[Top] [All Lists]

Re: [Amps] New NXP BLF578XR 1200W LDMOS FET is "indestructible"

Subject: Re: [Amps] New NXP BLF578XR 1200W LDMOS FET is "indestructible"
From: "Roger (sub1)" <>
Date: Sat, 23 Jul 2011 01:10:01 -0400
List-post: <">>
On 7/22/2011 4:16 PM, Paul Decker wrote:
> Over the years I've heard many people say that adequately cooling a 
> transistor is difficult because it is so small.  It's really difficult for me 
> to wrap my brain around that statement though after playing with the 
> 3cx400A7/3cx800A7 series tubes.  If one were to take the anode cooler off the 
> tube, they would end up with a piece of ceramic about 1.4" in diameter and 
> about .75" high.  This is on par with the size of these transistors.

Transistors are limited to a much lower temp than any part of a power 
tube, so even if you could get the same kind of connection, you would be 
unlikely to be able to keep the transistor cool enough.

Also there is a delta T within the transistor material and silicon has a 
poor thermal resistance compared to copper, brass, or Aluminum.

Transistors, including FETs are basically constructed on a Silicon 
substrate.  Insulating layers which are likely oxides, but may be 
nitrides or other compounds  are *difused* into layer(s) of Silicon. 
Capacitors may also be included in the power transistor's design.  
Connections to the Emitter, base, gate, source, drain...what ever they 
are called for that particular transistor are deposited onto the surface 
of the exposed portions of the transistor.   External connections to 
these "pads" may be soldered, or just pressure points.

The problem is that in general the poorest portion of thermal transfer 
from the Silicon die to the outside world is the transistor itself. 
Exotic heatsink transfer compounds can lessen the delta T between the 
device while heat spreaders can speed the spread of heat through the 
heatsink.  Thermal pipes/heatpipes actually work quite well, but they 
still have a limited capacity. Refrigerating the heatsink or heatsink 
and spreader also has distinct advantages and one major disadvantage 
which is condensation.  You can not cool the heatsink much prior to 
applying power to the device or you will get condensation, but the 
transistor has very little mass or thermal mass. This means the internal 
workings get hot fast with power, but the greater the temp difference 
between the transistor and the heatsink the faster it will dissipate 
heat and the cooler it will keep the transistor. Unfortunately there is 
a limit to how cold you can make the heat sink as the transistor has 
temperature limitations on the low side as well.

One problem with transistors has been the migration of the typing 
materials that create the gates and junctions so their performance 
deteriorates with time due to heat.

A tube will barely notice a heat spike that would destroy a power 
transistor under the best of conditions.

Great strides have been made in developing rugged transistors. The NXP 
BLF578XR appears to be an example.

So far high power transistor amps have required a lot of protective 
circuitry for the fragile and expensive transistors.
Theoretically a couple of transistors that could run the legal limit, 
key down, 24 X 7 could be manufactured for a fraction of the cost of an 
equivalent tube.

With bipolar transistors gain is inversely related to frequency so 
transistors rated for VHF and UHF have far too much gain to be stable on 
the lower HF range.  With these FETs I "would assume" (because I don't 
know for sure) that the 10 MHz lower limit is *probably* due to the gate 
voltage required at lower frequencies. It's likely that some one on here 
has a much better background with FETs than I and can give a good answer 
to that question.


Roger (K8RI)

>    It seems to me the dissipated power per area is roughly the same when 
> looking at the devices minus their coolers.  What would be interesting is if 
> a transistor manufacturer took a page from the tube world and integrated 
> similar cooling.
> _______________________________________________
> Amps mailing list

Amps mailing list

<Prev in Thread] Current Thread [Next in Thread>