Hello there. I am trying to add a compensation circuit very similar to the one explained here to a tube preamplifier based on a simple E88CC common cathode stage + cathode follower.
I am adding the compensation circuit between the balance and volume control. The thing is that I have high frequency attenuation.
Please see attachments.
If if bypass the balance control, the frequency response looks as it should in the LTspice simulation.
I believe this is an impedance mismatch problem but I cannot move the circuit after the cathode follower cause it works together with the volume control which is at the input.
NOTES: I dont want to move the volume after the cathode follower and also don't want to add another potentiometer for the compensation.
I also cannot use a four gang volume potentiometer.
I tried changing the balance pot value to a lower 10k, which seems to cure the problem but input impedance drops way too much to load sources properly.
Any ideas? THANK YOU!
I am adding the compensation circuit between the balance and volume control. The thing is that I have high frequency attenuation.
Please see attachments.
If if bypass the balance control, the frequency response looks as it should in the LTspice simulation.
I believe this is an impedance mismatch problem but I cannot move the circuit after the cathode follower cause it works together with the volume control which is at the input.
NOTES: I dont want to move the volume after the cathode follower and also don't want to add another potentiometer for the compensation.
I also cannot use a four gang volume potentiometer.
I tried changing the balance pot value to a lower 10k, which seems to cure the problem but input impedance drops way too much to load sources properly.
Any ideas? THANK YOU!
250k resistors, and a little capacitance causes high frequency roll off.
With a source impedance of R, when capacitive reactance load, Xc = R, there is a high frequency roll off.
An ECC88 gain that is at 90% of mu (u), has again of 30.
The grid to plate capacitance is 1.4 pF, and the Miller effect capacitance is 1.4pF x (1 +30) = 1.4 x 31 = 43pF.
Xc of 43pF at 20kHz is 185k.
A drive impedance R = 185k / 2 (92.5k), will be -1dB at 20kHz.
With a source impedance of R, when capacitive reactance load, Xc = R, there is a high frequency roll off.
An ECC88 gain that is at 90% of mu (u), has again of 30.
The grid to plate capacitance is 1.4 pF, and the Miller effect capacitance is 1.4pF x (1 +30) = 1.4 x 31 = 43pF.
Xc of 43pF at 20kHz is 185k.
A drive impedance R = 185k / 2 (92.5k), will be -1dB at 20kHz.
6A3sUMMER, thanks.
Yes. The thing is that I get way more attenuation in the HF than 1dB with the loudness circuit in. Please see first plot.
If I remove C1 and R6 at least it does not attenuate that bad, the circuit works for low frequencies and input impedance stays +100k
Any suggestion on how I can implement the compensation circuit?
BTW: whats the formula involved in this: ''A drive impedance R = 185k / 2 (92.5k), will be -1dB at 20kHz."
Thank you!
Yes. The thing is that I get way more attenuation in the HF than 1dB with the loudness circuit in. Please see first plot.
If I remove C1 and R6 at least it does not attenuate that bad, the circuit works for low frequencies and input impedance stays +100k
Any suggestion on how I can implement the compensation circuit?
BTW: whats the formula involved in this: ''A drive impedance R = 185k / 2 (92.5k), will be -1dB at 20kHz."
Thank you!
Last edited:
A single pole low pass RC has the characteristic that the -1dB point is one octave (1/2 frequency) versus the -3dB frequency.
They are -26 degrees phase shift at -1dB, and -45 degrees phase shift at -3dB.
In your first schematic, your signal comes from R3 and R4 in parallel, 55.6k Ohms to drive the next nework.
55.6k is loaded at high frequencies . . .
C1, R6, R1, and C2.
580pF, 120 Ohm, 2.7k Ohm, and 0.22uf to ground.
580pf is 13.7k Ohms at 20kHz
0.22uF is 36 Ohms at 20kHz.
55.6K is loaded at 20kHz by 13.7k, 120, 2.7k, and 36 Ohms = 16.556k Ohms
Attenuation at 20kHz is 16.6k/(55.6k + 16.6k) = 0.23. That is - 12.7dB
At 20Hz, 580pF is 13 Meg Ohms.
So the 55.6k is loaded by 100k +2.7k + 100k = 202.7k
202.7k / (202.7k + 55.6k), 0.785 attenuation; -2.1dB.
I am not sure what kind of loudness compensation you want.
Many old circuits used a pot with a loudness tap (4 leads total) and an R in series with a C from the loudness tap to ground.
Often, a switch was added in series with that RC, to be able to turn the loudness compensation on or off.
A simple, but reasonably effective loudness control, with only 4 parts per channel.
This gave a boost at low frequencies, but at higher frequencies, the response was flat.
More complex loudness compensation changed both the low frequencies and high frequencies, leaving the midrange alone.
Sorry I do not have any more ideas on this one.
For more than 2 years, I have not used a loudness control, bass control, treble control, etc.
My sound systems are affected by the power amplifier, that I make as flat as possible from 20Hz to 20kHz. Some fall off below 30Hz.
The other thing that affects the sound is which of many loudspeakers; far field or near field listening; position; and the room they are in.
The amplifier Damping factor also affects the loudspeaker frequency response.
The last thing, listening with the volume turned up does change the frequency response of our ears (and my ear).
I feel liberated to not be using loudness controli or tone controls of the amplifier. (I keep my amplifier free from such devices/circuits).
Your preference may vary.
If I find that a CD does not sound good on any of my varied systems, then I stop listening to that CD.
I am not going to try and compensate for what I consider to be a mistake in the production of that CD.
They are -26 degrees phase shift at -1dB, and -45 degrees phase shift at -3dB.
In your first schematic, your signal comes from R3 and R4 in parallel, 55.6k Ohms to drive the next nework.
55.6k is loaded at high frequencies . . .
C1, R6, R1, and C2.
580pF, 120 Ohm, 2.7k Ohm, and 0.22uf to ground.
580pf is 13.7k Ohms at 20kHz
0.22uF is 36 Ohms at 20kHz.
55.6K is loaded at 20kHz by 13.7k, 120, 2.7k, and 36 Ohms = 16.556k Ohms
Attenuation at 20kHz is 16.6k/(55.6k + 16.6k) = 0.23. That is - 12.7dB
At 20Hz, 580pF is 13 Meg Ohms.
So the 55.6k is loaded by 100k +2.7k + 100k = 202.7k
202.7k / (202.7k + 55.6k), 0.785 attenuation; -2.1dB.
I am not sure what kind of loudness compensation you want.
Many old circuits used a pot with a loudness tap (4 leads total) and an R in series with a C from the loudness tap to ground.
Often, a switch was added in series with that RC, to be able to turn the loudness compensation on or off.
A simple, but reasonably effective loudness control, with only 4 parts per channel.
This gave a boost at low frequencies, but at higher frequencies, the response was flat.
More complex loudness compensation changed both the low frequencies and high frequencies, leaving the midrange alone.
Sorry I do not have any more ideas on this one.
For more than 2 years, I have not used a loudness control, bass control, treble control, etc.
My sound systems are affected by the power amplifier, that I make as flat as possible from 20Hz to 20kHz. Some fall off below 30Hz.
The other thing that affects the sound is which of many loudspeakers; far field or near field listening; position; and the room they are in.
The amplifier Damping factor also affects the loudspeaker frequency response.
The last thing, listening with the volume turned up does change the frequency response of our ears (and my ear).
I feel liberated to not be using loudness controli or tone controls of the amplifier. (I keep my amplifier free from such devices/circuits).
Your preference may vary.
If I find that a CD does not sound good on any of my varied systems, then I stop listening to that CD.
I am not going to try and compensate for what I consider to be a mistake in the production of that CD.
Last edited:
You could try a 100 mH inductor in series with R1 and remove C1, or just remove C1 and accept there is no compensation at high frequencies. There shouldn't be anyway according to the ISO226:2003 equal loudness contours, as the high-frequency parts of them are essentially in parallel.
https://en.m.wikipedia.org/wiki/Equal-loudness_contour
https://en.m.wikipedia.org/wiki/Equal-loudness_contour
Thank you both! My circuit ended up like this.
Input impedance stays pretty close to 120k. The only bad thing is I need a 4PDT switch if I also want to switch the 150pF cap on and off to compensate the HF..
Input impedance stays pretty close to 120k. The only bad thing is I need a 4PDT switch if I also want to switch the 150pF cap on and off to compensate the HF..