Help me understand something about the output transformers in our tube amplifiers please. The response of most output transformers seem to look like this in general, with some degree of rolloff at both frequency extremes, especially below about 40Hz.
So how do we get anything resembling truly flat frequency response when the transformer sags at both extremes like this? Is the answer that negative feedback in the amplifier corrects the tansformers' frequency rolloffs at the extremes?
Another question: the inductance of a transformer controls the low frequency roll off, correct? I found this formula:
L = Z / ( 2π × f)
where Z is source impedance and f is the -3dB frequency
example: L = 6000 Ohm primary impedance/(2π × 40 Hz) = 24 Henrys is the required inductance for response to be down 3dB at 40Hz
Did I get that correct? If so, and we start looking at a situation like -1dB at 20Hz, we end up with huge values.
How do we calculate the -1dB point instead of the -3dB point? I am mathematically challenged and always have been.
And as if that were not enough, let's throw in any differences for single-ended vs. push-pull transformer response and roll off calculations.
Sometimes I think it's a miracle that any of these amplifiers make sound at all!
So how do we get anything resembling truly flat frequency response when the transformer sags at both extremes like this? Is the answer that negative feedback in the amplifier corrects the tansformers' frequency rolloffs at the extremes?
Another question: the inductance of a transformer controls the low frequency roll off, correct? I found this formula:
L = Z / ( 2π × f)
where Z is source impedance and f is the -3dB frequency
example: L = 6000 Ohm primary impedance/(2π × 40 Hz) = 24 Henrys is the required inductance for response to be down 3dB at 40Hz
Did I get that correct? If so, and we start looking at a situation like -1dB at 20Hz, we end up with huge values.
How do we calculate the -1dB point instead of the -3dB point? I am mathematically challenged and always have been.
And as if that were not enough, let's throw in any differences for single-ended vs. push-pull transformer response and roll off calculations.
Sometimes I think it's a miracle that any of these amplifiers make sound at all!
Last edited:
You have to take into consideration what the source impedence the tranformer primary is being driven by, the rp of the output tube.the lower the rp the more bass you will get out of your output transformer for a given primary xl at the freq of interest. Of course its never that simple.
The OPT primary impedance is a fiction, just a calculated number based on a particular load resistance and turns ratio. And, is separate from source impedance, which is a parallel of the output valve(s)' presenting impedance and the load, reflected back through the OPT.
OPT primary (also called "magnetizing") inductance appears as a parasitic load on the output valves, in parallel with an imaginary ideal transformer. So, the drill is to first calculate a "resistance" composed of the output valve(s)' source resistance in parallel with the reflected load "resistance" and then use your formula to determine the 3dB knee in response.
In real iron core transformers inductance varies greatly with frequency and signal level, and not monotonically, giving an extra level of difficulty, for which the judges will reward extra points.
All good fortune,
Chris
The low frequency response of an output tube and output transformer with no negative feedback, is due to a single pole high pass RL filter.
A single pole high pass filter below the -3dB frequency, is 6dB/Octave roll off rate.
Example, such as the plate impedance, rp, driving an output transformer's primary inductive reactance, XL:
The -1dB frequency is 2 x the -3dB frequency. ( -3dB @ 20Hz and -1dB @ 40Hz).
At 10Hz, it will be about -9dB.
And,
The -1dB frequency phase shift is about 27 degrees (Arc Cosine of 0.89).
The -3dB frequency phase shift is about 45 degrees (Arc Cosine of 0.707).
Cosines are your friend.
With no negative feedback, the high frequency roll off of a tube and output transformer is more complex:
Plate impedance driving distributed capacitance of the output transformer's primary
Leakage reactance from the primary driving the secondary
Usually, the high frequency roll off is 12dB per Octave for frequencies above the -3dB frequency.
Also, there often is a resonance effect at or near the -3dB frequency.
Peaks and Nulls have been known to happen at or near there.
Nice and neat things you can count on with a good resistor connected to the output transformer's secondary.
Not so fast . . .
What complicates all this even more is the nature of connecting the amplifier to a loudspeaker load.
Loudspeakers present all of these: pure resistive load, pure capacitive load, pure inductive load, RL load, RC load, and RLC load, depending on the frequency.
Any capacitive load or any inductive load (and RL, RC, and RLC loading) causes the "load line" on the plate curves to open up into an Ellipse.
And those loudspeakers inductive and capacitive loads do react with the output transformer's inducance and capacitance.
Some would have you believe that the audio world is ending, because of all these factors.
I have 5 vacuum tube amplifiers, and there is no negative feedback from the output transformer secondary (no global negative feedback).
The only negative feedback I use is:
From cathode to cathode (push pull and balanced);
Plate to Screen (push pull, balanced, and single ended).
Warning: I have a single 'jaded' ear.
Your Mileage May Vary.
Just design, build, and enjoy sitting back and listening to music on your vacuum tube amplifiers.
Think about the Music, Not the playback system.
Then you can criticize the conductor's too-fast pace, the microphone that is swallowed up by the saxophone's end-bell, and the very poor multi-track to 2 channel mix.
We all like perfection, OK, I get it.
But that clarinetist is using a # 2.5 hardness reed, and he needs to learn how to play on a # 1 hardness reed (a much smoother and pretty timbre sound). No Hi Fi, and no Stereo can fix his poor playing.
Have Fun!
A single pole high pass filter below the -3dB frequency, is 6dB/Octave roll off rate.
Example, such as the plate impedance, rp, driving an output transformer's primary inductive reactance, XL:
The -1dB frequency is 2 x the -3dB frequency. ( -3dB @ 20Hz and -1dB @ 40Hz).
At 10Hz, it will be about -9dB.
And,
The -1dB frequency phase shift is about 27 degrees (Arc Cosine of 0.89).
The -3dB frequency phase shift is about 45 degrees (Arc Cosine of 0.707).
Cosines are your friend.
With no negative feedback, the high frequency roll off of a tube and output transformer is more complex:
Plate impedance driving distributed capacitance of the output transformer's primary
Leakage reactance from the primary driving the secondary
Usually, the high frequency roll off is 12dB per Octave for frequencies above the -3dB frequency.
Also, there often is a resonance effect at or near the -3dB frequency.
Peaks and Nulls have been known to happen at or near there.
Nice and neat things you can count on with a good resistor connected to the output transformer's secondary.
Not so fast . . .
What complicates all this even more is the nature of connecting the amplifier to a loudspeaker load.
Loudspeakers present all of these: pure resistive load, pure capacitive load, pure inductive load, RL load, RC load, and RLC load, depending on the frequency.
Any capacitive load or any inductive load (and RL, RC, and RLC loading) causes the "load line" on the plate curves to open up into an Ellipse.
And those loudspeakers inductive and capacitive loads do react with the output transformer's inducance and capacitance.
Some would have you believe that the audio world is ending, because of all these factors.
I have 5 vacuum tube amplifiers, and there is no negative feedback from the output transformer secondary (no global negative feedback).
The only negative feedback I use is:
From cathode to cathode (push pull and balanced);
Plate to Screen (push pull, balanced, and single ended).
Warning: I have a single 'jaded' ear.
Your Mileage May Vary.
Just design, build, and enjoy sitting back and listening to music on your vacuum tube amplifiers.
Think about the Music, Not the playback system.
Then you can criticize the conductor's too-fast pace, the microphone that is swallowed up by the saxophone's end-bell, and the very poor multi-track to 2 channel mix.
We all like perfection, OK, I get it.
But that clarinetist is using a # 2.5 hardness reed, and he needs to learn how to play on a # 1 hardness reed (a much smoother and pretty timbre sound). No Hi Fi, and no Stereo can fix his poor playing.
Have Fun!
Last edited:
Negative feedback could be said to try to hammer down all proud nails to board level, but once board level is reached, it's reached.
IOW, there has to still be enough gain "around the loop" for feedback to have significant effect. At frequency extremes loop gain is falling the further away from the nice flat central region, so feedback has less and less effect.
All good fortune,
Chris
If you need more LF output, get a bigger iron.
If you want more HF response, get more complex and fancy winding sections, interleaving and section stacking for lower spread induction and lower interwinding C at the same time.
These are the two reasons why very good transformers are very expensive.
Use feedback for the final touches.
Jan
If you want more HF response, get more complex and fancy winding sections, interleaving and section stacking for lower spread induction and lower interwinding C at the same time.
These are the two reasons why very good transformers are very expensive.
Use feedback for the final touches.
Jan
Never Get Old,
Look at the frequency response graph you put in Post # 1.
Find the frequency where the low frequency is 3dB less than mid-band frequency *;
and find the frequency where the high frequency is 3dB less than mid-band frequency **.
Suppose you take the output of the secondary, and apply that signal back in the form of negative feedback to an early stage.
Then, adjust the amount of signal from the output transformer secondary gives a level of negative feedback until the mid-band frequencies are reduced by 3 dB (3dB of negative feedback).
The low frequency - 3dB , and the high frequency -3dB*, send 3dB less negative feedback to the early stage, which causes less negative feedback to be sent to the earlier stage; relative to the negative feedback that is sent to the earlier stage at mid-band frequencies.
That is as if you "hammered down" the mid-band frequency levels, but left the low frequency * and high frequency ** nearly the same as they were in the beginning (did Not "hammer down" the low * and high ** frequencies).
That is a simplified explanation of the global negative feedback.
Look at the frequency response graph you put in Post # 1.
Find the frequency where the low frequency is 3dB less than mid-band frequency *;
and find the frequency where the high frequency is 3dB less than mid-band frequency **.
Suppose you take the output of the secondary, and apply that signal back in the form of negative feedback to an early stage.
Then, adjust the amount of signal from the output transformer secondary gives a level of negative feedback until the mid-band frequencies are reduced by 3 dB (3dB of negative feedback).
The low frequency - 3dB , and the high frequency -3dB*, send 3dB less negative feedback to the early stage, which causes less negative feedback to be sent to the earlier stage; relative to the negative feedback that is sent to the earlier stage at mid-band frequencies.
That is as if you "hammered down" the mid-band frequency levels, but left the low frequency * and high frequency ** nearly the same as they were in the beginning (did Not "hammer down" the low * and high ** frequencies).
That is a simplified explanation of the global negative feedback.
That is absolutely perfect. I can understand that. This is why I had to hire a tutor weekly to survive physics in college! Thank you! I need plain (simplified) explanations like that as I obviously am still near the bottom of the learning curve here. I hope to not stay there forever.
This also makes sense, but I'd need an example to be able to perform the calculations myself.
Now that I have some more understanding of what is going on, I have more questions.
1) How can a transformer manufacturer test an output transformer in a meaningful way without it being installed in an amplifier?
2) How do we compare apples to apples and oranges to oranges when shopping for output transformers?
Someone posted this table here on the forum (I think). Does anyone know what book this is so I can read the rest of it? Books like this usually contain example calculations, and those always help me greatly.
Last edited:
In your case, you would pretend that the secondary load is a resistor of value marked on the OPT spec, so the OPT's contribution is its rated 6000 Ohms. Pentode output valves can be approximated, close enough for rock and roll, as infinite, so you end up with the numbers you originally calculated! This is also the worst case for primary inductance requirements. Triodes relax the requirements (from a frequency response standpoint) greatly.
That page is undoubtably from RDH4.
All good fortune,
Chris
Oh, how perfect!
Got it. Page 213 exactly as you said! I wish I had time to read and understand that whole book. I do want to read this part though.
Link for anyone who needs it. http://tubebooks.org/Books/RDH4.pdf
A subtler note about primary inductance: it appears as a parasitic load to the output valves, meaning it must also be driven (charged and discharged) in addition to the intended load (the speakers). To understand this graphically, the only way that really works for me, imagine the output valves' loadlines fattening into an ellipse and rotating clockwise. Neither of these things are good for output valve distortion, so less is more better. More inductance means less fattening and less rotating.
Generations have grown up on RDH4, but it's not an overnight read. Arf! The pre-war RDH3 is also interesting, and may be available in old paper copies affordably. It has information about iron-core inductors not present in RDH4 (post-war), although specific metals are surely a little out of date.
All good fortune,
Chris
Generations have grown up on RDH4, but it's not an overnight read. Arf! The pre-war RDH3 is also interesting, and may be available in old paper copies affordably. It has information about iron-core inductors not present in RDH4 (post-war), although specific metals are surely a little out of date.
All good fortune,
Chris
And now I come back to these:
That makes sense, but I don't recall "primary inductance" on any spec sheets. I'll look again. Maybe that's something studied only by transformer designers. When a transformer manufacturer just specifies "inductance" what exactly is it that they mean?
That makes sense, but I don't recall "primary inductance" on any spec sheets. I'll look again. Maybe that's something studied only by transformer designers. When a transformer manufacturer just specifies "inductance" what exactly is it that they mean?
As stated by @jan.didden, in my old Radio / Audio books, I also read :
1 - the bass extension is notably function of the induction of the primary in Henry and the importance of the primary current.
* Bigger and higher grade laminations will usually offer better results in that field.
2 - the treble extension is notably function of the primary to secondary leakages, the reparted capacities.
* Tight coupling, high insulation and interleaving of the windings will usually offer better results in that field.
3 - overshoots, uneven response, resonances are function of the quality of the windings itselfs.
* Precisely wound and deeply impregnated windings are a great improvement on that point.
There are a great number of other influential parameters to consider - not only in building an output transformer, but also in accordance with the type of tube used (triode, tetrode or pentode...) and circuit design (SE, PP, UL, distributed or cathode load, Circlotron...). I read this here and there in my vintage documentation, without noting which books / which pages, idiotely !
Yes @Never Get Old. already the point 1 give me sometimes strange results when performed at home, not always matching by far the manufacturer's specs - which the measurement methods are barely disclosed...
Building an Output Transformer is not an easy task at all, and I salute the DIYers who engage themselves in that path, and moreover, obtain significant results that meet their expectations !
T
1 - the bass extension is notably function of the induction of the primary in Henry and the importance of the primary current.
* Bigger and higher grade laminations will usually offer better results in that field.
2 - the treble extension is notably function of the primary to secondary leakages, the reparted capacities.
* Tight coupling, high insulation and interleaving of the windings will usually offer better results in that field.
3 - overshoots, uneven response, resonances are function of the quality of the windings itselfs.
* Precisely wound and deeply impregnated windings are a great improvement on that point.
There are a great number of other influential parameters to consider - not only in building an output transformer, but also in accordance with the type of tube used (triode, tetrode or pentode...) and circuit design (SE, PP, UL, distributed or cathode load, Circlotron...). I read this here and there in my vintage documentation, without noting which books / which pages, idiotely !
Yes @Never Get Old. already the point 1 give me sometimes strange results when performed at home, not always matching by far the manufacturer's specs - which the measurement methods are barely disclosed...
Building an Output Transformer is not an easy task at all, and I salute the DIYers who engage themselves in that path, and moreover, obtain significant results that meet their expectations !
T
Last edited:
They mean primary or 'magnetizing' inductance. How that's measured, at what level and at what frequency and with how much unbalanced DC current, is another matter. AFAIK, there's not even a recognized standard, but even if there were one, it would only be one data point on a 3D graph.
All good fortune,
Chris
Right now I am using two amplifiers that I am building as learning tools. One is an EL84 PP and the other is an EL34 SE. BOTH are pentode.
This is the first time I have ever looked at a single-ended design. The first thing I had to learn is what an air gap is and why we need one in a SE amplifier. This is an interesting thread about air gaps:
https://www.diyaudio.com/community/threads/gap-width-in-se-output-transformers.331641/
and
http://tubelab.com/articles/ideas/single-ended-output-stages/
And therein lies one of my (our) problems:
It seems to me that this is impossible for off-the-shelf transformers. In other words, trust the transformer designer to do it right and bigger usually is better but not always, up to a certain point of diminishing marginal return. Then test it in an actual amplifier and if it ain't right, either provide different transformer specs to the transformer maker or change the amplifier or both.
Last edited:
It is possible to dodge most of the impossible tradeoffs by using two cheap, crappy transformers instead of a single SOTA, super good one.
The low frequency is handled by a big lump of iron, which doesn't require complicated winding techniques to reduce the leakage inductance and could even be an ordinary 50/60Hz transformer, and the high frequency part is processed by a smallish ferrite transformer having negligible capacitance and leakage inductance.
The composite result is a very good transformer, and the additional costs and components add up to a much smaller cost than a very good, traditional transformer.
https://www.diyaudio.com/community/...and-into-two-transformers.286621/post-4611092
The low frequency is handled by a big lump of iron, which doesn't require complicated winding techniques to reduce the leakage inductance and could even be an ordinary 50/60Hz transformer, and the high frequency part is processed by a smallish ferrite transformer having negligible capacitance and leakage inductance.
The composite result is a very good transformer, and the additional costs and components add up to a much smaller cost than a very good, traditional transformer.
https://www.diyaudio.com/community/...and-into-two-transformers.286621/post-4611092