Output power vs. temperature

[KATEKEBO]

I have noticed (measured) that the output power of my MFJ-9420 xcvr is quite sensitive to the temperature.  When the radio is really cold (below 50 deg F), the measured output power peaks around 12W.  However, after warm-up, once the radio reaches room temperature (80 deg F), the power degrades to about 10W.  Finally, on really hot days, if the radio is about 100 deg F, the power degrades even further to about 8W.  In theory, both FETs and bipolar-junction transistors' gain increases with temperature, so the power should increase as radio warms up, not decrease.

Has anybody seen a similar issue with any other radio or specifically with the MFJ-9420?  Any ideas / suggestions?

S. Bucki
KD8KQH

[AC5UP]

...makes me wonder if there's a simple temperature compensation circuit designed to protect against running too hot that doesn't know where to plateau below room temperature.

Does the book make any mention of thermal protection?  It's not unusual to see a thermistor on a heat sink in a power supply or audio amplifier to either fold back the power or turn on a fan.

[KATEKEBO]

No, there is no thermal protection.  The MFJ-9420 is a VERY simple radio.
I wonder if the behavior is caused by a manufacturing / component defect.  The radio works OK, but it would be nice to find out what exactly causes this behavior.
S. Bucki
KD8KQH

[WB6BYU]

I'd look at the bias circuit.  We had a SSB transceiver kit where the idle current
varied significantly as the rig warmed up.  It used a "constant current" circuit
that relied on the voltage drop across a diode, and that wasn't stable with
temperature.

So one start would be to see how the bias voltage and/or idle current on the
finals varies with temperature.  (A hair dryer might provide a suitable hot
environment, and putting it in the freezer would allow you to measure the other
end of the spectrum.  Probably best to measure cold first, so the hair dryer
will help to remove the condensation in the process.)

[WB2WIK]

Maybe leave the rig in the refrigerator and remote the controls to the door?

[G3RZP]

'BYU is probably correct. The diode drop decreases with temperature at about, IIRC, 2mV/deg C.

[KATEKEBO]

WB6BYU and G3RZP - thanks for the input.  Indeed, there is a diode in the PA bias circuit.
So here is another question.  Would adjusting the bias voltage slightly up (from 0.6 to let's say 0.7 V) increase the PA output?  I have measured the bias voltage cold and warm, and it is 0.66 and 0.6 V, respectively.  What would the downside of increasing the bias voltage?

KD8KQH

[WB6BYU]

What would the downside of increasing the bias voltage?

Higher idle current, so higher power dissipation in the output transistor.

At low duty cycles this might not be a problem, but if dissipation is
already marginal it can cause the transistor to overheat.

The idle current is not a linear function of voltage.  The base junction
voltage drop of the final transistor is also about 0.6V, so increasing the
voltage a bit above that causes large increases in current - much more
than one might expect by a simple application of Ohm's law.

Probably the best approach is to adjust the bias for the most linear
standing current at normal operating temperature.  Or rebuild the
bias circuit to make it more stable with temperature.  Or just ignore
the difference and enjoy operating regardless of the output power.

And just because the bias voltage changes, that isn't necessarily
what is causing the difference:  you'd have to check that the drive
power is constant as well.  There are other possible causes for the
difference:  for example, a temperature sensitive capacitor in a high
Q circuit in a driver stage, or a marginal electrolytic capacitor.


You can investigate the source of the problem further by using
freeze-spray or a small source of heat (possibly a hair dryer with
a narrow output opening) to change the temperature in a small
section of the circuit and see if that makes a difference.  This was
how I found the bad capacitor that was causing frequency drift
in my 40m pocket CW transceiver.

[G3RZP]

There's also the point that as the ambient temperature goes up, the amount you can dissipate for a given rise in junction temperature goes down.

[KE3WD]

Not too hard of a job to redesign it to a well regulated - and easily adjustable - bias circuit and toss the cheap diode drop method. 


73

[G3RZP]

You need the bias temperature compensation, though. If you measure the no sig PA current just after switch on, and then again when the rig is hot, that will tell you if it's just PA biasing.