Date: Thu, 25 Jul 2019 12:42:53 +0100
From: Chris Wilson <dead.fets@gmail.com>
Message-ID: <1601026811.20190725124253@gmail.com>
To: Andy Talbot <rsgb_lf_group@blacksheep.org>
In-Reply-To: <CAA8k23T9n+vAR1R-7VTe9354WpySR9kw05+64cZExXh7nRByVA@mail.gmail.com>
References: <993240756.20190717143805@gmail.com> <5D2F677F.1090208@posteo.de> <CAA8k23T9n+vAR1R-7VTe9354WpySR9kw05+64cZExXh7nRByVA@mail.gmail.com>
MIME-Version: 1.0
Subject: Re: LF: DC restoration
Content-Type: text/plain; charset=utf-8
Sender: owner-rsgb_lf_group@blacksheep.org
Precedence: bulk
Reply-To: rsgb_lf_group@blacksheep.org
Content-Transfer-Encoding: 8bit


Hello Andy and Stefan,

Thanks  for  the  replies,  it's  a lot clearer now, I appreciate your
time!

Wednesday, July 17, 2019, 7:50:39 PM, you wrote:

> I assume we're talking about the series C from driver to gate, then
> a high value leakage R from gate to ground, and a reverse biassed diode in parallel with this.


> AC coupling without any DC restoration means the MEAN voltage on
> the gate is zero.  For LF transmitter uses, where the drive has a
> 50% duty cycle (or very close to) this is not too much of an issue
> as there will still be a significant positive swing.  So provided
> Vdd  / 2 of the driver stage is more than sufficient to fully
> turn-on the FET than AC coupling with just a C and leakage R is OK. 
> Transformer coupling - allowing galvanic isolation - works very well and that givs true AC drive.


> The problem comes in SMPSUs where the duty cycle is not zero.  
> Assume a 20% duty cycle and a drive waveform of 0/15V.   The mean
> voltage on the gate still has to  0V, so the get this from a 0/15V
> wavefomr means a positive excursion of 3/4 * 15V = 12V and a
> negative excursion of 3V.  That will still work OK
> But an 80% duty cycle is +3V and -12V that certainly won't.


> By adding a DC restoration diode, the negative excursion is clamped
> to around -0.6V (ish)) and the positive lifted to 14.4V (ish) WHATEVER the duty cycle.   


> SO even at 50% the drive amplitude is doubled.


> I think earlier IGBJT devices needed a negative voltage to turn off
> properly - I doubt that is the case with later ones.   And IGBJTs
> don't seem to have been adopted for LF transmitters anyway.


> I remember learning about DC restoration when I was quite young and
> being shown how TVs (monochrome at that time) worked.   It was a
> simple way of getting a constant black level on AC coupled video
> amps, by clamping the recovered sync pulses to 0V, then biassing to
> get black level, whnere teh video signal have a varying DC
> component.   Later designs clamped the back porch at true black
> level.  Exactly the same circuit arrangement


> Andy
> www.g4jnt.com


> On Wed, 17 Jul 2019 at 19:26, DK7FC <selberdenken@posteo.de> wrote:

> Hi Chris,
>  
>  Maybe it helps to show a concrete schematic together with your question.
>  I find reverse diodes quite useless, at least when MOSFETs are used. It
>  may be a different thing with bipolar transistors...
>  
>  73, Stefan
>  
>  Am 17.07.2019 15:38, schrieb Chris Wilson:
 >>
 >> Hello  rsgb,
 >>
 >>    Some Class D LF / MF amps use a diode to DC restore the waveform
 >>    after the FET driver chip to the gates. Why do they do this, I am
 >>    thinking the negative part of the waveform would turn OFF the gate
 >>    harder, assuming the positive part is adequate to turn the gates ON
 >>    hard enough? Is there any point in using the modern FET's designed
 >>    to have a negative rail to turn the gates off hard? Cree do the
 >>    Wolfspeed ones with this feature. I have also noticed that with amps
 >>    where the input to the driver is capacitively coupled adding a DC
 >>    restoration diode makes the waveform pretty much exactly 50 / 50
 >>    mark  space,  whereas without one this is not the case. Good idea to
 >>    use these restoration diodes there? Thanks.
 >>
 >>    
>  
>  


-- 
Best regards,
 Chris                            mailto:dead.fets@gmail.com