[ARC5] RF amp tubes...

[Kenneth G. Gordon]

On 13 Jun 2013 at 22:03, Brian Clarke wrote:

> Hello Ken,
> 
> There may be a few confounding factors here:

I don't doubt that in the least, Brian. As with many of us, I just "play around" 
doing "what ifs" to see what happens, then if I see some unusual results, dig 
in to find out why.

> 1. The G1 Voltage of the 12SK7 should be -3 and that of the 12SH7
> should be -1. 2.

OK. I'll check that. I assume this is for Class A?

> The gain of the 12SK7 is 2 mA/V and that for the
> 12SH7 is 4.9 mA/V, when run at the correct G1 Voltages.

Which is quite a big difference.

> 3. None of the
> 12S*7 family of pentodes was designed for low noise.

Yes. I suspected that. This was before that time that the majority of 
designers were working towards low noise. They wanted reliability first.

> 4. The 12SH7 is a
> sharp cut-off pentode, while the 12SK7 is a remote cut-off pentode.
> So, the 12SK7 will respond better to AVC than the 12SH7, which is
> better run flat out.

This particular model, an ARA 46104, has, essentially, no AGC, so that is not 
really an important factor here.

Which brings up a point: it has been my nearly constant experience that 
using a sharp-cutoff pentode in place of a remote-cutoff version makes 
almost no difference whatever in how a receiver operates in my particular 
operating environment here. 

I have not had as extensive experience in this area as some, but those 
receivers in which I have swapped tubes with much, much higher 
transconductances into circuits which originally used remote-cutoff tubes of 
much lower transconductance always resulted in (more or less) much lower 
noise, and much higher useable gain.

Of course, "higher transconductance and fewer grids equals less noise and 
more gain", but there is also the very real problem of parasitics and 
oscillation to contend with. Yet with most receivers with which I have tried 
this, the oscillations and parasitics never occurred.

I have become a very firm believer in using tubes with the highest 
transconductance that is possible to use in any given receiver's input and 
gain stages, and I worry about "sharp" vs "remote" cutoff absolutely last.

I also don't really believe in overcoming mixer-generated noise by using the 
output of an RF amp to "overcome" it. I much prefer mixers that generate the 
least noise possible. This is one reason why I have always been intrigued by 
the so-called "Pullen Mixer".

> One mod often made to Command receivers was to
> disconnect the 1st 12SK7 from both the AVC and RF gain controls to
> give better noise performance.

Yes. I have seen that, but never liked using that method. I prefer to have 
some control over unusually strong signals. I guess I like to be my own 
manual "remote cutoff" control.

> So, if you put a 12SH7 in place of the 12SK7 without altering the
> cathode resistor, it will be somewhat starved of plate current, which
> will lower the Johnson noise and the gain. And the increase in gain by
> a factor 2.45, according to my Valve book (at rated G1 Voltages), or 3
> according to your figures is possibly about right.

My figures are, of course, simply very crude estimates. I.e., I can HEAR the 
difference, but it is not accurately quantified.

> However, what
> happens if you lower the cathode resistor to give the 12SH7 its full
> head of steam?

I guess that is the next step. 

There is also the fact that these receivers always operated with 
less-than-optimal (according to the tube manuals) screen voltage. I suspect 
that there is a "balance point" where the best possible noise figure and gain 
is a combination of all these factors. 

I may connect a "power-pot" as a voltage divider, some time, to manually 
adjust the screen voltage, and take some measurements, to see if that will 
really make any significant difference. Personally, I doubt that it will, but I 
"need" to find out.

> I suspect that most of what you are seeing is the increase in gain -
> with a slight drop in noise because you are starving the 12SH7 plate.