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

Re: SSTC questions



Original poster: "Malcolm Watts by way of Terry Fritz <twftesla-at-qwest-dot-net>" <m.j.watts-at-massey.ac.nz>

Hi Jan,

On 4 Aug 01, at 6:33, Tesla list wrote:

> Original poster: "Jan Wagner by way of Terry Fritz <twftesla-at-qwest-dot-net>"
<jwagner-at-cc.hut.fi>
> 
> Hi,
> 
> some questions about SSTCs, again. ;o) It's about a full-bridge SMPS.
> 
> 1) which impedances must be matched, really? someone said that all
> should be matched. Does that mean
>  a) estimated streamer load input impedance

Well estimated load impedance. For air streamers there seems to be 
concensus in some quarters that a figue of 220k//1pF per foot of 
spark is about right.

>  => b) secondary output impedance

= the estimated load impedance. Calculate Zo for your coil - there is 
a formula on tssp somewhere or as a first approximation use 
SQRT(Ldc/Cmedhurst). The plug those values into the little formula I 
gave to find Zbase. 

>  => c) secondary unloaded base impedance

As you'll see, there is an impedance change once sparks start flying. 
I would forget about the unloaded output impedance and go for the 
loaded target.

>  => d) step-up transformer output impedance

Which = Zbase of the resonator as calculated from the considerations 
above.

>  => e) driver input impedance (Uin_min^2/Pin)

the combination of (d) and (e) will now give you a turns ratio for 
the driver (remembering that Zout/Zin = (Nsec/Npri)^2 

> This seems like quite much... Ok a) and b) probably will match
> themselves "automatically".

They are one and the same.

 But for the designing part, wouldn't it be
> enough to just match unloaded TC base input Z to the driver input Z
> with that step-up transformer?

If the goal is to deliver power to sparks then matching to a loaded 
(sparking) condition should be the goal shouldn't it?
 
> I figure that the transformer output Z should be equal to the TC base
> input Z, right? 

Yes.
 
> The other option would be narrowband 1/4wave transformer, i.e.
> T1_out_Z = SQRT ( Z_pri_in * Z_TC_in ). Intuition tells this might be
> the better way.

T1_out_Z = Z_TC_in.
 
> It'd be a big help if some SSTC-gurus could comment on which method is
> the correct one?
> 
> 
> 2) for a full-bridge, how do I calc the necessary commutating diodes'
> current handling capability, repetetive-pulsed, for the case when a
> large amount of the stored energy from the secondary is reflected back
> into the driver primary side?

Since primary current cannot change instantly, assume that all 
primary energy will backfeed into the power supply rails. Then 
(minimum) peak repetitive diode rating = peak primary current 
assuming all energy is bottled up in the transformer primary (safe 
assumption but of course your sparks are going to reduce this figure 
:)
 
> 3) primary current: the peak (sinusoidal) current is probably equal to
> the (square-wave) peak flat-top current?

You cannot have two waveforms present at once.
 
> 3) when the worst happens and considerable stored energy is reflected
> back to the driver primary and tries to charge the main storage
> capacitors, should there be some sort of overvoltage protection
> accross the step-up transformer primary winding? Like MOVs/VDRs? Or kW
> transient suppressors? Anyone found this to be necessary?

Your power supply filter capacitors should be sized such that they 
store considerably more energy than you deliver to the load per 
cycle. That will keep such a voltage rise to a value that is safe for 
the switches. The power supply cannot deliver more power after the 
first cycle if the filter cap is charged to a higher value than the 
rectified peak voltage. I don't think you will find this to be a 
problem. I would allow a 20% voltage margin for all transistors/FETs. 
The chosen current ratings of the devices should take into account 
the losses that will be incurred. For example, I was tempted to use 
some 2SC3997 (from memory) in a flyback charger. I pulled these out 
of some scrapped Apple monitors. Awesome ratings at first glance. 
250W Pdiss, 800Vceo, 20Acont. and I was proposing to run 18A peak 
repetitive. Unfortunately, Vce(sat) rises (exponentially) to about 5V 
at these current ratings and while decent heatsinking would take care 
of this, the power losses are more than I want in an efficient 
supply. For the toplogy I had in mind, the losses would have amounted 
to 180W peak (about 45W on average). To reduce this to a much more 
reasonable figure, I would have had to parallel two devices in each 
half of the circuit. That raises other problems, most notably extra 
components (double the number of transistors), ballasting resistors 
for current sharing and power losses in ballasts not to mention 
additional difficulties in accurately monitoring primary current. My 
solution to these problems was to bite the bullet and choose a couple 
of decent IGBTs to do the job far more cleanly.  
 
Regards,
malcolm