bias methods VS hiss/noise. schematics included.

[darron]

i've just finished and boxed an LFO circuit of my own design. turned up ALL the way i can hear more hiss in the signal than clean. I've isolated it down to a ge-transistor boost stage.

this made me start thinking of things that i could have done better to reduce noise, and in this case the only thing i could think of is the biasing.

schematics courtesy of AMZ ( www.muzique.com )

For convenience, I used biasing similar to this but using a multi-turn trimpot on both sides(voltage divider straight into signal):


as compared to this (voltage divider then smoothing then into signal):




has anyone noticed much difference in terms of noise/hiss (not hum)? another issue i suppose is that the trimmer, though high quality, is carbon comp, which i suppose wouldn't matter in theory with the smoothing. then again it might not matter at all.

thanks guys (:

[R.G.]

The first circuit you show is known in the trade as the "Stabilized Bias" circuit, or it was thirty years ago. It has the disadvantage that both of the input resistors contribute their thermal noise to the base of the transistor. This may or may not be bad, depending on the value of the resistors and whether or not they have excess noise.

Carbon comp has excess noise, carbon film has less, and metal film has no excess noise. All resistances have the same thermal noise based on the value of their resistance and temperature temperature.

The second circuit is the so-called "Noiseless Biasing" circuit; I believe that Jack drew that after I suggested the noiseless biasing circuit to zachary vex in this forum for biasing of MOSFET boosters. I've been a big proponent of the NB circuit. The NB circuit gets the noise down to one resistor's noise by shunting the noise from the two-resistor divider to ground at the two-resistor divider. It has low current noise because the current through it is as small as can be made.

All that being said, I think your problem is the germanium device, not your resistors. Germanium is not necessarily noisier than silicon, but germanium processing was crude, and that led to noise issues within the transistor. Silicon is much better processed, and suffers from many fewer chip issues because of that. So the germanium devices that we can actually get are noisier as a rule, and the older, the noisier.

Try subbing in a low noise silicon PNP for the germanium. 2N5087, 2N4250, BC559C or BC560C are designed for low noise. You may have to change the lowest resistor in the divider to raise the bias voltage. But this should get you some idea of the transistor's noise versus the biasing.

[darron]

A very thorough response R.G., Thanks!

You've pretty much confirmed what I thought to be true. I suspected it was almost entirely the transistor in that position, but you've also confirmed that the noiseless biasing in the future as well as removing carbon comp *can* help the situation. I won't bother replacing with a low noise PNP, I think I'll take the thing back to breadboard and move towards some sort of fet. That's a nice list for future reference though.


(:

[rogeryu_ph]

I built two of the AMZ Mosfet booster bias to voltage recommended but still i got discernible HISS 
What seems to be the problem? I even change the 10M carbon resistor into 1M metal resistor which someone i can't recall said to eliminate the hiss but still the same problem. I sub numerous BS170 available i had and even tried the several 2n7000 on my second built which i know is reversed but to no avail. Anyway i'm still using it but if i could eliminate the hiss i'll be much more delighted. It's a great booster which give me tube like overdriven tone when crank with my SS amp. I gave one to my friend and he's using it too. Can someone share if they have the same experience and how did they resolved the hiss problem that would be very much appreciated and thanks.

[R.G.]

It's physically impossible to eliminate all hiss. Hiss is a consequence of the thermal agitation of the charge carriers themselves, and the random nature of electrons arriving in amplifying devices. Thermal noise is proportional to resistance, so lower resistances at the same temperature will have lower noise. On top of the pure thermal noise, some materials have excess noise, above the thermal noise.

When you add active devices to the mix, it gets complicated. Contrary to not needing to match impedances for signal amplification, you do need to select impedances for lowest noise. Bipolars tend to be better than FETs at low source impedances, so moving coil phono cartridges to cite one example, need the lowest input impedance they can get. This took the form of many bipolar devices in parallel. At high impedances, FETs have better noise performance than bipolars, because they have vanishing low input currents, so current noise is avoided. If you really want to design low noise circuits, read Henry Ott's book, Noise Reduction Techniques in Electronic Systems. It is not a trivial subject, and not susceptible to learning only a few tricks.

[aziltz]

so in theory, replacing the trimmers in the ROG Amp-FET designs with fixed metal film resisters will decrease the noise floor?

[petemoore]

  In theory or in practice.
  But I had troubles with trimpots too.