Sorry for the delay in replying to this...
Eric wrote:
>?????I have been thinking about your original response to my request,
>and realize that I failed to see a good point that (I think) you were
>making. I wish to amend my original response as follows.
>
>??????I stand by my judgement about needing actual plate curves to make
>the final calculations and judgements about amplifier performance using
>a given tube.
Mostly; but see below.
>However, I have just realized that the basic
>inter-relations of PS voltage, voltage swing, and output power are
>really the same as with solid-state designs. That is, V-squared/2R
>still applies. The only difference is that our tubes can not be driven
>into hard saturation but only down to a few hundred voltages. As B+
>increases, this becomes a small percentage of the potentially available
>voltage swing, so we can still hope to achieve close to V-sq./2R.
>
>In summary, for a given power supply voltage and target output
>power, R is inherently fixed as V-sq./2P.
The ARRL Handbook gives a formula that seems to work rather better:
RL = Vp/(K*Ip) where K is a variable that replaces the "2" to account
for the larger conduction angles in class B, AB or A; about 1.6 often
seems to work for class-AB2 triode amps. You can estimate Ip from Po, Vp
and an assumed efficiency.
With a little help from the tube data - but without any characteristic
curves at all - these approximate methods will get an amp on the air.
>The only thing the tubes
>curves will tell us is that the tube is actually capable of doing it -
>that is, it can be driven to near-saturation. So as a useful rule of
>thumb for estimating a potential amps performance, I fully agree.
>
>??????My comments about power gain still stand, however. There are no
>basic rules about the power gain, grid current, etc. These are all
>unique to the physical design of each tube and must be all be
>determined from the documented tube specs. If they are not available,
>it becomes voodoo (oh-oh, there is THAT word again.)
No; it simply becomes a case for experimentation. The last I heard, that
was still a scientifically valid approach :-)
Even if you had a full set of characteristic curves, they still wouldn't
tell you the optimum locations for the two ends of the load line (zero
signal and instantaneous peak signal). Pappenfus, Bruene and Schoenike
have a nice little diagram showing all the conflicting factors that have
to be balanced using "engineering judgement". The curves will help you
stay out of the high current / low voltage area where the curves
compress and bend, but that's about all the useful help they can
provide.
You can only get *all* the necessary information - and especially the
information on IMD - by actually testing the unknown tube in an
amplifier... which is exactly how the tube manufacturers come up with
their recommended operating conditions.
Having adjusted the input network for optimum VSWR, and having adjusted
the output network for the optimum loading (balancing the conflicting
requirements of efficiency and linearity), then the amp is switched off.
The input and load resistances are found by connecting small trimpots
from cathode to ground and from anode to ground. These trimpots are then
adjusted to give a 50-ohm match at the other side of the input and
output networks - and then the ohmmeter gives you the design values for
the cathode and anode load resistances.
Fortunately all this has been done for the GS35b: with 3.5kV on load,
and expecting 1500W output, 1900 ohms is a good design target for the
load resistance, and the optimum input resistance is a touch under 50
ohms.
>I have also heard
>the 50 ohms Rin ?rumor for the GS35b,
>The Russian data sheet shows grid current
>curves which imply close to 50 ohm at lower power levels (<500W) but at
>1A. of peak plate current, Rin is dropping like a rock.
It certainly does change, but that's true of many tubes AFAIK. The
normal practice is to optimise the input match at full drive power, and
let the driver operate into a mismatched load only at lower output
levels.
> but the power gain figures I hear
>are all over the place.
Well, that depends on the competence of the experimenter. If someone can
only get 600W out of a GS35b, it isn't the tube that has a problem.
One of the main pitfalls is the facile assumption that a GS35b is a
"Russian 8877". Both tubes are triodes with a 1500W anode dissipation -
but there the resemblance ends. The 8877 is a high-mu triode that works
well at low anode voltages, and requires only a low bias voltage and
therefore quite low RF drive. The GS35b is the opposite in every
respect. It *demands* high anode voltage (it's useless below 3kV) and it
also needs high bias voltage and plenty of drive. If those requirements
are met, it goes well.
Its main practical drawback compared with the 8877 is that, for 1500W
output, the GS35b needs most of the drive power that your transceiver
can provide. That means you don't have so much opportunity to improve
the IMD performance of the transceiver finals by backing-off the RF
drive. Even so, you can expect typically 12dB gain at 1500W/3.5kV, and
1-2dB more at higher voltages and/or lower outputs (and those figures
are verified up to 432MHz).
All of that information has been derived experimentally without
reference to the characteristic curves - but you can now pretty much
infer what the curves must look like!
I really do feel that dismissing solid experimental data as "rumor"
comes awfully close to "not invented here". There are many hundreds of
GS35b amps working well in Europe, from Top Band to 432MHz and even
above. G0RUZ has built a dozen VHF/UHF amps, all with similar
performance, and the GS35b and its smaller brother the GS31 are to be
found in commercial amps covering HF, 50MHz and 144MHz.
Don't worry about the lack of characteristic curves - if you build it,
they will go.
--
73 from Ian G3SEK 'In Practice' columnist for RadCom (RSGB)
Editor, 'The VHF/UHF DX Book'
http://www.ifwtech.co.uk/g3sek
|