Noise Reduction in high Gain circuits - Metal Film?

[guitician]

Some "Noise" is part of a tube amps character

[ame="http://www.youtube.com/watch?v=0agVtbLcq-I"]YouTube- ROCK MY PLIMSOUL (1968) by the Jeff Beck Group[/ame]

[RudeBoy]

A minor comment -

One thing I've found interesting which could pertain to cathode resistor induced noise where the resistor is bypassed, is that the thermal noise power generated from a parallel RC is equal to kT/C. It is set by the value of the capacitor - the resistor doesn't enter in to it. The resistor affects the spectral distribution, noise equivalent bandwidth and so forth.

[kt77]

test

[raiken]

A minor comment -

One thing I've found interesting which could pertain to cathode resistor induced noise where the resistor is bypassed, is that the thermal noise power generated from a parallel RC is equal to kT/C. It is set by the value of the capacitor - the resistor doesn't enter in to it. The resistor affects the spectral distribution, noise equivalent bandwidth and so forth.

That's a very good point, but KTC noise is proportional to the inverse of the capacitor value as you noted, and gets very small for capacitors larger than 10nF or so, down in the sub-uV range. I don't think it is a contributing factor in the values we typically use in bypass capacitors.

Having said that, I think this is an excellent discussion topic that deserves further investigation. There are other noise sources in capacitors (especially electrolytic) that can contribute to the overall noise floor.

Randall Aiken

[kt77]

hey, I can't get page 2 to stay loaded...my browswer just keeps turning back to page one
wtf?

nevermind...it wuz Firefox

[mooreamps]

By the way, Gary, in case you're wondering why mean old Randall is picking on you in this post, it is because in nearly every post I read of yours, you come across as arrogant and condescending,

Respectfully,

Randall Aiken

I was not talking to Mr. Randall. I was just trying to answer the man's question. I am not responsible for his feelings. I do get short sometimes, so the point is taken. But, in my defense, it is not "horsesh*t"...


-g

[jaysg]

I can't stop laughing....

[mooreamps]

Whatever the theoretical rationale behind it is, Gary's original assertion (that fixed bias is inherently less noisy than cathode bias) should be simple enough for others to replicate. I hope to give it a go over the next week and report back, it would be useful for others to try it out also.
My initial idea is to use an AA battery to supply the fixed grid bias, and adjust the dc plate supply so that the standard 100k plate, 1k5/22uF cathode, 1M grid ref have identical operating conditions. Feed in a small (10mV? f=?) signal from a 50ohm source, scope the plate signal, and compare the noise on it with the 2 bias methods.
Can anyone suggest a better way of going about this?

This is how I set it up.

1. Input cap at 4.7 nf
2. Output coupling cap at 10 nF
3. Replace the cathode resistor and bypass cap with an ultra-bright LED.
4. Keep the rest the same.



-g

[mooreamps]

YouTube- Model gm-5 ; 5 watt Guitar Amplifer with Power Braking #1

Ya, I leave this video on YouTube for a reason. Not that's a particularly good video clip, nor that I know how to play a guitar. But, here is the reason....



To my knowledge, it was I who did the first "Octal Mod" on a Valve Junior. I know this because it was I who uploaded the block diagram on a site called ValveJunior.com. Then, I suspect it was others who reverse engineered my model 5 amp, and later on posted the "Octal Mod" on the other internet guitar amp forum.

So if you look at the "date" I uploaded this video, and then look at the "date" these schematics of the "Octal Mod" appeared on the other internet forum ; then explain it to me which one come first......... and then explain it me why I get so short sometimes with these people........




-g

[wizard333]

So back on topic, something Randall Aiken mentioned,

1) what brands of metal film do you find to be the best in terms of noise? I ask since you said in passing that they are not all equal.

2) While we're on the subject, should a thread start on ALL the ways to minimize noise in a high gain circuit? That seems to be their main shortcoming.

I'm talking about all the ways other than a) Filtering the crap out of the 1st stage so the amp feels solid state and horrible, b) loading so many bleeder caps that you pass most of your hiss to ground, giving the illusion of being 'quiet' but actually just making the high end and pick attack dull.

I.E. ways to do it without ruining the tone or feel of the amp, which ends up being what happens in most cases where I see it done commercially and the designers take the 'easy' way out.

[uvacom]

But, in my defense, it is not "horsesh*t"...

Okay, let's make the a priori assumption that reducing the value of the cathode resistor reduces the noise of that stage by 3dB as you say. Do you dispute Randy's assertion that using paralleled triodes reduces noise because the sum of the noise signal is uncorrelated while the sum of the applied signal is correlated? If so, on what grounds do you dispute this? If you do not dispute his assertion, then why is the reduction in noise not greater than 3dB, as both factors should then be contributing to the reduction in noise?

[raiken]

Randall - you'd be thought a visionary and perhaps a prophet if you just explained that the use of two devices aligned the electron flows and this caused an influx of quantum cosmic ray energy which resonated with the signal, giving a 15.2% push over the edge. Maybe put people on a waiting list for the newsletter which explained this.

You mean perhaps it needs a little olio di serpente?

Believe it or not, there is already a boutique amp company that has given a fancy name to paralleled input triodes.

Randall Aiken

[wizard333]

Here's another question, or series of questions:

1) If you wanted to lower the value of the 1st stage plate resistor to reduce noise,
a) given a 12AX7 in the 1st stage, 1.5k bypassed cathode resistor, starting from 100k and working backwards, how far can you take that and keep the bias reasonable? 82k? 68k? 47k?

b) "Safe" plate voltage here is <=250, or am I wrong on that?

c) Is it going to make enough audible difference to even mess with?

Obviously the 1st stage gain will go down, but its pretty easy to make up for it elsewhere.

[Steve Conner]

Guys, let's try to keep this to the actual topic.

I don't know which brands of metal or carbon film resistors are quieter. As far as I know, all modern designs of metal and carbon film resistors are quiet enough that their excess noise won't be an issue in even the highest gain amp. It's only the old carbon composition types, which had a great deal of excess noise because of the way the connecting wires were bonded to the resistor element.

Lowering the value of the first stage plate resistor won't change the signal-to-noise ratio one bit. If anything it'll make it slightly worse: Halving the resistance will cut the signal in half, but the resistor's own noise contribution will only be cut by the square root of two. The noise from the grid circuit will be cut in half along with the signal, so the SNR is no better in this respect.

Here's how the parallel triodes thing works: If you connect two triodes in parallel, the resulting tube has twice the transconductance, half the plate resistance, and a square root of 2 times (ie, 3dB) more internal noise, because noise voltages sum uncorrelated, as has been pointed out several times.

But because it has twice the gm and half the plate resistance, you can halve the values of the plate and cathode resistors, and still get the same gain and headroom. And halving the plate resistor also (roughly) halves the noise voltage seen at the plate. It was 3dB more than a single tube to start with, and halving it knocks 6dB off, so that's why the paralleled triodes end up 3dB quieter than a single tube.

[Old Tele man]

...somebody needs to show people one of the the old WWII "carbon microphones" that were commonly in use...just loosely packed carbon granules acting like a sound-modulated resistor.

...similar "results" can and are obtained from "old" composition resistors that have suffered long periods of vibration, both mechanical and acoustic.

...and, this is "in addition to" their thermal noise.