Mu amp cap values - do they matter? (mu = BSIAB gain stages)

[brett]

Hi
after a long time I'm finally building a BSIAB2.  The mu amps are a pair of 2N5457s and a pair of J201s.
The cap linking the source and gate of the 2N5457 that is the active load (ie the "top" one) is 0.068uF.  The corresponding cap for the "top" J201 is 0.1uF.  If I recall correctly, Jack Orman uses a very large cap in his version of the mu amp.

But I'm wondering whether the mu amp and the BSIAB2 can be at all sensitive to the difference.

My thinking is that both of the JFETs are biased by a pair of 1M resistors.  I figure that the JFET impedance is so high it can be ignored, so the imput resistance is about half a megohm.  With this much impedance, both 0.068 and 0.1uF caps have cutoff frequencies of about 5 Hz.  (Hence I regularly use smaller caps - e.g. 0.01uF - as inputs to JFETs)

So...
Is the difference between the caps meaningful? 
Does the low cutoff frequency associated with oversize caps allow for some desirable effect (holding a slow-release charge on the gate?   What sort of distortion would that provide?)

Any comments appreciated.

[DougH]

Yes, the size of those caps do make a big difference. They are usually chosen deliberately to tune the circuit in a particular way. I don't why the AMZ mu-amp caps were originally spec'd so high.

[intripped]

I'm sorry to resume this old thread, but this is exactly what I'd like to understand

Could someone please explain a bit more, what is the effect of the size of that capacitor?
Thank you

[robthequiet]

Here is a site that might help with the basics: http://www.falstad.com/circuit/e-capac.html

In a circuit such as the BSIAB where you have a capacitor joining gate and source, I would think of it in terms of feedback of a certain frequency to either damp or boost the response, kind of like a fixed tone control. But it's not necessarily that simple, as all of the components in the signal chain play a role in tone shaping, so there is a certain amount of engineering or educated guessing as to the values of the caps. Generally, small value caps work on higher frequency signals, whereas large values will generally be chosen to pass a broader range.

Here is a thread that includes some pretty good discussion/

[intripped]

Here is a thread that includes some pretty good discussion/

This thread is extremely informative, thank you a lot for this link!

[robthequiet]

You are very welcome. Glad to help.

[Ice-9]

R.G Keen has a good source of info on his Geofex site about the function and modding of these.

http://www.geofex.com/Article_Folders/modmuamp/modmuamp.htm

[Boner]

Sorry to bring up a dead thread but I looked for an answer and couldnt find it, so I figured I would ask wheere its most relevant



http://www.geofex.com/Article_Folders/modmuamp/modmuamp.htm

For the noiseless biasing scheme at the bottom, what exactly is the purpose the cap going from the bias point to ground? Is it to keep the bias point stable and filter out any power supply noise?

Also, what is a healthy range for the value? That was my biggest question. I figured in the high uF range?

[EBK]

That cap, as far as I can tell, is there to provide an AC ground to shunt noise, not to stabilize the DC voltage as a reservoir cap.  A MLCC, perhaps?  Maybe 100nF?  I'm half guessing here.

[Boner]

That makes sense.

AC ground in this case would be the halfway point between the two rails, IE Vbias?
And DC ground is zero...

In that posting the cap is tied to DC ground which confuses me.

[EBK]

That makes sense.

AC ground in this case would be the halfway point between the two rails, IE Vbias?
And DC ground is zero...

In that posting the cap is tied to DC ground which confuses me.

Here, I'm taking an analytical short cut.  I'm saying AC ground to mean that when we imagine that cap as a short circuit (i.e., as frequency goes higher, the cap's impedance goes down), that node is connected to ground through a low impedance path.  At very low frequencies (approaching 0, i.e., at DC only), we can view the cap as an open circuit, and under such conditions, that node would not be considered grounded.  Does that make sense? 

[PRR]

> keep the bias point stable and filter out any power supply noise?

Yes. (Here, "noiseless" is more like "buzz-less".)

So what's the cap value?

We have an R-C network to pass DC but block power supply crap. Thus it acts as a low-pass, a high-cut.

The filtering effect depends on the R, the C, and the frequencies we want to pass and reject.

As Audio Techs we should have a mental cheat-sheet of R and C values for common audio frequencies. To pass/block 20Hz, the bottom of the audio band, it looks like this:
10r +1,000uFd = 20Hz
100r + 100uFd = 20Hz
1K + 10uFd = 20Hz (15.9Hz but so what?)
10K + 1uFd = 20Hz
100K + 0.1uFd = 20Hz
1Meg + 0.01uFd = 20Hz

We see 10K resistors. Pencil the 1uFd value.

Power crap is usually 100/120Hz. 20Hz is 5:1 down so 1uFd against 10K is 5:1 reduction of power crap.

In a HIGH gain circuit, 5:1 reduction is not a lot. We may need much more.

In this case, 1uFd is an "awkward" value. Big for a film cap but small for an electrolytic cap. A 10uFd e-cap will be about the same size and price, and filter 10X better.

So I would re-pencil a 10uFd e-cap here.