One thing I haven't heard mentioned is that the huge capacitances of
power FETs are highly variable. They're essentially reverse-biased
junction capacitances, so they change dramatically over the period of
each cycle as the gate and drain voltages change. This tends to make
them lossy, and makes them difficult or impossible to absorb them into a
network. People have obviously found ways to live with the variable C,
but it's not a trivial problem.

Roy Lewallen, W7EL

On Wed, 14 Sep 2005 12:21:04 -0700, Roy Lewallen
wrote:

One thing I haven't heard mentioned is that the huge capacitances of
power FETs are highly variable. They're essentially reverse-biased
junction capacitances, so they change dramatically over the period of
each cycle as the gate and drain voltages change. This tends to make
them lossy, and makes them difficult or impossible to absorb them into a
network. People have obviously found ways to live with the variable C,
but it's not a trivial problem.

Roy Lewallen, W7EL

Switching power MOSFETS are the current version of the "sweep tube"
amplifier. Looking back those 6DQ5s and 6JS6s werent pretty either.

Roger that, they are no fun to drive. The miller capacitance is also
a similar factor and you have the same effect going on from the Drain
Source path becuase of the avalanche diode. I suspect those
combined effects are why the IMD for those single ended IRF510
amps are so pitiful.

Driving them as pushpull pairs helps somewhat as you can get one
and the other 180 out of phase to help reflect a passable reflection
back to the driver.


Allison
Kb1GMX

ha scritto:

On 14 Sep 2005 06:48:38 -0700, wrote:

Hello,


ha scritto:

On 14 Sep 2005 01:24:16 -0700, wrote:




Reading from the ARF488 datasheet (this mosfet is made for RF
amplification
up to 60 MHz or so) I see input capacitance of 1400 pf typical, and
150 pf output capacitance typical.


Yes that is true. However that input capacitance is close to what I'd
use to bypass a 50mhz circuit. It's near RF dead short unless you
try to absorb it into the feed network. Even then it's a very low Z.

Yes, I understand it. What isn't clear to me is why the ARF488 is
specified
for 60 MHz and has input capacitance of the one of an IRF840.
On QEX of may/jun 1999 you can find a 300W 50 MHz amplifier with
ARF488 push-pull.

I'll check that one out. I do not have QEX back to '99 so it may take
a while to get it.

I can send you both the QEX article and the ARF488 datasheet (also the
IRF840 one, but this is really easy to find in the internet). Just
drop me an email and I'll send them, my email address is the one I'm
writing from.


I suspect the ARF488 is a RF rated varient of the IRFs and may offer
other characteristice that are more complient at VHF. I cannot find
any data or even a reference to ARF488.

I am a 6m op that love to play on that band so curiousity always wins.

same here. I've found that my 10W CW aren't terribly effective even
with
the 8m boom yagi lately, so I'm thinking about something more powerful
while the sun is sleeping. Seems like the old days when I could hear
and work VK with a J-pole are gone.



One of those 4cx150s will do over 100W CW and the PS for that power
is small. Like you say something different is always a draw.

Also these tubes need forced air cooling and the power supply would
need
to live near the amplifier (unless I want to send a cable with a kV up
to the antenna pole), if the mosfet way fails I'll probably go with an
expensive
bipolar amplifier or put the amplifier in the shack.

73
Francesco IS0FKQ

On 15 Sep 2005 01:01:40 -0700, wrote:


ha scritto:

On 14 Sep 2005 06:48:38 -0700, wrote:

Hello,


ha scritto:

On 14 Sep 2005 01:24:16 -0700, wrote:




Reading from the ARF488 datasheet (this mosfet is made for RF
amplification
up to 60 MHz or so) I see input capacitance of 1400 pf typical, and
150 pf output capacitance typical.


Yes that is true. However that input capacitance is close to what I'd
use to bypass a 50mhz circuit. It's near RF dead short unless you
try to absorb it into the feed network. Even then it's a very low Z.

Yes, I understand it. What isn't clear to me is why the ARF488 is
specified
for 60 MHz and has input capacitance of the one of an IRF840.
On QEX of may/jun 1999 you can find a 300W 50 MHz amplifier with
ARF488 push-pull.

I'll check that one out. I do not have QEX back to '99 so it may take
a while to get it.

I can send you both the QEX article and the ARF488 datasheet (also the
IRF840 one, but this is really easy to find in the internet). Just
drop me an email and I'll send them, my email address is the one I'm
writing from.

I have the IRF840 datasheet. The ARF488 datasheet is not on the 'net.
Would like to see the QEX article. I have the arrl fowarding service.

KB1GMX at Arrl . net

Allison


I suspect the ARF488 is a RF rated varient of the IRFs and may offer
other characteristice that are more complient at VHF. I cannot find
any data or even a reference to ARF488.

I am a 6m op that love to play on that band so curiousity always wins.

same here. I've found that my 10W CW aren't terribly effective even
with
the 8m boom yagi lately, so I'm thinking about something more powerful
while the sun is sleeping. Seems like the old days when I could hear
and work VK with a J-pole are gone.



One of those 4cx150s will do over 100W CW and the PS for that power
is small. Like you say something different is always a draw.

Also these tubes need forced air cooling and the power supply would
need
to live near the amplifier (unless I want to send a cable with a kV up
to the antenna pole), if the mosfet way fails I'll probably go with an
expensive
bipolar amplifier or put the amplifier in the shack.

73
Francesco IS0FKQ

On 15 Sep 2005 01:01:40 -0700, wrote:

I can send you both the QEX article and the ARF488 datasheet (also the
IRF840 one, but this is really easy to find in the internet). Just

After looking at the data sheets and the apnote it confirms what I've
said.

The APT/ARF488 is a RF device and not close to the IRF parts. One
difference easily noted is the lack of the avalanche diode present
in the IRF parts. It's appearance spec wise is that of a low cost
flavor of currently know RF types without the high cost package.
I may add the ARF488 is unobtainium, apparently they have
discontinued it and gone to newer versions.

The apnote confirms the need for elaborate input match for a rather
low input impedence that is highly reactive. The 50Mhz amplifier
is not broadband and for 6m would be classified as very narrowband
with it's specs covering on the 50-51mhz range. Their use of
striplines to affect input match and output match contribute
to the narrowband behavour. It is a high power amp at 300W with
an IMD3 of -25dbm which is only ok, not great. You also need an
85V supply and have to deal with cooling both the supply and
the amp.

You could try an IRF parts in that circuit but read the apnote well
as it explains the problems and techniques applied to overcome
those problems at 50mhz. However the values required ot match
the IRF vs the ARF parts may vary greater than the design limits
resulting in poor performance.


Allison
KB1GMX

Hello,

first of all, I just reread my original post and I've written ARF488
instead
of the correct ARF448, sorry for the mistake.

ha scritto:

On 15 Sep 2005 01:01:40 -0700, wrote:



After looking at the data sheets and the apnote it confirms what I've
said.

The APT/ARF488 is a RF device and not close to the IRF parts. One
difference easily noted is the lack of the avalanche diode present
in the IRF parts. It's appearance spec wise is that of a low cost
flavor of currently know RF types without the high cost package.
I may add the ARF488 is unobtainium, apparently they have
discontinued it and gone to newer versions.

yes, indeed it's very difficult to find these mosfet today.


The apnote confirms the need for elaborate input match for a rather
low input impedence that is highly reactive. The 50Mhz amplifier
is not broadband and for 6m would be classified as very narrowband
with it's specs covering on the 50-51mhz range. Their use of
striplines to affect input match and output match contribute
to the narrowband behavour. It is a high power amp at 300W with
an IMD3 of -25dbm which is only ok, not great. You also need an
85V supply and have to deal with cooling both the supply and
the amp.

well, the real problem is difficulty in matching and poor IMD3,
we still are restricted to the 50-51 MHz part of the band and
I don't remember to ever have heard anyone over 50.300 MHz here
in EU.


You could try an IRF parts in that circuit but read the apnote well
as it explains the problems and techniques applied to overcome
those problems at 50mhz. However the values required ot match
the IRF vs the ARF parts may vary greater than the design limits
resulting in poor performance.

indeed, and the diode in the IRF part could be a problem, right?

I'll probably start to search a good bipolar transistor for this
project.

thanks

Francesco IS0FKQ

" bravely wrote to "All" (14 Sep 05 12:18:39)
--- on the heady topic of " switching power mosfet as RF amplifiers"

no From:
no Xref: core-easynews rec.radio.amateur.homebrew:87858

no On 14 Sep 2005 01:24:16 -0700, wrote:


ha scritto:


Good luck. Most of those devices will exhibit gain to amazing
frequencies due the fact that fets have no real limits like junction
transistors. What they do have that limits them is real world things
like lead inductance, Input capacitance, output capacitance and Drain
to Gate Miller capacitance all of which make it difficult to get power
in and out of them as frequency increases.

The input capacitance for the IRF510 is 135pf, the higher power parts
can easily be upwards of 1275pf for the IFR640 (Xc of around 2ohms).
Output capacitance around 400pf and feedback capacitance of 100PF
also for the 640.

Reading from the ARF488 datasheet (this mosfet is made for RF
amplification
up to 60 MHz or so) I see input capacitance of 1400 pf typical, and
150 pf output capacitance typical.
The IRF840 datasheet reports 1300 pf and 200 pf in/out capacitance
at the same frequency of the ARF488 (1 MHz). So what else affect
high frequency performance of these devices? Is the 50 pf difference
in output capacitance the "big" problem? Reverse transfer capacitance
of the ARF488 is 65 pf against the 18 pf of the IRF840.

no Yes that is true. However that input capacitance is close to what I'd
no use to bypass a 50mhz circuit. It's near RF dead short unless you
no try to absorb it into the feed network. Even then it's a very low Z.


Can one drive them with a grounded gate using a bipolar transistor at
the source?

A*s*i*m*o*v

.... - Grains Of Salt. Take As Needed With Above Message.

On Thu, 29 Sep 2005 13:35:14 GMT, "Asimov"
wrote:

no Yes that is true. However that input capacitance is close to what I'd
no use to bypass a 50mhz circuit. It's near RF dead short unless you
no try to absorb it into the feed network. Even then it's a very low Z.


Can one drive them with a grounded gate using a bipolar transistor at
the source?

Yes, but not with any significant power output and the gain would be
poor at high power. This is known as cascode configuration and is
used at low power (milliwatts) with good results.

Allison

Grounded gate? Not too likely with the IRF 510. You will need 5V
drive at least, at all full current! Since a grounded gate(base, grid)
amp has only voltage gain, you would have a theoretical gain of maybe
3db at 12V, 6db@24V, and so on,but given unity gain from the IRF 510 at
12V grounded source at VHF, maybe unity or worse real world at VHF
regardless of supply voltage.

The IRF 510 has a maximum peak voltage of 100, thus a maximum safe
supply of around 34. Of course, a tube with several hundred volts
supply(or several KV for a big one) is quite another matter, as are
those 1KV peak MOSFETS as seen in some Class E 160M/80M rigs. No idea
what input C is in the latter, though.

Lastly, at grounded gate you still get the same voltage-variable
135pf gate-source C as your input C, you just get out of the Miller C,
which is an estimated 22pf times 2 for an MF/HF rig with 12V supply and
12V drive, or times 3.5 for a 30V supply. Input loss gets damned high
in this capacitance, and this is why a device that(grounded-source) at
MF has 30dB gain has 10dB gain at VHF.