[Amps] Reducing heat in tetrode amps with EBS

[Jim Thomson]

Date: Fri, 31 Oct 2014 20:11:15 +1030
From: "Leigh Turner" <invertech at frontierisp.net.au>
To: "'Paul Christensen'" <w9ac at arrl.net>
Cc: amps at contesting.com
Subject: Re: [Amps] Reducing heat in tetrode amps with EBS

##  Paul,  you are not going from full class C cut off to normal operating ssb bias. 
That’s the old alpha bi-state ebs scheme.   With tri-state bias, its more like toggling 
between 20 ma to 180ma on a 8877.   Its only ever cut-off bias when in RX.  

##  the proposed digital delay scheme, on paper should work.  On some of my digital audio rack gear,
I can delay the audio output, in 1 msec increments....from 0- 2000 msecs.   I use a mixer, configured 
as a splitter..., with one channel used to feed..in real time,  low level audio to phone patch
on yaesu mk-v...with mic gain on zero.  This keys the vox..now..and of course the yaesu keys the 
amp key line.   Meanwhile,  the  2nd channel of the mixer drives the myriad of audio rack gear,
which includes the variable digital delay..then fed into the analog BM of the xcvr. This config
gives the relays  in the amp a headstart.   Works great for amps with slow mech relays. 
The mk-V and no doubt other xcvrs, already have a 0-30 msec digital delay on TX..but only for
CW.   The idea was again, to give the amps relays a head start.  The yaesus digital delay only works on
cw mode....so my experiment with delayed audio was done for ssb vox. 

##  Your proposed digital bias scheme appears to be the real ticket for a fully functioning ebs scheme.
I mean one that would toggle the bias between syllables, as well as words.  You really wouldnt even need 
a hang bias, or very little if any hang bias.  IE: a toggling bias that follows the audio waveform. 

##  But for any of these various ebs schemes to work correctly, they all require either an audio noise gate, or 
downward expander  in the audio line...esp if rf processing is used,  or audio compression.  Then you can set
the ebs trip point on the xcvr output  to something well below 1/2 watt.  Typ ham setups, with no noise gate,
have too much background noise, which trips the ebs..defeating its purpose.   Its either that, or the ebs 
threshold has to be set at a level just higher than the xcvrs  background noise output.   IE: the ebs
input threshold is now being used as a RF noise gate, which is a fubar way to go about it. 

##  On a related side note, Krell, who builds class A audio power amps, uses a sliding bias scheme on their amps
to reduce heat. Another audio amp maker,  Denon uses an optical encoder to switch the audio power amp  from class A 
to class B.  At low levels its run class A.  Exceed a given threshold and its now class B.  Continental, in dallas, who builds
100kw SW AM TX, uses a reduced controlled cxr scheme to reduce heat.  Its adjustable  from 0-1-2-3-4-5-6 db cxr redux.
It only works between words or pauses in the program material.   The listener sees a 0-6 db redux in cxr on his S meter
during pauses.   Saves $$$ in power costs.  The above schemes are all analog based. 

##  IMO, if the proposed digital bias ebs scheme cant switch between words, its really not worth it.   Ideally, if fast enough
it could switch between syllables.    Otherwise, conventional tri-state ebs, used in conjunction with an audio noise gate will be
as good as it gets.    

Jim  VE7RF     





Hi Paul,

These proposed EBS techniques you advocate using a short delay line on the
audio would work a treat...completely benign bias transitioning operation
that yields a useful dissipation / heat reduction.

Leigh
VK5KLT

-----Original Message-----
From: Amps [mailto:amps-bounces at contesting.com] On Behalf Of Paul
Christensen
Sent: Friday, 31 October 2014 7:53 AM
To: amps at contesting.com
Subject: Re: [Amps] Reducing heat in tetrode amps with EBS

One possible solution involves a digital delay line.   What is the duration 
of the typical bias settling period from full class C to third-stage class 
AB?  Perhaps 5-10 ms?  Faster?  Slower?

If audio was delayed just long enough for the amp to reach linear operation,

there would be no IMD grunge produced as a result of the EBS circuit. 
Coming out of EBS would require so-called "hang-bias," but even 10 ms is a 
very short period and not perceptible by the ear when monitoring in 
real-time.

Hang-bias is a software option in my S-QSK sequencing board and is 
adjustable from 0-255 ms in 1 ms increments.  Since amp control is managed 
in software, I may try experimenting with a delay line when the homebrew 
8877 amp gets closer to completion.

Paul, W9AC