[ARC5] Drift in BC-453 - stability in tube-type oscillators

[Kenneth G. Gordon]

On 13 Dec 2017 at 11:02, John Hutchins wrote:

> 
> Watching the conversation take shape over time:
> If the tube is the culprit in MO drift,  I may have miss remembered,  I read some where that 
> regulating the filament voltage to the M.O. 6J7 ... , and the Detector (12,6)K8 may help in 
> subduing the tube drift, do to changes in temperature do to voltage swing.   True or not true?
> Thanks
> Hutch

Well, in the case of our "ARC-5" receivers, the 12K8 is both the HFO AND the mixer. The 
RF amp is a 12SK7. The 12K8 is not really a "mixer" alone: it contains two tubes, the HFO 
part is a triode. The "mixer" part is a hexode. The cathode is common to the two sections 
though. The 12K8 should be more properly termed a "converter", I think.

Drift in tube-type oscillator circuits cannot, really, be separated from tube characteristics at 
the levels of stability we are discussing here. The temperature effects are only one of those 
things we have to keep in mind, although those are a major one. Changes in interelectrode 
voltages are also important.

Remember, we are HOPIING we can achieve drift levels of some milliHertz per DAY. This is 
NOT easy!!!! And that level is FAR more stable than any of us would need in normal CW or 
SSB operation.

For instance, in a real ECO, in which the screen grid acts as the "plate" of the oscillator, 
while the "electron stream" connects the output of the screen to the plate circuit for further 
amplification, the relationship between the screen voltage and the plate voltage effects drift 
to a marked extent. What happens there is that a change in screen voltage effects the 
frequency of the output in the OPPOSITE direction to a change in plate voltage.

So, there is a "sweet spot" in the relationship between screen and plate voltages which 
minimizes frequency drift due to changes, even very minor ones, in screen and plate 
voltages. An RSGB handbook I have here shows this relationship on a graph and it is very 
clear that those two effects "cross" at some point, one almost prefectly counteracting the 
other.

This is one reason that in really good ECOs BOTH the screen and the plate voltages are 
regulated.

However, in our 12K8 circuit, the HFO is a triode, and the ECO effect does not enter into it. I 
suspect that the reason the HFO's plate voltage was fixed so low (35 volts as I remember it) 
by ARC engineers was to minimize drift from that source.

There are several circuits in which the tube-effects are minimized. The Franklin is one such, 
as is the Vackar, and there are others.

Proper design and implementation of oscillator components, coil and caps, is also very 
important vis-a-vis stability. Both coils and caps can change values with changes in both 
temperature and applied voltages.

In an article in Electric Radio Magzine some time ago, an author worked very hard to 
minimize drift caused by temperature changes in the dimensions, even very small ones, of 
the oscillator tank coil. By use of some sorts of special coating techniques, he was able to 
reverse the normal direction of drift caused by the coil.

Minimizing temperature effects on the oscillator tuned circuit is, of course, a very important 
factor which must be considered, and finding them all and fixing them this is NOT easy.

There is a heck of a lot to stability in oscillators, especially tube-based ones. Solid-state 
ones at least have the advantage of having no nearby heat source, i.e., the filament.

And. yes, even very slight changes in filament voltage causes noticable drift at the levels of 
stability we are considering here. However, this level of drift would go completely unnoticed 
in normal operation. Therefore, yes, regulating the filament voltage was not necessary in all 
rigs we normally use with the usual modes.

Ken W7EKB 

---
This email has been checked for viruses by Avast antivirus software.
https://www.avast.com/antivirus