The switch killer

[Login to see usernames]

I'm the switch killer 

I found why i killed so many in this long way... I never actually knew what was really going on.

On short,

What happens is that when the primary coil of a transformer is switched on and than off, the polarity reverses and so it sums with the source voltage to charge the capacitance of the switch... if normal operation where maximum power transfer is the objective the rating of the transistor should be at least twice the voltage of the source.

But when you pulse the primary with Dc and than don't use all or most the power you are pulsing into it will appear on the primary as a high voltage summing forces with the sources voltage to destroy the switch.

There are means for snubbing this pulse that destroy the equipment, this is well known as snubbers, they provide a way to that high voltage power to be dissipated in a resistor at a lower voltage since a capacitor absorbed this power before. 

This limits the collapsing time.

As the faster is the collapse the higher the voltage generated, to effectively have a ultra high speed collapsing time would require little to no protection on the switch, nor across the primary for sake of clarity

But how to do this and don't kill the switch?

How many can we add in series? 10? 20? 30? ...?

It just takes some work and money...

[Login to see usernames]

The solution i found to be varistors. They are like zener diodes but are more powerful, there are other solutions however. but the idea is to create a system where the transformer magnetic field developed during pulse on, collapses as fast as it can as the switch turns of to reach the highest voltage possible. But how to make it controlled? I was trying with resistors but was a mess i keep burning...

 Well using varistors since they will clamp the voltage to a maximum set for the switch being used. Now the mosfets can run heavy inductive loads happy.

Now for saying i'm working on 0-100v  range for the primary with a T factor of 50 and allowing the voltage to grow up to 10x more than 100v

So 50kv within up to 5000 turns... primary with 100 turns...  so one volt per turn to 10 volt per turn during collapse since its limited by the varistors...

[Login to see usernames]

Actually another solution could be a non-dissipative snubber having a varistor to protect only the over voltages... a snubber winding is required however


I think i found a good solution for clamping the voltage of the transformer,

Using a 100v dc source with a capacitor in parallel for example you get two winding the primary and the clamp winding and a diode.

The primary is connected as usual, but the clamp winding is connected from ground to the positive thru a diode reverse biased...
When the switch turns off the voltage across the primary is not allowed to grow to infinity since the clamp coil will return the energy to the input circuit limiting the output voltage with little power consumed.

The secret to allow the field to collapse is to have a snubber coil smaller than the input coil by the factor of voltage amplification during pulse off time condition desired.

Of course snubbers and other surge protection is still required but theres never perfect coupling and theres also leakage inductance but  power dissipation on this will be minimal if the circuit is well designed.

Since the input capacitors are generally big they can re store lot of power with no problem...

[Login to see usernames]

I think i see why meyer wanted a big diode in the circuit check the attachment...

This circuit is design to be able to give voltage regulation of up to 5 % if current is up to 2A, this mean that when the field collapses the source will receive back a short burst of amps and as the capacitor is already charged the discharge becomes very fast.

So in the end the energy is sent to the coil and the energy that was not used by the secondary return to the capacitor as a small ripple. 

[Login to see usernames]

Circuit example, primary 100 turns 100v applied 10khz

If the transistor handles 1200v than you can allow the 100v to be increased by a factor of 12, but lets assume 10 for safety so the clamp coil should have 10 turns... As theres 10 x factor of transformation the amps in this coil are 10x greater so if you are using 5 amps in the primary on the clamp coil there must be 50 amps so the diode must sit on a heat dissipation metal. Of course the amps are very fast not really continuos thus the power dissipation is not very very high... however.

Now if the input capacitor is big like 100miliFaradays than it can handles more amps without much ripple.

the clamp coil return the energy to this input capacitor basically and since its always charged from the 100v source the voltage will vary within a certain amount, the ripple.   

[Login to see usernames]

http://www.ijmer.com/papers/Vol2_Issue5/DE2536193621.pdf

this show exactly the snubber winding i suggested

[Login to see usernames]

Hi Fabio,


So your goal is to re-use some lost energy?

Steve

[Login to see usernames]

Basically is just a way to dump lot of energy in the coils, and if its not used its just recycled like in resonance. This recycling clamp protects the circuit and limits the output voltage to a safe level for the switch. Is basically a non dissipative snubber so it simply don't waste too much energy like using only varistors and dissipative snubbers of any kind.

The input power in this way will probably be little lower than the power being pulsed because of this resonant like carachteristic

The important is to have the big capacitance in the source sequence,,,  This is the energy reservoir...