[NLRS] Transistor Idling Current for Class AB?

[Dr. Gerald N. Johnson]

On Sun, 2007-04-22 at 19:42 +0000, Scott wrote:
> 
> Another good technical question for the gurus of NLRS land...
> 
> I have an amplifier from a friend that uses three SD1477 trannys to give 
> in the neighborhood of 200-240W out.  The amp is set up for FM use, but 
> it is apparent that the circuit board was designed to allow modification 
> for linear use.  My question:  What would be a good reasonable Ic idling 
> current (no drive) for an SD1477? 

Often as much power as you are willing to dissipate at idle.

> Looking on line, I see some generic 
> stuff that looks like maybe somewhere around 200mA for an amp up to 
> about 25W.  It appears that each of these should put out about 80W, so 
> would 500mA be appropriate?  More or less?  What is the best way to 
> arrive at it experimentally?  I have a spectrum analyzer good to 26.5 
> GHz.  Would I set the radio up with a 1 KHz tone and 50 Watts out 
> driving the amp (SSB) and look at the spectrum analyzer?

No. Single tone won't do it. You need two tones equal level, then look
for the third and 5th order intermod products that for a class C amp
will be 15 or so and maybe 25 dB down. A really good solid state power
amp will only do 25 dB down on 3rd order and maybe 35 dB down on 5th
order.

Your first check can be with a single tone checking for P1dB, where the
amplifier compression has reduced gain by 1 dB. Often that's about all
the amplifier will do for peak power output and still be linear with the
best of forward biases.

After that you adjust drive and bias for the best compromise for power
output and minimized 3rd and 5th order intermod products. So you need a
narrow band spectrum analyzer with pretty good LO stability.

>  What would I 
> look for?  Or would on the air reports be adequate?

On the air reports would look for minimizing splatter preferably close
in splatter (which would be those third and fifth and higher order
intermod products). Again you have to adjust drive limit and bias so
with those interacting, on the air adjustments may take a lot longer
than using the spectrum analyzer if you can get it to resolve the two tone
intermod products. Though you could use a good receiver with a narrow IF 
filter and lots of attenuation to keep it from being overloaded. If you had
a step attenuator you might not need even a receiver calibration curve, 
just adjust the attenuation for a constant S-meter reading for each component
and compute levels from that.

To add complications, the significant base currents for transistors of
this power will demand good bias current and voltage regulation or the
bias supply will contribute to the distortion.

Another complication is that the load presented to the transistors may
be not optimum when optimized for class C.

Same procedure works for tubes, but usually the idle bias is set for
full plate dissipation with no output and the tube runs cooler with RF
coming out.

Some transistors may never be very linear.
> 
-- 
73, Jerry, K0CQ,
All content copyright Dr. Gerald N. Johnson, electrical engineer