an unmarked component on the AMZ Mu-Amp article

[mordechai]

The article I'm looking at is this one:

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

You can see that on the third diagram, the "SRPP amp with noiseless biasing", there is an unmarked cap to ground coming off the junction between the two 10K's and 4.7M resistors on the upper JFET.  What is the purpose of that cap, and what range of value is usually effective when applying it to this modification to a Mu-Amp set up?

[induction]

This is a simple power filter cap. This is how noiseless biasing, as the article says, 'offers the ability to shunt thermal noise from the junction of the two 10K's to ground'. The standard values of filter caps can be used here: 10u - 100u. Some go higher, some add a 100n in parallel to a larger cap.

The general idea here is that power supply ripple is much worse when it's fed into the input of a gain stage than when it's just used as V++ because the ripple gets amplified along with the signal.

This noiseless biasing scheme is very common and can be used in place of simple voltage divider biasing in most cases. It's especially useful to remember when adapting circuits for stompboxes that were originally designed for battery use only (on-board guitar circuits and older effects that plug directly into the guitar like orange squeezers and lpb's) and don't have ripple filtering included.

[mordechai]

Ah...so it's similar to what we find on the Beavis audio site where the DC 9V goes to a small value resistor, then a moderate sized electrolytic cap, and then a much smaller cap? 

[induction]

Sort of. What you describe is how you filter V++. Noiseless biasing allows you to apply the same idea to filter Vref. You have to add an extra resistor between the voltage divider and the jfet (or bjt or op-amp) input so that the filter cap doesn't suck all of the treble from your input signal to ground.

[mordechai]

You have to add an extra resistor between the voltage divider and the jfet (or bjt or op-amp) input so that the filter cap doesn't suck all of the treble from your input signal to ground.

So in the third diagram on the AMZ site listed above, would that resistor  be the 4.7M?

[induction]

I should have posted this link to the AMZ buffer article earlier. Look at the very first circuit snippet, and then look how it gets used in the rest of the article. Once you recognize it, you'll see it everywhere.

Edit: We posted at the same time. Yes the 4.7M is there to maintain the high impedance from signal to ground while allowing the voltage divider resistors to be lower in value.

[mordechai]

Induction, this is all very helpful and I'm starting to understand how this modification improves the performance of the MU-Amp structure.

I am curious about the function of the 1K resistor in the "fixed version" schematic.  Does it matter where the 1K is situated with respect to the Vout of the gainstage?  Take a look at the image I've drawn up below illustrating what I mean by my question:

http://s1130.photobucket.com/user/mordechaibenzev/media/MovedVout_zps761d6136.png.html?filters[user]=124527051&filters[recent]=1&sort=1&o=0

Would either option work?

[induction]

I'm confused about that resistor myself. I simmed the circuit in LTSpice, and I found the AC gain to be the same either way, but the DC bias of the output was higher when the output is connected to the source of the upper jfet. That makes sense because some dc will be dropped across that resistor.

My suspicion is that the output impedance will be 1k lower with the output taken from the source of the upper jfet rather than from the drain of the lower jfet because the signal won't have to go through that resistor. I couldn't find any evidence for that in the sims, though.

I'm suspicious whether I'm catching everything in the sim, because when I compared circuits with and without the 1k resistor I find the opposite of what Jack says in the article. He claims the 1k resistor increases the gain by a factor of 2, and I find that it decreases the gain by a factor of 2. So I'm not sure if I'm doing something wrong or if Jack got it wrong.

Maybe someone with more experience or theoretical understanding of this circuit will chime in.

[dwmorrin]

In general, the output of a source is low impedance, while the output of the drain will be higher impedance.
I could not see the photobucket image (link didn't work for me), but I think the answer is "yes", it matters where the 1k is situated.
As R.G. says, it "magically" allows the upper FET to be a low impedance source follower, while simultaneously providing an active load for the lower FET's drain.

I find FETs to be difficult to simulate, personally.  Sometimes the breadboard and scope can be much more educational.

[mordechai]

DW, let me try to embed the image itself:



You will see  (hopefully) that in the image on the left, which corresponds to what is there on Jack's site, the output is taken from above the 1K resistor, and thus from the source of the upper FET.  In the image on the left, you'll see that the output is taken from below the 1K resistor, and thus from the drain of the lower FET.  So, if I understand you correctly, the second picture (on the right) provides a scenario where there is a higher output impedance than in the picture on the left.  Is that correct?  If the output were going into another gain stage, what are the pros and cons of one scenario vs. the other?

[dwmorrin]

If you follow the phase of the signal, you'll see that it is the same phase, and should be about the same voltage across the 1k resistor.
If you look at it, scratch your head, pull out ohm's law... you'll eventually come to the conclusion that the 1k resistor "looks" like a much bigger resistance at ac.
This "big resistor" effect makes the lower FET a better voltage amplifier, and makes the upper FET a better current amplifier.
The upper FET's source output should have a low output impedance (RG says 200Ω), while the lower FET's output resistance is based on all that stuff connected to the drain (1k, FET, 4.7M resistor...).

Try running the output of the first picture into different load resistors (1M, 100k, 10k, 1k), and do the same for the second picture.  See if that makes any difference.
With the wide world of FETs, it may be the type of experiment you just have to breadboard to get the same results that RG got.

[mordechai]

I will indeed test out a few different scenarios here.  In your opinion, would changing the value of the 4.7M resistor be a worthwhile place to experiment?  I wonder if upping it to, say, 10M might have a beneficial effect on the output impedance of the lower FET's drain...

[dwmorrin]

In your opinion, would changing the value of the 4.7M resistor be a worthwhile place to experiment?  I wonder if upping it to, say, 10M might have a beneficial effect on the output impedance of the lower FET's drain...

In my opinion, it would have a detrimental effect to the output impedance of the lower FET's drain (assuming you're taking a voltage output there).  Raising that resistor could only raise the output impedance.  You want a low output impedance if you're working in a voltage drive world (which we are).
I think you're going around in a circle... re-read the original article.
I like that you haven't given up on the lower FET as an output - but why?  I think no matter how you slice it, the source output is going to end up better on paper.

[mordechai]

Honestly, it's mostly because I have a PCB project for a MOSFET boost similar to a SHO.  I saw Rick Holt's great "superheated SHO" project (which created a Mu-Amp section for the MOSFET in the circuit) and built it.  I loved what it did, but it increased noise quite a bit, so when I read Jack's article on silent biasing for Mu-Amps, I thought it would be worth exploring.  However, the PCB project has th output routed com the source of the MOSFET, so I am examining the options I have without hacking my PCB to pieces or having to purchase new materials.

[dwmorrin]

More gain=more noise.  When you're building voltage amps of any type, remember that "gain" is like zooming in with a microscope... all circuits have some inherent noise floor that you will quickly "zoom" down to.
Real "low noise" microphone preamps go through all sorts of hoops to try and maximize the signal to noise ratio.
While there are some tricks to know, you must always keep in mind that more gain will always bring the noise floor closer (all other parameters kept the same).

Even at the pickup level - I once had a guy in a store who bought a guitar with really high gain humbuckers (very overwound).  This was before I really knew electronics.  He brought the guitar back, complaining that it was noisy.  I had the guitar techs try every shielding/wiring trick, but the eventual consensus was simply that the higher gain/overwound pickups were inherently noisier.  (The guitar did not sound especially noisy to me... the guy was being very picky, and ended up keeping the guitar.)