Successful TPU-ECD replication !

[Earl]

Hi Harti,

this is a good idea only as long as you do not need fast switching.  The spec sheet only guarantees 1.75 mA output current.  A normal FET driver IC can deliver 3A, 4A, even 9A, so you can see that even putting all gates in parallel still delivers too little current to charge up the power FET input capacity of maybe 2nF.  In addition you have to get through the so-called Miller zone.

I prefer a fast, driver IC, but sometimes use the attached "cheap and dirty" circuit.  If you run the IC at 6V, then you can just get by with a "logic-level" FET.  Because the transistors operate as emitter followers, there are no storage charges to slow them down, but you lose 0.6V base-emitter.  Emmitter followers have very low output impedance.

Since radiant events are suspected to be caused by switching current on and off extremely quickly, the best is to turn the power FET on and off as fast as possible.  This can only be done by a driver IC, say operating at 10-15 VDC, spec'ed to charge and discharge the FET gate in 10 or 20ns.  But in practice you will not achieve this unless you use multiple SMD ceramic C's of different values.  IT IS NOT THE POWER SUPPLY THAT DELIVERS THE MULTI-AMP PULSES, IT'S THE C's AND THE C's ONLY.  The driver IC must be located directly at the FET's source and gate terminals, say closer than 1 ~ 4 mm NO MORE !!!

Nanoseconds = small
high currents = small
big = antenna

Having wires running all over your table is not how you do nanosecond switching.  But it makes a fantastic transmitter OF RF.

It is a science and an art to build fast switching circuits.

Regards, Earl

Why  are you using these expensive driver FETs ?

Why not just control the IRF 840 ?s gate with a parallel circuit
of 3 or 4 schmitttrigger hex inverter ICs like
MM74C14 Hex Schmitt Trigger  ?

Here is a PDF File.

Just put at least 2 or 3 of them parallel and there you go.
Should have enough punch to hit the IRF 840 gate fully open !

http://www.fairchildsemi.com/ds/MM/MM74C14.pdf

[hartiberlin]

Hi Earl,
many thanks for your experience and your circuit.
Maybe it is only required to have a fast OFF switching in this circuit,so
your circuit still needs a low ohmic resistor like 100 Ohms from Gate to ground ?

or is the lower PNP transistor enough to pull the gate down ?

Well, we are working here in the Khz ranges and I ask myself,
if we really need this fast Nanosecond switching ?

Would be intersting to see, if the TPU-ECD would also work with
being controlled by 3 different sine frequencies..

Regards, Stefan.

[Earl]

For a sine wave, I think it can safely be assumed that pure means low-distortion, low harmonics.  For a square wave or pulse, it is more difficult, but could be interpreted to mean either low-jitter (jitter = noise in effect) or high frequency-stability, low drift.  This might mean no R/C elements in the oscillator, only L/C elements permitted.  This rules out 555 timer circuits.

Regards, Earl

@ Duff

SM said "Pure frequency", not "Pure Sine wave".  I made that same false assumption in my thread, until re reading SM's words.

Cheers,
Bruce

@ Bruce & gn0stik

Thanks for the correction.

So minimum harmonic content  might equal less harmful effects to deal with.

-Duff

[Bruce_TPU]

@ Earl

SM suggested in his posts, to build a "mini tube amp" using Mosfets, to power the TPU, to start with.  He said it would be faster all around, to see something happen.  Can you explain as an electronic guy, how close to what is being done, is to a mini tube amp?  Is there any similarites?  What would the difference be?

I think this is an important post of SM's.

Thank you,
Bruce

[Earl]

Hi Harti,

As I understand Tesla, he sais both turn-on and turn-off should be as fast as possible and he tried everything he could to achieve this.

I personally do not like any resistance in the gate for switching applications.  My circuit is based on always having a "1" or "0" gate output.  Many gates do NOT have tri-state outputs (open circuit).  One of course always has to pay attention not to let a static charge zap the gate while handling/soldering.

The lower PNP transistor pulls the gate down quickly to ground since an emitter follower has low output impedance.

Very glad you asked the question about "only kHz range".  One time I told a group of assembled Engineers from one of the biggest, most well known manufacturers that were having trouble with their switching power supply that their "only kHz" power supply was a 1 GHz amplifier overdriven in the kHz range.  They all laughed.  Well their laughing stopped later when they realized that I was correct.  I saved the as* of the project manager.

Fast rise/fall times has advantages:  it may create radiant energy and the FET switching losses are minimized.  If there is too much EMC / EMI / EMV, then your circuit is too big.  All current loops must be physically as small as possible.

Also remember a "ground" only exists at DC.

Regards, Earl

Hi Earl,
many thanks for your experience and your circuit.
Maybe it is only required to have a fast OFF switching in this circuit,so
your circuit still needs a low ohmic resistor like 100 Ohms from Gate to ground ?

or is the lower PNP transistor enough to pull the gate down ?

Well, we are working here in the Khz ranges and I ask myself,
if we really need this fast Nanosecond switching ?

Would be intersting to see, if the TPU-ECD would also work with
being controlled by 3 different sine frequencies..

Regards, Stefan.