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

Re: Optimal Quenching

Subject: 
            Re: Optimal Quenching
       Date: 
            Mon, 17 Mar 1997 12:45:41 +1200
       From: 
            "Malcolm Watts" <MALCOLM-at-directorate.wnp.ac.nz>
Organization: 
            Wellington Polytechnic, NZ
         To: 
            tesla-at-pupman-dot-com


Hi John (Freau), all,
                        Firstly, John requested a description of my 
MOSFET gap setup. Briefly, I used a system resonant at about 180kHz 
(from memory - I don't have my notes here) coupled at k=0.1. The 
primary coil had a rather low inductance of just a few uH and Cp was
about 0.5uF. Explanation: with low primary voltages around 30 or so, 
I chose a rather large cap to get some energy in to make measurements 
easier. The gap consisted of 4 MTP3055E MOSFETs wired in parallel and
the primary alone gave a measured Q of 9. I was able to use the 
standard decrement formula because all losses were linear-resistive.
The low Q mirrored the appalling L/C ratio and L/R ratio. It was a 
substantial improvement on a single MOSFET alone though. That gave a 
Q around 4 if I remember correctly. Secondary output was monitored 
using a short piece of wire dangling from a scope probe several feet 
away from the coil. The primary was scoped directly across Lp on some 
occasions, and the "gap" on others. The "gap" was placed across the 
supply.
     The MOSFETs were turned on/off using a precision pulse generator 
with a 50 Ohm output impedance (10V capability) and the output was 
terminated at the gate-source with a 50 Ohm resistance to get good 
risetimes. Using this setup, I was able to investigate the effects of 
different quenchtimes, resonant charging of Cp using a choke 
(sometimes with a series diode) between the DC power supply and 
primary circuit, transfer times, class C operation (I could dump the 
cap at the resonant frequency by carefully adjusting generator 
frequency and dwell (pulse) time - very difficult). This was 
facinating though. With an impulse of constant (*not decrementing*) 
amplitude transferred to the secondary evey cycle with the right 
phasing, the output soared incredibly due to the high output 
impedance of the non-sparking secondary as the energy accumulated
on each cycle. There is no magic about where this kind of rise comes 
from. Because of the very low distributed Csec, any movement near the 
coil afffected this output greatly. Bringing a hand close reduced it 
to near zero as the secondary tune and hence relative phase wrt the 
primary changed.

     A disclaimer - the astute will note that there are parasitic 
substrate zeners present in the MOSFETs. This in no way affected the 
amount of energy transfer - they did affect low frequency primary 
behaviour with the charging choke as the primary cap and choke rang 
at a low frequency after the gap was cut off. It also meant that 
"quenching" the primary was less than clean on a negative half cycle
of oscillation, but the DC bias presented by the power supply 
overcame this effect substantially, esp. with an inductive feed from 
the power supply. I also used the power supply current limiting 
(current source characteristic) to good effect on occasions.

    All this is real - I took two rolls of film off the storage scope 
which showed everything captured in graphic detail complete with 
timebase and amplitude settings. I'd be happy to send some annotated 
photos to John if he wishes. I can't do this for everyone because of 
the cost sorry. There may be some of these on ftp sites as I have 
already sent some overseas.

    I am composing a piece to describe the physical detail of how the 
sidebands arise in the overcoupled system but this will take several 
days.

    A final note about the MOSFET experiments: anyone (everyone) who 
uses either transformers with leakage inductance built-in (neons) or 
near perfect transformers (pigs) with "current-limiting" inductors 
attached should know that current-limiting applies *only* to limiting 
the transformer current in the gap. It has exactly the opposite 
effect when charging the primary cap because the two form a resonant 
circuit of rather high Q. In many ways it is a Catch-22. There is an 
obvious need to reduce transformer discharge current in the gap (to 
zero if possible), and yet let the reactance of the primary cap at 
mains frequency control its charging current and hence final voltage.
The relevance of all this to my experiments? At low rep rates and 
with a charging choke between the supply and primary cap, the energy 
stored in the choke was able to charge Cp to near 60V from a 6V 
supply!!

Malcolm