MOSFETS and Solenoids

I made the board that has eight MOSFETS. I take two MOSFETS and put them in parallel. So I have four outputs. An output is going to power four solenoids. The solenoids take 26 amps.

I am attaching the spec sheet for the MOSFETS and also a schematic. In the schematic, the MOSFETS are in the top right area.

The solenoids are for four nitrous stages in a race car. For this race car, only two stages of nitrous. Here is my problem, on the racetrack I can't get the second stage to activate. Friday I went to the shop, we started the engine and activated the my controller without the nitrous bottle open. On the data recorder both stages activated and the battery voltage looked clean with two steps for each stage activation.

On the track the first stage activates and the second doesn't. When my board activates, the battery voltage gets upward spikes. So the final test run, the team used a mechanical push button so the driver could activate the second stage, in which it worked fine.

The only difference between the shop test and the race track is that the transbrake was not activated on the shop test. The transbrake is a solenoid that literally puts the transmission in low and reverse gear. This is what holds the car at the starting line until the driver releases the transbrake button and the car leaves. This release also activates my box that controls the activation of the nitrous stages by time.

This afternoon I am going to have the team do a shop test with the transbrake. So here is my question, why would the nitrous stage activate using the mechanical pushbutton and not the MOSFET? The wiring is essentially the same between the two scenarios. Hypothetical situation, if a circuit barely has enough ampacity, would a mechanical switch work and a MOSFET would not? Meaning does a MOSFET not like it when a circuit is starved for current? This is a 16v system

Thanks ahead for your input. I also attached the transistors I am using to turn on the ground to the Source leg of each MOSFET and a pic of the board with the MOSFETS and busars.

Still looking at the schematic but looking at the pic I would reflow those solder connections. Might not be making good contact between the Drain and bus.

Yea thats pretty ugly. What you see in the pix is actually excess solder from solder rod I melt onto the bus with hot air. I am using a hot plate and solder paste now to flow the solder. It works pretty nice and looks alot better.

Quote:

---

Originally Posted by mackrackit

Still looking at the schematic but looking at the pic I would reflow those solder connections. Might not be making good contact between the Drain and bus.

---

One thing I should add..I sent a board to a nitrous manufacturer. They put the board on their flowbench and the board activated four stages at the same time for 15 seconds several times.

I suspect the layout of components and traces. Can you post a pix of your board? A ground loop is possibly taking you down.

I am attaching a color and B/W PCB view from my software. The wiring on race cars aren't ideal. Battery ground is connected to the chassis. Then there are three or four other points on the chassis components are grounded.

I have one other team with the same problem and they are using a transbrake aswell. They are also activating first stage right after transbrake release.

I would prefer seeing just the copper layers without the overlaying idents. However, first impressions, GND signal traces are way too thin to be useful unless you have copper filled your board for the GND net. The way it is now, you will get huge surges on the GND line whenever a MOSFET switches. The GND trace, if not copper filled plane, should be at least as thick as the traces on Stages 5,6,7,8

I do have a ground plane. I took it out so you could see the traces easier. I am attaching a PDF with the ground plane.

Thanks
Toby

Toby

Couple of things I would check on.

1 - the trace going under the lower edge (going to R5) of the PCB. I do not know if it drives any heavy loads there.

2 - Vin to the MOSFETs - perhaps they are on the second layer and have thick tracks??

Jumper thick wires direct to the Vin lines till you find the cause. Jumper thick wires direct to the Ground points near the Mosfets till you find the cause. I suspect some heavy current switching on is ticking off your PIC to malfunction. I would typically see that the heavy currents flow first to the load switching devices(heavy loads) and then come back to the CPU(flea power)

Maybe someone else might see something I haven't caught

One other thing you might want to look into is the temperature of your board. The thermal enviroment under the hood of a car is quite different from the thermal enviroment of a lab bench. You might need more heat sink for your MOSFETS.

For what it is worth I agree with all the above comments, but I have a question or two that might help.

When the car is on the track are you using an alternator? In my day we did not. We put a couple deep cycle batteries in the back to help with traction and would run everything frm that. I am getting at two things. Maybe the battery is just low enough not to satuarate the FET or if you are using an alternator by the time the second stage kicks in it is giving "dirty" power.

If it works on the bench or sittng in the garage, like Tim was getting at, also look for heat or mechanical stresses.

One other thing I didn't see mentioned is the solenoids have to work harder when there is 900PSI of pressure behind it. This would require more current to open the solenoids. When you tested in the garage there was no pressure in the line. I don't know if you log battery voltage either, but with the load of fuel pumps and ignition amps under heavy load will drop the battery voltage regardless of using an alternator, since you said 16V I doubt you use an alternator.

Attached are two screen shots from our data acq system.

The first is labeled OctoShop.pdf.
There are two channels visible, engine rpm and battery voltage. Voltage is in green. Notice how smooth is it and the two steps down are the two stages activating.

The other is labeled OctoRun.pdf.
There are four channels, engine rpm, battery voltage, N20Fuel 1 and 2. The time is on the x axis. Right at 0 seconds the transbrake is released and the engine accelerates. Immediately stage one nitrous is activated and the solenoids open. This is validated by N20Fuel1 psi dropping and also battery voltage dropping. Two tenths later the second stage is activated but the solenoids do not open. This is validated by the battery voltage not dropping further and the N20Fuel2 not dropping.

Look at right before one second, the battery voltage drops and N20Fuel2 drops. This is the second stage finally opening. It doesn't stay open though. This happens two more times going down the track.

The differences between the OctoRun and OctoShop are the transbrake wasn't activated, the engine rpm is lower, and the car is not moving.

No alternator, just a pretty large 16v battery..around 45lbs.

I am not a solenoid expert but I was told that solenoids draw the same current no matter how hard they have to 'work' The current draw is a function of coil resistance and the input voltage?

The MOSFETS Source is a copper busbar 0.040" thick and 0.500" wide.

Tonite the team is going to start the engine and activate my box in the same manner its activated on the track with the transbrake release. I am going to have them do that with the engine at low rpm and then at high rpm. Its going to be either a problem with the transbrake and/or electrical noise from higher engine rpm. I will let you know what I hear.

I am going to draw up a board tonite that opto-isolates all inputs and outputs and uses an isolated DC/DC converter to help with the electrical noise side. I will post the schematic tonite.
Thanks for the input.

Here is the schematic using opto isolated input and outputs plus an isolated dc/dc converter. I used a 5v regulator in from of the converter because the 16v gets to around 20v with the charger.

Isolator and DC to DC (MEV3S0505SC) spec sheets

I'm not sure you're getting the full advantage an isolator has to offer when you tie the output side ground to the input side.

Ideally, you should have a supply for the output side (26Amps) that is isolated from the input side (5V side) via the isolators. Indirectly, it also means, you should ideally isolate the 2 supplies. Otherwise, the effects of the ground currents running into the low voltage side and disturbing the PIC will still continue to haunt you.

If it is the supply noise due to RPMs increasing, a simple half wave rectifier at the input of your 5V regulator should keep the circuit happy. I've seen this kind of problems when the alternator starts spiking the battery voltage at high RPMs.

Cheers

Try lowering the value of the gate drive resistor. Mosfets have a very high imput impedance and the gate-source charge needs to be thought of as a small capacitor. I have seen values less than 1k used here for mid frequency applications.
Also I do not see any flyback catch diodes. They release the back EMF energy of the solenoid as current/heat to ground. A high voltage spike on the drain will certainly make life for the MOSFET difficult as the built in flyback diode is configured for low side drive not high side as in your case.
If I was designing a driver I would put battery + to one side of the solenoid and put the N channel MOSFET from the solenoid to ground. The built in protection diode will then act as your protector diode and catch the back EMF spike. Alternatively a more efficient method is to recycle the energy through a snubber network.

Just my thoughts...
Tim.

I went with the P channel because thats the way the teams want it. Its more of a build what customers ask for deal. Right now the teams turn on and off the + side with relays.
I will give the lower gate resistance a go and see what happens.

Thanks to everyone. Once I get it sorted out I will let everyone know the results.

I am not too sure what you mean by input and output grounds tied together. Can you give me a location on my schematic?
How about a suggested part number for the half-wave rectifier?
Thanks
Toby

Quote:

---

Originally Posted by Jerson

I'm not sure you're getting the full advantage an isolator has to offer when you tie the output side ground to the input side.

Ideally, you should have a supply for the output side (26Amps) that is isolated from the input side (5V side) via the isolators. Indirectly, it also means, you should ideally isolate the 2 supplies. Otherwise, the effects of the ground currents running into the low voltage side and disturbing the PIC will still continue to haunt you.

If it is the supply noise due to RPMs increasing, a simple half wave rectifier at the input of your 5V regulator should keep the circuit happy. I've seen this kind of problems when the alternator starts spiking the battery voltage at high RPMs.

Cheers

---

Look at your opto isolators. You have tied the Led cathode to the same ground as the opto transistor ground. That is what I wish you could break apart and keep apart.

Do you have access to a scope? Check your gate to source voltage.
Does the drain voltage drop after the 1st solenoid energizes?
Ignitions generate incredible electrical interference. A high frequency lightning storm. The spark plug firing voltage can vary with engine load.
The interference could affect your mosfets and or PIC.
The electrical interference is both conducted and radiated into your circuits.
Proper Layout, shielding, twisted pair wiring, etc... may be needed.
Tires can also generate electrostatic electrical interference.
Try diagnostic LEDs to make sure your program is executing the code.

Good Luck

Thanks for pointing that out. Theoretically, if the board is completely isolated like the attached schematic I will be immune to electrical noise like markedwards is talking about or am I just minimizing it?

markedwards...I am pretty sure the PIC isn't resetting because the first stage stays engaged. When the PIC powercycles it goes through a start up sequence thats about ten seconds long including turning off all outputs. If its electrical noise its definately only at higher rpm. Testing in the shop everything looks fine.

I am getting a 4 channel analogue module for our data acq system. I am going to make some voltage dividers to log the voltage output of the MOSFETS with the 0-5v inputs.

Quote:

---

Originally Posted by Jerson

Look at your opto isolators. You have tied the Led cathode to the same ground as the opto transistor ground. That is what I wish you could break apart and keep apart.

---

Quote:

---

Originally Posted by Tobias

Thanks for pointing that out. Theoretically, if the board is completely isolated like the attached schematic I will be immune to electrical noise like markedwards is talking about or am I just minimizing it?

---

If noise is found to be the problem and it is really bad then a grounded metal enclosure might also help. Think Faraday Cage...

I didn't see any large capacitor for the battery + connection to ground.
With high frequency noise there is a significant impeadance between the battery and your device.
A scope will show you high frequency noise that your aquisition system
will likely average (filter) out. The aquisition system is similar to a DMM with a slower response. It will not show high frequency spikes or voltage dips.
Ignition and similar electrical noise can be conducted and re-radiated
inside the metal enclosure.

What is the solenoid's dc current draw?

There is a capacitor on the chassis. Its sold by the ignition company. Its pretty big, about half the size of your fist.

The solenoids draw 26 amps per stage. Some cars will have four stages. The fourth stage requires 14 amps.

Our data acq system only samples 50 times a second on the logger voltage. I did hook up my scope to VIN with the cap disconnected and its pretty scary. The cap really helped alot.

Quote:

---

Originally Posted by Tobias

Thanks for pointing that out. Theoretically, if the board is completely isolated like the attached schematic I will be immune to electrical noise like markedwards is talking about or am I just minimizing it?

---

The circuit now looks good. However, I have not checked it thoroughly, just a casual glance over. By isolating the 2 grounds and the 2 supplies, what you have done is eliminate the electrical intermingling of currents. The radiation aspect is what you now need to control. For that, everything you read in the rule book for HF design and high current design should apply.

I am willing to wager on you solving your problem by just separating out the 2 supplies and grounds like you have in the new schematic.

You need to ensure that the PIC does not brown out during the time one/maybe more MOSFETS are on and the engine is revving. This is the time you will see brownouts if your 5V power is not well regulated. Like I said earlier, a simple 1N4007 diode followed by a reservoir capacitor feeding your 5V regulator should keep the PIC happy. The best way to check would be to ensure that the circuit stays on for at least half a second when you remove power. Now, come whatever spikes due to the engine rev, you will still have the PIC running safely.