[Amps] Pure signal vs AMPs idle curent

[Manfred Mornhinweg]

Jim,

> For those of you folks that have pre-distortion capabilities,  like 
> the ANAN  xcvrs... and use an external tube linear amp. Has anyone 
> tried measuring the IMD results   with pure signal turned on....and 
> also varying the bias and idle current of a tube amp,  like a GG 
> triode ??

I haven't done it with a tube, but I have experimented this with solid
state amps. I don't think the results would be fundamentally different 
with tubes, but some differences caused more by the external circuitry 
than by the active device will probably show up.

PureSignal will keep the close-in intermod products, inside its 
correction bandwidth of roughly 40kHz, nicely down while you reduce the 
idling current, up to the point where you reduce it so much that that 
the amp has _no_ gain for the smallest signals (biased into cutoff). At 
that point PureSignal goes crazy, and the IMD shoots through the roof. 
It cannot deal with class C amplifiers yet, not even with pure class B. 
But it might, soon.

At the same time the IMD outside PureSignal's correction bandwidth goes
up unhindered, as you reduce the bias, and is worse than without PS. As 
you reduce idling current, this IMD typically becomes excessive before 
the point where PS goes crazy.

So you cannot go too far in reducing idling current, certainly not all
the way to cutoff, but you can get pretty close while maintaining an IMD
better than what a good class AB amp with high idling current and no 
predistortion can provide.

Typical numbers for my solid-state test amplifier could be roughly like 
these, and I hope the format of this table comes through:

	----- no PS -----	---- with PS ----
Ibias	IMD3	high IMDs	IMD3	high IMDs
=================================================================
2A	-28dB	-60dB		-60dB	-58dB
1A	-25dB	-55dB		-60dB	-53dB
0.5A	-20dB	-50dB		-60dB	-48dB
0.2A	-14dB	-45dB		-58dB	-38dB
	
In this table, "high IMDs" means the ones that fall outside of, but 
close to, the correction bandwidth of PS. Those are the most 
objectionable. I filled the table with typical values from memory, from 
when I made the tests some months ago, rather than with actual test 
data, which varies by band, drive level, etc. It shows the main trend.

Another thing I would like to mention in this context is that the 
reduction of IMD inside the correction bandwidth that you can achieve 
with the current version of PureSignal depends relatively little on the 
bias and on the drive level, as PS even corrects for considerable 
overdrive, but depends hugely on amplifier memory effects. The current 
PS algorhythm cannot correct for memory effects.

In conventional amplifiers the distortion coming from memory effects is 
usually swamped by distortion coming from small-signal nonlinearity 
(bias-related) and large-signal nonlinearity (saturation-related). But 
when you use PS to remove distortion coming from these two main causes, 
the previously hidden distortion from memory effects becomes dominating.

"Memory effects" refers to the case when the characteristics of an 
amplifier change according to what it was doing a moment ago. The main 
cause of memory effects is unstable supply voltages (main supply and 
bias). For lowest memory effects it's essential that all supply voltages 
be tightly regulated (rarely the case with tube amps). An electrolytic 
capacitor filtering the supply at the amplifier, with some inductance 
and resistance between that and the regulated supply, causes terrible 
memory effects! Because it generates a time-lagged voltage drop. Either 
the capacitor must be so large that the voltage on it remains highly 
constant over the whole envelope of the signal (hardly feasible in a big 
amplifier), or it should be removed. Small (ceramic) caps are enough to 
bypass RF currents, and the time constant they form with any resistance 
or inductance in the supply wiring is of negligible length.

Memory effects can also rise from temperature fluctuations inside 
semiconductors. I wouldn't expect this to be a problem with tubes, but 
at the same time I would expect bad memory effects with tube amplifiers 
coming from their typically unregulated plate supplies.

So, Jim, if you try driving a tube amp with your ANAN and you find that 
PS doesn't cause a dramatic improvement in IMD inside its correction 
bandwidth, you should suspect memory effects from unwanted supply 
voltage modulation as the main cause. And the bias should be adjusted to 
keep the IMD outside PS's correction bandwidth small enough. What 
exactly is "small enough" of course depends on how far the closest 
fellow ham is located, how good a ham neighbor you want tobe, and how 
likely that other ham is to come over with a shotgun!

Manfred


========================
Visit my hobby homepage!
http://ludens.cl
========================