Variable NFB, not quite science

Discussion in 'Shock Brother's DIY Amps' started by shortcircuit, May 25, 2019.

  1. shortcircuit

    shortcircuit Tele-Meister

    Posts:
    229
    Joined:
    Sep 9, 2017
    Location:
    North Carolina
    In preparing for my Marshall mashup amp, I settled on a PPIMV. A drawback of this setup is that the NFB voltage coming back to the phase inverter will be affected. I started toying around with the idea of a variable NFB setup with a 10K resistor in line with a 100K pot, so that as I turn down the master volume, I can adjust the NFB pot to compensate. I had built 2 other amps with a NFB pot, and I had put in a 0.82uF cap to block the DC from the phase inverter so the pot won't be scratchy. I didn't have any instability with this setup, but reading on this topic in various forums, the issue of phase shift crept up. There were mentions of lower frequencies going out of phase due to the capacitor in series with the resistor, hypothetically causing oscillations and other nastiness. Now, I have no formal training in electronics, and the last time that I did anything in a classroom with electronics was basic physics in college 30 years ago. I tried to read about phase shift in a R-C circuit, and my brain locked up. My takeaway was that voltage and current will be out of phase in a series R-C circuit, and it will get worse the lower the frequency. In a vacuum tube amp, however, there shouldn't be much current at all in the NFB circuit for that to have an effect, and I have 2 real-life examples.

    Anyway, tonight I desoldered the 0.82uF cap in one of the amps with the NFB pot and busted out the oscilloscope. Looking at the output of the phase inverter, I saw absolutely no difference in the waveform with the cap in circuit or out as I turned the frequency down on the signal generator. The output transformer crapped out at around 50Hz though. I didn't write any of this down, so this is not science.:p

    Any thought?
     
  2. Tom Kamphuys

    Tom Kamphuys Tele-Meister

    Age:
    39
    Posts:
    178
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    The important phase shift is between the input signal and output signal of an RC filter. The 'absence' of current is not important.

    A feedback circuit relies on the principle that a part of the output is coupled back and subtracted from the input. This improves e.g. linearity.

    If the phase of the signal is shifted anywhere in the loop, the subtraction is getting crappy and might even become addition (for a particular frequency range). If the gain is larger than one, the amplifier starts oscillation/screaming from just the noise (e.g. resistor noise).

    And while typing all this I started thinking :)
    The NFB should not contain any DC. It is taken from the output of the output transformer.
     
  3. shortcircuit

    shortcircuit Tele-Meister

    Posts:
    229
    Joined:
    Sep 9, 2017
    Location:
    North Carolina
    I wasn't blocking DC from the NFB signal. I was blocking DC from the phase inverter.

    I had looked at the NFB signal where it's injected into the phase inverter after passing through the resistor/pot and capacitor. As I turned down the frequency, the signal did not budge, where I had expected it to shift to the left on the x-axis. At 80 Hz, with 10K Ohm of impedance 0.82 uF of capacitance, there was supposed to be a -16 degree shift (I didn't calculate this myself. I used an online calculator:D).

    So why is there no shift? What am I missing?

    I'm not complaining or anything, as the amps work for me.;)
     
  4. robrob

    robrob Poster Extraordinaire Ad Free Member

    Posts:
    7,482
    Joined:
    Dec 29, 2012
    Location:
    United States
    You don't need a cap if you inject the NFB at the tail of an LTP phase inverter. Are you injecting it at the LTP grid?
     
  5. shortcircuit

    shortcircuit Tele-Meister

    Posts:
    229
    Joined:
    Sep 9, 2017
    Location:
    North Carolina
    The injection point is at the tail.:oops::lol:

    I had some scratchiness before I put in the cap. The voltage is not high, and shouldn't damage the pot, but I'm anal like that.

    Now that I think about it, this whole exercise is like a Seinfeld episode. Totally about nothing.
     
  6. Tom Kamphuys

    Tom Kamphuys Tele-Meister

    Age:
    39
    Posts:
    178
    Joined:
    Sep 18, 2018
    Location:
    The Netherlands
    Could you provide us with a drawing/schematic?

    And how did you look at the signal? E. g. If you use just one signal the scope will start at a certain signal level. No phase shift will be seen that way.
     
  7. shortcircuit

    shortcircuit Tele-Meister

    Posts:
    229
    Joined:
    Sep 9, 2017
    Location:
    North Carolina
    NFB.jpg

    I looked at the signal at the red oval and blue circles simultaneously. The top blue circle was in phase and the lower blue circle was out of phase as expected. The presence pot has a no-load mod and was turned off.
     
  8. robrob

    robrob Poster Extraordinaire Ad Free Member

    Posts:
    7,482
    Joined:
    Dec 29, 2012
    Location:
    United States
    Why two caps for NFB and presence? Connect the NFB to the south side of the .68uF presence cap and everyone's happy. The two caps will interact at certain presence and rudeness settings. They form a capacitive voltage divider.
     
    shortcircuit likes this.
  9. shortcircuit

    shortcircuit Tele-Meister

    Posts:
    229
    Joined:
    Sep 9, 2017
    Location:
    North Carolina
    Doh! Right in front of my face and I didn't see it. That's why the NBF pot got squirrely with the presence turned up. I'm just going to keep it simple and get rid of the 0.82uF cap and live with the little bit of scratchiness.
     
IMPORTANT: Treat everyone here with respect, no matter how difficult!
No sex, drug, political, religion or hate discussion permitted here.


  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.