is increasing as the (for example) I and E pieces are magnetically linked.
This #2 magnetic linking is seen (and was discussed by you CJ in another article) in the ferrite material cores, as you tighten the screw or clamp and align the pieces and clean the faces you get more inductance. ( note: If you tighten too much the faces deform and separate and your inductance reduces. It is amazing that ferrite can be flexed at all)
So I think to some extent it depends upon how you are tightening and the geometry and condition of the core.
If it is an IE core, and you clamp the I's closer to the E's or even clamp the lams closer together allowing I's to magnetically link to their cheek facing (stack adjacent) E's you get a better flux path and more inductance (the flux travels from E to I through part of the adjacent E).
If the E's are clamped tighter to the matching (same lam layer) I, the flux increases. If you get to the point where you effectively have a solid piece, and the flux path between the same layer IE cannot be improved (as if there was one solid one piece lam that looked like EI together (like an 8 or a B)) then clamping lams closer together doesn't link the I's and E's because they are fully linked, and all you get is more Eddy current and the inductance goes down. This could also happen with unwarped Lams with polished faces in perfect condition and alignment, where the flux path cannot be significantly improved by clamping.
Just a theory, but I think it matches the facts.
OK, as promised here is a plot of measured impedance against frequency. I would 100 turns round the bobbin, and used the same winding for both experiments... so DCR is the same in both cases.
Red is with 19 lams and is a gentle sliding fit into the bobbins. Blue is 20 lams, and is a tight squeeze to get that last lam in place....
The crossover comes around 300Hz. Below that, the 20 ohm stack gives higher impedance, and at higher frequencies the stack of 19 wins....
Good to see CJ in his Kayak!
Red is with 19 lams and is a gentle sliding fit into the bobbins. Blue is 20 lams, and is a tight squeeze to get that last lam in place....
The crossover comes around 300Hz. Below that, the 20 ohm stack gives higher impedance, and at higher frequencies the stack of 19 wins....
Good to see CJ in his Kayak!
