Hello Alan,
I have recently (this week) posted two e-mails on blacksheep, but
neither has appeared back in my mail box.
Is the RSGB reflector down, or is this a simple case of discrimination
against us "Murkans ? :-)
Best regards
André N4ICK
****************************
-------- Original Message --------
Subject: Re: LF: More on cores...
Date: Tue, 29 Mar 2005 15:00:31 -0500
From: Andre Kesteloot <[email protected]>
Reply-To: [email protected]
To: [email protected], [email protected]
References: <001301c533a6$e3f37c60$2102000a@ibm>
<004701c533c7$5f9b2a00$546ca450@preinstapdq0tf>
<000801c533ce$756d3ce0$2102000a@ibm> <[email protected]>
Marek SQ5BPM wrote:
> Hello all,
>
> I noticed that people are using much more turns than me in the output
> transformer, so I would appreciate any help in calculating the correct
> number of turns for the cores I am using here. Messing around the
> trafo would probably allow me to achieve more power than now (450W).
> Trying to go higher causes heavy ringing + FET heating. I've tried
> almost everything to reduce this (different combinations of Zobel
> network, decoupling, inductances in series with gates, screening,
> grounding etc.). In every case around 400W things start to go bad,
> 450W is a border of manageable heat and ringing. I am thinking about
> making a new FET driver in a way that allows regulation of the time of
> the FET on-state, leaving some rest period in-between.
In a single-ended Class C stage, we think of the load as the resistance
that will allow the collector to swing from the supply to
ground, hence R = V2/P and P =
(Vrms)2 / R (1)
This is a rms voltage, and since Vrms = (Vpeak / 1.414) we can replace
the value of Vrms in (1) hence
P = (Vpeak)2
/ (1.414)2 * R = (Vds)2 / 2R
When we have a push pull, we can consider it as being made up of two
halves, each supplying P/2.
Hence Zdd =
(Vds)2/2*(Pout/2)+(Vds)2/2*(Pout/2)=2(Vds)2/Pout (2)
So far, so good, for _sinusoidal_ signals.
Now it would seem to me that when we deal with class D circuitry, we
are dealing with (almost perfect) square waves, where average power,
r.m.s. power and peak power are all the same.
When a Power MosFet conducts, its internal resistance is so low that it
effectively brings the corresponding output transformer side to
ground. Because of the transformer action, the other drain is now at
2*Vdd, as can be seen clearly on a scope.
We can consider that each MosFet works 1/2 of the time from a drain
voltage that is actually (2*Vdd).
Hence the power switched into the core by both devices is
Pout = (2 *
Vdd)2 / Zdd .
When calculating the number of turns required for good transformer
action, is not unusual to consider each individual winding inductance as
having to be larger than 10 x the load and source impedance presented
to it.
Dye and Grangberg "Radio Frequency Transistors" is of course excellent,
but tends to focus more on wide-band RF amplifiers. For a very good
treatment of transformer-coupled class D power amplifier, you may want
to reference "Solid State Radio Engineering", by Krauss, Bostian and
Raab, pp.438-448.
Hope this helps
73
André N4ICK
