R.G.'s SRPP amp in BSIAB2?

[gaussmarkov]

for various reasons, the BSIAB2 caught my attention again and this time i noticed the relatively high values in the voltage dividers for the two mu-amps that the circuit contains.  this is because this time round i have already read R.G.'s note on noiseless biasing and the SRPP amp.    so now i am wondering:  has anyone tried replacing the two 1M resistor pairs with 10K's and a 500K coming out of the divider junction along with a cap to shunt thermal noise to ground?

i'd do this myself, but i am already over my head with other projects.  so i just thought i would throw it out there.    maybe the effect is not so useful as in boosters that are using 10M resistors in the divider?

--gm

[R.G.]

The thermal noise of any resistor is a function of its resistance and the absolute temperature. More resistance and higher temperature, more noise even at zero current.

Resistors also have current noise. This noise is a function of the resistance and the current. With zero current, there is zero current noise - only the thermal noise remains.

So if you have two large resistors across a power supply to generate a bias voltage, they generate whatever thermal noise they generate. However, if you have to meet some parallel combination of resistors, as you do to make a large input impedance, the resistors must each be twice as large as the impedance you want, and they have significant current running through them.

The noiseless biasing trick uses low resistance (and therefore lower thermal noise) resistors to generate the voltage, then shunts the noise to ground with a bypassing cap. The resistor which goes to the active element needs only be 1x the necessary input impedance, and it has only the input bias current running through it. The input bias current should be at least a factor of ten less than the current flowing through the biasing string so that the biasing string can be relatively independent of the bias current. So the thermal noise of the noiseless biasing resistor is half the thermal noise of either of the high impedance resistors it replaces, and a factor of square root of two lower than that because the thermal noise of the two resistors adds by RMS. It has less than one tenth of the current noise generated by its lower currents.

Is this worthwhile?

It depends. The more gain you have following the input, the more you need a low noise input. Thermal noise starts low, so the full gain of the following amp is applied to it. Guitar signal is clipped, so it only gets so big. The higher the gain following the input, the more the noise gains on the signal because it's not clipped.

[gaussmarkov]

R.G., thanks for the clarification.  given that the bsiab2 circuit is fairly high gain, putting in the SRPP amp could be a good thing.  on the other hand, i have not seen complaints about its noisiness.

here's a follow-up question:  what value cap would be good for shunting the noise?  something familiar like .1uF?

cheers, gm

[R.G.]

It depends on the value of the resistors making the bias source. You have to make that capacitor be a much lower impedance than the parallel combination of the two resistors at the lowest frequency your circuit sees; for most of us, this is in the range of 1Hz to 42Hz. So the cap needs to be big, up in the uF range at least. We also know that hiss is high frequency stuff, and that the impedance of an electrolytic rises again at quite high frequencies, so putting a 0.1uF monolithic ceramic cap in parallel with the electro will help eat hiss at the high end. You can't neglect either extreme of frequency.

[gaussmarkov]

It depends on the value of the resistors making the bias source. You have to make that capacitor be a much lower impedance than the parallel combination of the two resistors at the lowest frequency your circuit sees; for most of us, this is in the range of 1Hz to 42Hz. So the cap needs to be big, up in the uF range at least. We also know that hiss is high frequency stuff, and that the impedance of an electrolytic rises again at quite high frequencies, so putting a 0.1uF monolithic ceramic cap in parallel with the electro will help eat hiss at the high end. You can't neglect either extreme of frequency.

interesting.  i thought that hiss would be the only issue.  somehow that's what i imagined thermal and current noise would be.    R.G., thanks for following up.  i don't expect you to answer all of my questions, but i will keep posting in the hope that someone will.

i have also been puzzling over the importance of the "parallel combination of the two resistors" in the voltage divider.  R.G. mentions this in his article as well.  it's confusing to me because the resistors are not actually parallel.  do we treat them this way because both resistors in the voltage divider are connected to an AC ground, one +9V and the other ground?

thanks in advance, gm

[gaussmarkov]

i have also been puzzling over the importance of the "parallel combination of the two resistors" in the voltage divider.  R.G. mentions this in his article as well.  it's confusing to me because the resistors are not actually parallel.  do we treat them this way because both resistors in the voltage divider are connected to an AC ground, one +9V and the other ground?

thanks in advance, gm

anyone?

[Doug_H]

"R.G.'s" ?

http://www.tubecad.com/may2000/

[gaussmarkov]

"R.G.'s" ?

http://www.tubecad.com/may2000/

nope. not R.G.'s.  and he never claimed it was his either.  it was just my shorthand for referencing where i read about. 

[Doug_H]

Thanks, just wanted to be clear.

[brett]

Hi
I think that the bias resistors can be considered parallel if you consider that from the perspective of the bias point (ie where they are joined) their far ends are both connected to AC gound (V+ and DC ground are connected in terms of AC by a big cap).  Although an odd concept at first, it can be useful when designing circuits to remember that a capacitor-coupled DC supply has two AC grounds.
cheers

[gaussmarkov]

Hi
I think that the bias resistors can be considered parallel if you consider that from the perspective of the bias point (ie where they are joined) their far ends are both connected to AC gound (V+ and DC ground are connected in terms of AC by a big cap).  Although an odd concept at first, it can be useful when designing circuits to remember that a capacitor-coupled DC supply has two AC grounds.
cheers

cool. thanks brett!