Re: LF: W1VD amp help - IR2110 and timing

[<jrusgrove@comcast.net>]



Andy

 

Good idea on the multifilar winding.

 

>Everyone, stop, think. Stop clutching at straws

 

A fair number of these amplifiers have been built ... none of which exhibit the severe ringing that this particular one does. We're currently 'clutching at straws' to figure out what's different about this one.    

 

Jay W1VD

----- Original Message -----

From: Andy Talbot

To: LineOne

Sent: Tuesday, April 10, 2018 5:01 PM

Subject: Re: LF: W1VD amp help - IR2110 and timing

Everyone, stop, think. Stop clutching at straws

Ringing like that, showing several cycles of oscillation has to be  caused by a resonance at a higher frequency than the wanted one.

Resonances are caused by stray L and C.

Where is stray L coming from?

Leakage inductance.

It's classic - seen on any SMPSU.   You have to get your windings really well coupled.  A couuple of turns on a large torroid I'm sure can't be completely leakage-free.  Unless Ur were infinite.  So why not try multifilar windings ?

If you want 1+1:2 turns ratio, for example,  twist four conductors together and wind the resulting set the requisite number of turns.   Use two of the strands in series for the "2" secondary, and the other two as your 1+1 primary 

Multifilar gives you much tighter coupling

Andy G4JNT

On 10 April 2018 at 21:45, Chris Wilson <dead.fets@gmail.com> wrote:

Hello Paul, here's a link to my prototype build, the more I think back
to bad drain waveforms the more convinced I am that the major fix was
added  decoupling. I read Jay's very salient post about none inductive
probing,  but  for  clarity  my scope captures are with two X10 probes
with  the  hooked  spring  tips  on,  and the normal ground leads both
clipped  to the driver board PCB ground plane, next to one another, so
nothing careful done there to get those captures.

As  we  seem  to  be  running  short  of  ideas re the amp itself, bar
decoupling  and  short  gate  wires,  what about the LPF?? I used blue
polypropylene  (I think...) caps of about 1500V DC> Each section is in
its  own  totally sealed box, the mid toroid has two windows milled in
the box with gauze over them for a PC processor fan to blow air across
the  inductor. Each outer box feeds through to the mid one via drilled
holes  of  about  10mm  diameter  with  the  inductor  wires in Teflon
sleeving. Directly soldered. Caps are in each end box.

Decoupling caps are at http://www.chriswilson.tv/decoupling.jpg

they  are done on the actual board of the MKII amp ;) But functionally
there was no difference.

Tuesday, April 10, 2018, 9:04:45 PM, you wrote:

> OK, let's switch topics a bit.

> I have a serious drain circuit ringing problem that most builders
> apparently don't have. Many things change all but the first peak of
> the ringing, but that first one is stubborn. It almost seems like
> that first big peak is being "forced" to occur somehow.

> I believe someone mentioned, though I cannot find it now, that cross
> conduction (one phase starting to switch on before the other is
> completely off) would exaggerate ringing.

> I've become curious as to whether the timing in my amp is correct.
> No one has really commented on this aspect. My driver board is
> unique and it would not be at all surprising it it has stray
> capacitance in excess of most!

> Here are the two gate waveforms overlaid so you can see the crossing:

> http://www.n1bug.com/gates4-wdv13.jpg

> And here are the drains similarly overlaid to show the crossing:

> http://www.n1bug.com/drains2-wdv13.jpg

> *Is* this OK or could the timing be just a bit off? I lack the
> knowledge to interpret what I am looking at.

> *If* it could possibly stand some improvement, do I have control
> over it? I think the 1000pf caps and 15k resistors on the IR2110
> inputs would affect this timing but I haven't been able to work out
> exactly how or what to change if I wanted to experiment:

> http://w1vd.com/137-500-KWRev3.0.pdf

> Paul

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