MOSFETS do not switch off properly

[Hayato]

Hello,
I've been trying this for a while (about some weeks).
I built a mosfet driver to drive an ignition coil.

The first driver I've built was very simple, and the mosfet didn't turn off fast enough to generate the "kick effect/auto-inductance/back emf" (I don't know how do you call that ) so high voltage was not generated.

I could see that the MOST was not switching off, because I was using an 20A ammeter between the drain and the coil.

I tried with 2 mosfets, IRF3205 and IRF740.

Fail #1:
Sparks were not generated.
Mosfet didn't turn off.


Fail #2:
Some weak sparks were generated.
Mosfet didn't properly turn off.
(I could improve the performance by decreasing R2 value (10k) to 300R)


Fail #3:
Strong sparks were generated.
After a few seconds of working, the mosfet starts to conduct alone. So I have to turn off the supply to "reset" the circuit.


Any ideas?

 

[Hero999]

What voltage are you using for the signal?

Use a push pull trigger signal source like a 555 timer, and at least 10V for the signal and reduce the resistor in series with the MOSFET's gate to 10R or better still, get rid of it and use a small ferrite bead.

 

[Hayato]

What voltage are you using for the signal?

Use a push pull trigger signal source like a 555 timer, and at least 10V for the signal and reduce the resistor in series with the MOSFET's gate to 10R or better still, get rid of it and use a small ferrite bead.

Hello,
I'm using a 555 timer as signal source.
The signal is about 11V.

You mean, use a ferrite bead as a resistor? Or to make a "little transformer"?

 

[muki55]

Put MBR1645 or similar in the primary of the transformer (+ at + supply and - in the drain)

Industry Number : MBR1645
R-SCHOTTKY, 45VOLT, 16AMP, TO-220

 

[Hayato]

Put MBR1645 or similar in the primary of the transformer (+ at + supply and - in the drain)

Industry Number : MBR1645
R-SCHOTTKY, 45VOLT, 16AMP, TO-220

Nice idea.

To protect the MOST against back emf, isn't it?

But I still don't think that it's the cause of my problems.

 

[muki55]

Just try it!!!!!!!!!

and here's the working one but not mosfet output it's (2N3055) output.

 

[Hayato]

Just try it!!!!!!!!!

and here's the working one but not mosfet output it's (2N3055) output.

Hello,
I simply gave up using 2N3055, I had to parallel some of them, and they did not heat equally (I was using 0.5 ohm ballast emitter resistors).

 

[muki55]

Maybe this one will help. If not: Try tesla circuit. (Just joking) Here's my last one trying to help You. I'm Tryingggggggggggg. hu! hu! hu! hu! hu!

C1- 1uf/200v

 

[Hayato]

Maybe this one will help. If not: Try tesla circuit. (Just joking) Here's my last one trying to help You. I'm Tryingggggggggggg. hu! hu! hu! hu! hu!

C1- 1uf/200v

Hello,
Thanks for the helps, but I prefer to design the circuit myself. To take someone else job is not funny.

 

[Analog]

Hello,
I've been trying this for a while (about some weeks).
I built a mosfet driver to drive an ignition coil.

Fail #3:
Strong sparks were generated.
After a few seconds of working, the mosfet starts to conduct alone. So I have to turn off the supply to "reset" the circuit.
View attachment 9439

Any ideas?

I believe your magnetic core is reaching saturation because of asymmetric pulsing (only positive going pulses). And that's why it seems the switching device is always on.

To fix this, you must apply symmetric postive and negative going pulses to 'reset' the magnetic core of the transformer. Many times, you can also use a free wheeling diode to provide the reset. Another method would be to lower the duty cycle so that the transformer has sufficient time to dissipate its magnetic energy. When the switch opens, the transformer's primary will automatically generate a reverse voltage, but if not enough time is allocated before the next pulse comes in, it cannot fully discharge, and will eventually saturate - which you have witnessed first hand.

 

[ikalogic]

I've worked with some mofsets that needed -ve volatge to force them to switch OFF. if you have a pulses signal (0 - 5v) you can easily transform it into (10 , -10v) using a cheap 741 opamp. this is what i did, i was using the mosfet to drive a highly inductive load, and it wouldn't switch off, and that was the solution.

 

[Hayato]

I believe your magnetic core is reaching saturation because of asymmetric pulsing (only positive going pulses). And that's why it seems the switching device is always on.

To fix this, you must apply symmetric postive and negative going pulses to 'reset' the magnetic core of the transformer. Many times, you can also use a free wheeling diode to provide the reset. Another method would be to lower the duty cycle so that the transformer has sufficient time to dissipate its magnetic energy. When the switch opens, the transformer's primary will automatically generate a reverse voltage, but if not enough time is allocated before the next pulse comes in, it cannot fully discharge, and will eventually saturate - which you have witnessed first hand.

Hello,
I don't think that the core saturation is the problem, because it works fine with power transistors like 2SC4106.

But I'll try to add the diode.

Another thing that is driving me crazy is that I've paralleled 2 MOSFETS and they are heating unequally, when I say unequally is:
One is cold another is very very hot to the touch.


(CORRECTED THE SCHEMACTIC)

Another thing I've noted is that one of the MOSFETS is heating w/o signal applied.

I don't know what it is, because it's the second time it happens, I don't think I'm so unlucky to get 2 pairs of mosfets (each pair in a different radioshack) with 1 deffective mosfet on each pair.

 

[Roff]

Has anyone pointed out that collector and emitter need to be swapped on the BC327? Having said that, if you are generating the pulses with a 555, it will probably provide higher drive current than you will get from push-pull emitter followers, and you will get 1.4V more gate drive.
And what is the purpose of the zener?

 

[Hayato]

Has anyone pointed out that collector and emitter need to be swapped on the BC327? Having said that, if you are generating the pulses with a 555, it will probably provide higher drive current than you will get from push-pull emitter followers, and you will get 1.4V more gate drive.
And what is the purpose of the zener?

Hello,
What do you mean by swaped? Is this (for example):
BC337: EBC
BC327: CBE

I didn't do that because the datasheet, from fairchild, says that BC327 and 337 have the same leads configuration. (I thought that was strange, but it's the datasheet).

The purpose of the zener is to protect the gate from any voltage higher than 12V.

 

[Nigel Goodwin]

Hello,
What do you mean by swaped? Is this (for example):

No, on the circuit! - the two emitters need to connect together!.

 

[hjames]

Look at the class B push-pull amplifier and compare it to how you have your transistors hooked up. - Do you think the PNP will ever turn on?
https://en.wikipedia.org/wiki/Electronic_amplifier

If you say a the mosfets are heating up with "no signal", does that mean the gates are unconnected, or is the pulldown 1K resistor still connected?

 

[Hayato]

Ron, Nigel, James,

Sorry, I hooked the transistors right way, but when I drew the schematic I drew it wrong.

In the circuit I've built it is hooked right.

This is the correct schematic:


James,
When I say no signal is, that I'm not applying signal on Q5's base (BC337).

This is what happens, I apply some signal there, then I disconnect the Q5's base, the MOSFET keeps heating.

 

[Roff]

Your zener will never conduct unless your supply goes above about 13.4V. If it does, either Q3, or the zener, or both will overheat and possibly fail.
What about my other comment?

...if you are generating the pulses with a 555, it will probably provide higher drive current than you will get from push-pull emitter followers, and you will get 1.4V more gate drive.

Why do you need push-pull emitter followers? Are you using a CMOS 555? A bipolar 555 will source and sink over 200mA, IIRC.

 

[Hayato]

Your zener will never conduct unless your supply goes above about 13.4V. If it does, either Q3, or the zener, or both will overheat and possibly fail.
What about my other comment?

Why do you need push-pull emitter followers? Are you using a CMOS 555? A bipolar 555 will source and sink over 200mA, IIRC.

Ron, the main problem is that my mosfets are not turning off.

Well, the main purpose of the push-pull is to "pull down" the mosfet gate when no tension or low level is applied to Q5 base.
I think that a transistor conducting to the ground do a better job than a "pull down" resistor.

Well, the circuit is working "nice" now, look what I've done a few hours ago:
I connected the circuit in a tweeter with an ammeter, I've notice that it were flowing about 0.5A of current and Q2 (IRF740) were getting a little bit hot.
I removed it (letting Q1 only) and changed Q4 (PNP one).

The circuit worked fine, but when a spark, from the ignition coil HT terminal, leaked and hit the negative terminal...the remaining mosfet were destroyed.

Well, I have to discover what caused the Q2 to fail, with an ohmmeter I measured the resistance between all the terminals (G-S, G-D, D-S...), for my surprise it was about 50 ohms (!!!).
Q2 gate were leaking causing Q1 to conduct as well, causing malfunction in the whole circuit, and probably that was the cause of Q4 burning as well.

But I have no idea what caused Q2 to fail, maybe the it was first affected by the coil back-emf (this is my only guess).

Later I'll try to add some zener or varistor between the drains and sources to protect the mosts against back-emf.

 

[Hero999]

The IRF740 is rated for 400V so I don't think back EMF is your problem. I meant to replace the resistor with the ferrite bead.

However, the last circuit looks quite good to me.