I added a second GS-35b to my home-brew linear amplifier, in order to improve the performance on positive peaks.
http://mysite.verizon.net/sdp2/id20.htmlAlthough the GS-35 has more than enough plate dissipation capability to produce 1500 watts of peak output in AM mode (375 watts of carrier), I noticed that the output rf envelope seemed to be "flattening" on positive peaks.
The attached .jpeg shows that a Class AB2 (operating close to Class B) linear amplifier requires a peak plate current that is pi x as big as the average plate current. Thus with 800 mA of average plate current on 100% modulaton peaks (i.e., 400 mA of average plate current at carrier), I need more 2.5 amps of peak plate current.
In addition, the peak plate current occurs when the plate voltage is at the bottom of the rf cycle.. i.e., a small fraction of the 3000 volt B+ I am using.
Looking at the published curves for a GS-35b, I found that a GS-35b simply cannot deliver 2.5 amps of plate current, at that low a plate voltage, unless you assume some unreasonably high value of grid voltage and grid current.
I had room on the chassis (moving the existing tube socket a bit) to add a second GS-35b in parallel. My filament transformer (rated at 6 amps) was able to handle the second tube (each GS-35b requires 12.6 volts at 3 amps) without any heating problems.
The amplifier now works great at 1500 watts peak output, with excellent linearity on AM.
Stu