High Current MOSEFT Circuit Resistors

[Jaguarjoe]

Ballast resistors were prevalent during the days of points ignitions. The coil was designed to work at a lower voltage so it used a ballast resistor to limit current when the engine was running with at a full 12 volts. When the engine is cranking the 12 volt system could duck to as low as 6 volts. The ballast would then be bypassed to allow the full voltage to the coil even though that "full" voltage may only be 6 volts. The ignition switch did all of the work.

I like the no extra parts feature too.

Those parts that are limited to a 10 volt gate are meant to work at 5 volts. A voltage divider at the output of the driver would negate the advantage of using a driver. Putting the divider at the supply pin of the driver would be wasteful. Easiest way #1 is to use a 6 or 7.5 or 9 volt 3 terminal regulator. Easiest way #2 is to put 4 silicon diodes in series with the supply pin to the driver. Each diode drops about 0.7 volts. 4 would be a ~2.8 volt drop. They need to be at least 2 amp diodes. One 3.3 volt, 2+ amp zener would do the same thing with one part.

That 3040 part you found would work well.

I don't know if coil on plug (cop) or coil near plug (cnp) would help. Those coils are much smaller than a traditional coil so they don't pack the same wallop. COP/CNP was invented to eliminate the distributor and its drive mechanism.

A mosfet gate and an IGBT gate are identical.

I'm sure that Mouser, Newark and Digikey all sell mica insulators. Make sure you use thermal compound on every surface that conducts heat. Beside mica insulators you can buy silicone rubber insulators filled with some sort of heat conducting material. You won't need thermal compound then.

 

[Silntknight]

That's what I read also, so I should be fine if I take a coil from a newer car. Anyway, I have a 5V regulator on hand that I would love to use. Doesn't a regulator take some time to "regulate"? I guess I'm asking if it adds to switching time. I'd rather not order more parts, so I might go with several diodes (or a Zener if RadioShack has one) or the 5V regulator if that will work. I should probably not use capacitors to stabilize the voltage then...

DigiKey doesn't stock any of the ignition driver parts I've found, or even that you've found. I'll check Mouser and Newark.

 

[Silntknight]

IRGB14C40LPBF is in stock at Newark, so I might just get it. It clamps at 400V, so it's very close to the same. Also, the gate-emitter breakdown voltage is 12V, but the gate threshold is, at max, 2.2V, so I'll just use my 5V reg to handle that. Sound good?

 

[Jaguarjoe]

The IRGB part is great! A 5 volt regulator will be fine powering the gate driver. I don't recall how much current a 7805 can put out but it might need a heat sink.
Don't worry about 7805 speed, if you have enough capacitance at its output, the transient response it will be fine.
Get a coil from any mid 80's GM car or a Ford.

 

[Silntknight]

Mine can handle 1A easily, but a heatsink would be good. Because the tabs aren't electrically active on either, they will just go to the same heatsink (the big CPU one I have). Also, you mentioned capacitance. Should I have capacitors in my circuit or should I just go with the reg's capacitance? I was thinking that using capacitors would lead to the MOSFET being on for longer than I intended it to be because the cap's would drain into the MOSFET. The caps are used to regulate a continuous signal. I might actually scope this one to see the wave form with caps and without them. It could be really interesting.
I'll try to find one of those cars!

 

[Jaguarjoe]

Millions of cars to pluck a coil from.

Use Q = IT to find how much current you need from the 7805. Q is gate charge in nano coulombs (nc) listed on the mosfet spec sheet. I is required gate current that is also required driver current as well as required 7805 current. T = time desired to turn the mosfet on, your choice. You can juggle around and see how quick the turn on time will be with 1 amp flowing from the 7805. The power dissipated by the 7805 is P = EI = (Vin - 5v) * Iout. The key to minimizing Pd is to keep Vin as low as practical, the 7805 needs 2 volts to run, so Vin must be >7v. If you need more than 1 amp you can 1.5 amp 7805's or 3 amp LM350's.

Caps are reservoirs, they're good for the 7805 because they can supply current quickly, like when the gate driver snaps on- it'll need a lot of current NOW. You'll need many microfarads like 470 or so. The car battery is almost like that except it always has lots of current NOW! IIRC, the 7805 needs small caps around it- like 100nF or so to tame parasitic oscilations.

 

[Silntknight]

First, have you seen a pinout for this part? I'm guessing it is the same as every other TO-220 package, but I'd rather not guess. Anyway, the gate charge is 27nC. I'm just going to guess that the time interval is in the same unit scale (nano) as Coulombs is. That really doesn't make sense, but up to 270us is acceptable because I can easily compensate for that. More than a us and I might start to wonder if higher current wouldn't be a bad idea. The gate only needs an peak 10mA of current, with a hold of 1mA. I think I'll use some diodes to drop the current to 7V. Two or three should be fine for that, right? Also, shouldn't I^2 be somewhere in power dissipation? I know it was for MOSFETS.

So do you mean caps on the high voltage side or the low voltage side or both? Could I use many nanofarad caps because (AFAIK) lower capacity has a lower time constant? Say, 5-10 100nF caps? Of course, now that I re-read your post, I see that you said I should use 100nF caps. Wouldn't that keep the gate on longer than without caps? Also, I am still using the IGBT with a gate driver, so it has 1.5A (1A because of the regulator) available.

Random tangent, but why do inductors try to maintain current flow? Thinking about Ohm's Law, it makes sense that there is a voltage rise, but why does it try to maintain current levels?

 

[Jaguarjoe]

Q is in coulombs. 1nC is 10^-9 coulombs
I is in amps
T is in seconds. 1uS is 10^-6 seconds

Caps around the voltage regulator are not related to gate rise times. The object of the VR caps is to provide a clean stable voltage source.

Search for an LM7805 datasheet and it will show the pinout. It will also show the small caps around the regulator. These caps should be as close to the VR as reasonable.

Don't worry about getting to 7v until you know how much current you need. You may be OK at 12v.

P =EI = (I^2)R = (E^2)/R

When an inductor's windings charges with current going into it, it creates a magnetic field. When the current is rapidly shut off, the magnetic field collapses inducing a current back into the inductor's windings that is equal in amperage but opposite in polarity. The voltage across the inductor will rise as high as possible trying to maintain that amperage flowing. That's why big inductors like ignition coils need 400v igbts and zener diodes and little inductors like fuel injectors only need 47 volt zeners.

 

[Silntknight]

I meant pinout for the IRGB. I know the 7805. As for the regulator, if creating a stable voltage is the only concern, then I already have everything I need. I needed to make a 5V regulator circuit to power my sensors. I used 100uF, 10uF, and .1uF capacitors. Anyway, testing the circuit on a scope should help. Thanks.

 

[Jaguarjoe]

Look at page 11/12 of the data sheet for the pin #'s on the part and the lead assignments for those #'s:

https://www.electro-tech-online.com/custompdfs/2010/12/irgs14C40l.pdf

Gate = gate
Source = emitter
Drain = Collector

The tab is "hot"

 

[Silntknight]

Hmmm, just didn't see that the pin assignment was there too. It's a good thing I haven't placed my order yet because I'm also going to add some mica washers. Thanks!

 

[Jaguarjoe]

Don't forget the thermal heatsink compound. FWIW- its not the same as Silicone dielectric grease as used on spark plug boots.

 

[Silntknight]

I've still got plenty from when I built my computer. Anyway, back to the 5V reg, what capacitors do you recommend and where? Just a bunch (4) of low capacity (100nF) caps on both the input and output, two on each side?

 

[Jaguarjoe]

1- 100nF across the input, 1- 100nF across the output, both as close as practical to the chip. 1- 470uF across the output.

 

[Silntknight]

Ok, looks like I'll be taking another trip to DigiKey's page. While I'm ordering that, I might as well get parts for another project. Can I use two MOSFETS in parallel, but reverse oriented with rectifier diodes to function as an AC switch? I need to handle a large amount of current at 110VAC, and I'm pretty sure that setup would work. I'm thinking about just sticking with a logic level relay, but the MOSFET idea is just for information.

 

[Jaguarjoe]

Mosfets have "body diodes" that are a byproduct of their design. The diodes are opposite in polarity to the mosfet and are connected across the source and drain so they will always conduct heavily in one direction or cycle of the AC current. Two reversed in parallel will always conduct heavily in both directions. Not good.
A BJT or igbt are not candidates either.
A "triac" will work well though. They are made for switching full wave AC. An SCR (silicon controlled rectifier) switches half wave AC quite nicely. Crydom makes solid state switches, not too expensive and all the work (or fun?) of blowing up high current devices with relatively high voltages has already been done by them.
If you roll your own, use a triac driver like these: https://www.electro-tech-online.com/custompdfs/2010/12/FO2FFOD410.pdf they're about $4 each from Digikey. These are LED input to provide isolation and easy interface. The output is "zero crossing" to eliminate noise. Define the load, pick a triac, then pick the driver, Digikey will have everything.
What are you switching?

 

[Silntknight]

This is just for controlling a heating element. I'm not sure what its power rating is, but it runs on a normal wall outlet, so it can't be higher than 10A (1200W). I knew I could use a triac, but I wasn't sure how easy/hard that would be compared to a relay. If it really has an integrated opto, then that's probably the best solution.
Basically, I would run the triac the way I would run the relay, right? I'd just keep one of the wires connected normally and run the other one through this?

What I'll probably end up doing is just ordering all the parts from SparkFun. Maybe I'll wait until Free Day, but I'll probably be too impatient. I need to get a PICAXE, and one of their relays. They have a 30A model that I can use pretty easily.

 

[Jaguarjoe]

Don't get your hopes up for that free day. I tried last year at less than 1 minute after it kicked off and never got through for the rest of the sale. This year it'll prolly be worse now that more people know about it.

Yep, just wire it in series, just like a switch.

If its just on/off control, that Fairchild part will give you isolation from the mains power and also an LED that can be turned on by a uP or a switch & battery.

Here's an opto trigger for 65 cents and a 15 amp 600v triac for $1.35:

https://www.allelectronics.com/index.php?page=search&search_query=triac&x=22&y=7

 

[Silntknight]

Yea, I ordered the parts just yesterday. I decided not to wait.

 

[Silntknight]

Is the 470uF really necessary? I feel like it would be the largest component on the board and fairly hard to squeeze in.