Sa67 tubes

Different tubes have different in harmonic distortions (IHD) because of metallurgy and shapes of elements. This is because they radiate the electrons similar to an antenna and something similar to Fourier harmonics are propagated. But its not Fourier harmonics in that its not a flat calculation. I have named it myself, but academia and the science community haven't proved with blind physical cross examination and  experimentation, yet. Therefore, its not adopted as mainstream electronic theory. However, this school of thought has been around, but only accepted by engineers who designed tubes in the 50s and 60s.

There are difference reasons for tubes to have noise. One is poor storage, that is a form of oxidization internally. That changes the impedance drop/conduction of the tube and this loss of gain is hiss. One is handling noise from elements getting misaligned from it being sat down on a hard surface, or put in a drawer to roll around, or just tossed into a bag. Tube wear, which is the loss over time  of conduction that causes a de-focusing effect of the electrons emanating from the cathode. Circuit drift noise when the tube changes parameters over time and loses its signal to noise ratio due to loss of negative resistance in the grid.

If you build with or change out to nos tubes you have to either burn off internal oxidization or make sure the supplier has done this. Which is easy, you make a test circuit of just the heater on a some box mounted socket ( I Just use a metal J-box mounted tube socket) with banana plug terminals that go to the cathode and one on the plate and apply just the heater voltage for a week or when resistance reading from cathode to plate is near a new one. 

You see, the U67 amplification circuit is what I call an unprotected grid type. Where the tube circuit is not stabilized independently from the input source element.  So, you have a limited range when either devices drifts. Another disadvantage is since there is no grid stop, it has no RFI suppression. The advantage is the input impedance never becomes mismatch if the source impedance changes (if it drifts in temperature). I haven't observed it physically if the impedance of the diaphragm changes with operating temperature for this to be significant. (I doubt it because the de-emphasis in the negative feedback would have to vary at the same time) The other reason why some would leave it out is because Johnson noise was an issue with the type of resistors that were available in that era of design. Now these days that is not an issue. That is one reason it is an obsolete design.

Now let's look at the output.

The other reason why its an obsolete design is that its made for impedance standards that are not really compatible to modern designs.

R11, R10, and R12 are the load resistors in the circuit  which gives you the generic 600 ohm loading but feeds forward as an output impedance of 200 ohms. That was designed for I/o compatibility. I imagine there was a test version that went to a 600 ohm transformer. Now these days, I would add feed forward impedance to match modern preamps and complete the circuit at the new impedance standard of around 2.5K (2.2K). This would insure that no mic pre would generate excessive noise due to impedance mismatch.

So I would add two 1K resistors (matching by buying like 10 and sorting the 2 best) and change the output impedance. I would use low insertion loss resistors like these: mou.sr/3NEBdQW

Another interesting note is that this is a pre-terminated device, which the converter likes to see instead of a voltage out or transformer secondary.