[Amps] Additional comments, re GU-74B/4CX800A

[Jim Garland]

Thanks for your comments, Leigh. I agree with you about the benefit of
single tube amps vs multiple tubes. However, I bought a dozen GU-74Bs back
when they were dirt cheap, along with three sockets and chimneys, so I need
to do something with them! 
73,
Jim W8ZR

> -----Original Message-----
> From: Leigh Turner [mailto:invertech at frontierisp.net.au]
> Sent: Friday, October 26, 2012 5:06 AM
> To: 4CX250B; amps at contesting.com
> Subject: RE: [Amps] Additional comments, re GU-74B/4CX800A
> 
> 
> Hi Jim,
> 
> Instead of a 3-holer GU74B amp why not use a single large GU84B tube with
> its 2500 Watt rated plate dissipation?
> 
> I reckon if you need to use more than one tube to meet a given Po target
in
> a QRO amp, then you've arguably selected the wrong tube :-)  An exception
> might be if one has a glut of smaller tubes available in the junk box.
> 
> Yes indeed the use of degenerative RF negative feedback via an un-bypassed
> cathode resistor will reduce the need for such a high value of ZSAC and
yet
> still achieve good IMD performance.  The use of a properly designed and
> implemented tri-state EBS will significantly reduce plate dissipation with
> SSB speech and the bias transition artefacts will be virtually
undetectable.
> 
> 
> One will learn a lot by carefully studying the ACOM-2000 amp schematics;
> their designers have skilfully balanced all the parameters to optimise
> performance very nicely around the GU74B tubes.
> 
> Cheers,
> 
> Leigh
> VK5KLT
> 
> 
> -----Original Message-----
> From: Amps [mailto:amps-bounces at contesting.com] On Behalf Of Jim Garland
> Sent: Friday, 26 October 2012 5:25 AM
> To: amps at contesting.com
> Subject: [Amps] Additional comments, re GU-74B/4CX800A
> 
> This has been an interesting thread on the GU-74B. Since I kicked off the
> discussion with my question about grid current ratings for the tube, I
> thought it might be helpful to explain my context.
> 
> One problem I've had in designing appropriate amplifier parameters is the
> variation in published tube specifications. As noted by others, the
original
> GU-74B data sheet gives maximum ratings as follows:
> 
> 
> 
> max plate dissipation - 600W
> 
> max plate voltage - 2000V
> 
> max screen grid voltage - 300V
> 
> max plate current -750 mA.
> 
> 
> 
> When Svetlana rebranded the tube as a 4CX800A, they upped the published
> maximum ratings, presumably because they figured the commercial and ICAS
> services were not as stringent as the mil-spec service the tube was
> originally designed for:
> 
> 
> 
> max plate dissipation - 800W
> 
> max plate voltage - 2500V
> 
> max screen grid voltage - 350V
> 
> max plate current - 800 mA
> 
> 
> 
> The ambiguity in tube ratings is also reflected in the way commercial
> amplifiers use the tube, with most manufacturers (i.e., Acom, QRO) pushing
> the screen voltage to about 350V and the plate voltage to 2500V,
evidentally
> with no ill effects on tube performance or tube life. At this high screen
> voltage, both the operating bias and resting plate current is quite high.
> 
> For example, QRO specifies about 500W of resting plate dissipation per
tube,
> with a claimed operating bias of -70V. High resting dissipation is a
common
> problem with many tetrodes. One solution is to bias the tube nearly to
> cutoff with no speech, and then lower the bias when RF is detected at the
> grid. This is done by Acom and also Alpha (in the 8410), but at some cost
of
> circuit complexity. There is also the possibility that bias switching
> artifacts might be audable in the transmitted signal.
> 
>                 Svetlana recognized this problem and proposed using
cathode
> degeneration (e.g., inserting a 25 ohm resistor between the cathode and
> ground) to reduce the resting dissipation to a reasonable value. This is
the
> approach I'm planning to use in my homebrew amp (three GU-74Bs), along
with
> reducing the screen voltage to about 250V. Here are some typical operating
> parameters predicted from the tube constant current curves, for a plate
> voltage of 2500V, screen voltage of 250V, and grid bias of -40V, but no
> cathode degeneration. Values are per tube.
> 
> 
> 
> Grid current (mA): 32.3
> 
> Screen current (mA): 11.7
> 
> Plate current(Amps): 0.600
> 
> Input power (Watts): 1490
> 
> Output power (Watts): 997
> 
> Plate Dissipation (Watts): 502
> 
> Efficiency: 66.9%
> 
> Plate load (ohms): 2160
> 
> Grid Swing (Volts): 52.0
> 
> Resting Dissipation (Watts): 250
> 
> Drive Power (W): 1.68
> 
> 
> 
> What's interesting about these results is that the amplifier isn't
operating
> in a linear regime, even when each tube is dissipating 250W of resting
> power! (If it were, the theoretical efficiency would be closer to 61-62%).
> Another interesting result is that appreciable key-down grid and screen
> currents are drawn, although both are well within tube limits (2W grid
> dissipation and 15W screen dissipation). I don't know how much these
results
> would change by adding some cathode resistance, although obviously more
> drive voltage would be required. The bottom line, I guess, is that there's
> no easy way to rein in the resting dissipation of these tubes while still
> preserving linearity. One either puts up with the heat or else devises a
> tiered electronic bias circuit, a la Acom's.
> 
> Incidentally, although some folks have asserted that the GU-74B life
> expectancy will be extended if the tube is not pushed to, e.g., 1000W
> output, I don't see why that is necessarily true. As the above numbers
> indicate, at 1000W output, the tube is only dissipating 500W, and the
other
> parameters are well within maximum ratings. Assuming adequate cooling is
> supplied and that the filament voltage is maintained, I'd think the
biggest
> killer of tube life would be drawing excessive cathode current. At 0.6A
> cathode current, the tube is being operated conservatively and still
> producing 1000W of RF.
> 
> 73,
> 
> Jim W8ZR
> 
> _______________________________________________
> Amps mailing list
> Amps at contesting.com
> http://lists.contesting.com/mailman/listinfo/amps
> 
>