Hello friends!
I built the classic 6v6 from Dynaco, I really liked the process and I'm learning a lot.
I was looking at several posts here in the forum about NFB or FB in general.
I read some articles from audionote, as well as from Japanese amp builders about the FB, and realized that they don't like it very much.
And that's where my curiosity comes in.
I read here, I don't remember where, that ultralinear OPTs are a kind of local FB, how true is that?
My query is:
I was doing some simulations in LTspice, with and without FB (the one in the schematic).
https://diyaudioprojects.com/Schematics/DIY-Push-Pull-PP-6V6-Tube-Amplifier/
How accurate is this representation in LTspice in terms of reality?
Do my valves and transformers behave the same?
and if so, I leave you 2 tests that I did, one with FB and the other without FB.
What happens with the signal, is it what happens in reality?
I notice that without the FB, I turn up the volume of the amp and it seems like it is screaming in my face, it makes sense.
As I told you, I am learning from you all the time and I want to continue doing so.
Thank you!
I built the classic 6v6 from Dynaco, I really liked the process and I'm learning a lot.
I was looking at several posts here in the forum about NFB or FB in general.
I read some articles from audionote, as well as from Japanese amp builders about the FB, and realized that they don't like it very much.
And that's where my curiosity comes in.
I read here, I don't remember where, that ultralinear OPTs are a kind of local FB, how true is that?
My query is:
I was doing some simulations in LTspice, with and without FB (the one in the schematic).
https://diyaudioprojects.com/Schematics/DIY-Push-Pull-PP-6V6-Tube-Amplifier/
How accurate is this representation in LTspice in terms of reality?
Do my valves and transformers behave the same?
and if so, I leave you 2 tests that I did, one with FB and the other without FB.
What happens with the signal, is it what happens in reality?
I notice that without the FB, I turn up the volume of the amp and it seems like it is screaming in my face, it makes sense.
As I told you, I am learning from you all the time and I want to continue doing so.
Thank you!
Your amplifier was 'shouting' without [global] negative feedback; Right?
Are you willing to try a simple modification?
Willing to have a little less power output?
Willing to listen at lower power?
1. Disconnect the global negative feedback.
2. Disconnect the 6V6 screen from the Ultra Linear tap; connect the screen to a 100 Ohm resistor, and the other end of the 100 Ohm resistor to the Plate.
3. Repeat for the other 6V6.
Note: it is all easily changed back, if you do not like it.
I hope your 6V6 tubes are very well matched. The self bias is common to both tubes, if they are not well matched, the output transformer may be prematurely going into saturation.
Have fun listening!
Are you willing to try a simple modification?
Willing to have a little less power output?
Willing to listen at lower power?
1. Disconnect the global negative feedback.
2. Disconnect the 6V6 screen from the Ultra Linear tap; connect the screen to a 100 Ohm resistor, and the other end of the 100 Ohm resistor to the Plate.
3. Repeat for the other 6V6.
Note: it is all easily changed back, if you do not like it.
I hope your 6V6 tubes are very well matched. The self bias is common to both tubes, if they are not well matched, the output transformer may be prematurely going into saturation.
Have fun listening!
I've found NFB isn't needed or doesn't improve things much on small low powered SE amps, but it does improve frequency response & THD on a PP amp. If you want to run the amp without NFB it would mean changing the circuit to have less gain by changing the ECC83/12AX7 for an ECC81 or ECC88 say, or depending on how much you need to drive the OP stage an ECC82 or 6SN7. I haven't done any calcs so these are off the cuff ideas.
As the previous poster said, the 6V6's share a common cathode RC, it would be better to use two 470r's perhaps instead.
Andy.
As the previous poster said, the 6V6's share a common cathode RC, it would be better to use two 470r's perhaps instead.
Andy.
6A3sUMMER
I'm going to try with those changes, without FB and with those resistors between screen and anode.
The problem is that I have some inefficient speakers, 87db 6ohm JBL A130.
The simulations I did in LTspice, if I'm doing them right, show something similar to the ones I did without FB but in FFT the even harmonics are shown better.
I'm going to try and tell you
I'm going to try with those changes, without FB and with those resistors between screen and anode.
The problem is that I have some inefficient speakers, 87db 6ohm JBL A130.
The simulations I did in LTspice, if I'm doing them right, show something similar to the ones I did without FB but in FFT the even harmonics are shown better.
I'm going to try and tell you
Diabolical Artificer
because?
amplifies the input signal better?
I tried with 1k and it gives a better graph
Diabolical Articifer,
Interesting.
You said:
"I've found NFB isn't needed or doesn't improve things much on small low powered SE amps".
1. Did you try that on triode output tubes?
2. Did you try that on pentode or beam power tubes that were operated in pentode or beam power mode?
3. Did you try that on pentode or beam power tubes that were triode wired pentode or triode wired beam power tubes?
Generally, I found that se triode, se triode wired pentode, and se triode wired beam power tubes worked best without global negative feedback.
Did you try that on pentode or beam power tubes that were in ultra linear mode?
I found the results of 1. and 3. above to be better than 2. above.
Ultra linear seemed to be be middle performing, between 1. & 3., and 2 above.
Interesting.
You said:
"I've found NFB isn't needed or doesn't improve things much on small low powered SE amps".
1. Did you try that on triode output tubes?
2. Did you try that on pentode or beam power tubes that were operated in pentode or beam power mode?
3. Did you try that on pentode or beam power tubes that were triode wired pentode or triode wired beam power tubes?
Generally, I found that se triode, se triode wired pentode, and se triode wired beam power tubes worked best without global negative feedback.
Did you try that on pentode or beam power tubes that were in ultra linear mode?
I found the results of 1. and 3. above to be better than 2. above.
Ultra linear seemed to be be middle performing, between 1. & 3., and 2 above.
because? I just doubled Rk which is 250r x 2 = 500r ( because the current is halved), then used the nearest common R value, so 470r. That & 470r is used in loads of EL84/6V6 PP amps. This means matched valves are not needed. I don't do simulations, I build an amp, test it, then listen to it. I've found that after spending hours testing an amp, my ears are the final arbiter & I can't hear any difference with NFB applied. My ears are nearly 60 years old, your ears will be different.
6A3s etc ... "Interesting.
You said:... etc" No, haven't done most of that, just built a few SE amps from scratch using bits of paper & experience & after farting about for hours tweaking & testing I found I liked the sound of the amp in question (low power SE amp) better without NFB. I used to spend hours running tests, prototyping and mucking about, who's got the time & what does all this data mean in the end? I like building amps and mucking about with valves, then listening to them. Couldn't give a monkeys left elbow if triode/UL/Schade wotnot has 0.01% better THD figures. If it sounds good to my ears, I'm happy. Sorry.
Andy.
Diabolical,
"Sorry"?
You have nothing to be sorry about.
You mostly answered my questions . . .
I believe you prefer se amplifiers without negative feedback, Right?
You said: "I found I liked the sound of the amp in question (low power SE amp)".
What kind of output tubes did you use for that/those se amplifier(s)?
Triodes?
Pentodes?
Beam Power tubes?
Thanks!
"Sorry"?
You have nothing to be sorry about.
You mostly answered my questions . . .
I believe you prefer se amplifiers without negative feedback, Right?
You said: "I found I liked the sound of the amp in question (low power SE amp)".
What kind of output tubes did you use for that/those se amplifier(s)?
Triodes?
Pentodes?
Beam Power tubes?
Thanks!
Can you give me the sine wave data and the simulation commands you used?
Maybe it changes due to the definition of the transformers and valves, I am not very expert in LTspice
It's an English thing, if someone runs you over, breaks both your legs & puts you in hospital for 6 weeks, we'd apologise for dinting their car. Hyperbole, but still true. I was also apologising for my non technical approach to valve amp design.
To answer your question the last SE amp I built used a 6SL7 front end with 6V6 OP valves. I made the OP switchable between pentode and triode mode. The no NFB applied to both configurations. Before that I designed & built a SE amp to replace the gubbins in a Lowe radio/amp using a PCL82 for a mate to replace the EC85/ELL80 OP stage, amongst other stuff.
The DIY valve amp builders community seems to fall into three camps, broadly speaking EG 1) Stick to a known published design 2) The simulators, understanding every aspect & simulating the design before building or prototype a design for three years 3) The bodgers, tinkerers & winging it brigade: lets zap an EF80 with 1500v for a laugh typa ting.
Have fun, Andy.
Please find attached the simulation file, including the tube (Ayumi's) and push pull transformer (Robert McLean's) models I used. I'm no LTSpice expert either, but there is plenty of good information here.
Good classification. I put myself in class 2), but with a bit less then 3 years between prototypes
. For my first amp I followed approach 1) though.Simulations. Great for many. I hate software.
I get much of my ideas from published designs (Caution: some are known to be good; and some are published but absolutely will not work as drawn).
Many ideas come from other 'disciplines', RF, Ultrasound Dopplers, Spectrum Analyzer and other Test and Measurement circuits, and many other circuits as well . . .
Then I try and combine various of those design ideas, often in funny ways.
Tinker, modify, fix, yes. Starting about 1959.
Later, fixing transmitters, receivers, and radars on a US Naval destroyer . . . without the correct parts (you can do a lot when you are under pressure).
I get much of my ideas from published designs (Caution: some are known to be good; and some are published but absolutely will not work as drawn).
Many ideas come from other 'disciplines', RF, Ultrasound Dopplers, Spectrum Analyzer and other Test and Measurement circuits, and many other circuits as well . . .
Then I try and combine various of those design ideas, often in funny ways.
Tinker, modify, fix, yes. Starting about 1959.
Later, fixing transmitters, receivers, and radars on a US Naval destroyer . . . without the correct parts (you can do a lot when you are under pressure).
It sounds like you have vast experience 6A3sUMMER!!!!
I don't have that much, nor do I have an oscilloscope, nor do I have such good ears, but I love electronics and experimenting, I want to get the best I can with what I have.
Thank you for so much help, I'm going to try the mods you gave me, knowing that I win and lose with the different options
I don't have that much, nor do I have an oscilloscope, nor do I have such good ears, but I love electronics and experimenting, I want to get the best I can with what I have.
Thank you for so much help, I'm going to try the mods you gave me, knowing that I win and lose with the different options
Hi, what is your opinion on how pentode, UL and triode react to difficult loads (as most speakers are) considering that the triode has no knees on the curves, UL has a generally wide knee (so the loadline has similar behaviour on a good range of loads) and pentode has a sharp knee (so variation in load means more/less power and hitting g1=0 on the horizontal section or vertical section.
Thanks,
Roberto
zintolo,
Knee?
I try and keep my amplifiers signal input level so that they never get in the portion of the plate curves that has the knee.
I am not trying to get maximum power, I am trying to get finesse.
I either turn the volume control down, or bring out a more powerful amplifier if I need louder.
In general, the control of a loudspeaker is mostly dependent on the damping factor of the amplifier output stage.
Generally difficult speaker loads need higher damping factor than easy speaker loads.
Whether an amplifier is single ended, or push pull, and if there is no negative feedback . . .
The damping factor of amplifiers is generally . . .
With Pentode or Beam Power modes, low damping factor
With Ultra Linear Pentode or Beam Power tubes, medium damping factor
With Triodes, triode wired pentodes and triode beam power tubes, higher damping factor.
Example of a loudspeaker and a single ended amplifier:
8 Ohm loudspeaker:
8 Ohm, 40 Ohm maximum at bass resonance (1 frequency for closed box, 2 frequencies for ported box).
8 x 5 = 40; 5:1 impedance ratio (5 times higher load impedance, 5,000 Ohm primary "becomes" 25,000 Ohm primary
Pentode / Beam Power Tube . . .
Gm = 6000uMhos (6mA/V on g1)
RL = 5,000 Ohms with 8 Ohms on 8 Ohm secondary tap
g1 to Plate Gain = 6mA/V x 5000 Ohms = 30 . . . 8 Ohm load, gain is 30
g1 to Plate Gain = 6mA/V x 5000 Ohms x 5 times higher load impedance = 150 . . . 40 Ohm load; gain is 150
Suppose the g1 bias is 20V, then a 10V signal to g1 is OK, Right?
No, not when the loudspeaker impedance at bass resonance frequency is 40 Ohms.
If the signal to g1 is 10V, we have 150 x 10 = 1500V signal on the plate (can not even get there, the B+ was only 300V, the plate only goes to 600V) . . . never got to 1500V, Not OK.
With a 10V signal to g1, and the loudspeaker 8 Ohms at many frequencies, the tube plate would go to 10 x 30 = 300V higher than the 300V B+; the plate would go to + 600V, all is OK.
You get the idea, an essentially unloaded pentode or beam power tube can not control the loudspeaker very well, and the output stage itself is uncontrolled, unless negative feedback is applied to correct the signal output level.
I hope that answers your questions.
Knee?
I try and keep my amplifiers signal input level so that they never get in the portion of the plate curves that has the knee.
I am not trying to get maximum power, I am trying to get finesse.
I either turn the volume control down, or bring out a more powerful amplifier if I need louder.
In general, the control of a loudspeaker is mostly dependent on the damping factor of the amplifier output stage.
Generally difficult speaker loads need higher damping factor than easy speaker loads.
Whether an amplifier is single ended, or push pull, and if there is no negative feedback . . .
The damping factor of amplifiers is generally . . .
With Pentode or Beam Power modes, low damping factor
With Ultra Linear Pentode or Beam Power tubes, medium damping factor
With Triodes, triode wired pentodes and triode beam power tubes, higher damping factor.
Example of a loudspeaker and a single ended amplifier:
8 Ohm loudspeaker:
8 Ohm, 40 Ohm maximum at bass resonance (1 frequency for closed box, 2 frequencies for ported box).
8 x 5 = 40; 5:1 impedance ratio (5 times higher load impedance, 5,000 Ohm primary "becomes" 25,000 Ohm primary
Pentode / Beam Power Tube . . .
Gm = 6000uMhos (6mA/V on g1)
RL = 5,000 Ohms with 8 Ohms on 8 Ohm secondary tap
g1 to Plate Gain = 6mA/V x 5000 Ohms = 30 . . . 8 Ohm load, gain is 30
g1 to Plate Gain = 6mA/V x 5000 Ohms x 5 times higher load impedance = 150 . . . 40 Ohm load; gain is 150
Suppose the g1 bias is 20V, then a 10V signal to g1 is OK, Right?
No, not when the loudspeaker impedance at bass resonance frequency is 40 Ohms.
If the signal to g1 is 10V, we have 150 x 10 = 1500V signal on the plate (can not even get there, the B+ was only 300V, the plate only goes to 600V) . . . never got to 1500V, Not OK.
With a 10V signal to g1, and the loudspeaker 8 Ohms at many frequencies, the tube plate would go to 10 x 30 = 300V higher than the 300V B+; the plate would go to + 600V, all is OK.
You get the idea, an essentially unloaded pentode or beam power tube can not control the loudspeaker very well, and the output stage itself is uncontrolled, unless negative feedback is applied to correct the signal output level.
I hope that answers your questions.
Hi @6A3sUMMER ,
you are right, I didn't explain my question well.
I talked about the knee because I was focusing on how the different loadlines (or better the different ellipses) at different frequencies affect the sound on peaks from an harmonic content point of view.
PS
Next question will then be how systems recover from peaks: sometimes even great systems have bad behaviours on peaks and songs recorded with great dynamics could need too much power for just small details, or live with that fault.
you are right, I didn't explain my question well.
I talked about the knee because I was focusing on how the different loadlines (or better the different ellipses) at different frequencies affect the sound on peaks from an harmonic content point of view.
PS
Next question will then be how systems recover from peaks: sometimes even great systems have bad behaviours on peaks and songs recorded with great dynamics could need too much power for just small details, or live with that fault.
zintolo,
1. The beauty of graphs for those who are "visual analytical":
Loadlines; straight and eliptical
Take a tube datasheet's graph of plate curves versus grid voltage steps.
The plate curves in certain areas may be Ohmic, knee, straight, evenly spaced, un-evenly spaced, and have kinks.
Draw a loadline that goes from 0V plate and lots of plate current; to 0mA plate current and lots of plate volts.
Straight loadline:
The straight loadline within a certain range of plate voltages is fairly linear. Beyond the linear range, at very low plate voltages and very high plate voltages, there is non-linearity.
The straight loadline may cross through the low voltage Ohmic; knee; un-evenly spaced; and kink regions of plate voltage, that is the cause of non linearity.
Elliptical loadline:
Draw a loadline that goes from 0V plate and lots of plate current; to 0mA plate current and lots of plate volts . . .
And Now, draw an ellipse around that straight loadline.
Do you notice that now, the ellipse crosses in more extreme areas, the linearity with the ellipse versus plate voltage is lost, versus the same range of plate voltage and the straight loadline.
The elliptical loadline causes both Harmonic and Intermodulation distortion; and occurs whenever the loadline is other than straight purely resistive.
Resistive loadlines have those distortions, but often the elliptical loadline has more distortion.
Loudspeakers can have frequencies where they are capacitive, inductive, RC, RL, RLC, and R. The R is a non-elliptical load line,
The only other non-elliptical load line is when RLC is at Resonance; both parallel RLC, and serial RLC.
2. You said:
A. "how systems recover from peaks: sometimes even great systems have bad behaviours on peaks and songs recorded with great dynamics could need too much power for just small details, or live with that fault."
Can we have a little more description of what kind of overload you are talking about?
Core Saturation; RC coupling bias shift; RC self bias bias shift; local and/or global negative feedback overload/sticking non corrective region, or what other overload did you have in mind?
And, what did you mean by saying: "could need too much power for just small details"?
Did you mean when the overloading bass drum signal prevents the piccolo signal from being heard because it is buried in the bass overload signal?
Did you mean when the playback system has a woofer or midrange cone that has an excursion that is beyond the linear maximum excursion of the loudspeaker driver?
. . . Or what did you mean?
3. To one extent or another, in general, we do have to live with the fact that a playback system is less perfect than being at a live concert.
1. The beauty of graphs for those who are "visual analytical":
Loadlines; straight and eliptical
Take a tube datasheet's graph of plate curves versus grid voltage steps.
The plate curves in certain areas may be Ohmic, knee, straight, evenly spaced, un-evenly spaced, and have kinks.
Draw a loadline that goes from 0V plate and lots of plate current; to 0mA plate current and lots of plate volts.
Straight loadline:
The straight loadline within a certain range of plate voltages is fairly linear. Beyond the linear range, at very low plate voltages and very high plate voltages, there is non-linearity.
The straight loadline may cross through the low voltage Ohmic; knee; un-evenly spaced; and kink regions of plate voltage, that is the cause of non linearity.
Elliptical loadline:
Draw a loadline that goes from 0V plate and lots of plate current; to 0mA plate current and lots of plate volts . . .
And Now, draw an ellipse around that straight loadline.
Do you notice that now, the ellipse crosses in more extreme areas, the linearity with the ellipse versus plate voltage is lost, versus the same range of plate voltage and the straight loadline.
The elliptical loadline causes both Harmonic and Intermodulation distortion; and occurs whenever the loadline is other than straight purely resistive.
Resistive loadlines have those distortions, but often the elliptical loadline has more distortion.
Loudspeakers can have frequencies where they are capacitive, inductive, RC, RL, RLC, and R. The R is a non-elliptical load line,
The only other non-elliptical load line is when RLC is at Resonance; both parallel RLC, and serial RLC.
2. You said:
A. "how systems recover from peaks: sometimes even great systems have bad behaviours on peaks and songs recorded with great dynamics could need too much power for just small details, or live with that fault."
Can we have a little more description of what kind of overload you are talking about?
Core Saturation; RC coupling bias shift; RC self bias bias shift; local and/or global negative feedback overload/sticking non corrective region, or what other overload did you have in mind?
And, what did you mean by saying: "could need too much power for just small details"?
Did you mean when the overloading bass drum signal prevents the piccolo signal from being heard because it is buried in the bass overload signal?
Did you mean when the playback system has a woofer or midrange cone that has an excursion that is beyond the linear maximum excursion of the loudspeaker driver?
. . . Or what did you mean?
3. To one extent or another, in general, we do have to live with the fact that a playback system is less perfect than being at a live concert.
Last edited:
That may a negative for some, a positive for other listeners. I prefer studio albums to live albums. Some may strive for as little colouration of the recorded medium, others welcome the colouration that a SE amp, the sound of an LP or magnetic tape has: it's a subjective subject.
After designing building a few amp and mucking about with with DIY audio for a while I've found the audible differences between the various topography's of OP stages is subtle, and any perceived difference between a triode/UL/pentode/BT stage, is a fleeting & temporary phenomenon. The same applies to any perceived differences between amp A & amp B.
Loadlines & test result data often is meaningless compared to when you listen to an amp but it can tell you how an amp behaves under extreme conditions, with difficult loads say or at higher power outputs, but again it's not often we get the chance to really crank up the volume unless we have deaf or no neighbores.
Andy.