Superfast Power MOSFETs for a Linear Amp?

[Paul Burridge]

Mosfets in the MRF5xx series (511, 521 for example) have been used
up to the 10 meter band with good results. A pair of them can give
at least 50w pep output. Depending on the input/output circuitry
used and the transistor they require 12-28v power supply. Layout
is somewhat critical.

THanks, very interesting. I wonder if anyone's tried paralleling up
half a dozen of 'em for more power?

 

[Paul Burridge]

BTW I have a bunch of 813's I'd be willing to sell. Someone make me
a good offer on a lot of 5 of them. (Used, but don't look too bad).

Nice. Got any KT88s?

 

[Paul Burridge]

Paul:

There is an interesting two part article in QST May & June 1997
describing 300 W and 500 W output Class E (more efficient, non-linear)
amplifiers using a single IRFP440 (300 W) or IRFP450(500 W) at 7 MHz.
Since you are in the UK, there is a brief mention in Radcom (Technical
Topics) August 1997.

Ordinary garden-variety power FETs can also be used for HF linear
amplification using appropriate bias settings - though exactly how
linear they can be in practice I don't know.

Thanks, Steve. It's obviously doable. I haven't checked a Power FET
data sheet yet, but what's the betting they don't quote output
admittances for these devices? :-(

 

[John Larkin]

If the drain load is resonated and you want some control over the resonant
frequency (i.e. you have external to the mosfet tuning Cs) I don't see much
discontinuities in inductors current. Of course when all the stuff is well
"wired" and that's an entirely different matter, isn't it Paul ? ;-)

John, can you suggest some refs that nicely snap ?

I think that most mosfet substrate diodes are now designed to have
soft recovery, so that they won't snap and make a horrible dv/dt. I
blew up a lot of early Motorola mosfets in an h-bridge motor driver...
had to switch to darlingtons, and only figured it out later.

Anything with a p-i-n structure has a chance of being a snap diode. I
think it needs a hyperbolic doping profile or something to work well.
1N4005-7 types work, but usually only after a brief forward bias, not
DC. Somebody told me that many varicaps snap, but I haven't verified
that. Specifically-designed PIN diodes (the kind used in RF switches
and attenuators) don't snap, as they are doped to have very long
recovery times.

We tested over 60 different TO-220 power diodes to find the best
high-voltage drift step-recovery part. Then we found something else,
much better and more repeatable, but that's still a secret.

John

 

[Highland Ham]

THanks, very interesting. I wonder if anyone's tried paralleling up

half a dozen of 'em for more power?

==================================
Mind in-&output capacitances ,which add-on when parallelling.

Frank GM0CSZ / KN6WH

 

[Tim Wescott]

==================================
Mind in-&output capacitances ,which add-on when parallelling.

Frank GM0CSZ / KN6WH

Not to mention that "critical layout".

 

[Paul Burridge]

==================================
Mind in-&output capacitances ,which add-on when parallelling.

Indeed, but there must be some current-pumping circuit that might
assist here?

 

[Walter Harley]

THanks, very interesting. I wonder if anyone's tried paralleling up
half a dozen of 'em for more power?

Modern commercial radio transmitters (including FM band, 88-108MHz) by,
e.g., Harris use MOSFETs. So clearly it is possible to get up to 25kW. I
believe they also use them in 100kW AM band transmitters. A bit of
meandering on Harris' web page will tell you more.

 

[R J Carpenter]

Modern commercial radio transmitters (including FM band, 88-108MHz) by,
e.g., Harris use MOSFETs. So clearly it is possible to get up to 25kW.
I believe they also use them in 100kW AM band transmitters. A bit of
meandering on Harris' web page will tell you more.

I rather doubt that there are any tube-type 50 kW AM broadcast band
transmitters sold these days. The 50 kW solid state Harris unit seems to be
extremely widely used. Such transmitters use lots of hot-swappable
modules - and can operate quite well with a few modules removed. At least
some of these high-power AM rigs are essentially huge D-to-A converters.
Look in the IEEE Transactions on Broadcasting for some ideas.

The 88-108 MHz units are conventional amplifiers, made of a number of
modules and combiners.

The trick in the near future is adding IBOC digital to these rigs without
non-linearities and unacceptable mixing products. Although "digital", the RF
is of course analog in nature. I gather that solid state rigs accommodating
IBOC were at the NAB show a couple of weeks ago.

When WTOP was running IBOC tests on 1500 kHz I don't know whether they were
using their main solid state 50 kW Harris rig or the tube Continental 50 kW
backup.

 

[Tam/WB2TT]

Modern commercial radio transmitters (including FM band, 88-108MHz) by,
e.g., Harris use MOSFETs. So clearly it is possible to get up to 25kW. I
believe they also use them in 100kW AM band transmitters. A bit of
meandering on Harris' web page will tell you more.

Their AM transmitter combines the outputs of 60 some modules to get the
instantanous PEP they need. All modules are fed the same square wave signal,
and modules run class E (I think). They don't say what the sampling rate is,
but I would guess ~20 KHz. There is a WLW related web site that has more
info than the Harris site. If I run across it again, I will post it.

Tam

 

[Paul Keinanen]

Directly parallelling the semiconductors have several drawbacks. First
of all, the capacitances are in parallel thus limiting the frequency
response. Also load sharing between the semiconductors can be hard to
achieve.

Modern commercial radio transmitters (including FM band, 88-108MHz) by,
e.g., Harris use MOSFETs. So clearly it is possible to get up to 25kW. I
believe they also use them in 100kW AM band transmitters. A bit of
meandering on Harris' web page will tell you more.

They are complete amplifier modules with well specified output
impedances that are combined. If e.g. Wilkinson dividers/combiners are
used, these have a limited bandwidth, but this is usually not a big
issue in broadcasting, in which the frequency remains the same and if
changes are needed, the relative frequency range is limited.

However, Wilkinson dividers and combiners would not be suitable for
1.8-50 MHz amateur linear amplifier due to the huge relative frequency
range.

Paul OH3LWR

 

[eatitcold]

Hi all,

I was thinking about having a go at contriving (I won't say
"designing" for obvious reasons) a class 'C' RF amp using MOSFETs
instead of the usual BJTs/toobz. They seem - on the face of it at
least - ideally suited to the task. I'm just a bit concerned about
whether even the fastest ones would be fast enough, even given
adequate gate drive. I'd be surprised if they weren't good for at
least a few Mhz., but am quite frankly clueless as to MUF. Anyone
know?

mosfet amp stuff;
http://www.qsl.net/ab4oj/icom/ic7800/7800txdev.html
http://www.icomamerica.com/amateur/amps/default.asp

 

[Leland C. Scott]

Look at this example for starters.

http://www.picks.force9.co.uk/throb.htm


--
Leland C. Scott
KC8LDO

Wireless Network
Mobile computing
on the go brought
to you by Micro$oft

 

[Roy Lewallen]

I've successfully used a complementary-symmetry driver for an MRF510 at
20 and 15 meters, class C, 5 watts out. The driver is essentially
digital, being driven to the rails.

Roy Lewallen, W7EL