FFT Distortion Signatures
Hello All,
@ petertub
Yes I did do the other half and other tubes. I will post more.
@ rongon
I will do other Anode resistors and higher B+ voltages. Now the line filter has only 400 Volt capacitors. 600 volt capacitors are on order. Plus I am still poking around to see any patterns and or surprises, and then I will fill in a spread sheet with noise voltages all done with the same test configuration. After that I will look more closely at different circuits and distortion signatures.
20 to 20
Thanks I neglected to tell about the output. There is a 0.68uf GE Polypropylene capacitor followed by a 1meg resistor to ground plus the analyzer is 100K + a few pf capacitance. First off I do not want too much load to confuse the EIN measurements. Later there will be real world loads added.
Today I plotted various used 12AU7 types most all were dreadful in terms of noise, even a “premium” pair found on ebay.
I plotted various New Old Stock 12AU7 types using the same test circuit, the same signal input voltage and output voltages very near 2 volts produced completely different looking Distortion Signature FFT plots. It is no wonder people “hear” differences when they roll tubes in and out of their equipment. The Phillips ECG Blue label 5814A’s (from a freshly opened case of 100) produced a string of harmonics out past H10. This is the tube that I adjusted the B+ voltage to null the 3rd Harmonic to near zero. The Sylvania 5963 produced 2nd and 3rd Harmonics only, a lot neater and less cluttered look than the Phillips ECG Blue label 5814A’s. THD’s measured much the same, it is the FFT Distortion Signatures that have differences.
See the attached plots.
Comments?
DT
Hello All,
@ petertub
Yes I did do the other half and other tubes. I will post more.
@ rongon
I will do other Anode resistors and higher B+ voltages. Now the line filter has only 400 Volt capacitors. 600 volt capacitors are on order. Plus I am still poking around to see any patterns and or surprises, and then I will fill in a spread sheet with noise voltages all done with the same test configuration. After that I will look more closely at different circuits and distortion signatures.
20 to 20
Thanks I neglected to tell about the output. There is a 0.68uf GE Polypropylene capacitor followed by a 1meg resistor to ground plus the analyzer is 100K + a few pf capacitance. First off I do not want too much load to confuse the EIN measurements. Later there will be real world loads added.
Today I plotted various used 12AU7 types most all were dreadful in terms of noise, even a “premium” pair found on ebay.
I plotted various New Old Stock 12AU7 types using the same test circuit, the same signal input voltage and output voltages very near 2 volts produced completely different looking Distortion Signature FFT plots. It is no wonder people “hear” differences when they roll tubes in and out of their equipment. The Phillips ECG Blue label 5814A’s (from a freshly opened case of 100) produced a string of harmonics out past H10. This is the tube that I adjusted the B+ voltage to null the 3rd Harmonic to near zero. The Sylvania 5963 produced 2nd and 3rd Harmonics only, a lot neater and less cluttered look than the Phillips ECG Blue label 5814A’s. THD’s measured much the same, it is the FFT Distortion Signatures that have differences.
See the attached plots.
Comments?
DT
That's precisely why I didn't waste my time trying to produce extended FFTs for my book; once you get past the 5th harmonic or so, whatever happens is unique to the exact tube you're using and therefore useless as general information. A lot of people don't seem to appreciate this and I was criticised for it. Haters gonna hate! 😀
If we can reduce the noise.
Merlin,
My hat is off to you.
It is my impression that in the frequency domain large sample sizes and averaging will unmask low level harmonics that we Sapiens will always be unaware of in the real time domain.
Reducing the noise is much more important than chasing low level high order harmonics.
DT
Merlin,
My hat is off to you.
It is my impression that in the frequency domain large sample sizes and averaging will unmask low level harmonics that we Sapiens will always be unaware of in the real time domain.
Reducing the noise is much more important than chasing low level high order harmonics.
DT
Last edited:
FFT Distortion Signature at 12 mA and -1.81 grid Volts is much improved
Hello All,
Interesting stuff, playing with these tubes over the years, building power supplies and drawing load lines, despite serious effort the outcome can have some random effects.
Now with some bench instruments I can measure the results.
I put together a spreadsheet and stepped through 10 volt increases in the B+ power supply voltages and recorded the noise at the output of this 12AU7 bread board. I noted that the noise at the output decreased at each 10 volt increase in the supply voltage up to ~ 12mA current measured as voltage drop across the 300 ohm cathode resistor. Above ~12 mA the noise began to creep back up. I did the same procedure with a 150 ohm cathode resistor and saw the same result with the lowest noise at ~12 mA across the cathode resistor. For this 5814A Philips tube at 12 mA the EIN calculated to be 1.24 uV, the gain was 14.38. The volts across the 150 Ohm cathode resistor measured 1.81Volts. Volts between the anode and cathode measured 131 Volts.
The over the previous test at 12 mA and -3.5 grid volts. For 2 Volts output +6dBV, 2nd H -54dbr, 3rd H -88dBr, 4th H -114dBr and 5th and higher H’s below -120dBr. THD+N ~0.20% at 2.0 volts output.
See attached FFT plot.
I measured several other 12AU7 type tubes, the lot of Phillips 5814A peformed best for noise and distortion.
Comments?
DT
Hello All,
Interesting stuff, playing with these tubes over the years, building power supplies and drawing load lines, despite serious effort the outcome can have some random effects.
Now with some bench instruments I can measure the results.
I put together a spreadsheet and stepped through 10 volt increases in the B+ power supply voltages and recorded the noise at the output of this 12AU7 bread board. I noted that the noise at the output decreased at each 10 volt increase in the supply voltage up to ~ 12mA current measured as voltage drop across the 300 ohm cathode resistor. Above ~12 mA the noise began to creep back up. I did the same procedure with a 150 ohm cathode resistor and saw the same result with the lowest noise at ~12 mA across the cathode resistor. For this 5814A Philips tube at 12 mA the EIN calculated to be 1.24 uV, the gain was 14.38. The volts across the 150 Ohm cathode resistor measured 1.81Volts. Volts between the anode and cathode measured 131 Volts.
The over the previous test at 12 mA and -3.5 grid volts. For 2 Volts output +6dBV, 2nd H -54dbr, 3rd H -88dBr, 4th H -114dBr and 5th and higher H’s below -120dBr. THD+N ~0.20% at 2.0 volts output.
See attached FFT plot.
I measured several other 12AU7 type tubes, the lot of Phillips 5814A peformed best for noise and distortion.
Comments?
DT
- Status
- Not open for further replies.