Best practices for preamp stages

[Malcolm Irving]

If I correctly understand thermal noise (which is questionable...) the lower the resistor's value, the less thermal noise it generates. Since the resistor in question is at the input to the first stage, whatever thermal noise it generates is subject to the highest gain the amp is capable of. It makes sense that you would want that resistor to be the lowest value possible without introducing other issues such as stability and RF admittance. ....

Yes, when the input is shorted by the input jack socket, or by turning the guitar volume to zero, the first grid stopper is usually the main source of hiss.
Merlin derives some figures for this in the 2nd edition of 'Designing Tube Preamps for Guitar and Bass' on pp 64-66.
I think with a 10k grid stopper the noise which the grid stopper resistor generates is less than the noise generated in the first triode.
It's difficult to say how low you can take the grid stopper resistance, unless you know the transmission frequency of the Radio Station you need to block.
I believe some high gain amps do away with the first grid stopper altogether and use a ferrite around the wire to try to keep out RF.

[martin manning]

So this is a problem only if there is an unused input which is grounded but still mixed into the audio path?

[Malcolm Irving]

The high gain amps with no grid stopper are just single input amps. Sorry I can’t be specific about which ones I mean, but I seem to remember coming across it.

Edit:
Ah. Maybe I misunderstood your point. When you have the guitar plugged in with its volume up, you still get the same hiss from the grid stopper, plus any hiss from the guitar. But, the DC resistance of a guitar pickup is fairly low (about 3k to 8k if I remember right). The inductance and capacitance of the pickup don’t generate any hiss. If you have the guitar volume up half way there is extra hiss from that pot. Minimum hiss is when the guitar volume is at zero, with another local minimum when the guitar volume is full up.

[Malcolm Irving]

Just measured the DC resistance at the output jack of a Gibson with one humbucking pickup switched on. With the guitar volume at zero, there was 70 ohms (should be zero – but perhaps the pot is not perfect). With guitar volume on full, I read 7k3. Finding the position of the guitar volume that maximised the resistance, I read 81k.
I think this resistance would be added to the grid stopper resistance to find the thermal noise source, which depends on the square root of the total resistance.

It also depends on (square root of) absolute temperature (deg. K) but it’s not worth freezing your socks off just to reduce the hiss a tad.

[JazzGuitarGimp]

This is common knowledge, but in case there is someone new to the subject reading this: Metal Film resistors have the lowest noise coefficient of the available types of resistors, and so this first gain stage grid stopper resistor is a good candidate for Metal Film. I have also heard (but don't know if it's valid) that the larger the resistor, the lower the noise. So, for instance 1/2W is preferred to 1/4W, 1W is preferred to 1/2W. This may be nothing more than the resistors' "square root of absolute temperature" (as stated by Malcolm) - I assume the cooler the resistor, the lower the noise. So maybe there is no point in using a higher wattage part at the first stage grid stopper as a 1/4W part won't heat substantially from the tiny signal voltage across it.

[Malcolm Irving]

Good point. According to Merlin again (this time his new ‘Designing High Fidelity Valve Preamps’ pp 66 and 198), resistors have two types of noise. Johnson (thermal) noise which is unavoidable, and excess noise caused by spontaneous fluctuations in conductivity, which is proportional to current. Wire wound resistors have zero excess noise. Then in order of increasing excess noise: metal film, metal oxide, carbon film, and finally carbon composition.
I’m no expert on this, but I would have thought that fluctuations in conductivity would be highly localised, and therefore the excess noise produced might have some dependence on current density as well as current. If so, that might help to explain why higher wattage resistors have lower noise, since the cross sectional area of their resistive material would generally be greater (I think).
Higher temperature in low wattage resistors would also be a factor.

[roberto]

The main difference I see in your design, related to normal operations, is that usually the second triode has opposite phase (so gives negative feedback to the other stage), while here it gives positive feedback being in phase with the other triode.