Date: Thu, 26 Nov 2009 09:57:29 +0100
To: "rsgb_lf_group@blacksheep.org" <rsgb_lf_group@blacksheep.org>
From: Rik Strobbe <rik.strobbe@fys.kuleuven.be>
In-Reply-To: <C1D951F883084B1C9F845B4EDA782028@df2py>
References: <BLU146-DS158D4B7A1E9F536E4A2D0CF9C0@phx.gbl>
 <C1D951F883084B1C9F845B4EDA782028@df2py>
Mime-Version: 1.0
Message-Id: <20091126085721.A229931E702@smtps01.kuleuven.be>
Subject: Re: LF: instability on MosFet amps
Content-Type: multipart/alternative;
	boundary="=====================_4350009==.ALT"
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org


--=====================_4350009==.ALT
Content-Type: text/plain; charset="iso-8859-1"; format=flowed
Content-Transfer-Encoding: quoted-printable

Dear Wolf, LF-group,

to my experience most FET's are killed by=20
over-voltage (spikes at the drain). Exceeding=20
current or power seems less an item.
The voltage spikes are load dependent and thus=20
can be caused by antenna mismatching. A R-C=20
series network from drain to ground is a good remedy.
I have played around with class-E PA's are they=20
are very fascinating. But although very simple to=20
build you need at least a good oscilloscope to tune them correctly.
On 500kHz I reached 92% for a 100 W PA. Most of=20
the remaining 8% was dissipated in the coils,=20
rather than in the FET, so improving the coils could even raise the effici=
ency.
But compared to a class D PA I found the class E=20
PA rather sensitive toward load.
For a class D PA the main concern about the load=20
(antenna) is that it is resistive (antenna in=20
resonance) and "more or less" 50 Ohm. My class D=20
PA ( 82% efficiency) seems to be happy with any=20
resistive load in the 40-70 Ohm range. The class=20
E PA on the other hand really needed a 50 Ohm resistive load.

Regarding split filters: I tested them years ago=20
on 137kHz and found out that they work well but -=20
at least in my case - they were quite pernicious=20
for the PA efficiency that dropped from about 85% to about 75%.
I didn't investigate the reason for that (as the=20
R-C networks at the drain solved the spike=20
problem). A wild guess would be that the harmonic=20
power that is reflected by a LPF (but absorbed in=20
a split filter) somehow is "recuperated" in the PA.

73, Rik  ON7YD - OR7T




At 16:25 25/11/2009, you wrote:
>Hello group !
>I found that almost all amp configurations=20
>generate harmonics. In case of a homebrewed HF=20
>amp, with a push =96pull design, I had to build=20
>hi-pass filters with a resistive termination, so=20
>called split filters, for each band. Without=20
>dissipating the harmonic energy the stage will=20
>have to cope with excess voltages and thus bury the MosFets .
>My last amp stage for the DI2BJ-beacon is=20
>followed by a hi-pass that opens at 750 kcs .=20
>This way the first harmonic and all the higher=20
>ones go into thermal dissipation.
>A T-type adjustable low-pass is used to match and reduce harmonics furthe=
r.
>Have fun                     de wolf  df2py

--=====================_4350009==.ALT
Content-Type: text/html; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable

<html>
<body>
Dear Wolf, LF-group,<br><br>
to my experience most FET's are killed by over-voltage (spikes at the
drain). Exceeding current or power seems less an item.<br>
The voltage spikes are load dependent and thus can be caused by antenna
mismatching. A R-C series network from drain to ground is a good
remedy.<br>
I have played around with class-E PA's are they are very fascinating. But
although very simple to build you need at least a good oscilloscope to
tune them correctly.<br>
On 500kHz I reached 92% for a 100 W PA. Most of the remaining 8% was
dissipated in the coils, rather than in the FET, so improving the coils
could even raise the efficiency.<br>
But compared to a class D PA I found the class E PA rather sensitive
toward load. <br>
For a class D PA the main concern about the load (antenna) is that it is
resistive (antenna in resonance) and &quot;more or less&quot; 50 Ohm. My
class D PA ( 82% efficiency) seems to be happy with any resistive load in
the 40-70 Ohm range. The class E PA on the other hand really needed a 50
Ohm resistive load.<br><br>
Regarding split filters: I tested them years ago on 137kHz and found out
that they work well but - at least in my case - they were quite
pernicious for the PA efficiency that dropped from about 85% to about
75%.<br>
I didn't investigate the reason for that (as the R-C networks at the
drain solved the spike problem). A wild guess would be that the harmonic
power that is reflected by a LPF (but absorbed in a split filter) somehow
is &quot;recuperated&quot; in the PA.<br><br>
73, Rik&nbsp; ON7YD - OR7T<br><br>
<br><br>
<br>
At 16:25 25/11/2009, you wrote:<br>
<blockquote type=3Dcite class=3Dcite cite=3D""><font size=3D2 color=3D"#00=
0080">
Hello group !<br>
I found that almost all amp configurations generate harmonics. In case of
a homebrewed HF amp, with a push =96pull design, I had to build hi-pass
filters with a resistive termination, so called split filters, for each
band. Without dissipating the harmonic energy the stage will have to cope
with excess voltages and thus bury the MosFets .<br>
My last amp stage for the DI2BJ-beacon is followed by a hi-pass that
opens at 750 kcs . This way the first harmonic and all the higher ones go
into thermal dissipation.<br>
A T-type adjustable low-pass is used to match and reduce harmonics
further.<br>
Have
fun&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp=
;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;&nbsp;
de wolf&nbsp; df2py</font></blockquote></body>
</html>

--=====================_4350009==.ALT--