Transfoless Valve Pre: Reducing Noise [Edit: Reducing Hum]

As a follow on from analag's design and comments, I found a few interesting points I wanted to share. You may know this already..... it comes from Radiotron 4th Ed page 784. I'll paraphrase and those who want it can look it up.

When no stepup transformer is used at the input, Langford-Smith suggests the following and more:

Complete electrostatic shielding of first valve, components and wiring, with grid and plate circuits shielded separately.

Best to put PT on separate, preferably non magnetic chassis.

Heater leads twisted, carefully placed away from sensitive components, and shielded with braid..... AC .

Keep input, grid connection, valve, cathode circuit close together.

Orientate valves to position of least noise.

Use RF oscillator with a freq of 30/40 kcs for heater or other usual heater tricks or use DC.

Electrostatic shielding in PT is reqd.

There is more there but I thought this was interesting.... even in other applications.

Isnt grid noise the real problem here?

None of the things above solves this problem, i think.

They are however good guidelines to proper layout and wiring in tube equipment. :thumb:

mmmh.... maybe I should have called the topic "reducing HUM".... Good point Kit ops

The main sources of noise are given as: noise in the plate load resistor, valve noise, and noise in the grid circuit. Modern resistors can be chosen for low noise eg metal film.... he recommends wire wound in the cathode circuit....Valve noise can be reduced by using a valve with high mutual conductance eg 12AY7

So as Kit said, grid noise is the problem.............."the limit on practicable amplification is placed by the noise level caused by the thermal agitation in the grid circuit of the first stage together with some valve noise..."p 788

It seems to me we should be able to achieve a Signal to Noise level of say 60dB without an input transformer, based on the figures Langford Smith gives...... surely this is enough for quality recording. Yeah.... I know people like to see figures of 90dB plus, but is this really necessary? Oncce the recording has been mastered, the bottom dB's get chopped off anyways ....... or the noise is masked by sound and instruments......

I'm just thinking aloud here....

[quote author="mikka"]mmmh.... maybe I should have called the topic "reducing HUM".... Good point Kit ops

The main sources of noise are given as: noise in the plate load resistor, valve noise, and noise in the grid circuit. Modern resistors can be chosen for low noise eg metal film.... he recommends wire wound in the cathode circuit....Valve noise can be reduced by using a valve with high mutual conductance eg 12AY7

So as Kit said, grid noise is the problem.............."the limit on practicable amplification is placed by the noise level caused by the thermal agitation in the grid circuit of the first stage together with some valve noise..."p 788

It seems to me we should be able to achieve a Signal to Noise level of say 60dB without an input transformer, based on the figures Langford Smith gives...... surely this is enough for quality recording. Yeah.... I know people like to see figures of 90dB plus, but is this really necessary? Oncce the recording has been mastered, the bottom dB's get chopped off anyways ....... or the noise is masked by sound and instruments......[/quote]

60dB is about the s/n of a Type I cassette with Dolby B, on a very good machine. Not great. And the s/n gets worse as you multiply tracks.

Making a transformerless mic preamp from a tube is tough. If a mic is 150 ohms, then a circuit which adds 3dB of noise to its inherent noise will also have an equivalent noise resistance of 150 ohms.

If you bypass the cathode resistor it doesn't contribute to noise, so we can ignore it. That leaves the tube's own noise and the plate resistor. Let's say the tube has a gain of 30x with a 47k plate resistor. The plate resistor will have an equivalent noise resistance of 47k/(30^2), or 52 ohms. For the total noise resistance to be 150 ohms, that means the tube's own noise resistance must be about 98 ohms. That means, in turn, a transconductance of about 26mmhos (mS), which is very, very high. Tubes with high enough mu and low enough plate resistance to have a gain of 30x with a 47k resistor while still having a gm as high as that are very thin on the ground. And that's for a 3dB noise penalty. Okay for high-output condensers, maybe, but not for a dynamic or ribbon.

Peace,
Paul

Thanks for your comments, Paul. I'm here to get edjicated ... it's a slow process...... more self edjication needed. :roll: