[General] MOSFET switching instead of Darlington

[Instructor]

Good Day. I need to switch from a Darlington for a car ignition coil to MOSFET.
The Darlington needs a huge Heat sink while the MOSFET trial did not.
I just want to confirm the circuit and checking if there might be any error with the MOSFET. Can you please help?
Attached are both circuits, the old and the new.
Thank you.

 

[dick_freebird]

I don't know how big that FET is, but expect that 6.8K
pulldown will leave you a very leisurely turnoff and a weak
spark. Your NPN buffer is OK (maybe) for turnon strength
but has nothing for turnoff. I'd play with values there
(if you don't just want a proper, 59-cent power MOSFET
driver) and see what maximizes spark length, total energy,
whatever the care-about is. And check stabilized FET and
coil temperatures as well, to see if anything reliability-
impacting is there.

There is an unlabeled resistor on the emitter of that NPN
which forms a gate voltage divider. You want the gate driven
to spec-recommended levels (10V). With the follower drop
and a 12V supply you are already getting close, not much
room for any divider drop.

Now I have to wonder, why the CD4017 at all. Other than
trying to get a narrow duty cycle? You could do that by
timing values on the 555 and use a common-emitter or
FET-driver inversion to flip the usual >=50% to <=50%
(check the timing-component info for figuring duty along
with fosc).

 

[FvM]

Driving the MOSFET from 555 output through 100 ohm would surely give better performance.

 

[asdf44]

Why a 400V mosfet and 220V zener in a 12V application?

 

[Instructor]

Thank you for your quick reply.
First, I am using 4017 to get the 50/50 cycle for the switching.
Concerning the 6.8k pull down, I got this result from a search on the web. More than 4 to 5 samples of driving IRF740 or 640 used this component. Another used a 12V Zener parallel to a 2.2k resistor.
A third system used two parallel transistors, one PNP connected to negative, and one NPN connected to positive for switching the MOSFET.
I am truly now in the trial phase and requesting opinions concerning the higher output current value for the system.
I also used parallel MOSFETs, the result was 10 times additional current but very narrow spark.
This system gave me a spark of 2.5 cm width.
The empty resistor is according to my calculation should not exceed 33 ohms.
I appreciate every opinion you offer and I you advise another system that improves the output, I will me much pleased to try it.
Another important thing, this system has to run for 40000 hours non interrupted.

 

[Instructor]

Dear Sir, an ignition coil provides a return high voltage to the system.
In order to protect the system, I used 1N5408, but I read that a Zener is more effective for protection, that is why I decided to transfer to more protection because as I mentionned above, I need 40000 hours of continuous work from the circuit. I am also using a 7812 for protecting the 555 and the 4017.
And thank you for your reply.

 

[betwixt]

Instructor: you need to stop thinking about voltage and start thinking about time. The voltage at the output is largely proportional to how fast you switch the current in the primary of the ignition coil. A MOSFET is a good choice because it has low resistance when turned on properly (= more voltage for the coil and less heat) but it also has a large input capacitance at it's gate pin. In order to quickly turn it on and off again you have to raise and lower the gate voltage quickly and that involves charging and discharging that gate capacitance. For best results you should ensure the current available at the gate is as high as possible, even though the static current needed by a MOSFET is very small. Your attempts with parallel MOSFETS prove this point, you did give it more ability to draw current through the coil but the extra capacitance made the rise and fall of that current much slower, hence the weaker spark.

You should consider either a dedicated gate driver IC or a pair of small BJT in simple push-pull configuration to drive the gate, it will give you far better results.

The 1N5408 is OK at the relatively low switching speed although a faster switching device might be better. A Zener can be used to protect the MOSFET if necessary but bear in mind it will still conduct like a normal diode in forward direction.

asdf44: the voltage at the drain of the MOSFET is not 12V. It will switch from nearly 0V to possibly several hundred volts due to the collapse of the magnetic field in the ignition coil.


Brian.

 

[Instructor]

More trials. I tried directly through 1k ohm resistor straight from leg 12 from 4017, and the result was similar to the old system, with much less heat dissipation. Could I improve the gain on the coil for at least twice, may through different MOSFET or a parallel or series MOSFET circuit?
Thank you all for your help

 

[betwixt]

Read my previous post. The 1K resistor is limiting the rate the gate capacitance can be charged and discharged. If you do not want to build a proper driver circuit, reduce the resistor to 10 Ohms.

Brian.

 

[Instructor]

According to your instructions, I believe I have one of two choices:
Either using directly from 555 through a small resistor like 100 ohm, or using a dedicated gate driver, IC or transistors. If transistors work fine, it would be an easier solution. But if ICs provide higher gain, I will then change the 4017 to TC4426 serie. And should change the diode 1N5408 to NTE588 or LT3434 fast high voltage switching diodes.
Please, I would appreciate all help concerning improving the circuit, including higher possible gain, even through parallel or serie MOSFET. Thank you all, again and again.
Fellows, I am a PhD in Motion graphics, and if anyone needs help concerning any media sector, I would willingly provide my help for free in return.

 

[FvM]

You are designing a driver for an existing ignition coil. If I understand right, the motivation to replace the darlington with a MOSFET is mainly reduction of on-state losses. The darlington circuit already limits the output voltage by the Vce breakdown voltage of both transistors. In this relation, the MOSFET circuit can easily achieve better results, if it avoids to make switch-off worse than the already super-slow triple darlington circuit.

On the other hand, I guess that there's no benefit in making switching considerably faster than a small fraction of the ignition coil self resonance frequency. In so far, don't over-engineer the design.

 

[Instructor]

Dear FvM, I am only trying to get the best results, regarding bad results from the previous circuit based on overheating during the hot summer periods.
I need perfect system, with minimum errors and maximum output. Cost is not an issue here because I am payed well for this perfection, so must do it. I appreciate your opinion, but also need more detailed technical information for improvement. Please provide me with as much as you can so I can decide the best for the system functionality.
Thank you for your opinion and waiting for more please.

 

[FvM]

My suggestion is to try with the basic 555 + series resistor (may be 50 or 100 ohm) + IRF740 setup.

 

[betwixt]

The details on making it more efficient are here: www.ti.com/lit/an/slua341/slua341.pdf It relates to power supply switching but the calculations are the same.

The BJT driver is detailed on the second schematic on this page: https://electronics.stackexchange.com/questions/96776/555-to-drive-a-mosfet
Note the comment under the schematic! You can adjust the 555 pulse width and speed so the 4017 isn't needed at all.


Brian.

 

[dick_freebird]

The 555 always has a greater than or equal to 50% output-high
duty cycle. In general automotive ignition uses a small dwell and
a long period between - no more current in the primary than is
needed to fully energize the core (short of saturation, barely).
This volt-seconds number, which is intrinsic to the transformer,
should determine your "on" time. Anything more is only wasted
(at best).

Since the parallel MOSFETs are showing you Cdg and/or Coss
sensitivity, I might suggest you look at RF LDMOS (if you can
find any at the kickback peak voltage rating) or maybe something
cute like a GeneSiC SJT (a silicon carbide quasi-BJT, real fast,
purportedly the lowest conduction losses in the wide bandgap
power device family, reliable past 300C including repetitive
avalanche - not that their 600V+ devices are likely to see
that, but you might be able to forego the power zener and
then not cut the crest off the flyback) simply substituted for
your present Darlington lower device.

I might also suggest that you change from a time and voltage
control mode (not that there looks to be any control per se)
to an emitter-current-terminated dwell - wait for the current
(across a low value sense resistor) to rise to a threshold that
suits the primary inductor saturation limits, snap it off at that
point, outwait the secondary spark extinction, and repeat -
a self-relaxation-oscillator, which might be as simple as a
sense resistor, a couple of common-emitter gain and timing
stages leading back to your Darlington base. I've sone this
as simply as a wirewound resistor and a 12V NC relay (but
with no regard to longevity - I expect the chatter arcing
would fry it eventually although a proper snubber might
extend that). At any rate the control ICs seem to be more
complexity than is warranted.

 

[Instructor]

Dear fellows, I tried the system directly through 100 ohm resistor from 555 and the result was an output current of 64 watts across the coil pins.
I tried trough CD4017, and the result was 10 watts.
I raised the resistance to 1k with 555, but still the same. I reduced the frequency to minimum possible arc length, 1 cm, and the current was reduced to 50 watts only.
I need an output current of 18 watts, so how can I reduce it from 555 or raise a bit the 4017 voltage.
I though of using a totem pole on 4017, but I need to be able to control the output, not get the results of 555. Please I admire your opinions.
Thank you so much.

 

[Instructor]

Dear Brian, I already checked that page before getting into edaboard forum. As I just mentionned, I need a compromise, that even frequency adjustment isn't sufficient for this case. Reducing frequency reduces arc length. Through the Darlington, I had the exact spark length. With direct 555 to MOSFET, the spark jumped out of the coil edges to hit the metallic chassis. Reducing the frequency reduced the spark length, and barely the current. With 4017, the current is reduced to 1/2 my request with the exact spark length I need. I am looking now for something to control the MOSFET gate voltage, not the frequency.
Thanks for the links.

 

[dick_freebird]

Back up a bit. First, you seem not to get that coil on and
off time, transistor determined turnoff / risetime at the
primary, are your real "knobs". When you complain that
jacking frequency around isn't getting you what you want.
that's because it has nothing to do directly with the
outcome, only by also modifying the coil dwell time as
part of modifying the whole cycle.

Learn what the coil can take before it saturates the
core. Learn to make one pulse that is correct. Then
figure out how fast you can apply those pulses before
something goes into float. 1, 2, 3.

 

[Instructor]

Dear Sir, when I return from work tomorrow afternoon , I will start by first adjusting the frequency of the 555. If no luck, I will try to adjust the current across the MOSFET gate.
If I doesn't provide with what I need , I am going to use a totem pole on 4017 output instead, and amplify the signal through supply resistors. If that doesn't work out, it would mean I have still a lot of additional researches to do, even though I downloaded and read all what is available on the web concerning the ignition coil MOSFET circuits.
Thank you

 

[betwixt]

I agree with Dick_freebird.

You are missing the point about the 4017. All it does is divide the frequency from the NE555 by two and it produces a near square wave output. The drawback to the 4017 is it has very limited output current at pin 12 and you really need as much as possible to drive the MOSFET. Halving the NE555 frequency, by changing the capacitor value will give you the same frequency, same square wave (check the data sheet for the wiring configuration) but about 20 times as much drive current - and you use fewer components!

A square wave probably isn't optimal anyway, if you keep the MOSFET or Darlington conducting for longer than the dwell time it will simply waste power as heat and reduce the circuits efficiency. As you do not need frequency control or duty cycle control, you could try making them both adjustable. You should be able to find the combination that gives most efficient spark production. In a similar application I was able to do produce continuous sparks with a MOSFET driving the coil and it barely got warm even without a heat sink.

Brian.