DIYbanter - View Single Post

Dave Platt[_2_]

In article ,
wrote:

On your point of 'even a new tube'.... there are two such tubes out there, being the 19T8 and the
6AQ8 that can be quite annoying. At this moment, I keep three (3) Dynaco FM3 tuners, and I have a
total of five (5) 6AQ8 tubes available to me. Of the five, two work in all tuners. Two work in two
tuners, one works in only one tuner. All three of them test fairly close together on the Hickok
539B.

And, I have two radios using a 19T8 in the FM section. And four (4) such tubes. None of them work
in both tuners. All of them work in one-or-the-other tuner.

Go figure.

From what I've read, this is an "often-told tale" - not at all
uncommon. "Synergy, or lack thereof" seems to be the issue in a lot
of cases.

That is to say, the radios aren't identical. Differences in the
characteristics of the other tubes, drifted resistors which affect
biasing, etc. make them different, so what the radio may "want" from
the tube in question differs.

Some radios (some circuits, rather) are more tolerant of tube
parameter variations. Others are finicky... sometimes people take
shortcuts with e.g. biasing setups, which means that the tube must
fall into a particular range of its possible parameter-space in order
to work correctly. Worst-case here is if you see a schematic with a
lot of "selected" tubes or other components... that can mean that the
circuit is sensitive to component value variation. A batch of tubes
which all read "good" on the tester (and which all fall well within
the manufacturer's specs) may or may not work.

Also, the testers tend to pick one particular set of operating
conditions (e.g. plate voltage) for each model of tube. These
conditions may differ significantly from the conditions under which
the tube actually operates in a particular model of radio.

So, there's a lot of justice behind the "The best tester for a tube,
is the device in which the tube was being used" philosophy.

It would be interesting to know how the 19T8 and the
6AQ8 tubes you have, actually measure out on a good firebottle-capable
curve tracer... see just how their parameter sets cluster, and
correlate that to the radios in which they do and do not work
properly.