High Gain - Cold Bias

[Malcolm Irving]

Would it be fair to say that solid state circuits are generally less noisy than tube circuits, for a given amount of gain? Solid state circuits, being relatively high current and low voltage, use lower (ohms) valued resistors, which produce less noise. I don’t know if an individual transistor produces less noise than an individual tube, for a given amount of voltage gain?

I’m just wondering if this might be part of the reason why some guitarists prefer a clean running tube amp (to get the slight distortion, sag, etc. that we call ‘tube sound’) plus solid-state pedals for the heavily distorted sounds?

[trem]

Clipping to a low level such as with diodes does not improve signal to noise ratio. In property designed and set up sequence of amplifier stages, the SNR is determined by the signal and the first stage..... Of course, in most guitar amps, the series input resistor contributes more noise than the tube. The guitar itself also might be noisier the the tube.

If Merlin explains Why The Resistor makes noise, i have both of his Pre Books, and i can read it there.
Otherwise.....How does a resistor add to the noise floor.?
I am assuming you are talking about the "typical" 68k resistor at the In Jacks.?
Thank You

[Mike Sulzer]

I think it would be fair to say that in nearly all applications, the best SS circuit, when designed with low noise as the priority, is quieter than the equivalent tube circuit. But often it is no better or even worse as their are other priorities. In a circuit designed for low noise, it is the amplifying device that contributes most of the noise since an effort has been made to keep the noise from passive components low. Guitar amps have in general not been designed with low noise as the priority.

Would it be fair to say that solid state circuits are generally less noisy than tube circuits, for a given amount of gain? Solid state circuits, being relatively high current and low voltage, use lower (ohms) valued resistors, which produce less noise. I don’t know if an individual transistor produces less noise than an individual tube, for a given amount of voltage gain?

I’m just wondering if this might be part of the reason why some guitarists prefer a clean running tube amp (to get the slight distortion, sag, etc. that we call ‘tube sound’) plus solid-state pedals for the heavily distorted sounds?

[Mike Sulzer]

If Merlin explains Why The Resistor makes noise, i have both of his Pre Books, and i can read it there.
Otherwise.....How does a resistor add to the noise floor.?
I am assuming you are talking about the "typical" 68k resistor at the In Jacks.?
Thank You

In a resistor there are many free electrons banging around randomly. Of course, there are an equal number of positive charges that are not free to move much; these are the atoms the electrons left. If each electron stayed very close to a positively charged atom there would be only an extremely low electric field since the + and - charges would very nearly cancel. But the electrons are banging around, and so there is a time varying random electric field that translates to a random time varying voltage across the resistor. You can think of this as a random voltage source in series with a noise free resistor.

That means if you put two resistors in series you get a higher voltage since the two voltage sources add. However, it is not twice as much because the sources are random and independent of each other. It is bigger by the square root of two. (or twice the voltage squared) But two resistors in series are no different than one resistor of twice the value. So a higher values resistor has more noise. For many purposes you can think of the noise from a tube as equivalent to that of some resistor. A good 12AX7 is equivalent to about 1K if I remember correctly. That means that 68K in series with the input is a lot worse!

[trem]

OK...Thanks.
I am not being lazy, i could look it up, but.....all i REMEBER Merlin saying was that for RF Suppression, that 68k resistor could be reduced A Lot, at least 50%.
It would still be effective and would also be less noisy. I always wondered WHY it would be less noisy.
I do not think he ever addressed it. But Merlin probably assumes a level of understanding, that readers of his books will have, and that level is above my own.
Thanks Again

[Leo_Gnardo]

I am not being lazy, i could look it up, but.....all i REMEBER Merlin saying was that for RF Suppression, that 68k resistor could be reduced A Lot, at least 50%.

If the 68K value is taken from typical Fender preamps, consider there are actually two 68K resistors there, in parallel. So 33K should be the expected value, and you can work down from there to the point RF starts to become a problem then back up one step and select that resistor.

[eschertron]

OK...Thanks.
I am not being lazy, i could look it up, but.....all i REMEBER Merlin saying was that for RF Suppression, that 68k resistor could be reduced A Lot, at least 50%.
It would still be effective and would also be less noisy. I always wondered WHY it would be less noisy.
I do not think he ever addressed it. But Merlin probably assumes a level of understanding, that readers of his books will have, and that level is above my own.
Thanks Again

I hit that chapter in Merlin's preamp book A LOT. Because I need A LOT of help in understanding (like the help I get in this thread here )
Basically, my takeaway is that at 10k, the resistor noise is less than (swamped by? See? I still need more familiarity) the tube noise, so a smaller resistor is futile. Apparently, 10k is still effective for EMI blocking and oscillation suppression.

[Mike Sulzer]

I hit that chapter in Merlin's preamp book A LOT. Because I need A LOT of help in understanding (like the help I get in this thread here )
Basically, my takeaway is that at 10k, the resistor noise is less than (swamped by? See? I still need more familiarity) the tube noise, so a smaller resistor is futile. Apparently, 10k is still effective for EMI blocking and oscillation suppression.

I have seen that 10k recommendation, but 12AX7 noise is less than that of a 10K resistor. Merlin might be allowing for so called flicker noise which increases a lot at low frequencies. This can be a problem in a tube phono preamp because of the huge gain at low frequencies due to the equalization. Low frequencies are not emphasized in a guitar amp. I went into this noise issue in great detail in this thread:http://music-electronics-forum.com/t41822/

[loudthud]

Clipping to a low level such as with diodes does not improve signal to noise ratio. In property designed and set up sequence of amplifier stages, the SNR is determined by the signal and the first stage: the noise from the first stage and the signal are amplified together, and so the noise from the first stage is amplified more than the from any other stage.

Clipping diodes do not reduce noise directly, but they do lower the clipping level referred to the input. As a consequence, you do not need as much gain to get an equivalent clipping level at the input compared to the no clipping diode preamp. Often you can get by with one or two fewer stages of gain so it seems like lower noise.

[Chuck H]

I was going to say that ^^^^^^^^^^^^^^^^^^^^^

WRT clipping diodes you CAN achieve clipping without added gain. Whatever gain is needed for recovery not withstanding. But in the case of the cold clipper it must still produce gain such that the stage is driven into cutoff. Whatever input noise is with that gain is still amplified analogous to whatever gain is necessary for that. I think that's what Mike was saying.

[Mike Sulzer]

If you clip to a lower level, you have to amplify back up afterwards. The total gain on the noise comes out the same.

Clipping diodes do not reduce noise directly, but they do lower the clipping level referred to the input. As a consequence, you do not need as much gain to get an equivalent clipping level at the input compared to the no clipping diode preamp. Often you can get by with one or two fewer stages of gain so it seems like lower noise.

[d95err]

If you clip to a lower level, you have to amplify back up afterwards. The total gain on the noise comes out the same.

I don't agree. For example, amps with an effect loop usually have to attenuate the signal a lot to get it down to a level suitable for effects. With clipping diodes after the final gain stage of the preamp (like with the "Jose" master volume mod, as used by e.g. Cameron and Friedman), you can use less attenuation before the loop. No compensating gain needed.

[Mike Sulzer]

Amps have to work when you are not using an effects loop. That is just another distraction from the main point. Explain how clipping at a lower level results in a better SNR. Explain why SNR is not determined by what happens back at early stages of the preamp.

I don't agree. For example, amps with an effect loop usually have to attenuate the signal a lot to get it down to a level suitable for effects. With clipping diodes after the final gain stage of the preamp (like with the "Jose" master volume mod, as used by e.g. Cameron and Friedman), you can use less attenuation before the loop. No compensating gain needed.

[d95err]

Amps have to work when you are not using an effects loop. That is just another distraction from the main point. Explain how clipping at a lower level results in a better SNR. Explain why SNR is not determined by what happens back at early stages of the preamp.

I'd say the general principle is that the more total gain the amp has, the more the critical noise from the first stage is amplified. In addition, more stages mean more noise. For example, adding 30dB of gain and then attenuating by 30dB adds more noise than not amplifying the signal at all.

The topic concerns high-gain amps such as the Soldano SLO or Peavey 6505+. So, we don't have to consider non-master amps that need to blast the PI and poweramp to get distortion. On the contrary, poweramp distortion is usually not part of the desired sound for modern high-gain amps.

Many of these amps attenuate the signal from the high-gain channel at the end of the preamp, for various resaons. For example, to match the level with the clean channel or to interface with reverb circuits, effect loops, etc. In addition, there is a master volume that attenuates the signal futher (maximum master volume is often not the "sweet spot" for these amps). Therefore, we can get away with a diode clipped signal without having to compensate with extra gain (depending on the type of diodes used).

Let's say we have a typical 3-stage preamp. To turn it into a high-gain monster (more clipping/distortion/sustain) we can either: Increase the amount gain of the existing stages, add more gain stages or use clipping diodes at the end. Increasing the gain or increasing the number of stages also increases noise. Adding clipping diodes adds no noise.

I have built a 3-stage amp with switchable clipping diodes á la Jose. If you engage the diodes and turn the gain down to get the same level of distortion/clipping/saturation, you can hear that the level of noise is lower. However, I usually prefer the sound with gain up and diodes off. I can live with the added noise

[Malcolm Irving]

Let’s compare two simple cases which just reach clipping in the pre-amp.

Take the guitar input signal to be 100mV peak-to-peak, and assume that the PI and power stage run clean and require 5V peak-to-peak into the PI to achieve full output.

Case A (with diode clipper): gain required before diodes to reach clip 1.4/0.1 = 14, gain of clipper stage = 1, gain required after clipper 5/1.4 = 3.57. Overall gain of pre-amp = 14 x 1 x 3.57 = 50

Case B (with triode stage clipper): gain required before clipper 5/0.1 = 50, gain produced by clipper = 40, gain required after clipper = 5/200 = 0.025. Overall gain = 50 x 40 x 0.025 = 50

Same in both cases! This supports Mike’s point of view (I think).

The resistive attenuation required after the triode clipper in case B will add a little noise but this will be negligible compared to input stage noise.

(I’ve assumed that the diodes clip at 1.4V p-p and that the triode clipper requires 5V p-p input to clip, has a gain of 40 and produces 200V p-p output.)