[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [TCML] DRSSTC weird waveform behavior



hi Justin,

On Sun, Jul 3, 2011 at 11:38 PM, Justin B <boshjack@xxxxxxxxxxx> wrote:

>
> Steve, When I do my scope work, I have the probe connected to the OUT (pin
> 3) from the CM300.  The black clip on the probe goes to the other primary
> connection. I use an isolation transformer so I can do this without any
> issues of that black clip, blowing up my bridge.  So that being said, I have
> no real mains ground being connected to the bridge, to ground any of the 0V
> lines if thats what you ment by "Wheres your ground clip connector"  If you
> ment the probe, other connection of the primary, which if im not mistaken is
> the middle of the voltage doubler.  So, as your schematic shows this as a
> neutral, well, I have an isolation transformer, so theres no real ground.
>

This is all fine, though you do need to be careful about the scope probe
inductance picking up voltage induced by the large fields near any of the
primary conductors.  This induced voltage is usually 90 degrees out of phase
and can make your voltage measurements (particularly the parts that should
look flat) look funny.  Anyway, you are going about things "correctly" from
my point of view.


>
> I did find I can adjust the voltage waveform left/right by having the
> resistor and inductor in series. and have the two items, in parallel to the
> CT.  I was able to shift it, and knock down ringing.  But, I figured i
> couldn't get it fully down due to lack of variable inductor.  So, before I
> call anything else an issue, I have one on order that will do the job.
>  Hopefully things will quiet down, and fix my cross sectional issues too.
>  The real issue comes down to the fact that im using the old slow CM300s, so
> I need to play with delay time, which I can only do by slowing the gate by
> adding more resistance.  Obviously too much is bad too.
>

I think you said you had a resistor of several K ohms with the inductor?
 This is far too large for practical tuning.  For example, i use about 50
ohms with a 50uH inductor for obtaining enough phase lead for CM300s.  Any
more resistance that the network looks too resistive and you dont get enough
phase lead.  So unless you are using a really huge inductance here, than
anything over a few hundred ohms is gonna look mostly Resistive, and not
produce much phase lead at all.  Id suggest getting something like a 200 or
500 ohm pot in there, and at least 50 to 100uH inductance.  You know, you
can actually calculate the phase angle if you review how complex numbers and
phasors work (or look it up).


>
> So, right now im awaiting to test that, see if the variable inductance can
> do anything different, since the pot was able to shift the waveform.  But if
> im doing isolation, DOES my bridge need an actual mains ground at some point
> on the 0V connections?
>

It shouldnt be *necessary* to have a variable L, as long as you get the
right combination of R and L (again, read up on phasors) to set enough
leading voltage.

And no, your bridge does not require mains ground.

Steve


>
> > Date: Sun, 3 Jul 2011 14:36:10 -0500
> > Subject: Re: RE: [TCML] DRSSTC weird waveform behavior
> > From: steve.ward@xxxxxxxxx
> > To: tesla@xxxxxxxxxx
> >
> > With regards to the LR network on the feedback CT output: if you are
> using
> > clipping diodes, you will get non-linear behavior of the circuit, and it
> > will no longer provide a constant phase lead as it is intended to because
> > the current will start flowing through the diodes rather than the linear
> L
> > and R components, and this will cause the behavior to change.  The series
> > R/L network across the CT output causes the voltage developed across the
> CT
> > output to lead the current seen at the input of the CT.  This is quite
> > different than how i used CTs for my controller which your coil is based.
> >  In that case the voltage was simply in phase with the current, and
> produced
> > whatever voltage necessary to allow that current to flow in the clipping
> > diodes.  There is no way to add Ls and Rs to this circuit and have it
> work
> > properly.
> >
> > Also, given the waveforms, it does look like an awful big voltage drop on
> > the bridge output, i suspect either a lot of stray inductance with the
> > measurement reference (where is your ground clip connected??), or a lot
> of
> > stray inductance from the IGBTs to the main bus storage capacitance, in
> > which case the bus inductance is really *part of* your primary coil
> > inductance, and thus shows some voltage drop.
> >
> > Finally, and getting back to the original issue at hand.  Im not sure if
> i
> > ever posted the final conclusion on these voltage spikes due to late IGBT
> > switching.  Yes, i did experiment with more gate turn on resistance to
> slow
> > down the transition, and this was helpful.  But in the end it seemed that
> > these voltage spikes, caused by free-wheeling diode recovery, were so low
> in
> > energy that they posed no real threat to the IGBT.  So from then on, i
> used
> > a moderately low 5 ohm gate resistance and kept the switching speed up.
> >
> > After seeing Finn's "Prediktor" controller use the lead compensator so
> > nicely, i developed my own, though have yet to publish it in any formal
> way:
> >
> > http://www.stevehv.4hv.org/leadcomp/UD2_1revB/
> >
> > And now the switch timing can be controlled much better, allowing the
> IGBT
> > turn on to happen really at zero current, which keeps the wheeling diode
> > recovery from causing big voltage transients.
> >
> > Steve
>
>  _______________________________________________
> Tesla mailing list
> Tesla@xxxxxxxxxx
> http://www.pupman.com/mailman/listinfo/tesla
>
_______________________________________________
Tesla mailing list
Tesla@xxxxxxxxxx
http://www.pupman.com/mailman/listinfo/tesla