Amps
[Top] [All Lists]

[AMPS] splatter

To: <amps@contesting.com>
Subject: [AMPS] splatter
From: Peter.Chadwick@gpsemi.com (Peter Chadwick)
Date: Tue, 10 Jun 1997 11:07:00 +0100
Hi guys,

Been following the discussion for some time. I'll start by being
controversial - when you run grid current in AB1, it's not AB1 anymore,
it's AB2. AB1 is defined as 'no grid current', so you don't need a stiff
bias supply. AB2, you do. If you can't guarantee (by Rich's increased
bias voltage, ALC, black magic or whatever) NO grid current, I submit
that you need a stiff bias supply. However, NO grid current can mean up
to a few tens ( I don't think hundreds) of microamps. Some years back, I
worked on the design of some professional PA's, and we found better
consistency by using either a regulated bias and regulated screen, or
unregulated bias and unregulated screen - which is what you'd expect,
since the supply variations tend to cancel themselves. But don't
regulate just one of them. This suggest that a really stiff bias supply
probably means a regulated screen supply, and for some tubes, the ideal
is a shunt regulated supply, although as has ben said, in many cases,
zeners are adequate. (They really improve 6146 PA's, helping prevent
plate current runaway)

As far as the 'audio sounds OK, so I can't be splattering' discussion
goes, some tests have shown that you can get up to intermod products
17dB down before the audio starts to sound poor. Since infinite clipping
gives 3rd order products 13dB down, the in channel audio quality is (as
so many of you have said) no indication whatsoever of the adjacent
channel performance.

I've been using a different method of evaluation. Feed white noise (
well, pink, since it doesn't need to go above 4 KHz or below 50Hz) into
the SSB transceiver. Load up to  full PEP, using an oscilloscope to
check that the noise peaks don't exceed the full PEP, (use the drive
control) , and then use a spectrum analyser (mine's an  HP 141T, with
the 110MHz plug in, but it's good enough) to evaluate the width of the
signal. I'm the guy Ian G3SEK referred to with the 4-250's in AB2 with a
stiff shunt regulated bias supply, and a series screen regulator (never
seen reverse grid current on 4-250s, but there is 10mA bleeder in case),
which uses a 6L6G that my father paid the then equivalent of 50 cents
for in 1936 - I want his money's worth out of that tube! - and this is
how I established that AB2 was better than AB1 for linearity - in my
case - not necessarily anyone else's. This is of course, passive grid
with 200 ohms of shunt resistor and a 4:1 step up wideband transformer
and no ALC - or neutralising. (The measured reverse isolation is over
40dB, so I don't see that there's enough feedback to make neutralising
necessary, although in theory you could get about 1.4 degrees pk  of
phase distortion). Using this noise measuring method, the signal is
about 2700Hz wide at -3dBPEP, about 5KHz wide at -40dBPEP, and  about
10KHz wide at -60dBPEP,  but we're beginning to hit phase noise limits
there. Even allowing for the fact that a spectrum analyser can lie like
an HR manager if overdriven, I think that this an interesting and fairly
delicate test for those that can do it. It relates more nearly to
reality than a two tone test. (Ian, I've got the 1200MHz plug in for the
Spec An as well if you want to borrow some time). Incidentally, this
linear uses 8.2ohm  resistors in the cathode returns of the 4-250s, with
RF chokes across them so you don't instantaneously shift the bias point
with the RF - this negative feedback gets rid of the excess grid volts,
improves linearity, and needs at least 1watt carbon resistors if they're
not to get brown over the years - even though they're in the airstream
from the fans blowing over the filament seals. I use two muffin fans,
one blowing each tube base, and a big fan blowing from the top over the
anode seals. HV is about 3100volts at full load, output under the noise
test 1KW PEP, average (on the Bird thruline, dependent on speech
processor setting) about 400w.

Now, what sort of signal would an AL1200, or an Alpha or a QRO produce
on this test? Anyone able to find out?

How about suggesting an Amplifiers forum session at Dayton next year?
Any interest in taking part?

73

Peter Chadwick G3RZP

--
FAQ on WWW:               http://www.contesting.com/ampfaq.html
Submissions:              amps@contesting.com
Administrative requests:  amps-REQUEST@contesting.com
Problems:                 owner-amps@contesting.com
Search:                   http://www.contesting.com/km9p/search.htm

<Prev in Thread] Current Thread [Next in Thread>