Moses
Well-known member
Im now *slightly* confused about the LL7902.. In order to convert stereo to MS, would I need one or two of these transformers? I may have misunderstood, but I thought only one would be required?
Mo
Mo
Transformers..
- Thread starter Moses
- Start date
Help Support GroupDIY Audio Forum:
Jonte Knif
Well-known member
You need two.
There is no fundamental difference between these Lundahl trannies we have "analyzed" They all got so called C-cores, which contain two sets of coils on the opposite sides of it. We have called them (the sets) bobbins or coils. Winding refers to one part of the coil. There are two windings per bobbin in some, four in 7902. If I only had a picture now!
Hmm. Look at the datasheet . There is only one core, denoted by the two vertical lines. The windings for coil 2 are presented by dotted lines, to make it clear (or confuse you?) that they belong to another coil.
Why is this important? When combining and connecting windings you usually want to connect some winding from coil 1 with another from coil 2. This produces nice symmetry. Look at the connection alternatives, and draw some of them to the transformer symbol and you'll see. L2 and R2 are good to look at.
In MS coding you would put all primary windings in series and two sets of secondary windings, also pairs of series connected windings. So, basically L4/R2 from the alternatives.
If you'd prefer lower copper losses you'd have series/parallel primary and parallel secondaries.
This is getting complicated to explain. Draw the connections from the data sheet and meditate a while.
You can also think like this: In many (good) transformers there are multiple secondary/primary layers, but you don't see them from outside or in the data sheets. Lundahl has the habit of taking all winding ends to pins. Not always, but sometimes, like in 79xx. This gives freedom and alternatives, but causes confusion. I like it a lot! Some of my products simply would not exist without them.
If I would build an MS box for regular mixing tasks, I wouldn't mind using something else, be it Lundahl or other good stuff, but for mastering studio purposes only 7902 (or 7903) will qualify. 7903 is sometimes very handy, it is actually a hybrid between 7901 and 7902 as you can see from the copper resistances.
-Jonte
There is no fundamental difference between these Lundahl trannies we have "analyzed" They all got so called C-cores, which contain two sets of coils on the opposite sides of it. We have called them (the sets) bobbins or coils. Winding refers to one part of the coil. There are two windings per bobbin in some, four in 7902. If I only had a picture now!
Hmm. Look at the datasheet . There is only one core, denoted by the two vertical lines. The windings for coil 2 are presented by dotted lines, to make it clear (or confuse you?) that they belong to another coil.
Why is this important? When combining and connecting windings you usually want to connect some winding from coil 1 with another from coil 2. This produces nice symmetry. Look at the connection alternatives, and draw some of them to the transformer symbol and you'll see. L2 and R2 are good to look at.
In MS coding you would put all primary windings in series and two sets of secondary windings, also pairs of series connected windings. So, basically L4/R2 from the alternatives.
If you'd prefer lower copper losses you'd have series/parallel primary and parallel secondaries.
This is getting complicated to explain. Draw the connections from the data sheet and meditate a while.
You can also think like this: In many (good) transformers there are multiple secondary/primary layers, but you don't see them from outside or in the data sheets. Lundahl has the habit of taking all winding ends to pins. Not always, but sometimes, like in 79xx. This gives freedom and alternatives, but causes confusion. I like it a lot! Some of my products simply would not exist without them.
If I would build an MS box for regular mixing tasks, I wouldn't mind using something else, be it Lundahl or other good stuff, but for mastering studio purposes only 7902 (or 7903) will qualify. 7903 is sometimes very handy, it is actually a hybrid between 7901 and 7902 as you can see from the copper resistances.
-Jonte
Entropia Cub
Well-known member
Kiitos Jonte for all those replies, a lot of knowledge inside. Even if I'm a bit disappointed & embarassed by the trust you revealed : I've bought two LL1540 for the coding part of a future M-S matrix and two LL1517 for the output decoding huhuhu... Bleh, my bad : I've just realised that an input of a M-S matrix is also an output for the device inserted ! :
I'll find another task for these transformers anyway !
A bit more in the topic, how could I detect a too big copper loss with transformers in a given application ? Is there a kind of proportion like Z application / Z tx and if this is bigger than 5 (by example), I should avoid this transformer in this application ?
Kiitos again for your advices !
I'll find another task for these transformers anyway !
A bit more in the topic, how could I detect a too big copper loss with transformers in a given application ? Is there a kind of proportion like Z application / Z tx and if this is bigger than 5 (by example), I should avoid this transformer in this application ?
Kiitos again for your advices !
CJ
Well-known member
the ll79 series, well, they are problem solvers, designed for high level, not exactly high fidelity.
i did not care much for the sound of the 7903.
and with more core, you have more core loss, as you need more magnetizing current to get it off the ground, which eats your signal.
one of the reasons that the inductance does not exactly divide by 4 when you use half the winding is that you are hitting the core in a fashion that causes it's perm to go up a bit, depending on the type of lams.
long tall sally is just a term made up by Doc Hoyer, only authorized Mac and Marantz winder alive.
as a props thing, he must be 87 this year.
uses super glue, why didn't i think of that?
here is what is meant by a "long tall sally", this is your typical LL construction, but don't forget the glass C cores, that is their thang right now.
7903>
more pics if interested:
this 10k rise is just like the peerless input...
http://www.vacuumbrain.com/The_Lab/Lundahl/7903_plot.jpg
computer windings are tight>
http://www.vacuumbrain.com/The_Lab/Lundahl/7903_guts.jpg
i did not care much for the sound of the 7903.
and with more core, you have more core loss, as you need more magnetizing current to get it off the ground, which eats your signal.
one of the reasons that the inductance does not exactly divide by 4 when you use half the winding is that you are hitting the core in a fashion that causes it's perm to go up a bit, depending on the type of lams.
long tall sally is just a term made up by Doc Hoyer, only authorized Mac and Marantz winder alive.
as a props thing, he must be 87 this year.
uses super glue, why didn't i think of that?
here is what is meant by a "long tall sally", this is your typical LL construction, but don't forget the glass C cores, that is their thang right now.
7903>
more pics if interested:
this 10k rise is just like the peerless input...
http://www.vacuumbrain.com/The_Lab/Lundahl/7903_plot.jpg
computer windings are tight>
http://www.vacuumbrain.com/The_Lab/Lundahl/7903_guts.jpg
> "Long Tall Sally" How did it get the name? (I guess it is obvious to native speakers...)
"Long Tall Sally" is a rock and roll 12-bar blues song recorded by Little Richard and everybody else.
http://en.wikipedia.org/wiki/Long_Tall_Sally
http://www.youtube.com/watch?v=QFL047fmsgg
http://www.youtube.com/watch?v=WhxxQn5HsQs
http://www.youtube.com/watch?v=6ztELZl3UBg
http://www.youtube.com/watch?v=rcI6_EonmGA
I like to think I am a native speaker but I don't really know what it is about.
"Long Tall Sally" is a rock and roll 12-bar blues song recorded by Little Richard and everybody else.
http://en.wikipedia.org/wiki/Long_Tall_Sally
http://www.youtube.com/watch?v=QFL047fmsgg
http://www.youtube.com/watch?v=WhxxQn5HsQs
http://www.youtube.com/watch?v=6ztELZl3UBg
http://www.youtube.com/watch?v=rcI6_EonmGA
I like to think I am a native speaker but I don't really know what it is about.
CJ
Well-known member
well, we could have called it the boney maronie if you wanted.
equal time for the hillbillys, like larry williams
sorry larry, a russian folk dancer was all i could find.
leakage, ok, a lot of theory being put forth, but what is it really?
well, technically, it is (1-k^2)/L, where k is the couplig factor, which is always less than 1, unless you are using liquid nitrogen and mu metal in outer space at near the spped of light, ok
so what the heck good is a formula with k if we do not know what the hell k is or how to determine it?
well, that is why we have labs, to put formulas out, leakage bassed on something we know, like coil geometry/
well, all you need to know is N^2, thats turns squared, the most dominant term in the equation, as it is quadratic.
so instantly, we see that a 600 ohm primary, which is never over 600 turns unless it is a wallmart special, and if it is a tamura 80 core, more like 100 turns, so compare that with a 3000 turn secondary and you see that a 600 ohm is no biggy, there is leakage, but nothing compared to the 10 k stuff.
leakage can be minimized by sectioning, but in a 100 turn section, you only need 2 layers, so whats the use?
geometry is also a factor, two balanced coils on a lohng, thin core will provide the least leakage.
so the lundahl has two things going for it, coil geometry and few turns due to the swedish steel. pr whatever they mix, i think everybody in the universe is using super q by mag met, except maybe oliver, who probably has an illegal foundry in his basement, those dang germans...
ow with all this talk, you could have A/B ed the xfmr, one primary vs two, about 100 times over,
i'm jus sayin...
equal time for the hillbillys, like larry williams
sorry larry, a russian folk dancer was all i could find.
leakage, ok, a lot of theory being put forth, but what is it really?
well, technically, it is (1-k^2)/L, where k is the couplig factor, which is always less than 1, unless you are using liquid nitrogen and mu metal in outer space at near the spped of light, ok
so what the heck good is a formula with k if we do not know what the hell k is or how to determine it?
well, that is why we have labs, to put formulas out, leakage bassed on something we know, like coil geometry/
well, all you need to know is N^2, thats turns squared, the most dominant term in the equation, as it is quadratic.
so instantly, we see that a 600 ohm primary, which is never over 600 turns unless it is a wallmart special, and if it is a tamura 80 core, more like 100 turns, so compare that with a 3000 turn secondary and you see that a 600 ohm is no biggy, there is leakage, but nothing compared to the 10 k stuff.
leakage can be minimized by sectioning, but in a 100 turn section, you only need 2 layers, so whats the use?
geometry is also a factor, two balanced coils on a lohng, thin core will provide the least leakage.
so the lundahl has two things going for it, coil geometry and few turns due to the swedish steel. pr whatever they mix, i think everybody in the universe is using super q by mag met, except maybe oliver, who probably has an illegal foundry in his basement, those dang germans...
ow with all this talk, you could have A/B ed the xfmr, one primary vs two, about 100 times over,
i'm jus sayin...
![Soldering Iron Kit, 120W LED Digital Advanced Solder Iron Soldering Gun kit, 110V Welding Tools, Smart Temperature Control [356℉-932℉], Extra 5pcs Tips, Auto Sleep, Temp Calibration, Orange](https://m.media-amazon.com/images/I/51sFKu9SdeL._SL500_.jpg)
Jonte Knif
Well-known member
hehe, well they do solve problems, but I think you'd better just call them general purpose trannies. About the sound I will not start to argue. Didn't work for you for what ever reason.
Your measurement sure looks bad. Could you shed some light on the conditions, because I have not been able to get anything like that. Low source, high load? Everything balanced? I would like to duplicate it.
I have used 7903 as line input in several combinations, and with right impedances it provides ruler flat resp from 10 to 100k. (-0,5dB)
But:
We have been talking about MS coding. And I boldly claimed that 7902 is superb. (forget the 7903) High impedance trannies are not good all around things, they have to be properly terminated and hence work best as line input.
The problem with an MS box is, that you never know the source and load impedance it might some day "see". So even 7902 should be used with some parallel connections to make it really all around. I took a series of measurements with AP, going through sources from 40 to 600 Ohms balanced, and loads 600 ohms and 100k, balanced. If you are interested in the curves I can post them but they look really boring.
All measurements are without any zobels. So, with some work they can be improved if really needed. (I doubt) Capacitive load was quite low, only 1 meter cable and the pretty high C of AP input.
Primary was always L2. Secondary R2 or in coding 1+1.
Bass is good, no surprices there.
Worst case treble was of course with 40Ohm source and 100k load. +1.2dB @ 120kHz. (AP doesn't go higher) Good enough?
With 600 Ohm load the treble rolls of nicely, -1.3 dB @ 120k.
600 Ohm / 600 Ohm, perfect.
Two examples in MS coding.
40 Ohms, 100k. +1.4 dB @ 120k
150 Ohms, 600 Ohms, -0.5dB @ 120k
I hope this comforts people a bit. Still marginally worse results can be achieved with unbalanced source or load.
-Jonte
CJ
Well-known member
i have a plot of the 7903 that is flat, but i had to really damp it down, as that thing is a stiff source once you get it going.
you can make any xfmr run flat, just use a low level and a zobel, so in order to distinguish the character of the coil/core assy, the tests are first done unloaded.
the 30db is nice, and lundahl is always conservative with thier specs, but you do pay a price.
maybe it was too clean, thats it! not enuff distortion on the bottom end.
i do not use log paper on those graphs, so the peaks get multiplied by ten, or shrunk by ten, see, log paper was for engineers who wanted to hide transformer peaks.
you can make any xfmr run flat, just use a low level and a zobel, so in order to distinguish the character of the coil/core assy, the tests are first done unloaded.
the 30db is nice, and lundahl is always conservative with thier specs, but you do pay a price.
maybe it was too clean, thats it! not enuff distortion on the bottom end.
i do not use log paper on those graphs, so the peaks get multiplied by ten, or shrunk by ten, see, log paper was for engineers who wanted to hide transformer peaks.
Moses
Well-known member
Excuse my silence!! Im still here, just trying to ingest all the information that you guys are sharing (Thanks!!)...
Mo
Mo
CJ
Well-known member
another tidbit, if you have a 1:1 600:600, you want the pri coupled the best, this will minimize leakage, so in a 4 section xfmr, 1:1, triad uses section 2 and 3 as their primary, which has better symmetry, not as close to the core, so you do not get as good as linkage as the 1st winding, but slot 2 and 3 together is better than 1 and 4, because section 4 will have the worst coupling, as it is located 3 miles away from the center leg, which is coupled best to the inner windings.
the forth winding is usually the dog, it has a little less inductance than the rest, the most leakage, the most dcr, the most capacitance, the most distortion, it is really the black sheep of the 1:1, so if you want to avoid a winding on a F core, make it #4.
less distortion with the two primaries close to each other, a little better shielding from noise, as you can put a copper shield in there, which they do, and the dcr is more even, which is good for longitudal noise in a balanced system.
now i remember, that is how you get different inductance division other than 4 when you measure coils in an F core situation, dig this,
if you measure 1 and 2 together, you will get the highest henries.
if you measure 3 and 4, you will get the least.
so if you measure 3 and 4 and get 8 henries, you might get more than 2 henries from the first section, as it has the best coupling being close to the core, least dcr, etc.
so you might get 2.25 henries instead of 2.
the shields and un used copper windings, since copper is non magnetic, do not help the inductance figures of windings located outside their proximities either.
the forth winding is usually the dog, it has a little less inductance than the rest, the most leakage, the most dcr, the most capacitance, the most distortion, it is really the black sheep of the 1:1, so if you want to avoid a winding on a F core, make it #4.
less distortion with the two primaries close to each other, a little better shielding from noise, as you can put a copper shield in there, which they do, and the dcr is more even, which is good for longitudal noise in a balanced system.
now i remember, that is how you get different inductance division other than 4 when you measure coils in an F core situation, dig this,
if you measure 1 and 2 together, you will get the highest henries.
if you measure 3 and 4, you will get the least.
so if you measure 3 and 4 and get 8 henries, you might get more than 2 henries from the first section, as it has the best coupling being close to the core, least dcr, etc.
so you might get 2.25 henries instead of 2.
the shields and un used copper windings, since copper is non magnetic, do not help the inductance figures of windings located outside their proximities either.
Marik
Well-known member
CJ said:
I am yet to hear a good sounding DU core transformer, including venerable 2480. For sound my favorite is hi Ni torroidal ribbon core. I got a few 8:1 here, primary is sectionized, L=80H DCR=70.5 Ohm, and sec. DCR=9 Ohm. Transparency and amount of details is quite amazing.
Best, M
Jonte Knif
Well-known member
Marik:
Do you have some suggestions for MS? Toroidal ribbon core sure would be magnetically pretty ideal.
To derail further: (why on earth did we start to talk about sound?)
Every persons opinion is important whether his work is welding, mixing or mastering, but I'm just wondering that are those mastering engineers I know and work with half deaf because they really like gear with tons of Lundahl DU trannies in them, and feel that they are very detailed. And mastering folks work all the time bypassing individual pieces of gear, so I have the feeling that they have to notice if there is something "wrong" The comparison is always very direct and fast.
I would very much like to see a lot of well performed ABX tests about trannie sounds. They are just a bit time consuming, but provide information instead of opinions. (not that opinions wouldn't be fine, but at least some opinions can often be interpreted as "stuff X is crap, stuff Y gets you into audio heaven" which is a bit sad)
We did one test, comparing normal C core Lundahl to amorphous C core. Same model, only different core. The copious samples were delivered to 4 persons to be listened blind and 2 persons were able to get very consistent results. (Amorphous beats silicon iron according to our test, but that is not the point now.)
An essential part of DIY is not objective, off course, and I guess presenting tables of ABX results for every component would decrease creativity and fun. That is not the spirit of DIY. But very often I wonder why some people here are so hostile towards audiophools not noticing that they are not very far from that world _in terms of the methodology_ used to provide information.
And now I notice that I'm really starting to sound like sales representative to Lundahl, which I'm not I just happen to have a lot of experience of them. And I know they are not made to "cost no object" standards, and can be beaten (but often are not). Just very fine cost conscious engineering.
peace,
Jonte
Do you have some suggestions for MS? Toroidal ribbon core sure would be magnetically pretty ideal.
To derail further: (why on earth did we start to talk about sound?)
Every persons opinion is important whether his work is welding, mixing or mastering, but I'm just wondering that are those mastering engineers I know and work with half deaf because they really like gear with tons of Lundahl DU trannies in them, and feel that they are very detailed. And mastering folks work all the time bypassing individual pieces of gear, so I have the feeling that they have to notice if there is something "wrong" The comparison is always very direct and fast.
I would very much like to see a lot of well performed ABX tests about trannie sounds. They are just a bit time consuming, but provide information instead of opinions. (not that opinions wouldn't be fine, but at least some opinions can often be interpreted as "stuff X is crap, stuff Y gets you into audio heaven" which is a bit sad)
We did one test, comparing normal C core Lundahl to amorphous C core. Same model, only different core. The copious samples were delivered to 4 persons to be listened blind and 2 persons were able to get very consistent results. (Amorphous beats silicon iron according to our test, but that is not the point now.)
An essential part of DIY is not objective, off course, and I guess presenting tables of ABX results for every component would decrease creativity and fun. That is not the spirit of DIY. But very often I wonder why some people here are so hostile towards audiophools not noticing that they are not very far from that world _in terms of the methodology_ used to provide information.
And now I notice that I'm really starting to sound like sales representative to Lundahl, which I'm not
I just happen to have a lot of experience of them. And I know they are not made to "cost no object" standards, and can be beaten (but often are not). Just very fine cost conscious engineering. peace,
Jonte
Marik
Well-known member
Jonte Knif said:
What is the source impedance and what an overload (esp. on low frequencies) requirements? To do 10K with low copper losses will be hard. It can be wound quadrifilar, or sectionized, or both. For high levels most likely will need to be whether low Ni or M6.
I actually use MS a LOT and personally prefer active solution. The main reasons are low cost, and ability to change the "width". Or passive Neumann resistive circuit (if you don't need balanced) might be also a good and cheap way of doing it.
Most of the time we know good or bad only in comparison. Indeed sound is a very subjective thing and is rather an opinion, and of course, if the person in his/hers life heard only Lundahl and/or Cinemag then the opinion worth nothing. However, if a person has good ears and is in habit of making LOT"S of comparisons and supports the opinion with a measured data, then it can have a lot of value. When we talk transformers something like impedance and phase curves, their uniformity, etc. etc. etc. can give a very good information and hint as for the actual sound.
I cannot understand what they mean by "same model, only different core". Do they have the same winding? but then amorthous has much higher permeability than silicon, so the same winding will give much higher inductance, and with that core it'll have lower overload. Or they adjust for those? but then they will need to have less turns (then they can use thicker wire, with less losses), much bigger core, etc. Enough to make a difference by itself, but definitely, amorthous core has different sonic signature. I just cannot see how it is annealed, as they first do the bobbins and then later wind the ribbon strip through them... ??? ??? ???
Best, M
Jonte Knif
Well-known member
Hi Marik,
I was just thinking if you have some particular manufacturer and model in mind, because there are people around here making their step into passive MS-coding, and we might help them.
They have to judge themselves what is ultimately the goal, but I would like to see fairly low copper losses, and immunity to different sources/loads. 7902 performer well in those respects when parallel/series configured. Headroom, well, as much as possible...
Yes.
Overload is lower, yes. And also due to the lower max flux.density, something like 1.4T compared to 2.1T. In single ended trannies the inductance will actually be lower, because the air gap has to be wider. But we had PP stage.
In this particular test overload was of no concern, there was a lot of margin. And people mainly commented things which I associate technically to high frequencies.
No, that would indeed make the test invalid. I was trying to objectively find out if I can offer AM cores as an alternative, because they do cost quite a lot. Of course it is totally subjective what they are worth but at least they make a difference. And note: I could not measure a difference. Null tests revealed that something funny is going on, but at a fairly low level.
??? No. All their C-cores are first wound, then cut, then precision ground (I guess annealing comes at this point, never thought about it) and simply put together with the coils. I have noticed some variation in inductance due to imperfect surface mating, but it is marginal, so I guess they have pretty good machines to bring the halves together. Oh yes, and actually they put an air gap to all their C-core trannies to allow for some DC unbalance. In this case several mA. This might actually explain why I didn't notice a significant difference in inductance because the air gap dominates.
-Jonte
I was just thinking if you have some particular manufacturer and model in mind, because there are people around here making their step into passive MS-coding, and we might help them.
They have to judge themselves what is ultimately the goal, but I would like to see fairly low copper losses, and immunity to different sources/loads. 7902 performer well in those respects when parallel/series configured. Headroom, well, as much as possible...
Yes.
Overload is lower, yes. And also due to the lower max flux.density, something like 1.4T compared to 2.1T. In single ended trannies the inductance will actually be lower, because the air gap has to be wider. But we had PP stage.
In this particular test overload was of no concern, there was a lot of margin. And people mainly commented things which I associate technically to high frequencies.
No, that would indeed make the test invalid. I was trying to objectively find out if I can offer AM cores as an alternative, because they do cost quite a lot. Of course it is totally subjective what they are worth but at least they make a difference. And note: I could not measure a difference. Null tests revealed that something funny is going on, but at a fairly low level.
??? No. All their C-cores are first wound, then cut, then precision ground (I guess annealing comes at this point, never thought about it) and simply put together with the coils. I have noticed some variation in inductance due to imperfect surface mating, but it is marginal, so I guess they have pretty good machines to bring the halves together. Oh yes, and actually they put an air gap to all their C-core trannies to allow for some DC unbalance. In this case several mA. This might actually explain why I didn't notice a significant difference in inductance because the air gap dominates.
-Jonte
Marik
Well-known member
Jonte Knif said:
There are no torroidal transformers readily available, either for MS or microphone outputs/inputs. If there is enough interest I can run those.
If the sound quality and lowest copper losses are your goals then you should define what's the "immunity to different sournces" is. It is one thing to run it from emitter follower, or opamp, and completely different from say, 6SN7. Also, "as much headroom as it is possible" doesn't tell a thing. Sure enough, you can use a huge M6 core and then you have lotsa headroom, but then again, at the expense of other things. As always, everything depends where you want to compromise.
But generally speaking, if you can live with something like 3-5V RMS on 20Hz then something like Pri. L=20-50H with well under 100 Ohm DCR should not be a big problem.
Ah, you are talking about gapped stuff.
Did you measure impedance and phase response across the bandwidth?
The bigger iron, yes... maybe. However, the small signal ones with Amorthous core are wound with continuos strip.
Best, M
Jonte Knif
Well-known member
I measured only the primary inductance and then freq response to 120k when installed in the amp. Nothing to tell about. If there was a 0,2dB difference that might as well have been caused by minute differences in coils, normal to manufacturing tolerances. I'm pretty sure I should have more accurate system than portable system one to trace something that is going to on on the molecular level in the core.
Actually I have a couple of different trannies here in both versions at the moment and could do more tests. Perhaps I can find something.
We have been discussing general purpose MS matrices. I would say, connected to "professional line level gear" So, source can be anything up to 600 Ohms and input anything down to 600 Ohms. Electrostatic shield would perhaps not be needed. I dunno.
Interesting info about the smaller AM cores. I can not imagine how that is made. Any pictures?
But yes, the bigger (i.e. the tube amp interstage/output) are all cut, they all have a small gap for PP and bigger for SE.
Actually I have a couple of different trannies here in both versions at the moment and could do more tests. Perhaps I can find something.
We have been discussing general purpose MS matrices. I would say, connected to "professional line level gear" So, source can be anything up to 600 Ohms and input anything down to 600 Ohms. Electrostatic shield would perhaps not be needed. I dunno.
Interesting info about the smaller AM cores. I can not imagine how that is made. Any pictures?
But yes, the bigger (i.e. the tube amp interstage/output) are all cut, they all have a small gap for PP and bigger for SE.
Marik
Well-known member
Jonte Knif said:
OK, we want a source impedance 600 Ohm (although most likely folks will be running it off the op-amp), then for the worst case let's assume the -1db@10Hz. Then we've got L~20H.
IMO, if the transformer can take some 2-3V RMS@20Hz before saturation then it should be more than enough for MOST purposes and applications, so we can use medium size core, which although expensive, is not that terrible.
Because of high permeability of the core and relatively low inductance we won't need too much of the winding and capacitance should not be of issue, so we can use quadrifilar winding. I will need to make a new shuttle slider, but no biggie. The lower turns will allow using larger gage wire. With something like 32AWG we could probably get something in 10-15 Ohm range per winding. Since it is a 1+1:1+1, besides MS it can be used as 1:1, 1:2, and 2:1 fashion.
With potting the transformer dimension will be something like 3/4"x1 5/8"
Sorry, no pics. I got this info from Kevin Carter (he is a US Lundahl distributor). They use proprietary technique, with in-house custom built machinery.
Best, M
CJ
Well-known member
that glass stuff has it's own airgap, they are quite spongy, the big outputs i saw at the booth, made you want to squeeze them like a loaf of wonder bread, but the boss was there, grimacing over an AES prime rib sandwich, and we all know what those are like.
