From: Rik Strobbe <rik.strobbe@kuleuven.be>
To: "rsgb_lf_group@blacksheep.org" <rsgb_lf_group@blacksheep.org>
Thread-Topic: LF: PA problem
Thread-Index: AQHTWmJpyXQDHC+Ik0GMqONVBYc8sA==
Date: Fri, 10 Nov 2017 20:30:24 +0000
Message-ID: <1510345776321.99412@kuleuven.be>
References: <22d6c712-43af-3a5d-099e-adb285f086a1@n1bug.com>
 <7d1279b6d1a84b692d281d48901f2b23@xs4all.nl>
 <a6d30d5a-dbb9-de7c-bb78-23e76a77fc58@n1bug.com>
 <5A03089B.6060600@posteo.de>,<c9986850-af9f-4a41-6457-06677be0fc63@n1bug.com>
In-Reply-To: <c9986850-af9f-4a41-6457-06677be0fc63@n1bug.com>
Accept-Language: nl-BE, en-GB, en-US
Content-Language: nl-BE
MIME-Version: 1.0
Subject: Re: LF: PA problem
Content-Type: text/plain; charset="iso-8859-1"
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Content-Transfer-Encoding: 8bit

Hello Paul,

I just had a look at the circuit (http://qrp-labs.com/ultimate3/u3mods/lfamp.html).
One end of the output transformer connected is to the +13.8V should be "cold" (no RF).
As already mentioned regulated power supplies often behave strange to reverse (RF) currents.
To rule out any hickups of the power supply I would suggest to run the PA on a large (car) battery or an unregulated power supply (just transformer + rectifier + large C of at least 10000uF). Now the voltage at one end of the output transformer should be (almost) pure DC.
If the PA run fine (expected output power and efficiency) now you should focus on AC decoupling the power supply (proper shoke with serveral uF of good caps at both ends).

73, Rik  ON7YD - OR7T

________________________________________
Van: owner-rsgb_lf_group@blacksheep.org <owner-rsgb_lf_group@blacksheep.org> namens N1BUG <paul@n1bug.com>
Verzonden: vrijdag 10 november 2017 19:44
Aan: rsgb_lf_group@blacksheep.org
Onderwerp: Re: LF: PA problem

Hi Stefan,

Your answer is not late at all. After working very hard on antennas
the past days I am just now getting back to the little PA.

About RF choke in the supply line: This is something I don't
understand about most of the LF/MF PA designs I have looked at. I
don't see such a choke in any of them! If I do see one, it is just a
few uH. Why don't all LF/MA PAs need a big choke in the +DC lead?

So today I put in a new FET. I set the bias so it draws 10 mA at
idle. Several people suggested I start there. With my Fluke DMM
between the power supply and the PA I got 20 watts RF output but
efficiency was very poor. The FET was generating a lot of heat!

I removed the DMM but left everything else exactly as it was. Now I
had only 17 watts RF output but the FET was running cold. This is
very similar to results I obtained some days ago.

So I started to probe around with the scope. I see a somewhat
distorted sine wave at the gate, a much greater amplitude distorted
sine wave at the drain, same at the PA RF output. I see a nice clean
sine wave at the other side of the LPF. But I see a lot of RF (36
volts peak to peak) at the +13.8V power connection to the PA! Surely
it is not supposed to put 36V peak to peak RF on the power supply!?
That seems crazy to me.

So either it does need a big choke there, or something is wrong with it.

I didn't do any experiments with different bypass capacitors yet. I
do not see any signs of oscillation with my 100 MHz scope but who knows.

73,
Paul N1BUG


On 11/08/2017 08:37 AM, DK7FC wrote:
> Hi Paul,
>
> My answer may be a bit late.
>
> It sounds to me like a EMC issue. A RF choke is needed in the supply
> line. 1 mH may be a good choice. And i would add a 1000 uF electrolytic
> capacitor with a low internal resistance (the good ones :-) ) near the
> Drain, in parallel to the 1 uF capacitor.
>
> A few more arguments: It is not ideal to drive a class E with a linear
> signal. It is better (regarding efficiency) to use a hard keying by
> using a MOSFET driver such as ICL7667, TC4422 or so. A class E is not a
> design for linear modes, it is for a constant output power, good for
> QRSS/DFCW, CW, WSPR, OPERA, EbNaut...
>
> I would recommend to build a class D PA. You can use the existing
> components. A class D is much less critical regarding matching (SWR) and
> the efficiency can be in a range of > 90% at LF as well...
>
> 73, Stefan