High Current MOSEFT Circuit Resistors

[Silntknight]

Ok. Anyway, I'll download Spice (provided it's free) and test various configurations. If I can live with 60V 30A MOSFETS for both ignition and solenoid/fuel injector circuits then I'll go with those because the lower resistance should mean less heat. In that case, I'll use 50V Zeners. Thanks everyone!

 

[Jaguarjoe]

You can't use a 50 volt zener and a 60 volt MOSFET for the ignition driver. The high voltage occurring at the secondary winding is generated by the back emf pulse of the primary. This pulse needs to be as high as practical. When loaded with a 20kv spark from a spark plug the back emf pulse is somewhere around 300 volts. With no spark plug it could be twice as high. Stick with the 500 volt MOSFET and the 400 volt zener. If the Rds-on of the MOSFET is too high try the MJ10012 or the MJ10016 Darlington. They have about 2 volts across C-E when on hard. If you use the 400 volt 10012 you zener ought to be 350 volts or so.

60 volt MOSFETS with 50 volt zeners are great for injectors and solenoids.

I don't think Spice is free. LT Spice is though. It was a gift from Linear Technologies.

 

[Jaguarjoe]

Huh! You are right. Damn new technology gets me every time.

 

[Silntknight]

Ok, I'll stick with higher voltage components for the ignition driver. I checked the MJ10012 and it said it was intended for ignition drivers. Would I have to do more than just swap it out? In other words, I have the MOSFET circuit drawn-up. Would I have to change that at all to accommodate an NPN transistor? Heat shouldn't be a problem. I've done a lot of metalworking, so I'll build some custom heatsinks for them. I'll use LTSpice to see what the times are like using a 60V NMOS and 500V NMOS with corresponding Zeners (47V and 200V). I remember you said that turn off would be faster with a higher voltage tolerance.

Not sure what you meant with your last comment...

 

[Silntknight]

How in the world do I use LTSpice. I need to simulate a 500V 5A NMOS, but the component isn't in the library and I can't seem to save changes to the MOSFET library to create another component. How did you do this!?!?

 

[Jaguarjoe]

200 volts is a tad low for ignition driver.
No, transistors and MOSFETs are not interchangeable. Transistors are current devices, MOSFETs are voltage devices. With a transistor, you feed a little current into the base then the collector will allow a larger current to flow. With a MOSFET, you apply a voltage to the gate then the MOSFET's drain to source channel opens up allowing current to flow.
All transistors have a gain (or beta or Hfe) which is Ic/Ib. This is about 10 for many transistors in saturation (on real hard). You want 5 amps of coil current so for an Ic of 5 and a beta of 10 you will need 500ma of base current. I don't know what's driving your MOSFET but I can almost guarantee it won't put out 600ma. You can use an IGBT too. It is a big power transistor with a MOSFET front end. It handles high current with a small voltage drop and it is easier to turn on like a MOSFET. This is a good example:

https://www.electro-tech-online.com/custompdfs/2010/11/NGD15N41CL-DPDF.pdf

My last comment in my last post was meant to be a reply to post #40. I hit reply instead of reply with quote.

 

[Silntknight]

Sorry, I meant 400V. Anyway, By interchangeable, I simply meant I can drop the transistor in the spot of the NMOS without changing the circuity. I know it is a current device. I think I'll stick with an NMOS just so that I don't have to worry about the current stuff, though it should be easy because I know that the pin is 3.3V and I need 500mA of current, so the resistor is 6.6 Ohms, so a 5 ohm resistor would work. That aside, do you know how I can simulate a 500V NMOS in LTSpice, I can't find a library component to match my components.
Closing note, I know about IGBTs also, but they are much more expensive and the are better for much higher current applications. I read a whole article on choosing the right one, and MOSFETS seemed to work better.

 

[Jaguarjoe]

though it should be easy because I know that the pin is 3.3V and I need 500mA of current, so the resistor is 6.6 Ohms, so a 5 ohm resistor would work.

I'm reading between the lines here, but is this "pin" on a 3.3 volt microprocessor? If it is, it won't put out 500ma. Maybe 20-25ma per pin max with a chip total of maybe 100ma max. This means that if you have ten outputs each with a 25ma load, you can only energize 4 at one time (4 * 25 = 100)
You will have an extremely hard time driving a MOSFET with 3.3 volts. About the best you can do is a 5 volt logic level MOSFET. Generally, they will have an "L" in their part number. Regular MOSFETs require around 10 volts or so. There are MOSFET driver chips that help with this.
DO NOT mount the ignition driver anywhere near the microprocessor. 400v spikes in the ignition system will drive it crazy.
If this is in an automobile, make sure that the power supply is protected from the notoriously spikey 12 volt system in the car. Depending on the particular 12 volt system there can be "load dumps" which can generate 40 to 80 volt surges. IIRC, an LM2930 is a protected regulator.
I can't help you with SPICE.

 

[Silntknight]

It is a 3.3V micro. It can put out 800mA of current. Also, the NMOS I have been looking at all have a Vgs of 3.0 min and 5.0 max. That is for the ignition driver. The solenoid MOSFETS are (likely) 60V 30A logic level MOSFETS. So you're saying that I shouldn't have my ignition driver about 2 inches from the micro's board? You're really screwing with my design! I guess I could mount it right on the ignition coil...

I'll be connecting the micro to a nice 9V battery, or something like it. I do intend to keep the circuits separate. I guess I could use opto's at this point to make the system safer. Anyway, this is a very small automobile. In fact, it has one cylinder. It seats one person, isn't road legal (no lights, airbags, etc.), and definitely wouldn't be considered a car.

 

[Silntknight]

If I move all the drivers off the daughter-board, then I can use a 4-channel opto-isolator with some transistors to restore the logic (since an opto-isolator inverts the logic?) to control the MOSFETS, if they truly run on 10V. Then, all I need is a 9V battery (convenient, eh?) to control the MOSFETS gates, which are connected through the opto-isolator. This gives me double protection for my board. Then, the MOSFETS directly control the 12V driver rails. Sound good? I hope so because I might just do it.

Also, this would give me more space for Zeners so I would just use 500V MOSFETS with 400V of Zeners across them.

Another edit, should I be sending power or signals over long wires (2ft). If I choose to send power, then I'll keep all the drivers on the daughter-board. If I send signals (12V through an opto-isolator), then I will put the drivers on small breakouts mounted right on the device they control.

 

[Jaguarjoe]

Opto's on the uP board with drivers near their devices receiving signals from those Opto's sounds good.

What brand of uP puts out 800ma? Thats like 2.5 watts, thats a lot.

 

[Silntknight]

Ok, but does an opto invert the signal? I found this diagram for an existing opto that would suit my need (https://www.electro-tech-online.com/custompdfs/2010/11/Optoisolator-v12.pdf), but I'm not sure how the transistors invert the signal to correct it. Anyway, the one I am using is a 4-channel opto, so If I added two channels to the diagram, replaced the transistors with the MOSFETS I am using, and kept the MOSFET circuit the same as it is now (in other words, scrap whatever the diagram in the link shows after the transistors), would it work? My guess is yes, so I'll go ahead and make a new schematic.

The Maple can support 800mA.
leaflabs.com

 

[Jaguarjoe]

The Sparkfun part does not invert. Your plan to replace the xistors with MOSFETS sounds ok if your HV supply does not exceed the maximum MOSFET gate voltage and the "on" voltage across the phototransistor is low enough to guarantee the MOSFET is turned off.

The spec sheet for that Cadillac processor says each output is good for 25ma with a chip total of 150ma. Nothing on that chip will handle 800ma. I looked at the board and don't see anything that could handle 800ma either. Unless there's more to it that I can't see, I would call Mr. Maple and ask him what that 800 number refers to. I would also ask him for a complete schematic of that board too so you can properly interface to it. Why do you need so much horsepower for a one lung engine?

 

[Silntknight]

I'm actually very confused by the diagram and what you said. It shows an HV line connected directly to the transistor base and the collector of the opto. Wouldn't that just short the transistor on? The HV supply is just a 12V car battery, which shouldn't exceed the MOSFET gate voltage max (30V) regularly. Also, what do you mean by "the 'on' voltage across the phototransistor is low enough to guarantee the MOSFET is turned off"?

If I can't end up using an opto and I simply connected the pin to the MOSFET gate, would the microcontroller and the MOSFET need to have the same ground? If so, both have to run on the car battery. In that case, I don't have any protection from voltage spikes from the car battery. The best I could do is place a Zener across my regulator just like I did across my MOSFETS. This one would breakdown at 20V because my 5V regulator can only handle up to 30-ish.

If it can't handle it, I assume there is a power transistor there somewhere. Otherwise, I don't know. I can also get a schematic pretty easily. It is posted online. Anyway, not sure why I need it.

Why do you think I'm after horsepower? I said "injector/solenoids" because the devices are all fuel injectors but only 1 is being used as such.

 

[Silntknight]

I'd like my circuit to be like this:

 

[Jaguarjoe]

As it stands, the opto will turn the MOSFET on with whatever current it can deliver but there is nothing to shut the MOSFET off. Right now when the opto turns off the gate will be floating because there is nothing to take it to gnd. The gate is a capacitor which must charge to turn on the MOSFET and discharge to turn it off. For short risetimes this requires a lot of current for a short time to charge the gate cap, like an amp or maybe more. For quick fall times the gate needs to be taken to gnd. A lot of current will flow out of the cap so that the gate's connection to gnd must be a very low resistance for a short time. Looking at the Sparkfun part, the transistor will turn on and if its saturated will provide a high current low resistance path to gnd to shut the MOSFET off quickly but only has its collector resistor to turn the MOSFET on. This will take a long time. You can buy small gate driver chips that have a transistor to turn the gate on hard and another one to turn the gate off hard.

Powering off of a lone car battery is fine. Put 500nf bypass capacitors across each coil/MOSFET circuit, from +12 to gnd as close to the coils as you can get. The biggest contributor of garbage to a car's electrical system is the mechanical alternator's inability to keep up.

When I was referring to horsepower, I meant the uP, not the motor.

What are you doing for sensor inputs? Throttle position, crank position, cam position, and a temperature or two? N-alpha, S-D, or MAF/VAF?

 

[Silntknight]

What you are saying is that the SparkFun part uses the transistor to invert the logic? When the opto is "on" then the transistor drains to ground which turns the MOSFET off, right? That means that when the opto is off the MOSFET is on. If this is true, I need the opto to be on whenever the microcontroller ISN'T sending a signal.
Anyway, about gate drivers, if I used one, would I just connect it to the OUT from a NPN transistor, with the base connected to the collector of the opto and to +HV, like the SparkFun circuit? Then could I use this (https://www.electro-tech-online.com/attachments/gate-driver-jpg.31391/) without the resistors? Actually, I'd still have the farthest right resistor, 100 ohms, but it would be in front of the entire circuit, on the PWM in line.
By the way, would it be good to have a Zener across the gate to ground? You said the car battery can spike to 80V, which could destroy the MOSFET. I could use a Zener to clamp it at 20V.

I don't have an alternator on this. Remember, not a real car. Still, the cap can reduce noise from the engine.

For sensors, I have a Hall Effect sensor on the crank flywheel, a MAP sensor, an IAT sensor, and a "throttle position sensor" (it isn't really sensing a throttle position because it controls the throttle position). I am using an electrical valve train, which is what the solenoids are for. This means that I don't need a cam position sensor.

 

[Jaguarjoe]

You are right about how the Sparkfun part works.
The gate driver circuit you have shown sure is simple almost too simple. As long as you can afford to lose 0.7 volts when its on and have an extra 0.7 when its off. Look at these:

http://www.datasheetcatalog.org/datasheet/microchip/21422b.pdf

You can get inverting or non inverting units in little 8 pin packages, they come in pairs, 2 bucks apiece. Plug and play.

I didn't say the car battery can spike, I said the electrical system in a car can spike. You don't have an electrical "system" to worry about. The battery by itself will be fine. Just be careful with all that current available. Short circuits will be spectacular!

If you can control intake valve stroke I think you can eliminate the map sensor and throttle entirely. BMW does this, I think its called valvetronic. If not, does the throttle positioner have a feedback pot? I think you'll need that because you can lose counts on a stepper motor drive . A map sensor on a one lung engine is hard to implement because of the hugh impulses present.

Did you ever find out what the 800ma number means on the uP?

 

[Silntknight]

I am fine with a gain/loss of .7V. How do you know I will lose .7V just by looking at the circuit? I would run 12V through it, so losing .7V still gets the gate a hard on. .7V is way to low for the gate so it is hard off. If you think it will work, I am fine with just building a small circuit like that. In that case, how do I create inversion? Would I use another transistor in front of the driver like the SparkFun part?
To decide whether or not I should get inverting/non-inverting, I need to know if ALL optos invert the signal. I'd prefer to be able to just buy the part.

Well, without a "system" I guess I'm safe, but I'll still use an opto and keep the circuits separate. Shorts could be exciting with 165 CCA pumping through the boards. I'll throw in some fuses.

I control the stoke PAST the normal 180 degrees. I need a MAP because I need to know the pressure in the cylinder, which is isobaric to the manifold pressure because the engine works against it. The "throttle" is actually the valve timing. I don't have a throttle plate. I'm not sure what I need feedback for. I use a sliding pot to control how long the intake valve stays open after 180 degrees. No stepper motors are involved. Also, I don't think BMW does this because it would imply that they use modified Miller cycle engine, which require a supercharger (normally) to function. The MAP shouldn't be that bad. I'll have large dips during the intake stroke, but the MAP will equalize to the cylinder pressure during the expulsion during the "first" compression stroke. Then, when the valve closes, the pressure is the same as the manifold pressure. Easy system for stoichiometry.

No, I don't really need 800mA so I never bothered. I can check if you'd like me to.

 

[Silntknight]

How should I connect the opto to the driver? Like this?