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

Re: DC power supply again




From: 	Malcolm Watts[SMTP:MALCOLM-at-directorate.wnp.ac.nz]
Sent: 	Thursday, September 04, 1997 3:12 PM
To: 	tesla-at-pupman-dot-com
Subject: 	Re: DC power supply again

Hi Harri,
 
> From:   Harri Suomalainen[SMTP:haba-at-cc.hut.fi]
> Sent:   Thursday, September 04, 1997 6:36 AM
> To:     Tesla List
> Subject:    Re: DC power supply again
> 
> On Thu, 28 Aug 1997, Tesla List wrote:
> > > Assume the gap fires, primary resonates and transfers only partially the
> > > energy during 1st half cycle to the secundary. Then, the primary current
> > > after resonance to 0 volts will start going partially to the cap (charging
> > > it negative!) and partially through the flyback secundary and diode.
> > > All the energy left in the cap (now negative) will have to be dissipated
> > > somewhere. Possibility of distruction? Have you thought about this?
> > 
> > Indeed yes. I have already successfully run coils from such a supply. 
> 
> Humm.. very interesting. Does the gap quench after the 1st half cycle
> (ie. when the cap has resonated to 0 volts) ?

No. It behaves as any other coil. You cannot quench at the primary 
zero-crossing unless the primary energy energy has mostly gone to the 
secondary because the mag field surrounding the primary is mostly 
coupled to the primary at moderate k's. That energy keeps primary 
current a maximum at that time. Perhaps you meant quench at the end 
of the first transfer? I've tried active quenching in the past 
and it hasn't really worked. I now just rely on the output streamers 
getting rid of the system energy efficiently. As well as that, it now 
seems that those rare long sparks rely on the oscillations hanging 
around long enough to establish long air streamers. To add to the 
mix, there might also be a subtle interaction going on between the 
beat frequency and resonant frequency such that the right combination 
of the two does the trick. This thing seems to be as much about air 
ionization and stepped leader characteristics as it does about coil 
behaviour :(  A nasty mix. 

> Have you checked that
> perhaps? I find the reason making the system never loaded by a negatively
> charged cap (after resonating) extremely important. It is the key to
> keeping the device in working condition! (conserns about the second
> setup below)

Yeah, right :)  I took a complete set of pictures off the scope of 
this system. Richard Hull either has or shortly will have them if you 
would like to contact him for a copy. 

> > The supply behaves like a current source of sorts and the gap shunts
> > the output while conducting so the rectifiers need not hold off more
> > than that required by the flyback design alone.
> 
> Humm.. Are you using a setup like this? This would seem the case you
> are talking about, right?
>                       cap
> +hv ------+-----------||------)
>           V                   )
>           ^   <-gap           )
>           |                   )
> gnd ------+-------------------)  <-- primary
> 

Exactly.

> The problem I was talking about appears with a setup like this:
>                 gap
> +hv -----+------><-------)
>          |               )
>          = cap           )
>          |               )
> gnd -----+---------------)  <- primary
> 

I wouldn't use a circuit like that on _any_ system.

> Unfortunately I thought the thing narrowly and was thinking only about
> the later type of primary circuit. Silly me.. :( At quick look the
> second one seems to work properly in your type of setup.
> 
> 
> > In fact using inductive storage is _exactly_ the right topology for 
> > the job. You will instantly blow a forward or transformer type 
> > converter if it has no current limiting and if it does, you are 
> 
> Certainly. I was all the time thinking on current limiting by perhaps
> with a secundary choke (like in the normal forward topology) or by
> some limiter in the primary side (like inductors in resonant topologies).

Fair deal. I might in fact have to do that (go to a bridge with 
charging choke) to get to 2kW because the necessarily high secondary
inductance throws a limitation on charge time depending on cap size. 
According to my calcs, I would be limited to 700W or so using a 50nF
primary cap with the original design. As luck would have it, I ran 
out of wire when the secondary was only half complete so I can do 
better but at the expense of half the original output voltage :(
     Any choke _must_ be in the secondary side or you throw a huge 
voltage burden on the secondary windings.
  
 > I'm not very fond of flybacks becouse they usually need larger
> transformers. They also induce more stress to the primary switching
> element(s). I'd go for some other topology but that's just me and
> my personal preferances.

The transformer design I've come up with is about 6" diameter by 
around 3 - 4" high. A lot of core gapping is needed. I am going to 
use IRF840's. That allows a lot of lattitude for changing to an H-
bridge if I wish because I chose the flyback voltage = rectified peak 
mains (about 330V). The compromise in choosing that value is that the 
output rectifiers have to hold off around 40kV (I'm building a 50kV 
stack - or was going to until the wire problem appeared). This design
is fine for about 2.3kW including losses.

> > likely to suffer serious gap quench problems. I have total confidence 
> > in my choice based on experiments I have already conducted.
> 
> That is certainly great to hear. SMPSUs would certainly be nicer than
> bulky polo bigs etc :) I'd just love to hear more on your findings as
> your project advances more!

I'll post. I am going to mount the core halves on paxolin this 
weekend. If I choose to go to H-bridge, I simply remove the core 
gapping and operate at a frequency that gives B(pk) of 250mT. I am 
designing two printed circuit boards for the two topologies to allow 
a quick change. In the H-bridge configuration, I can then wind chokes
to give the correct resonant charging conditions for the supply 
operating frequency.

> > I see simplicity as the best approach. I looked at multiple windings, 
> > multiple transformers and it didn't look good. With good design, 
> > switchers can be made pretty efficient with regard to heat 
> > generation.
> 
> Yap, 90% (or even better!) is very realistic nowadays.

Malcolm