Re: electronic PWM variac

["Tesla list" <tesla@xxxxxxxxxx>]

Original poster: Sparktron01@xxxxxxxxxxx 

I'm sorry I've not jumped into this thread, on a one week business trip,
family stuff, and getting ready for Ed Wingate's Teslathon.

Feedback on this thread through 8/23/2005...

1.  Snubbers:
     There is "a lot" of information on snubbers, primarily directed toward
     SCR/Triac 50/60Hz forced commutation.  I built one using a 33-100 ohm
     power resistor and 600V 1-10nF PP snubber capacitor.  At 12kHz, the
     IGBT within DC bridge (+/-) was seeing a significant reduction (1/3
     of original value at least) of non snubbered switching spike by use
     of RC alone.  I was running a NST basically unloaded running into
     a 1nF capacitor into a spark gap (I was trying to blow up the NST,
     cap, or IGBT).  Well after 4 HOURS of continuous running (300W
     measured, 9kV, 30 mA NST), the IGBT + heatsink was about
     130-140F, the transformer was warm... and everything was STILL
     RUNNING AND SPARKING...  :^D

2.  Gate Resistor
     I'm using a HUGE gate series resistor (1000 ohms).  Yep I'm throwing
     away ENORMOUS amounts of energy (<10W at equivalent 1kW load
      base),  and significantly reducing switching Ldi/dt (and
      attendent snubbing requirements)  to boot.

      I think I'll take heat (which I can eaily control with a larger
      heatsink or worse case force air cooling) versus weird random
      IGBT failures that will take a 100Mhz (minimum) scope to find the
      glitches.  With a 20-40Khz carrier frequency, even 2 uS of turn
      on/off time will dramatically reduce EMI/EMC issues with
      modest increases in device heating.  You are controlling a 60Hz
      load, not a DRSSTC at 250kHz!!!!!

3.  Low Pass Filter (LPF)
     I have a signifcant number of 8.2uF / 330VAC PP capacitors out of
     a 60Hz 2kVA CVS.  Running a Butterworth LPF simulation in EWB
     came up with from source end 8.2uF across line to neutral, series
     L of ~ 500uH (at rated load) and 16.4uF (2 caps) as final output
     shunt C prior to load.  HF component was attenuated 40dB at a
     minimum (Fo of PWM 20kHz).  500uH inductor is only "difficult/tricky"
     design hurdle...

4.  Controls
     I was hoping to use current feedback control, however it is
     conditionally stable (i.e can run away and become unstable) without
     slope compensation, and with a surging load (i.e. big load step changes).
     A TL 494 in voltage mode control (with independent current limiting
     ballasting mode) is a much simplier, proven, robust design; and far
     easier to compensate.  Even with a Fco of 500Hz, PWM single ended
     control  is significantly faster current limiting then manual control
     or an R/L ballast ...  ;^)

Regards
Dave Sharpe, TCBOR/HEAS
Chesterfield, VA USA


> Original poster: Jim Lux <jimlux@xxxxxxxxxxxxx>
>
> At 11:38 AM 8/18/2005, you wrote:
> >Original poster: Brady Hauth <bhauth@xxxxxxxxx>
> >
> > > Has anyone tried to turn one of these into an "electronic ballast"
> > > that synthesizes the effect of series inductance into a SG coil?
> >
> >I believe this is how fluorescent light ballast works, and that runs
> >about $1/watt, which as far as I'm concerned is ridiculously expensive
> >when you use a MOT for virtually all ballast needs,  shorting the
> >secondary or tapping the primary as necessary.
>
> Unless you're ballasting a 20 kVA pole pig and aren't interested in
> carrying around another 100  plus pounds of iron and copper.
>
> A PWM primary controller also potentially could implement some
> interesting voltage/current characteristics not possible with a
> conventional R/L ballast.
>
> Even more interesting might be a three phase configuration driving a
> DC coil with a triggered spark gap.
>
>
>
>
>