 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
In Reply to: Re: PP Iron for SE - repost of a 1996 rec.audio.tubes by "Paul".... posted by Steve Eddy on February 18, 2004 at 22:35:08:
But here you're talking about a pulse, which is unipolar. We're dealing with AC waveforms.first we have to agree music is not sinusodal, then i have to ask so what if we are unipolar or AC... in the case of the pulse core, after a singlke pulse without the -Bdc applied, the core would remain at the +Br, but when the -Bdc is added the core then returns to the bias point rather than the Br. would this be any different if AC were used instread of a pulse?
(se)With a DC offset, while you produce a field, the energy stored, the energy which is responsible for dematetizing, doesn't get returned to the system until the offset is removed.
but if we assume the loop momentarily ends up at a -Br point during a musical passsage, the bias point will demagnetize and remagnetize the core to the Bdc point. the question becomes what happens if the signal ends on a +Br point, then i am not clear and what oyu say may be correct.
(se)That notion is easily dispelled by the fact that as you increase DC offset the transformer becomes more and more nonlinear.
there are a lot of "ifs" and assumptions in that statement. i don;t know if i woukld call it a fact (unless you assume) that Bmax=Bsat then i'll agree, but when Bmax is a fraction of Bsat wouldnt; be suprised to see the most linear part of the loop being somewhere other than (0,0) as an origin.
(SE)He said that as you increase saturation you get closer and closer to an air core transformer.
yes, and whats more linear than air?
(Se)That's NOT the same as saying you will have ideal linearity. Indeed, as you get closer and closer to saturation, distortion increases.
but above the knee of the loop the core is quite linear (your perm is very low, but thats a different issue)
(se)Take one of your transformers and remove the core and tell me how linear it is compared to with the core. :)
it will be perfectly linear without the core, i think thats the point :-) your perm will be 1, but thats a different issue.
sure it may be picking an nit, but i think its an important nit to understand. adding a magnetic core is adding nonlinearity, much of the time the increase in perm more that offsets any nonlinearity introduced by the core hence we see lots of iron cored (or nickel :-) outputs and very few air cored ones.
dave
dave
Follow Ups:
first we have to agree music is not sinusodal, then i have to ask so what if we are unipolar or AC... in the case of the pulse core, after a singlke pulse without the -Bdc applied, the core would remain at the +Br, but when the -Bdc is added the core then returns to the bias point rather than the Br.It will no more return to the bias point than an unbiased transformer would return to 0,0.
The DC just introduces an offset.
but if we assume the loop momentarily ends up at a -Br point during a musical passsage, the bias point will demagnetize and remagnetize the core to the Bdc point.
No, it won't. The magnetic field due to the DC current never collapses. It magnetizes the core only once and that's it. After that, any magnetization or demagnetization of the core is brought about by the signal and everything else happens as usual but it happens about the Bdc point rather than about the 0,0 point.
there are a lot of "ifs" and assumptions in that statement. i don;t know if i woukld call it a fact (unless you assume) that Bmax=Bsat then i'll agree, but when Bmax is a fraction of Bsat wouldnt; be suprised to see the most linear part of the loop being somewhere other than (0,0) as an origin.
I haven't seen any instance where increasing DC current in the transformer increases its linearity. Except in the instance where a DC current in one winding is used to offset a DC current in the other winding.
yes, and whats more linear than air?
A vacuum. :)
Think about it though. Take one of your transformers and remove the core. Is it more linear without the core than it is with the core?
but above the knee of the loop the core is quite linear (your perm is very low, but thats a different issue)
Well sure. Once you saturate the core, it's not doing anything after that so it can't very well be nonlinear.
And yes, the permability drops and you have the equivalent of an air core transformer. But again, remove the core of one of your transformers and tell me how linear it is.
Without the high permeability of an unsaurated core, the coupling between the two windings drops considerably and distortion increases considerably as well. The primary inductace drops just as considerably and your low frequency response goes out the window.
it will be perfectly linear without the core, i think thats the point :-) your perm will be 1, but thats a different issue.
Let's see if it's perfectly linear without the core. Again, remove the cores of one of your transformers and see if it's at least AS linear as it was with the core.
sure it may be picking an nit, but i think its an important nit to understand. adding a magnetic core is adding nonlinearity, much of the time the increase in perm more that offsets any nonlinearity introduced by the core hence we see lots of iron cored (or nickel :-) outputs and very few air cored ones.
The increase in permeability also more than offsets the nonlinearity of piss-poor coupling. :)
se
(SE)It will no more return to the bias point than an unbiased transformer would return to 0,0.then what is he point of the application of a -Bdc in pulse core use? the use of the -Bdc is to combat remenance which is exactly the reason we never return to the (0,0) point. i wish i could find some better references on it, but it is a documented practice, it just doesn't see widespread uses since an airgap just happens to be cheaper.
(se)No, it won't. The magnetic field due to the DC current never collapses. It magnetizes the core only once and that's it. After that, any magnetization or demagnetization of the core is brought about by the signal and everything else happens as usual but it happens about the Bdc point rather than about the 0,0 point.
but what happens when the signal is removed? with a Bdc point the core remains at that point. without the Bdc "reference" the core ends up at a point based on its previous history, it could be (0,1000) it could be (0, 10,000) it just moves about.
(se)I haven't seen any instance where increasing DC current in the transformer increases its linearity.
well can we agree that the nonlinear factor in a transformer (wrt the core) is the variability of the perm?
(se)Think about it though. Take one of your transformers and remove the core. Is it more linear without the core than it is with the core?
depends what you look at... i understand the simple answer, but that possibly overlooks the big picture.
(se)Well sure. Once you saturate the core, it's not doing anything after that so it can't very well be nonlinear.
right... it might not do the job as intended, but remember we are already breaking rulse here and finding good results (or rumor has it :-)
(se)And yes, the permability drops and you have the equivalent of an air core transformer. But again, remove the core of one of your transformers and tell me how linear it is.
very linear... but again that depends on the winding geometry, and we are talking mor about core behavior... two very different (but equally important) issues.
(se)Without the high permeability of an unsaurated core, the coupling between the two windings drops considerably and distortion increases considerably as well.depends what you look at... consider that many cores cease to be effective above a few khz so the perm is gone anyways.
(se)The primary inductace drops just as considerably and your low frequency response goes out the window.
sure, but again we know that and take that into consideration when looking at the overall design... it has little to do with the concept at hand...
(se)Let's see if it's perfectly linear without the core. Again, remove the cores of one of your transformers and see if it's at least AS linear as it was with the core.
what do you define as linear??? and again it depends on the design. lets take a 1:1 bifilar tranformer and i'll say it will be much more linear without the core than with it.
dave
ds says < < lets take a 1:1 bifilar tranformer and i'll say it will be much more linear without the core than with it.> >Well dave this is the same as saying a hitter can swing faster without a bat in his hands. The only problem is he will not hit very well. Your coreless trans will be useless in any amplifier application. I guess it will take after it's designer.
regards
Bob
then what is he point of the application of a -Bdc in pulse core use?I don't know. I'm not familiar with the technique and what the reasoning is behind it.
the use of the -Bdc is to combat remenance which is exactly the reason we never return to the (0,0) point.
Is it or is that just a guess on your part?
I don't see how the DC bias can combat remnance as the restorative force of the field produced by the DC doesn't get returned to the system in order to restore anything.
. i wish i could find some better references on it, but it is a documented practice, it just doesn't see widespread uses since an airgap just happens to be cheaper.
Well you haven't given any reference so far. :)
but what happens when the signal is removed?
Same thing that happens when the signal is removed from a non-biased core except where it ends up will be offset depending on how much of a bias there is.
with a Bdc point the core remains at that point. without the Bdc "reference" the core ends up at a point based on its previous history, it could be (0,1000) it could be (0, 10,000) it just moves about.
The core would only remain at the Bdc point if the Bdc point were such that the core always remained in saturation.
Tell me, where does the restorative force come from which brings the core back to the Bdc point once the signal is removed?
well can we agree that the nonlinear factor in a transformer (wrt the core) is the variability of the perm?
Well I guess it would have to be wouldn't it?
depends what you look at... i understand the simple answer, but that possibly overlooks the big picture.
What I'm looking at is non-linear distortion. So what "big picture" are you referring to which would have a transformer which has considerable non-linear distortion but does not distort?
very linear... but again that depends on the winding geometry, and we are talking mor about core behavior... two very different (but equally important) issues.
Ok. I agree that a core which always remains saturated would not cause distortion.
But so what? If the core's saturated, it's not going to behave anything like it would without it being saturated so what's the point?
depends what you look at... consider that many cores cease to be effective above a few khz so the perm is gone anyways.
But my system doesn't include speakers that are made up of nothing but a tweeter.
sure, but again we know that and take that into consideration when looking at the overall design... it has little to do with the concept at hand...
Ok...
what do you define as linear???
I define as linear that which is linear. Do we have to consult Webster's on that? If so, how 'bout this one:
"having or being a response or output that is directly proportional to the input."
and again it depends on the design. lets take a 1:1 bifilar tranformer and i'll say it will be much more linear without the core than with it.
Ok. But of what practial use would it be?
se
(ds)the use of the -Bdc is to combat remenance which is exactly the reason we never return to the (0,0) point.(se)Is it or is that just a guess on your part?
Not a guess… I have only seen a few references to this, nothing that goes deeply into it since even though it works, there are other "better" (read cheaper) ways to do it.
(se)I don't see how the DC bias can combat remenance as the restorative force of the field produced by the DC doesn't get returned to the system in order to restore anything.
Ok simple test… bias a core with a Bdc, now hit the core with a negative unidirectional signal to take it to the –Br point. Now apply a very small signal, where will the minor loop center on? The Bdc point or the –Br?
(se)Well you haven't given any reference so far. :)
nor have you :-), the only place we seem to disagree is on what happens with the remenance when there is a Bdc present. And that is a subject I have never seen mentioned specifically anywhere.
Reuben lee touches on the remenance in pulse cores, but that’s about it.
(ds)but what happens when the signal is removed?
(se)Same thing that happens when the signal is removed from a non-biased core except where it ends up will be offset depending on how much of a bias there is.
OK lets try this… it might get confusing but try to follow.
When we have a minor loop centering upon Bdc, and if suddenly the signal is removed say at the Bmax of the minor loop, we know the core will return to the vertical axis of Hdc. Now is it your contention that due the the remenance, the core will actually remain magnetized to the new point equivalent to Brdc which will be the point in the loop where the top of the minor loop crosses the Hdc axis and not at the original point Bdc.
(se)Tell me, where does the restorative force come from which brings the core back to the Bdc point once the signal is removed?
If we are still talking about remenace here, remember that remenace only exists because of a DC offset to begin with. It’s a catch 22, most if not all things we look at wrt tubes uses symmetrical cyclical AC which has no DC component, but what about music? We say its AC but dosn’t it also have a DC component?
(se)But my system doesn't include speakers that are made up of nothing but a tweeter.
This is what always gets us into trouble, you can’t talk in general concepts then refute the concept with a specific case. The points you are making only hold true when you attach numbers to the concept. What I am suggesting is that we first agree on the concept, then we will know how to properly attach the numbers and draw the correct conclusions. This all came about by placing DC through a nongapped transformer. Look at the graph of the incramental permeability of a core vs. a dc magnetizing force. Now if we are talking linearity, I will have to say that the perm is most linear with a higher value of H. Now if we are talking values its obvious that the H=0 gives you the biggest numbers but is decidedly less linear.
What confuses things is when someone puts in their own numbers and draws what I see as an incorrect conclusion. The higher but less linear perm often makes a complete circuit more linear, but that’s a number thing and not a conceptual one.
(ds)and again it depends on the design. lets take a 1:1 bifilar tranformer and i'll say it will be much more linear without the core than with it.
(se)Ok. But of what practial use would it be?
That isn’t the point now is it?? Its nice that we have made it this far into the discussion without introducing specifics (numbers) now all of the sudden you want to say if there isn’t enough inductance, the bass will roll off? And if the bass rolls off the circuit is non-linear?
Fine… I agree but it has nothing to do with anything I have been trying to talk about. I could easily state that just wind your bifilar air cored transformer with lots of inductance but then we would just rehash the obvious and get into the choice of compromise :-)
in any event... lets get back on track...
looking at the above graph, can we agree that increasing the value of H increases the linearity of the perm.
since perm relates directly to inductance, if we assume that the final result gives us adequate inductance (judged by ear) could the increased linearity possibly explain why some people have found good results with PP iron SE??? (please note we have no idea what value of H they are using so we have to assume there is enough inductance to give the desired bass response since no mention is made of the roloff.)
dave
in any event... lets get back on track...Ok. I've got a full day of painting ahead of me so I don't have the time to address the rest of your post right now anyway. :)
looking at the above graph, can we agree that increasing the value of H increases the linearity of the perm.
That's not how I see it.
Looks to me that it's most linear at Hdc = 0 and gets progressively more nonlinear as Hdc increases.
se
how do you figure that???dave
how do you figure that???I guess I could ask you the same thing. :)
Both the x and y axes in your plot are logarithmic. That being the case, as far as I'm aware, a linear function will produce a straight line. And looking at the plot, Hdc = 0 is the closest approximation to a straight line and would therefore be the most linear.
Here, I overlaid some straight lines on the plot to make it more obvious.
se
to be fair i chose around 12KG which is a fair bit less than Bmax.what next... you can only use a 3KG signal for it to be linear.... if i only had a 3KG max flux on the core i would use nickel :-) and then of course gap it to get the linearity rather than use a dc offset.
dave
OK so the increase in perm is linear, but who cares?to me a linear perm would be a straight horizontal line.
dave
OK so the increase in perm is linear, but who cares?Who cares? I thought you did. You were the one talking about linearity, no?
to me a linear perm would be a straight horizontal line.
A linear permeability would be a straight line. Period. Doesn't matter if it's horizontal or not. And while the plot with the highest Hdc is the more horizontal of the bunch, it's not the straightest of the bunch and therefore is not as linear.
Hdc = 0 is the most linear.
se
you are completely missing my point.when we (at least me) refer to linear perm, it is in reference to it being the same value wrt AC excitation any derivation from that is a nonlinearity. fabricate two inductors based on the curves one at h=0 the other at H=10.
now use these inductors to form a simple LR divider. which one will more closely follow what an ideal (linear) inductance will give?
to me what i consider linear perm is the same as linear inductance, a situaiton where the value does not change with applied voltage or frequency...
dave
dave
you are completely missing my point.Ok...
when we (at least me) refer to linear perm, it is in reference to it being the same value wrt AC excitation any derivation from that is a nonlinearity.
Then I'd say you're using the wrong terms to describe what you mean.
Seems that instead of linear you mean static and instead of nonlinear you mean dynamic.
And with respect to the graph you posted, Hdc = 0 is less static but more linear while Hdc = 10 is more static but less linear.
In any case, what's your point? What's the "problem" you're trying to address here? Or more to the point, exactly what effect on the signal are you trying to minimize?
se
OK its a mixup in terms.so when people refer to linear inductance, they really mean static inductance?
and when compared to the static (ideal horizontal line) which is more linear? not to mention it seems to (incorrectly) be the trend in audio to look only at the big numbers, so if we look at the "linearity" and i mean "linearity" from 0-15KG which is more linear?
as for the point, i was suggesting one possible reason why a core with a DC offset might sound better.... and the suggestion is because the inductance is more "static" :-) my whole motivation for this is based on several empeircal reports of improved behavior from a DC offset, and i just don;t buy the easy answer "you like distortion"
dave
so when people refer to linear inductance, they really mean static inductance?No, they mean an inductor which has linear behavior.
and when compared to the static (ideal horizontal line) which is more linear?
Again, the more linear is that which most closely approximates a straight line.
If you have a perfectly straight horizontal line and a perfectly straight line at say 45 degrees, they're both equally linear.
If you have a perfectly straight line at 45 degrees and a less than perfectly straight line more horizontally, the line at 45 degrees will be the more linear.
And again, going back to your graph, Hdc = 0 is the more linear than Hdc = 10.
as for the point, i was suggesting one possible reason why a core with a DC offset might sound better....
Ok. And I'm simply saying that DC offset on the core increases nonlinearity and subsequently distortion.
...and the suggestion is because the inductance is more "static" :-) my whole motivation for this is based on several empeircal reports of improved behavior from a DC offset...
Ok. But I haven't seen any evidence that the "improved behavior" is any sort of objective technical improvement. It's been well-established for perhaps a century now that DC on a transformer's primary or secondary increases signal distortion.
...and i just don;t buy the easy answer "you like distortion"
Why? That might well be the answer. If you're going to reject that possibility out of hand, then you're obviously more concerned with confirming some preconcieved belief than getting at the truth.
That being the case, I think I'll just go back to enjoying listening to music with a bit of DC bias on my transformers and not really care why it sounds better to me.
Have a great weekend!
se
(se)Ok. But I haven't seen any evidence that the "improved behavior" is any sort of objective technical improvement. It's been well-established for perhaps a century now that DC on a transformer's primary or secondary increases signal distortion.well established by whom? what has been established is the reduced perm, hence the reduced inductance for a given source increases distortion if you don;t account for the reduction in perm.
(se)Why? That might well be the answer. If you're going to reject that possibility out of hand, then you're obviously more concerned with confirming some preconcieved belief than getting at the truth.
no, it just makes no sense to me.... having too little inductance causing distortion makes sense, and that could explain a pleasing distortion, but discussion of linear perm is much more "pure" and untainted from the numbers that limit an arguement.
i really don't care, like you i enjoy what i like and feel no need to make excuses, and won;t apaologize for what i like, but that doesn;t stop me from trying to figure out why.
(se)That being the case, I think I'll just go back to enjoying listening to music with a bit of DC bias on my transformers and not really care why it sounds better to me.
its this percise belief that made me want to have the discussion with you in the first place.... thanks.
dave
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2025, Audio Asylum, all rights reserved ]
