[Amps] Power Transistor Question -- 2SC5125

[Manfred Mornhinweg]

Hi Peter, Bob, and all,

> Having said all that, in my ham gear, I've had more semiconductor 
> failures in the last 20 years than tube failures. Possibly part of 
> this is that I have considerably more semiconductors than tubes,

Sure that's part of the reason, but indeed tubes are quite reliable. One
of my hobbies is restoring antique radios. When I get a 70 year old,
nonfunctional radio, most typically I have to replace a lot of
capacitors, some resistors, much of the cabling, rewind one or two 
transformers, clean many contacts, but the tubes usually are still fine! 
I have several radios of that period which still have the full original 
tube outfit and work well! Only the tuning eyes are very weak in most 
old radios. In them, it's mostly the phosphor coating that gets weak, 
rather than the cathodes.

On the other hand, when servicing the gear of fellow hams, I have had to
replace more RF power tubes than RF power transistors. In a properly
designed radio, the transistors are essentially indestructible, because
of SWR and thermal protections. Tubes instead are typically used without
  any protections. All it takes is loading up the wrong way, and the
tubes will blow up. I have quite a collection of 6146 and 6DQ5 tubes
with craters in them, and even a few with molten plates!

I also have a couple of gassy 3-500Z tubes, that will arc over
internally as soon as you switch on the high voltage. All these are
EIMACs! I usually replace them by less expensive brands, which seem to
have less tendency to becoming gassy.

But overall, yes, tubes are more reliable than most people think.

Semiconductors are extremely reliable, when properly applied. The
problem is that there are SO MANY in any modern radio! When a typical
transceiver contains something like a few million transistors (most of
them in ICs), the likelyhood of at least ONE of them to fail is high.


Bob,

> 3.    Junction temperature is NOT heatsink temperature! Junction to 
> case thermal impedance might be in the 0.5 degC/W area

The specs for the transistor in question don't tell the thermal
resistance, but it's easy enough to infer it from the data given: 170
watts maximum dissipation. It's a gold-ceramic capsule, so the silicon
will be rated to work at 200 deg C. That means 175 degrees drop for 170
watts, which makes the junction-to-case thermal resistance a tad more
than 1 degC/W. Yes, for the two transistors used in that radio, the
combined thermal resistance is indeed close to 0.5 degC/W. But let's me
more scientifically correct, and replace those degrees Celsius by
Kelvin! So we have about 0.5 K/W.

> then add case to heatsink of 0.5 to 1.5 degC / watt

Given the size of the mounting surface, and the fact that no insulation
is needed between it and the heatsink, a case-to-heatsink thermal
resistance of 0.3 K/W per transistor should be about right. Makes
0.15 K/W total for the two.

> then add heatsink to ambient of (highly variable) 5 degC/W up to 15 
> degC/w 

The heatsink used on an HF radio needs to be A LOT better than that! 
With the fan running, it should be around 0.5 K/W, if the radio is rated 
for 100% duty cycle transmission at full power, like many are.

 > for a total of 6 to 18 degC/W. Now pump 100W of thermal
> energy through that impedance and you get a very, very high juntion
> temperature. Ouch.

Of course, with that high thermal resistance, it would be impossible to 
run at 100 watts! With the more correct numbers, a total 
junction-to-ambient thermal resistance of about 1.15 K/W results. 
Running at 100 watts dissipation, the junctions would be 115 K above the 
ambient. At 25 degrees Celsius ambient temperature, the junctions would 
be at 140 deg C. That's fine. But now add a few more watts sent by the 
driver transistors into the same heatsink, and the end effect is that 
the junctions will be running near 150 degrees Celsius at prolonged 
key-down 100 Watt operation.

> Interesting engineering question that has kept me involved in fun 
> stuff over the years.

Have a look at my thermal design page:

http://ludens.cl/Electron/Thermal.html

Manfred.

----------------------------
Visit my hobby website!
http://ludens.cl
----------------------------