I was typing too.
I re-edited my post 160. It now answers post 157.
As to what new capacitance, it assumes that the original values were correct and stable.
Making other changes, such as individual stage frequency responses, output transformer selection, very large gain changes, or very large negative feedback dB changes can require re-adjustment of the R X C product.
I would expect the 12AU7 to be better for the gain you want, and for linearity, than using a 12AT7 input stage.
I re-edited my post 160. It now answers post 157.
As to what new capacitance, it assumes that the original values were correct and stable.
Making other changes, such as individual stage frequency responses, output transformer selection, very large gain changes, or very large negative feedback dB changes can require re-adjustment of the R X C product.
I would expect the 12AU7 to be better for the gain you want, and for linearity, than using a 12AT7 input stage.
thanks for the correction! Seems like other methods (strapping outputs as triodes for example) which reduce gain of a given stage, if that stage is inside the FB loop, would have a similar effect then, correct?
Open Loop Gain: Gain with No negative feedback
Closed Loop Gain: Gain with Global negative feedback.
As long as the Open Loop Gain of an amplifier is quite a bit greater than the Closed Loop Gain of an amplifier, the gain will be dictated by the negative feedback ratio (usually set by resistors, with capacitors for high frequency compensation). So, a slight to moderate reduction of gain inside the loop will not change the gain of the amplifier outside the loop.
Closed Loop Gain: Gain with Global negative feedback.
As long as the Open Loop Gain of an amplifier is quite a bit greater than the Closed Loop Gain of an amplifier, the gain will be dictated by the negative feedback ratio (usually set by resistors, with capacitors for high frequency compensation). So, a slight to moderate reduction of gain inside the loop will not change the gain of the amplifier outside the loop.
Oops,
I incorrectly stated that "reducing the gain of an internal stage would reduce the global negative feedback". What I meant to say was that the dB of negative feedback would be reduced.
If open loop has gain of 100, and closed loop has gain of 10, the negative feedback is 20dB. If you reduced one stage's gain to 1/2 of before, then the open loop gain would be 50. The closed loop gain would still be 10. Now the negative feedback would be 14 dB.
I incorrectly stated that "reducing the gain of an internal stage would reduce the global negative feedback". What I meant to say was that the dB of negative feedback would be reduced.
If open loop has gain of 100, and closed loop has gain of 10, the negative feedback is 20dB. If you reduced one stage's gain to 1/2 of before, then the open loop gain would be 50. The closed loop gain would still be 10. Now the negative feedback would be 14 dB.
Diyengineer,
For my latest use of a 12AU7 input stage/driver stage, I did the following:
I wanted to save filament power, if I could get just one section of the 12AU7 to do what I needed it to do. I wanted a low enough plate resistance, low plate dissipation, and enough low distortion voltage swing when loaded by 270k Ohms. I wanted to only wire one half of the filament, if possible.
I did not want to have to use two individual self bias resistors, and two bypass caps. Paralleled sections would have required that for current sharing and best performance. So, I picked one triode, and the corresponding filament.
I grounded the plate, grid, and cathode of the other (unused) triode. This amplifier was a mono block.
I used an IXYS current source and set it to about 6.7mA. It was the plate load of the triode. This drove a 0.1uF coupling cap, and a 270k Ohm grid resistor of the next stage. I went to the 12AU7 tube curves (this is key to seeing the approximate results), and determined that I wanted about 3V of self bias, at 6.7mA of current. The load line I drew on the curves was 270k Ohms. 3V/0.0067A = 448 Ohms. I used a 470 Ohm self bias resistor. I got about 3.15V self bias. I bypassed the self bias resistor with a cap. The input signal can swing from about +3V to 0, to -3V. The plate is driving the high impedance current source and the 270k Ohm grid resistor.
I did not check the gain, but I bet it is somewhere between 15 and 20 (12AU7 with high resistance load), It was able to easily drive the output stage grid with low distortion.
Looking into the cathode, we see 1/Gm, + Rp load/u 1/2600 uMhos + 270K/20; 385 Ohms + 13.5k Ohms. This is so high of a cathode impedance, we only need to bypass the 470 Ohm self bias resistor. If Xc of the bypass cap at 20Hz = 47 Ohms, the frequency response will be degraded by far less than 1 dB at 20Hz. Xc of 170uF = 47 Ohms. Use a 200uF bypass cap.
Hopefully that will show you how I sometimes design an input/driver stage. I do not always do these things the same way, just been doing it so many times over the years.
I try to keep circuits as simple as I can. I am not a fan of SRPP (I have used it). And I do not use pentodes, except as triode wired output stages (pentode and beam power tubes in 'triode' mode).
For my latest use of a 12AU7 input stage/driver stage, I did the following:
I wanted to save filament power, if I could get just one section of the 12AU7 to do what I needed it to do. I wanted a low enough plate resistance, low plate dissipation, and enough low distortion voltage swing when loaded by 270k Ohms. I wanted to only wire one half of the filament, if possible.
I did not want to have to use two individual self bias resistors, and two bypass caps. Paralleled sections would have required that for current sharing and best performance. So, I picked one triode, and the corresponding filament.
I grounded the plate, grid, and cathode of the other (unused) triode. This amplifier was a mono block.
I used an IXYS current source and set it to about 6.7mA. It was the plate load of the triode. This drove a 0.1uF coupling cap, and a 270k Ohm grid resistor of the next stage. I went to the 12AU7 tube curves (this is key to seeing the approximate results), and determined that I wanted about 3V of self bias, at 6.7mA of current. The load line I drew on the curves was 270k Ohms. 3V/0.0067A = 448 Ohms. I used a 470 Ohm self bias resistor. I got about 3.15V self bias. I bypassed the self bias resistor with a cap. The input signal can swing from about +3V to 0, to -3V. The plate is driving the high impedance current source and the 270k Ohm grid resistor.
I did not check the gain, but I bet it is somewhere between 15 and 20 (12AU7 with high resistance load), It was able to easily drive the output stage grid with low distortion.
Looking into the cathode, we see 1/Gm, + Rp load/u 1/2600 uMhos + 270K/20; 385 Ohms + 13.5k Ohms. This is so high of a cathode impedance, we only need to bypass the 470 Ohm self bias resistor. If Xc of the bypass cap at 20Hz = 47 Ohms, the frequency response will be degraded by far less than 1 dB at 20Hz. Xc of 170uF = 47 Ohms. Use a 200uF bypass cap.
Hopefully that will show you how I sometimes design an input/driver stage. I do not always do these things the same way, just been doing it so many times over the years.
I try to keep circuits as simple as I can. I am not a fan of SRPP (I have used it). And I do not use pentodes, except as triode wired output stages (pentode and beam power tubes in 'triode' mode).
Last edited:
I know what you meant though. The feedback factor (beta) as in
Aclosed = Aopen/(1+Beta*Aopen)
is set by the feedback voltage divider ratio. That doesn't change since the resistors are fixed. But (I think now summarizing your point), if Aopen is reduced it affects how much feedback is applied, which tends to stabilize Aclosed, if Aopen >> Aclosed.
Last edited:
kward,
You are correct, I left off the unity factor. I tend to think more simply, so often forget about the fine points. It is kind of like if you keep the ratios big enough, the 1 is effectively almost dropped out of the result.
I generally think of negative feedback as being effective, as long as: 1. There is not so much negative feedback versus the gain/phase of the amp (including the output transformer, and the reactive speaker load) that the amp becomes unstable. And 2. As long as there is a reasonable amount of open loop gain, so that the feedback resistors reasonably accurately set the gain. Very low open loop gain, and closed loop resistors that 'set a high gain ratio' does not work either. You can not have an open loop gain of 14dB, and set the closed loop gain resistors to 24dB.
But you know all that.
I use Triode output devices; and Pentodes and Beam Power output devices that are wired in Triode Mode. And with these output stages, I do not use negative feedback.
You are correct, I left off the unity factor. I tend to think more simply, so often forget about the fine points. It is kind of like if you keep the ratios big enough, the 1 is effectively almost dropped out of the result.
I generally think of negative feedback as being effective, as long as: 1. There is not so much negative feedback versus the gain/phase of the amp (including the output transformer, and the reactive speaker load) that the amp becomes unstable. And 2. As long as there is a reasonable amount of open loop gain, so that the feedback resistors reasonably accurately set the gain. Very low open loop gain, and closed loop resistors that 'set a high gain ratio' does not work either. You can not have an open loop gain of 14dB, and set the closed loop gain resistors to 24dB.
But you know all that.
I use Triode output devices; and Pentodes and Beam Power output devices that are wired in Triode Mode. And with these output stages, I do not use negative feedback.
In my post # 170, I could have saved some calculation work and save a part, and eliminated the 200uF bypass cap altogether with minimal change of the results. If the plate load had been a 20k Ohm resistor, that would have needed a bypass cap. Cathode impedance 385 Ohms in series with 20k/20 = 1k Ohm.
trobbins,
Thanks! I think many of us can often do a better job. I will try to remember to do more schematics for some of my discussions.
Wow! I see schematics without fuses. And I see them without bleeder resistors. It takes more work to do ground nodes, but is very important for several reasons, including hum reduction, and ground loop reduction.
Others will learn if we do it right.
Thanks! I think many of us can often do a better job. I will try to remember to do more schematics for some of my discussions.
Wow! I see schematics without fuses. And I see them without bleeder resistors. It takes more work to do ground nodes, but is very important for several reasons, including hum reduction, and ground loop reduction.
Others will learn if we do it right.
The aim is for the OP to update their schematic relevant to the present topic of discussion, which has now turned (after 146 posts) to gain structures within a GNF configuration. Always good to have an accurate reference point for discussions.
Apart from the actual monoblock, and a desire to modify it, it would be good to appreciate what is wrong or below-par with the present total sound system such that focus has turned to the final amp monoblock(s).
Also an appreciation of what performance the present monoblock provides would be good to have in front of the forum, such as if the OP has made any measurements such as frequency response, output signal level versus THD, signal to noise, or what intentions there may be about peak signal level the OP wants to reach, or what signal is used as input and whether that is bandwidth limited in any way. I can appreciate from posts so far, that there may be little to go on here, but just in case
Apart from the actual monoblock, and a desire to modify it, it would be good to appreciate what is wrong or below-par with the present total sound system such that focus has turned to the final amp monoblock(s).
Also an appreciation of what performance the present monoblock provides would be good to have in front of the forum, such as if the OP has made any measurements such as frequency response, output signal level versus THD, signal to noise, or what intentions there may be about peak signal level the OP wants to reach, or what signal is used as input and whether that is bandwidth limited in any way. I can appreciate from posts so far, that there may be little to go on here, but just in case
Hello,
I received all the parts to try to reduce the gain, I’ll see what happens this weekend.
But I noticed another small issue that is only showing up in one of the mono blocks. Please note both amps are identical. I did some searching around and couldn’t find anything.
I noticed that the left channel starts to distort much sooner than the right when the volume is turned up to a “medium” amount. The right block doesn’t start to distort until much louder. Also I noticed that when no input signal is present and the amp is sitting idle, the speaker will slowly oscillate in and out slowly, sometimes pretty far excursion. It occasionally will go quick making a small pop sound.
I’m assuming the early distortion and speaker oscillating might be related.
As a test I, disconnected the NFB to see if that was causing it but it still did it.
Any ideas of what might cause this?
Thanks a lot
I received all the parts to try to reduce the gain, I’ll see what happens this weekend.
But I noticed another small issue that is only showing up in one of the mono blocks. Please note both amps are identical. I did some searching around and couldn’t find anything.
I noticed that the left channel starts to distort much sooner than the right when the volume is turned up to a “medium” amount. The right block doesn’t start to distort until much louder. Also I noticed that when no input signal is present and the amp is sitting idle, the speaker will slowly oscillate in and out slowly, sometimes pretty far excursion. It occasionally will go quick making a small pop sound.
I’m assuming the early distortion and speaker oscillating might be related.
As a test I, disconnected the NFB to see if that was causing it but it still did it.
Any ideas of what might cause this?
Thanks a lot
I guess you have tried swapping the valves from one amp to the other?
The other thing it is hard to see from the pictures (all blue wires) is how you have the output valve cathode resistor / capacitors wired. They should be 'paired' outer two together and inner two together. (Not left two and right two which is correct for the input grids.)
The other thing it is hard to see from the pictures (all blue wires) is how you have the output valve cathode resistor / capacitors wired. They should be 'paired' outer two together and inner two together. (Not left two and right two which is correct for the input grids.)
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.