From: "James Moritz" <j.r.moritz@herts.ac.uk>
Organization: University of Hertfordshire
To: rsgb_lf_group@blacksheep.org
Date: Tue, 29 May 2001 11:59:40 +0000
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII; format=flowed
Content-Transfer-Encoding: 8bit
Subject: Re: LF: Field effect versus bipolar transistors
In-reply-to: <001001c0e821$a6c3f9c0$a1d51ad4@w8k3f0>
Message-ID: <E154hD7-00043H-00@gemini.herts.ac.uk>
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Sender: <majordom@post.thorcom.com>

Dear Dick, LF group,

Not too sure about linear amplifiers, but FETs are certainly better 
for the switching - type LF PAs that are currently popular. The 
reason for this is similar to those given for the use of MOSFETs in 
switching power supplies, which have much in common. These are:
-Switching speed an order of magnitude greater, giving lower 
switching losses
-The base current drive waveform must be carefully controlled for 
optimum performance from bipolars, while MOSFETs only require 
an "on-off" square wave from a low impedance source. 
-MOSFETs do not suffer from  "secondary breakdown" 
phenomena, which means that the transient voltages and currents 
in a bipolar PA need to be more carefully controlled to prevent 
failure due to the formation of hot spots within the transistor. This 
makes MOSFETs more robust against spikey switching waveforms.

The situation seems more complicated with "linear" amplifier 
designs. Significant problems arise because a linear amplifier 
inherently dissipates power, wheras most modern power 
MOSFETs are designed to operate as high efficiency switches 
with low power dissipation. This results in devices with high current 
and voltage ratings being produced in small packages, which are 
unable to dissipate the heat produced in a linear application, and 
so must be operated at much below their ratings. This in turn 
produces problems with biasing - a much-vaunted advantage of 
MOSFETs is the negative temperature coefficient of the drain 
current vs. gate bias voltage characteristic; the current falls as the 
device gets hotter, so supposedly ensuring stable biasing, easy 
parallel operation etc, etc. In fact, this only applies when the 
device is operating relatively close to it's maximum current; when 
used at lower currents, as you are forced to do by the power 
dissipation issue, the tempco is in fact positive, and thermal 
runaway  can occur as with bipolars. The Mosfets designed for 
audio PAs are much better in this respect, but cost more, and are 
only available in relatively low ratings.

Also, there seems to be a dearth of reasonably high frequency, 
high power, low cost bipolar devices, whilst MOSFETs get cheaper 
all the time, and are inherently fast enough without needing to go to 
expensive, specialised RF devices

I think the main problem with MOSFET PAs at the moment is that 
not all design issues have been addressed properly yet. For 
example, the transformer push-pull linear and class D designs are 
prone to produce big voltage spikes, which mean death for any PA 
device. The Decca circuit and derrivatives demonstrate that robust 
and reliable designs can be achieved, however.

Cheers, Jim Moritz
73 de M0BMU