[Amps] 3.5 kV 2A REGULATED Power Supply: Schematic ?

[R.Measures]

>R. Measures wrote:
>>>So, the operating condition definitely change during the step response test,
>>>but what's the effect on intermods?  Not much.  Intermod performance doesn't
>>>depend on plate voltage very much at all (for the tetrode case we thinking
>>>about here).  And the grid bias changes a little bit around the AB1
>>>operating point, but that's well known to be a non critical setting.
>>
>>**  It seems to me that a sag in bias potential would simply increase 
>>ZSAC slightly -- which would make little difference in IMD if the ZSAC 
>>was OK to start with.
>
>There's some confusion here. On the one hand, the effect of grid bias on 
>ZSAC and intermodulation is not critical within quite a wide range. On 
>the other hand, that statement also involves a hidden assumption that 
>the grid bias remains constant wherever you set it.
>
**  The grid does not and can not draw current when a competent operator 
sets the grid potential just above the level where it can draw 
grid-current with the max PEP drive that's available.  This is the second 
step in adjusting an AB1 amplifier -- the first step being: tuning out 
the grid's XC with the grid roller-L.  (L1 in 
http://www.somis.org/D.a.05.GIF)


>If the impedance of the grid supply is very high, say 10K, then even 1mA 
>of grid current will make the grid bias 10V more negative. 

**  This scenario is possible only when the bias is adjusted by an 
operator who is not competent.  

>This equates 
>to a sudden decrease in gain at the onset of grid current, so the 
>input/output power characteristic is no longer linear - it has a sharp 
>kink, and that's bad news.
>
>When you think of the gain characteristic as a mathematical power series 
>(see 'Care & Feeding', Terman etc) it is substantially linear, but it 
>also contains some curvature due to higher-order terms which imply 
>distortion. Gentle curvature across the whole plot implies low-order 
>distortion with IMD products close to the signal. But a sharp kink in 
>one place can *only* be created by quite large high-order terms, whose 
>IMD appears as wideband splatter.
>
>
>(Yes, some 6146 amplifiers do have high-impedance screen and grid bias 
>supplies, but there's no magic involved. 

>What *is* generally involved is 
>negative feedback and ALC, which Collins relied on heavily to improve 
>IMD.)
>
**  Indeed
>-- 
>73 from Ian