#12 - Superb info.

What type/material of capacitor or is this important? Is this to be a polarized cap or not?

The highest rated aluminum cap I see is at 630V. 450V is the only ones I see stocked.

The ceramics go way up in the rated voltage but they seem so tiny. Could I use a small surface mount for our task?

The micas also have a high voltage rating but I think they are polarized. As I mentioned above, I am not sure if we can used a polarized cap.

The rate I fire at is limited by the valves and I fire at most once every 1/3 of a second, simulating an automatic small arms weapon.

 

I'm thinking a poly film capacitor like they use in cars.

http://www.mouser.com/Passive-Compo...Z1yzvvz1Z1yzbq84Z1yz61yoZ1z0wxey&Ns=Pricing|0

Those are rated at 1600V but you can probably get away with 1200V or 1000V because the transfer of energy will not be perfect and I hope you will not run the coil up to 8 amps.

 

.

When you say I need a path to ground for the cap, we are talking about parallel across the cap/coil correct? In this case, what about when my FET is off and there is no ground?


http://www.freeimagehosting.net/l1m7p


In circuit #1 - When the fet is off, there is no path to ground for the resistor
In circuit #2 - There is always current flowing thru the coil


Could you help me understand how to achieve the path to ground for the cap? Thanks!

 

And don't forget, I never said anything about your shorting out transistor. I'm just telling you how to control the energy dump.

Like this:

 

You could eliminate the spark part of your problem pretty easy. Use a GM HEI module and a switch like shown in this link; http://home.comcast.net/~loudgpz/GPZweb/Ignition/GPZheiModForPoints.html This would eliminate all your capacitor and spike problems, since the module does all of that internally.

But anything you do is eventually going to need to address overheating the coil from the power being on all the time. In an ignition system this is called dwell time.

A logic circuit that would trigger the HEI module after the gas valve closes would be the next step. That way all of the firing steps are tied together.

 

I've certainly gotten alot of info to ponder over the last 24 hours.

If I have say 12V/1500ohm as suggested, I have 8mA of current

I have some concerns about keeping the current flowing at all times. First, my system runs off a battery, I guess for the 33AH battery I use 8mA is somewhat minor but I do have to consider the rest of the projects constant draw (control board, radio). Second, Is 8mA enough to create the field, so that when power is removed it will spark inadvertently?

I dont think heat will be an issue for a constant 8mA as the coil is pretty beefy plus it is attached to my metal frame for sinking heat. Although, after you make enough explosions, the frame gets pretty warm.

Is there any major reasons to get away from the diode and use the RC snubber? I suppose if we are talking 1KV+ coming out of the primary, my diode is SEVERELY underated. I spec'd it at 250V blocking volts @ 40A because I assumed the primary voltage would never exceed 250V. I have order a ton of parts today to play with. I purchased several fast diodes with ratings >1KV and >8A, perhaps this will solve the failure. If at all possible I would prefer for the solution to be able to use my existing board since I dont want to have to pay setup fees, etc again to fab a new board. The only thing that makes me wonder here is that, I rarely seem to have to replace the diode, only the FET. I guess that if it cant get enough of the energy to ground quick enough, it overtakes the FET's measly internal diode and bakes it.

I did also order some high voltage caps as suggested and some resistors to try as well.

In the meantime, I will add a diode inline with spark FET1 gate so if the gate, upon failure, is outputting a voltage, it will not activate the valve FET2.

Of course, now that I am wanting a failure for testing.... It isn't failing. Typical.

 

There are other ways to quench the kick that lands in the capacitor. You might try a 50 volt MOV and a 100 volt capacitor and see if it still sparks when the first inductive kick is merely routed to ground.

Of course, all this time I am assuming you are using a mosfet that can survive 1000 volts...

 

Of course, all this time I am assuming you are using a mosfet that can survive 1000 volts...

Ummm, perhaps I have overlooked that. After all the discussion about the kick back, diodes and caps, I didn't ever consider my FET is rated at 100V 12A.

I am assuming this has to be changed. lol

RFP12N10L

 

I am going to order some
APT4M120K

tommmorow.

Rated at 1200V but only 5A Continuous. I will be pulsing so I should be good up to 15A.

 

No wonder you're shorting out mosfets. Dumping that much energy into "noplace" will knock holes in them right quick. Shortbus in post#25 is looking real good right now.

 

Shortbus in post#25 is looking real good right now.

My biggest concern is space with that solution. I see they only run about $25 so thats not a big deal. I have some stock at my local autozone so I may get one to play with.

I would however prefer to use the FET and DIODE (or RC snubber if need be) instead if any way possible. I have ordered a pretty decent amount of diferent components to play with that will be here early AM. Unfortunately I did not order any FETs. I suppose I will do that tommorow and have them by Monday.

Today I added my diode to the FET1 gate, and of course, I couldnt MAKE it fail. I also put the scope on the primary of the coil and I saw that most spikes were 50V, but occassionally there would be 150V to 500V and at one point, I may have saw almost 1KV. Damn, I never expected it to be that high.

I did expect the diode across the Source/Drain, being so fast, would absorb that spike and the FET would never experience it. And perhaps that is still the case but the diode itself is way underated. That problem will be solved tommorow. Am I misled by thinking this? I mean the time to activate for the diode is teeny tiny and should dump that energy before it affects the FET in theory? I have seen countless sample coil drivers and most have FET/BJTs that have way smaller ratings than mine so it confuses me.

I really appreciate all the time and expertise you guys have donated to my cause, especially #12. You are all appreciated.

 

Checking the resistor across the capacitor, to drain the cap to 10 volts before the next pulse:
R = -t/C Ln V2/V1
R = (-1/3 second)/.47e-6 X (Ln 990/1000)
R = 7.09 e5 x .01
R is equal to or less than 7.09k
That reduces the 8 milliamp drain.
12V /6.8k = 1.76ma

Assuming a .47uf capacitor, You can set the drainage to 31 volts, theoretically, but with 3 quick pulses, the 10 voltses will accumulate to 30 volts in 3 energy dumps, so I used 10 volts in the math.

The moral of the story? Do the math.

The RC time formulas are here:http://forum.allaboutcircuits.com/blog.php?b=485

 

Use an IGBT designed for driving Ign coils, all you need is whats in the circuit. But it needs to be switched with a pulse, as it will over heat if left on too long. It will also require a heatsink. This is what i use in an Ign coil tester ive built.

 

Use an IGBT designed for driving Ign coils, all you need is whats in the circuit. But it needs to be switched with a pulse, as it will over heat if left on too long. It will also require a heatsink. This is what i use in an Ign coil tester ive built.

Wow so no external components at all? Interesting, any idea why the voltage and current ratings are lower than my fet yet this one works?

 

In answer to your question, these devices are designed with in built protection from the HV spikes generated by the coil in the primary windings. Here is another option using a Ford/Bosch IGN module that has built in current limiting. This is for times when an engine stops & the module is swtched on, it actualy turns off till it sees another pulse. It is driven by a Hall device but it can also be trigered by grounding the terminal ive marked as Trig. This module does need a heat sink & the heat sink & ground terminal need to be common conected. Here in Australia these distributers are in early fuel injected V8 F series trucks & also moreso on 6cyl EA to EL Falcon cars. I get these distributers for free from a scrap yard, I remove them before the vehicles are crushed.

 

Nobody has mentioned that you can limit the current through the coil by using a short "on" pulse.

You can use math to figure out the "dwell" time that will get the primary up to, say, 4 amps. That will cut the dump energy by a factor of 4. (energy = 1/2 L (I squared)).

From the definition of inductance, 1V/1H = 1A/1Second
IL/V = Time
4(.0069)/12 = .0023 seconds

Still, this is a very long time for a car engine. This coil is probably designed to work a lot faster than that, thus even less current than my first guess.

I'm thinking about: a trigger pull starts 1/2 of a 556 timer that stays on for .7 seconds. That circuit enables the other half of the 556 timer to deliver some 1 millisecond pulses with 332 milliseconds of "off" time between pulses. Then the first timer, being a "one-shot" configuration disables the second timer and refuses to start until the trigger is pulled again.

Do the math to see how much current will be flowing at 1 millisecond, then you will see that the dump energy will be well within the voltage range described in post #33

 

Yet another option a Mitsubishi J121 Ign module usualy found on a heat sink on the side of an Ign coil in older Mitsubishis & some Mazda utes. These need to be driven by pulses aswell or they will over heat.

 

I am hoping to avoid the ignition module route (space and rewiring concerns). I feel somewhat confident that my safety circuit, as far as the trigger signal is concerned, is working rather well. Trigger signal from the controller has never gotten stuck on since I implemented the 555 non-triggerable timer there.

I use 1M and 0.1uF to get a trigger on time of 0.110s. I have some limitations, as this also fires my gas valves so I cant make it too short. This 110mS then is my dwell time if I understand all of you correctly. That's forever longer than the time you calculated for 4A.

If I transpose your formula to I = (TV)/L, (0.110s * 12v) / 0.0069H -> I get I=191.3A. What am I doing wrong here?

I like your idea #12 about the 555 pair. My only concern, is that the last time I tried to put the 555 on my FET board it didnt work properly, because of what I am assuming is the EMI.

If I could get the 555's to operate properly I could take advantage of your suggestion and also have my timing paradox for the valves solved since I could have a separate length trigger signals, one longer for the valves that is the master for the spark.

The IGBT is interesting and I actually do have some BU941ZP 'coil driver' NPN darlington transistors. I never used them because I thought the ratings would never work, I guess I didnt consider the clamps, etc to make them suitable for the application. I think I will try one today.

Thanks All!

 

I know the standard wisdom is using 555 or multiple 555 circuits for this stuff. But in my on going education on this type of thing, there are other things available. And they still use the RC timing calculations that the 555 does.

The 74xx123 and 74xx423 are similar to the normal 555. but are retriggerable and have adjustable out put pulse capability. I used the xx in the part numbers because they are available in a few of the logic families, LS, HC, ect.

 

The 74xx123 and 74xx423 are similar to the normal 555. but are retriggerable and have adjustable out put pulse capability. I used the xx in the part numbers because they are available in a few of the logic families, LS, HC, ect.

Absolutely, when I mentioned my 555 circuit earlier, it is actually a '123

I am playing with one of the new diodes I got in this morning and one thing I notice is that my spark is waaay to hot. It is solid white. The plug stopped working and I pulled it out and noticed it had fused a thread from the electrode to the arm. Crazy. I knocked the thread that was shorting it and it's good again, but I still say that spark is running too hot.

I am thinking of bumping my 1.2ohm ballast resistor up to at least 3 ohms.