Many ultra-linear amplifiers historically have used global negative feedback to lower output impedance and further linearize the amplifier output.
Global NFB is not mandatory with ultra-linear operation, that said I've not personally seen or worked on any amps that use UL without global NFB, but that doesn't mean they aren't out there. I think it is safe to assume there are UL amps without global NFB. (I find it hard to believe someone hasn't recently built such an amp.)
Global NFB is not mandatory with ultra-linear operation, that said I've not personally seen or worked on any amps that use UL without global NFB, but that doesn't mean they aren't out there. I think it is safe to assume there are UL amps without global NFB. (I find it hard to believe someone hasn't recently built such an amp.)
Being as KT88 & KT66 produce giant amounts of IMD, it's no suprise that untrained ears have a struggle perceiving phenomena that laboratory equipment & FFT can show up much better.
Once you know where to look, it becomes easier to listen, a bit like demonstrating how Keroes "ultralinear" compresses dynamic range.
You only can know when you listen to the original sound recording, then play it back like through a distorting mirror. The AU7 being a particularly non linear device, makes more of the mirror obvious.
I suspect, "tube rolling" people only are listening to different levels of distortion, but convince themselves that some distortion is more or less pleasing than others because of masking effects inherent in compressed audio, which ALL audio commercial recordings are.
I have a masking test on my test CD, (with phase change) during the test.
It's quite extraordinary how small phase changes mask and unmask particular otherwise hidden frequencies in a real hifi/speaker combo.
Last edited:
I guess lovers of 300B, 2A3, etc, have not seen the high distortion that they produce in an oscilloscope either, but for some reason there are many enthusiasts of these old and low power amplifier designs .....
I bet it is not due to lack of training, rather the opposite!
Once again, it is the repeated history, the instruments against what we perceive through our sensory / auditory senses .....
I started with valves, switched to SS and went back to my first love.
I have never heard better than these days, despite my limitation of 14 KHZ (*), it really doesn't seem relevant to me what the instruments say.
(*) And it is in my plans to add a supertweeter !
Why use a separate super-tweeter with your speakers? - Audio Affair Blog
I bet it is not due to lack of training, rather the opposite!
Once again, it is the repeated history, the instruments against what we perceive through our sensory / auditory senses .....
I started with valves, switched to SS and went back to my first love.
I have never heard better than these days, despite my limitation of 14 KHZ (*), it really doesn't seem relevant to me what the instruments say.
(*) And it is in my plans to add a supertweeter !
Why use a separate super-tweeter with your speakers? - Audio Affair Blog
I can do you a double blind test of frequencies over 20khz. (My little daughter can hear them).
If I switch in the HF and switch it out, without you knowing I am doing it, I can prove you 100% wrong about being able to detect it.
You will fail.
As for the high distortion of those old 300B/2A3 etc, I don't like mud, least of all mud which adds to yet more of the mud at LF.
Some people evidently do, as well as being unable to tell the difference between the same live recordings rendered in mp3 and PCM HR...preferring the mp3 version each time.
Double blind testing is a bitch.
It really is not fun, for those claiming things which are inaudible but plainly the result of Bias.
I remember removing a Hi end shop nasty room resonance with a DSP.
The owner of the hi end audio prefered the sound of his crappy room without correction despite a massive gain in clarity when we corrected for it.....it removed his "emotion".
(I have been begging him to instal some room acoustic treatments for years, even doing a measurement study).
A friend noticed it immediately, but didn't own the shop,- and instantly liked it.
haha!
It removed layers of MUD, but some people love mud, and you can't convince them otherwise, just like the high end speaker cable fad, and all the other "power conditioners" and other "hi end" nonsense.
If I switch in the HF and switch it out, without you knowing I am doing it, I can prove you 100% wrong about being able to detect it.
You will fail.
As for the high distortion of those old 300B/2A3 etc, I don't like mud, least of all mud which adds to yet more of the mud at LF.
Some people evidently do, as well as being unable to tell the difference between the same live recordings rendered in mp3 and PCM HR...preferring the mp3 version each time.
Double blind testing is a bitch.
It really is not fun, for those claiming things which are inaudible but plainly the result of Bias.
I remember removing a Hi end shop nasty room resonance with a DSP.
The owner of the hi end audio prefered the sound of his crappy room without correction despite a massive gain in clarity when we corrected for it.....it removed his "emotion".
(I have been begging him to instal some room acoustic treatments for years, even doing a measurement study).
A friend noticed it immediately, but didn't own the shop,- and instantly liked it.
haha!
It removed layers of MUD, but some people love mud, and you can't convince them otherwise, just like the high end speaker cable fad, and all the other "power conditioners" and other "hi end" nonsense.
Last edited:
kevinkr,
Yes, there are push pull UL amplifiers out there that do not use any other negative feedback.
I recently designed and built a 7591 push pull, a 6L6 push pull, and the only feedback they used were the 40% ultra linear taps.
I also designed a KT66 single ended, and the only feedback it used was the UL tap (50%).
But then I added output tube plate to driver tube cathode negative feedback.
Yes, there are push pull UL amplifiers out there that do not use any other negative feedback.
I recently designed and built a 7591 push pull, a 6L6 push pull, and the only feedback they used were the 40% ultra linear taps.
I also designed a KT66 single ended, and the only feedback it used was the UL tap (50%).
But then I added output tube plate to driver tube cathode negative feedback.
I have a version of Gingertube's EL84 "Baby Huey" running for the last 13+ years.
It uses UL feedback and plate-to-plate feedback. The additional global negative feedback from OT secondary was not necessary and I removed it after testing/listening.
Another case of UL without global feedback (but with additional Schade-style feedback).
I like it.
Cheers,
Martin
It uses UL feedback and plate-to-plate feedback. The additional global negative feedback from OT secondary was not necessary and I removed it after testing/listening.
Another case of UL without global feedback (but with additional Schade-style feedback).
I like it.
Cheers,
Martin
Back to topic:
The non-linearity of the ECC82 can be dealt with to a certain extend and this tube has useful properties. So, I would not condemn it as useless for Hifi.
However, since there are many more linear tubes available, I wouldn't know why to use it as first choice (except maybe its wide availability - but that would be more of a concern in commercial products, not DIY).
I would leave it for guitar amps - where I like this tube and don't use it at its most linear operation point on purpose.
The non-linearity of the ECC82 can be dealt with to a certain extend and this tube has useful properties. So, I would not condemn it as useless for Hifi.
However, since there are many more linear tubes available, I wouldn't know why to use it as first choice (except maybe its wide availability - but that would be more of a concern in commercial products, not DIY).
I would leave it for guitar amps - where I like this tube and don't use it at its most linear operation point on purpose.
It's like comparing oranges and apples.
Ultralinear is negative feedback.
6L6 do not react well to UL, for the simple reason the close alignment of the screen grid.
807 + versions of it are near identical.
The KT66 is a close cousin of the 6L6 so would behave the same, as it was developed originally with a top cap (KT8C).
They are always betrayed by the extremely small screen grid current under full load, so much so I use a gas discharge shunt stabiliser for the screen supply.
The 7591 is a totally different animal.
massively hi gain, and a screen grid that takes gobs of current.
You will extinguish a big shunt stabiliser the moment you start driving a pair hard.
Being as it appears the 7591 would react well to negative feedback of that sort, it's impossible to compare the 2 applications of NFB, because they are not even remotely comparable.
Gillespie says the same.
There are good and bad applications of NFB in general especially that used on screen grids in UL applications.
I guess that's why the best 6L6-oid applications run them up in the Va, 600-850V zone and apply loop feedback, so there's a 400-500V differential between screen & anode supply.
Unless I'm reading the schematic wrong (which is entirely possible) , Tubelab_com's SSE doesn't use global feedback. Just ~6dB of cathode feedback on the output stage.
sarcastic1,
I am sorry you understood what I said.
I said: Yes, there are push pull UL amplifiers out there that do not use any other negative feedback.
Simply put, the UL amplifiers use negative feedback in the form of the Ultra Linear output stage.
But some of those same UL amplifiers do not have any negative feedback of any other kind (no additional kind, such as: global, cathode, plate to plate, plate to control grid, unity coupling, Triode wired mode, etc.
Just saying
I am sorry you understood what I said.
I said: Yes, there are push pull UL amplifiers out there that do not use any other negative feedback.
Simply put, the UL amplifiers use negative feedback in the form of the Ultra Linear output stage.
But some of those same UL amplifiers do not have any negative feedback of any other kind (no additional kind, such as: global, cathode, plate to plate, plate to control grid, unity coupling, Triode wired mode, etc.
Just saying
Yes too much NFB kills the sound dead as dead can be.
Mcintosh is the perfect example of all the bells and whistles, NFB in all directions, laboratory distortion figures so low they claim it on everything they write, yet a stultifying awful sterile sound that makes you want to run away,
Mac are actually just like one of the others I auditioned, the hideous EAR yoshino V12 with the ghastly looking bulgycases, and the blingly chromy JADIS amps in general with their ginormous but largely badly made and inefficient transformers and PSUs.
Mcintosh is the perfect example of all the bells and whistles, NFB in all directions, laboratory distortion figures so low they claim it on everything they write, yet a stultifying awful sterile sound that makes you want to run away,
Mac are actually just like one of the others I auditioned, the hideous EAR yoshino V12 with the ghastly looking bulgycases, and the blingly chromy JADIS amps in general with their ginormous but largely badly made and inefficient transformers and PSUs.
The 12AU7 was designed primarily as an RF type. That's where you see the majority of 12AU7 applications: PP, Class C drivers/frequency multipliers, oscillator/buffers, product detectors, and quasi-digital applications like astable multibibrators, Schmidt Triggers, R-S latches, flip-flops. If you compare dara sheets, the u-Factor plots are loopy and get worse the lower the VPK. The 12AX7 has a u-Factor plot that's nearly billiard table flat, and that's what you want to see for low distortion audio.
That doesn't mean t that 12AU7s have to sound bad, it just means you'll have to work harder to find a good audio loadline. Also, keep that VPK up.
I remember when researching RF-tubes the common comment was that the old 12AU7s were good for RF and AF but the modern tubes couldn't handle the same RF thus only good for AF.