> Bill,
> Have you ever measured the Zero Signal Ic with temp
> changes Bill..?
Sure. Hasn't everyone?
Verification is as simple as heating a rectifier
diode with a soldering iron while watching the
resistance across it. More heat--less resistance.
When applied to a 2SC2879 transistor, more
heat means more B-E current which leads to
more C-E current. A good transistor, a power
supply, a resistor, and a soldering iron is all you
need to verify it directly. I don't expect you to
believe me. I fully realize that you don't know
anything about me or what I've done. I've known
about this forum for quite awhile, but have only
recently subscribed to it. I fully expect to have
to prove everything I say here. I don't have a
problem with that.
> Or have any data/examples to reference..?
Look at Application Note EB-63 in the Motorola
RF Device Data Book. That's arguably the design
that got the ball rolling as far as that technology
is concerned. You'll notice that the bias circuit
has a diode that is thermally connected to the
heatsink. The bias voltage is equal to the voltage
drop across the diode--which decreases with heat.
The idea was that the diode drop, and therefore the
bias, would stay in step with the drop across the
RF power transistor B-E junctions therefore maintaining
the operating class.
Application Note AN-762 builds upon the EB-63
design and is applicable to sideband. Its bias circuit
employs a regulator to deal with supply variations, but
it's still based on the drop across a transistor junction
that is thermally associated with the RF power transistor
flanges.
> Transition from "a little dirty" to "clean" with temp seems
> a bit much.
The increase in idle current with increases in temperature,
and therefore the movement in operating class toward Class
A, will depend on flange temperature regulation and bias circuit
design. As you progress through this you'll see that those
variables don't have standardized designs associated with them.
I stand firmly behind the quote below. The use of undersized
heatsinks with no complimentary forced-air assistance, and
oversimplified (a nice way of saying misguided) bias circuits
that overbias the transistors when they are at room temperature,
are the rule rather than the exception when it comes to homebrew
amps because those variables can be neglected that much and
the amp will still work for awhile. In the quote below I'm
assuming that the rest of the amp is properly designed. That
means enough degenerative feedback and isolation between
the input and output signals that the amp is stable regardless
of operating class (which is rarely the case). When that's the
case the spectral purity will improve as the amp's operating
class moves from Class B to Class AB. In most cases the
distortion and related spectral anomolies change from one
type to another. The change in operating class decreases
the crossover distortion and related in-band spectral impurities
while simultaneously increasing the small signal gain which
increases the harmonic content. The good news is that the
latter is easier to deal with through the use of filters, but there's
a limit. In my opinion, it's best not to tempt fate by overbiasing
the amp in the first place.
--
> skipp
> -
> [snip]
> >>>>When adjusting the bias for a 2sc2879 amp operating
> >>>>at 15v, what
> -
> Remember that unless you're using a temperature
> compensating bias circuit the resting collector current
> will rise as the heatsink, flange, and B-E junction(s)
> increase in temperature. In a nutshell, what is Class B
> and a little dirty at room temperature can quickly become
> Class AB and clean after the first QSO--especially with
> 15 volts on the collector(s).
> --
> -------------------------------------------------------------------
> -=[Bill Eitner]=-
|