Hi,
I purchased a couple of the ubiquitous "electric arc lighter" devices off of Amazon because I was curious how the designers were able to pack a 3.7v to many kV step-up circuit in a package the size of a ballpoint pen.
I tested out the device and indeed it is able to make a ~4 mm sustained arc.
After opening up the device for inspection I found that the parts count is low and the circuit seemingly simple. Many of the components appear to be for the USB charging circuit.
An unlabeled 8 pin chip appears to be the MCU. I measured a square wave coming off of pin 8 at ~17 kHz with a duty cycle of ~72%. This goes through a 10 ohm resistor to pin 1 of what almost certainly is a MOSFET (marked BN5T10AG 23335G). Pin 2 of the MOSFET goes to the (- marked) terminal of the transformer. A 10k ohm resistor connects MOSFET pin 2 to MOSFET pin 3 and battery ground. Battery positive passes through to the (+ marked) terminal of the transformer.
This appears to be a simple flyback circuit.

Next, I wanted to see if I could make my own driver circuit and bypass the marked-but-still-unknown-quality components. I decided to mimic the circuit as close as possible. I setup an STM32 to produce a square wave at the same frequency and duty cycle as the mystery MCU. I used a IRF832 for my MOSFET and included the 10 ohm and 10k ohm resistors in the same configuration as the original circuit. I powered the circuit with the same 3.7 li-ion battery.

My homemade driver does not cause any arc on the transformer output. I have been through the wiring and checked the signals on the scope but I am not sure what is going on. Any ideas on what I might be missing?

 

how exactly did you connect it?
STM32 is an 3.3V controller. the highest output voltage you can hope for is a little bit less than 3.3V.
the transistor you use needs at least 4V to turn on. see the problem?

 

how exactly did you connect it?
STM32 is an 3.3V controller. the highest output voltage you can hope for is a little bit less than 3.3V.
the transistor you use needs at least 4V to turn on. see the problem?

Thanks for the reply!
I am using IRF832 which according to the datasheet has a V_gs(th) of 2-4 volts, so I think the 3.3v should be perfect. To test the circuit, I have also substituted the transformer for a resistor+LED and observed different brightness levels when I change the duty cycle of the input signal.

I scoped the various signals; here is a comparison:

• Original driver circuit square wave signal:
  • 

• My driver circuit square wave signal from STM32:
  1. 

• Original driver circuit signal after MOSFET with transformer disconnected (scope probe - on MOSFET drain, scope probe + on battery +):
  • 

1. My driver circuit signal after MOSFET with transformer disconnected (scope probe - on MOSFET drain, scope probe + on battery +):
   1. 

The timing on the "after MOSFET" scope captures is the same but the signal profile is different. Could this be causing my issue? Is there any different MOSFET characteristics that might only be evident when the transformer is connected? I am afraid to connect the scope to the MOSFET output when the transformer is connected because I would expect some big voltage spikes.

 

exactly 2-4V. it is a range... so your transistor may be one that needs 4V. and that is the minimum needed to barely turn it on. You do not want that, in this case transistor need to work as a switch so you need to drive it hard, not tickle it.
post your circuit...

 

(scope probe - on MOSFET drain, scope probe + on battery +)

What kind of probe you have and why are you connecting it like that?

normal scope probe have a tip used for probing and the alligator clip that should be connected to reference point. and that should be the negative terminal of the battery.

 

exactly 2-4V. it is a range... so your transistor may be one that needs 4V. and that is the minimum needed to barely turn it on. You do not want that, in this case transistor need to work as a switch so you need to drive it hard, not tickle it.
post your circuit...

Oh, I guess I misinterpreted 'max' as absolute maximum to not exceed... but I now see that the absolute max is 20v. I will give this a try.

What kind of probe you have and why are you connecting it like that?

normal scope probe have a tip used for probing and the alligator clip that should be connected to reference point. and that should be the negative terminal of the battery.

Well I'm using a normal scope probe. I was thinking that since the gate is switching the low-side of the circuit that I should clip there.

 

I've switched over to an Arduino with 5v GPIO to generate the square wave signal to exceed the max V_gs(th) of the MOSFET. Still no luck getting the transformer to throw an arc.

Another thing I tried was connecting my Arduino pulse to the MOSFET circuitry on the original as-purchased device. This did work and produced a nice crisp arc. Therefore, my issue is almost certainly something wrong with my MOSFET or MOSFET circuit arrangement.

I've tried a few different MOSFET's and the latest scope images are captured with IRF740.

Here is my up to date circuit diagram.


I've taken new scope images and I can't see many differences in the signals.

1. Original driver circuit square wave signal
   1. 

1. My driver circuit square wave signal from Arduino:
   1. 

1. Original driver circuit signal after MOSFET with transformer disconnected
   • 

1. My driver circuit signal after MOSFET with transformer disconnected:
   1. I notice on this one that there is that big negative spike right at t=0. I'm not sure what this is or if it is significant, but it is a difference between the original and my signals.
   2. 

 

IRF740 datasheet
https://www.vishay.com/docs/91054/91054.pdf

Gate-source threshold voltage VGS(th) V DS = VGS , I D = 250 μA 2.0 - 4.0 V

So somewhere between 2 to 4 volts the mosfet will conduct a minimum of 250 uA. You likely need a lot more current.
This "threshold voltage" is where the mosfet just barely starts to turn on.

 

IRF740 datasheet
https://www.vishay.com/docs/91054/91054.pdf

Gate-source threshold voltage VGS(th) V DS = VGS , I D = 250 μA 2.0 - 4.0 V

So somewhere between 2 to 4 volts the mosfet will conduct a minimum of 250 uA. You likely need a lot more current.
This "threshold voltage" is where the mosfet just barely starts to turn on.

I don't follow. The Arduino GPIO is 5v and capable of sourcing 20mA continuous current. Shouldn't this be well within the "on" region of the mosfet?

 

The Arduino GPIO is 5v and capable of sourcing 20mA continuous current. Shouldn't this be well within the "on" region of the mosfet?

No, Vgs of 10 volts is into the "well on" for that mosfet. See the datasheet above for the IRF740 page 3 Fig 3 Typical Transfer Characteristics

Likely the transformer needs 100 mA or more, possibly 1A to generate a good spark. Acording to this the IRF740 at 25 C with
5 volts on the gate barely gets to 100 mA (and this is typical not minimum, your IRF740 might be a bit worse).

On page 3 "Drain-source on-state resistance" is stated at Vgs of 10 volts -- that's a long way from 5 volts.

As an example, compare it to an AO3402 mosfet datasheet:

https://aosmd.com/sites/default/files/res/datasheets/AO3402.pdf

Ignoring that this is surface mount, it's specified to work at Vgs 2.5 volts (with < 85 mOhm resistance). On page 2
this is specified at 2A of drain current -- that's a big difference compared to the IRF740.

 

I see. Thanks for the clarification. I've done some more reading and I now understand the point you are making. I either need to increase my square wave signal voltage to >10v or find a logic level MOSFET with an R_DS(on) rating for V_GS of 5v.
Logic level MOSFETs seem to be common but not at the high V_DS ratings that I expect to need to protect the MOSFET from inductive spikes from the transformer. My reading suggests that the usual snubber diodes negatively impact arc quality so you are better off getting a MOSFET with a high V_DS.
Obviously the mystery MOSFET in the original circuit is able to be switched with V_GS of 3.7v and stay alive without an obvious snubber circuit. Any suggestions on a substitute MOSFET?

 

Probably a big portion of the problem is way too much resistance in that power loop: BATTERY, COIL PRIMARY, and the TRANSISTOR. Most likely the original had pulses of quite high current with a very sharp rise and fall, and probably the copy has many inches of wire while the original had only a very short run of wire.
More than almost everything else, High current and fast rise times are needed to produce high voltage spikes.
My guess is that in the spark lighter the rate of change is faster. possibly a lot faster. AND, probably the current loop resistance is different.

 

Update!
I got a hold of an IRLZ44N logic level MOSFET and stuck it in the circuit... and it works! Thanks for everyone's patience in explaining these elementary MOSFET principles.
The notable IRLZ44N specs are:

1. Rds_on: 0.025Ω when Vgs=5v, Id=25A
2. Id_max = 47A
3. Vdss = 55v

It will be interesting to see how this MOSFET holds up given its relatively low voltage rating. From my reading I would expect transient voltage spikes to kill it. I've found a few different options with higher Vdss ratings I could order but I'm not super sure how high of a rating I should aim for... how big should I expect these transient spikes to be?

 

Those voltage spikes are proportional to the change in current per time. Not nearly as fast as a mechanical contact make and break.

 

Update!
I got a hold of an IRLZ44N logic level MOSFET and stuck it in the circuit... and it works! Thanks for everyone's patience in explaining these elementary MOSFET principles.
The notable IRLZ44N specs are:

1. Rds_on: 0.025Ω when Vgs=5v, Id=25A
2. Id_max = 47A
3. Vdss = 55v

It will be interesting to see how this MOSFET holds up given its relatively low voltage rating. From my reading I would expect transient voltage spikes to kill it. I've found a few different options with higher Vdss ratings I could order but I'm not super sure how high of a rating I should aim for... how big should I expect these transient spikes to be?

did you change anything else in the circuit? im attempting to re-create this

 

did you change anything else in the circuit? im attempting to re-create this

Nope, switching to a logic-level MOSFET with the appropriate Vgs did the trick!