The quality of the output transformer must be put in the right perspective; it is part (albeit a very important one) of the tube amplifier. Then we have to deal with the quality of the music source, and last but definitely not least the loudspeakers and the room.
That said, in my opinion there are three elements which play an important role in (output) transformers:
1. Core geometry; especially the difference in efficiency between EI, c-cores and toroidal cores.
The geometry of c-cores and toroidal cores is better w.r.t. alignment of grain orientation. You can search for and check specifications and find out that c-cores and toroidal cores have a better power / weight ratio. In audio transformers this means less windings which is always beneficial (capacitance).
2. Thickness of laminations; the 20-25 mu thin "tape" of amorphous and Finemet cores results in far less loss in the higher frequencies (because of lower eddy current) which is not "special" because these cores are made for high frequency applications. For audio transformers the low eddy currents is good for transparency and small signal behaviour.
3. Permeability. High permeability is good to keep the number of windings low for the needed inductance (low frequency bandwidth) and a lower number of windings is again good for low winding capacitance.
Other differences between these core materials like maximum excitation (T) have to deal with appropriately in designing audio transformers.
Now with all this we can wind silicon steel, amorphous, nickel and Finemet transformers with very comparable frequency responses, and yet they sound "different" like amplifiers, loudspeakers and so on with comparable specifications can sound "different".
Equal value resistors and capacitors can sound "different".
An audio transformer likewise has it's sonic character.
That said, in my opinion there are three elements which play an important role in (output) transformers:
1. Core geometry; especially the difference in efficiency between EI, c-cores and toroidal cores.
The geometry of c-cores and toroidal cores is better w.r.t. alignment of grain orientation. You can search for and check specifications and find out that c-cores and toroidal cores have a better power / weight ratio. In audio transformers this means less windings which is always beneficial (capacitance).
2. Thickness of laminations; the 20-25 mu thin "tape" of amorphous and Finemet cores results in far less loss in the higher frequencies (because of lower eddy current) which is not "special" because these cores are made for high frequency applications. For audio transformers the low eddy currents is good for transparency and small signal behaviour.
3. Permeability. High permeability is good to keep the number of windings low for the needed inductance (low frequency bandwidth) and a lower number of windings is again good for low winding capacitance.
Other differences between these core materials like maximum excitation (T) have to deal with appropriately in designing audio transformers.
Now with all this we can wind silicon steel, amorphous, nickel and Finemet transformers with very comparable frequency responses, and yet they sound "different" like amplifiers, loudspeakers and so on with comparable specifications can sound "different".
Equal value resistors and capacitors can sound "different".
An audio transformer likewise has it's sonic character.
today i have mounted the Monolith magnetics Sumit S9 in place of the Lundahl LL1664 . i have sold the 1664AM not because i did"t like their sound balance too clear for me ..
First thing to say with the Monolith is how big and heavy they are compare to the lundahl . so i have to rebuild my amp to fit them
first minutes on , they sound totally different than the LL1664 . the load is not the exactly the same 3.3K vs 3K i feel a little less gain .
sweet sound , a little softer too but refined highs with good details .
it's late i can't push hard because the family is sleeping. so i will make an update later. anyway i'm happy for the moment nothing disturbing with theses black boxes
First thing to say with the Monolith is how big and heavy they are compare to the lundahl . so i have to rebuild my amp to fit them
first minutes on , they sound totally different than the LL1664 . the load is not the exactly the same 3.3K vs 3K i feel a little less gain .
sweet sound , a little softer too but refined highs with good details .
it's late i can't push hard because the family is sleeping. so i will make an update later. anyway i'm happy for the moment nothing disturbing with theses black boxes
9h30 A.M it"s time for heavy listening . this Monolith S9 sound good ! i'm not a believer of burn-in in electronic at least a small warm up for the tubes so what i heard now is what i can expect from these transfo.
is it better than the Lundahl ? i don't know but i will not go back .
i like very much the LL1664 for the tonal balance , the strong bass and the detailed highs .
i already prefer the Monolith S9 for the smoothest and refined highs , the separation of instruments and the controlled low end.
but what i notice now with more volume is the "easy" distortion less sounding . impressive with complexes part of music .
so stop the Blah blah talking after the amorphous adventure , l will sell my LL1664 as well and keep the Monolith S9
so what i heard now is what i can expect from these transfo.is it better than the Lundahl ? i don't know but i will not go back .
i like very much the LL1664 for the tonal balance , the strong bass and the detailed highs .
i already prefer the Monolith S9 for the smoothest and refined highs , the separation of instruments and the controlled low end.
but what i notice now with more volume is the "easy" distortion less sounding . impressive with complexes part of music .
so stop the Blah blah talking after the amorphous adventure , l will sell my LL1664 as well and keep the Monolith S9
... after the amorphous adventure , l will sell my LL1664 as well and keep the Monolith S9
You have added so much to this discussion. You have all our thanks for shelling out the money to do the experiment! More money than some of us would have put into an OPT. But the results seem to show that you have to throw money at OPTs.
If you could tell us more about the differences between the regular LL1664 and the S9 that would be really useful. It would help us see what we would be getting for half the price.
andy , i'm not very good to explain sound subtleties in english . i have a very special amplifier , it's a E55L spud with minimum parts in the schematic . ALPS RK50 passive preamp ( grid choke , grid Bias lithium Cell) LCLC ( MKP PIO) tube power supply . so every minor change of component become dramatic and a real nightmare . sometimes you want to put it in the trash , that was what i feel the Amorphous LL1664. but i can't say that if i didn't like it in this amp i would not like it in another tube Amp . and i don't want to bash lundahl transformer too , the normal LL1664 is real performer , cheap and far better than the James 6123 .
for the Monolith S9 i didn't know this company except on some french forum where they have not good press with some stupid guru. i always prefer buy and see and for once i'm happy with my choice . they will stay in this amp for sure and end of the stories ( for the output transformer.)
for the Monolith S9 i didn't know this company except on some french forum where they have not good press with some stupid guru. i always prefer buy and see and for once i'm happy with my choice . they will stay in this amp for sure and end of the stories ( for the output transformer.
)An externally hosted image should be here but it was not working when we last tested it.
I'm not against Chinese at all. But at the same time, I'm happy when I see fellow Europeans doing things well! In these difficult times I am a strong supporter of Europe. In the UK we have to vote whether to stay in Europe this year. I'm a definite "yes". As Juanitox says "French is not a crime"!
i'm not sure to follow you in that way ..
An externally hosted image should be here but it was not working when we last tested it.
9h30 A.M it"s time for heavy listening . this Monolith S9 sound good ! i'm not a believer of burn-in in electronic at least a small warm up for the tubes
is it better than the Lundahl ? i don't know but i will not go back .
i like very much the LL1664 for the tonal balance , the strong bass and the detailed highs .
i already prefer the Monolith S9 for the smoothest and refined highs , the separation of instruments and the controlled low end.
but what i notice now with more volume is the "easy" distortion less sounding . impressive with complexes part of music .
so stop the Blah blah talking after the amorphous adventure , l will sell my LL1664 as well and keep the Monolith S9
I'm glad you like them.. Be sure to let Yves know your impressions. Your impressions parallel mine when I switched from Lundahl ITs to the MM ITs. They are much more linear than what they replaced, and have considerably flatter frequency response as well. Mine are open frames installed inside the amplifiers, eventually I will replace the input transformers in the amps with MM as well.
I am using all MM iron in my latest DHT line stage design with EML 20AM - should be interesting when I get the prototypes finished.
I think you will be surprised to discover that performance probably does improve with use. I don't know the physics but it may have something to do with domain orientation in new cores - purely magnetic in nature.
My MM IT were built to meet a specification so in fact these outperform the LL1635 they replaced having much flatter response down to 20Hz (they actually get there, the 1635 were pretty rolled off below 30Hz, more than -2dB down) As I remarked they are much larger than the Lundahls they replaced, have twice the primary inductance and with bifilar windings less leakage inductance as well. Overall amplifier FR measurements resulted in much flatter frequency response. (Better than -1dB 20Hz - 20kHz with the better IT)
When I upgrade the input transformers I'll do some measurements.
Since influence the sound is difficult to determine in a single measurement, I try to measure my amps in every way I know, to try to understand better sound differences (I need not say that often it ends unsuccessful ...).
But ... an easy measurement to perform in tube amps, and I almost never see been done, and more importantly, I've found it the ultimate explanation for the difference in some particular cases, it is the measurement of output impedance in all frequencies.
How to do it? Very simple, for those who have a sound card and the ARTA / LIMP: measures to the secondary output transformer as if measuring a speaker. This will show you the output impedance in a wide frequency range. I include here a measurement that I did with my 300B amplifier with custom made toroid OT (forgive me, in case I forgot to save with the scale to 10R, it would be easier to read). Note that the final audio freq, have a change in the ri and a consequent defasage in the damping factor, which no longer operate at 0 °. In fact, it is almost 40° in 20kHz. And that this transformer has only 3mH leakage inductance re 3k5 in the primary!
High leakage trafos will have much more than 45° in HF.
I noticed, several times, that amplifiers with high phase in HF DF sometimes sound more harsh, and amplifiers with zero or little lag in this area tend to sound smoother and softer. I do not know why (amplifier, tweeter interaction?), Or if it was just my luck ...
But this is only ONE parameter among many others to be considered, and most time supercedes this parameter.
The cure? only NFB, local or global, or an dream trafo...
I have an PP amp with PL509 with almost zero degree in 20kHz but it uses CFB directly at speakers balanced output (another custom made OT).
).But ... an easy measurement to perform in tube amps, and I almost never see been done, and more importantly, I've found it the ultimate explanation for the difference in some particular cases, it is the measurement of output impedance in all frequencies.
How to do it? Very simple, for those who have a sound card and the ARTA / LIMP: measures to the secondary output transformer as if measuring a speaker. This will show you the output impedance in a wide frequency range. I include here a measurement that I did with my 300B amplifier with custom made toroid OT (forgive me, in case I forgot to save with the scale to 10R, it would be easier to read). Note that the final audio freq, have a change in the ri and a consequent defasage in the damping factor, which no longer operate at 0 °. In fact, it is almost 40° in 20kHz. And that this transformer has only 3mH leakage inductance re 3k5 in the primary!
High leakage trafos will have much more than 45° in HF.
I noticed, several times, that amplifiers with high phase in HF DF sometimes sound more harsh, and amplifiers with zero or little lag in this area tend to sound smoother and softer. I do not know why (amplifier, tweeter interaction?), Or if it was just my luck ...
But this is only ONE parameter among many others to be considered, and most time supercedes this parameter.
The cure? only NFB, local or global, or an dream trafo...
I have an PP amp with PL509 with almost zero degree in 20kHz but it uses CFB directly at speakers balanced output (another custom made OT).
Attachments
Last edited:
Hi Alexandre,
thanks for the post! So, just to make things clearer. Do you connect the souncard to the output of the amp as in figure 3.1 of the manual?http://www.artalabs.hr/download/LIMP-user-manual.pdf
and the amplifier's input is shorted to ground?
many thanks, obrigado, Erik
thanks for the post!
So, just to make things clearer. Do you connect the souncard to the output of the amp as in figure 3.1 of the manual?http://www.artalabs.hr/download/LIMP-user-manual.pdfand the amplifier's input is shorted to ground?
many thanks, obrigado, Erik
Hi Erik,
Sorry about my long delay.
Yes, in my testing gig I used this circuit as per fig. 3.1, but only because my Asus Xonar has a decent hphones output. Some audio boards struggle to deliver full output even in 100R, and introduce errors.
Is good practice to include input clamping to not damage the audio board if some signal is applied accidentally in the amp's input (eg.: some grounding disconnection) when the audio board is connected to the amp in this manner (measuring the output Z as a speaker).
After the reading node (left in soundcard) is necessary to use good gauge wire to not introduce impedance errors, or create a overlay with the residual shorting measurement to subtract from actual measurement. And is good to maintain short wiring due to inductance involved in low impedance world.
The amplifier input is shorted to ground to minimize noise.
Sorry about my long delay.
Yes, in my testing gig I used this circuit as per fig. 3.1, but only because my Asus Xonar has a decent hphones output. Some audio boards struggle to deliver full output even in 100R, and introduce errors.
Is good practice to include input clamping to not damage the audio board if some signal is applied accidentally in the amp's input (eg.: some grounding disconnection) when the audio board is connected to the amp in this manner (measuring the output Z as a speaker).
After the reading node (left in soundcard) is necessary to use good gauge wire to not introduce impedance errors, or create a overlay with the residual shorting measurement to subtract from actual measurement. And is good to maintain short wiring due to inductance involved in low impedance world.
The amplifier input is shorted to ground to minimize noise.
Last edited:
I only call ths very old quote because I played for a while with toroidals.
My own conclusions (and some people feel free to disagree):
Toroidal core specs calls for PP applications. Is easy to obtain outrageous freq. response with PP designs. Is not so with SE designs. LinuksGuru is correct. Is a PITA for making one homogeneous GAP in toroidal core (is a hard material to saw, and destroy some blades), and most of time results in an uneven cut, and one needs to discard the core
(is almost impossible to trim that; the core fells apart in trying fast process here). So one can need carefully make one magnetically insulating material for the GAP, and needs to be very stiff and stable to support the windings pressure in order to not alter the µE (the resulting inductance+max iDC allowed). Finally other core factors, so good to PP designs (like magnetic lenght), prove a liability to SE, like the necessity to different core+copper quantity and designs to allow for DC (far worse than E-I), and the not simmetrical drive removes one toroidal advantage, etc, etc.For me is a lot of work to results in an expensive and heavy SE piece, and same design in C core results in far easy building process and lower weight.
In my 300B amp I use one toroidal SE I've tried. The sound is good but I haven't compared to commercial models. It have good specs but smaller commercial E-I cores can obtain same or better results. In stark contrast, my PP prototypes are a breeze with notable results (but not optimized like Plitrons), difficult to achieve with same sized E-I or even C core.
Moving up to 5K I did a direct comparison in my 4P1L SE amp between the LL1660/80 configured as 5K and the LL1682 which would be 8K into 8 ohms.
I preferred the LL1682, which had more air and more transient detail - more resonance on double bass and tom-toms in the drumkit. Better vocals. Not a huge difference - the LL1660 was certainly smooth and had marginally better bass, but overall not as crisp as the smaller LL1682.
You could extrapolate that the LL1663 and 1664 would probably sound a little better than the LL1660. Smaller design, simpler build without all the extra windings.
Still haven't heard any Tango, Tamura or Monolith..... I'm making do with Lundahl.
I preferred the LL1682, which had more air and more transient detail - more resonance on double bass and tom-toms in the drumkit. Better vocals. Not a huge difference - the LL1660 was certainly smooth and had marginally better bass, but overall not as crisp as the smaller LL1682.
You could extrapolate that the LL1663 and 1664 would probably sound a little better than the LL1660. Smaller design, simpler build without all the extra windings.
Still haven't heard any Tango, Tamura or Monolith..... I'm making do with Lundahl.
I'm using Mark Audio Alpair 10s - full range 6".
https://www.madisoundspeakerstore.c.../markaudio-alpair-10-gold-6-full-range-gen-2/
Amp has a 4P1L output stage, and input stage is either 4P1L or 26. All in filament bias. 2-stage amp.
https://www.madisoundspeakerstore.c.../markaudio-alpair-10-gold-6-full-range-gen-2/
Amp has a 4P1L output stage, and input stage is either 4P1L or 26. All in filament bias. 2-stage amp.