Question about coupling capacitors and filtering

Discussion in 'Tube Audio' started by Vez, Oct 10, 2017.

  1. Vez

    Vez New Member

    Messages:
    40
    I hope this is a stupid question but, are any other frequencies blocked by coupling capacitors that are lower than frequencies blocked from a previous stage's coupling capacitor? In other words, if the first stage couples to the second stage with a .01uf, is it not pointless to expect larger capacitors in any subsequent stages to filter no more than the .01uf?
    Does the NF line introduce lower frequencies that were previously filtered to ground?
    Thank you,
    Rick
     
  2. gadget73

    gadget73 junk junkie Subscriber

    Messages:
    32,798
    Location:
    Southern NJ
    it depends on the grid resistor too. It forms a high pass filter with the cap and the resistor value. Basically the higher the grid resistor, the lower the cutoff frequency is for a given cap value. Often the stages need different rolloff points in order to keep it stable. Its not really because of the roll-off though, but because filters introduce phase shift. Make it all line up and it may oscillate horribly. There is a formula for crunching it all out if you really want to know

    Fc = 1/2πRC

    that gives the -3db point. Personally I cheat and use calculators online that let me put in the cap value and resistor.


    Negative feedback helps compensate for any loss in low frequency because of cap values but it may not be able to completely overcome it. Again, depends on specific values. If the caps are just way too small, all you'll get is a bunch of added distortion as the amp attempts to overcompensate.
     
    ferninando likes this.
  3. Palustris

    Palustris Well-Known Member

    Messages:
    663
    You need to look at the situation logically. Picture a machine that drops jelly beans in a rainbow of colors onto a conveyor belt and that conveyor belt drops the jelly beans into a bucket. Now as the conveyor belt passes me I remove all the red jelly beans, how many red jelly beans are in the bucket?
     
    CopperWizard likes this.
  4. kward

    kward AK Subscriber Subscriber

    Messages:
    3,002
    Location:
    Utah
    It is usually better (but not always) to put the filter with the highest rolloff frequency on the output stage.

    As a practical example (with made-up numbers): an amplifier has two R/C couplings, one after the gain stage with a -3dB point at 1 Hertz, and the other after the inverter with -3 dB point at 15 Hertz. In this example, farther down the amplification chain is where the more dominant (15 Hertz) rolloff filtering occurs. The rolloff frequencies of these two filters are nearly 4 octaves apart (4 doubles of frequency), so that the phase shift incurred from each filter has less tendency to overlap. Phase shifts are additive. If each filter phase shifts the signal by 90 degrees at 100 Hz let's say, then the resultant signal will be phase shifted by 180 degrees at 100 Hz. When the feedback signal phase shift exceeds 180 degrees at a given frequency as compared to the input signal to the stage where the feedback is connected, and has gain greater than unity at that frequency, the amp will oscillate.

    Of course, if there's no feedback applied, then you can phase shift all you want, and the amp won't oscillate. It only becomes a concern in feedback systems.
     
    Last edited: Oct 11, 2017
  5. BinaryMike

    BinaryMike Pelagic EE Subscriber

    NFB systems compare the input signal to the output signal and then drive the amplifier with difference signal, which has the effect of boosting previously rolled-off portions of the audio spectrum at both ends. This effect can be quite dramatic with high loop gain, so in the sense that otherwise inaudible LF content may be restored to its rightful place in your ears, NFB does indeed reintroduce signal content that might otherwise have been lost to filtering, be it intentional or unintentional.
     
  6. gadget73

    gadget73 junk junkie Subscriber

    Messages:
    32,798
    Location:
    Southern NJ
    I typed up a big long thing using an example out of an amp I own, then lost it.

    Nutshell, the Pilot SA-260 has a 3hz -3db point on the voltage amp and a 6hz -3db point feeding into the output tubes. The "middle", since its a paraphase, is also 3 hz.

    Increasingly I like the idea of direct coupling as much as practical. It gets rid of the phase shift concerns.

    There are also sometimes cap choices made to work with limitations in other components. If the output transformers on a little amp are completely done by 50 hz, there is little point in amplifying signal below that point. Stuff in consoles sometimes had an LF roll-off because the turntable sits basically right above the speakers and acoustic feedback needs to be considered.
     
    Last edited: Oct 11, 2017
  7. Vez

    Vez New Member

    Messages:
    40
    Thanks for your comments Gaget.
    What's wrong with having hi voltage on the grid if it's cathode biased?
    I have an old Ampeg "flip top" guitar amplifier I thought it was interesting in that it's tone stack was after V2. V2, a twin tripod, had its first plate directly coupled to its second grid which was a cathode follower, then the tone stack and the second triode had no plate resistor. I guess in taking to signal from that high voltage plate would make for the tones stacked seeing a high impedance output and would lose gain as the signal went through the tone stack?? So it takes the signal from the cathode.
    (I think if you click on image you should get the entire photo. More images here: https://goo.gl/photos/YvGrdEdkxQ4PFDTt5 ).

    [​IMG]
    [​IMG]
    [​IMG]
     
  8. gadget73

    gadget73 junk junkie Subscriber

    Messages:
    32,798
    Location:
    Southern NJ
    Nothing especially wrong with it, but it moves the cathode voltage up a bunch, and then you get into considerations about heater-cathode voltage differences. With a bunch of tubes on one string you end up picking a compromise voltage that tries to keep everything happy even if its not absolutely ideal for anything. If that Ampeg were mine for example, I'd probably connect the center of the hum pot to the cathode of the 6L6 tubes. That should put it 20 odd volts positive to ground, reducing the H-K voltage on V2B by the same amount.

    In the Ampeg, the cathode follower setup was probably to provide the tone controls with a nice low impedance source. I don't see anything there that strikes me as being difficult to drive, but it won't hurt either. Of course it could have simply been that they wanted to use a 6SL7 for gain in that spot and decided to not waste the extra triode.
     
  9. BinaryMike

    BinaryMike Pelagic EE Subscriber

    Probably the original idea was to simply prevent diminished gain in the first half of that 6SL7, by buffering its load.
     
  10. Audiovet

    Audiovet Well-Known Member

    Messages:
    590
    Location:
    Pretoria, South Africa
    Vez,

    Stability with NFB is a whole field on its own. If interested you will probably have advantage from reading articles on the same. If the basic conditions are taken into account, there should be no reason to need d.c. coupling with all its added complications. (It gets tricky, but I have managed to get stability with 4 x l.f. cut-offs in the circuit at 28 dB of nfb.) Modern capacitors are physically small and do not need a lot of mounting space.

    In that sense (n.f.b. stability) one would need some cut-offs to be quite lower than what is required for listening. (One must not overlook the output transformer, which also constitutes one cut-off pole.) But as said, one will need to know which Nyquist stability criteria are involved. One will need a low frequency signal generator and oscilloscope to check for stability.
     

Share This Page