Power control of the EMS system

[shameem]

Just my 2 cents - IMHO the coil/inj +12v feed should not be in the same line as FreeEMS box due to flyback issues - i found out (by accident) that the best flyback damper is the battery itself.... In one of my friend's MS setup - i routed the feed to a separate relay directly to the battery and all the weird spikes and reset problems vanished just like that......

[Fred]

I think I understand what you mean and I think I agree with you. That diagram was supposed to show the control flow and connections more than how it is physically wired. In that respect it is slightly inconsistent. I might attempt to do a better diagram later with the correct quantity of relays and correct wire routing. I may try to do a ground one or even include a ground one in this diagram if it will fit cleanly too.

As you say, grounding and power feeds are the most important aspect of wiring up an EMS box. If you saw the wiring on my truck you would laugh as it is major overkill, however it started first try (once I rearranged the coil wiring to be correct) and I had no noise issues at all.

Fred.

[jharvey]

Hump hump, or was it bump, I get so confused some times.

I think the freeEMS should have multiple ground wires of varying lengths. I'd say at least 2 perhaps 3 grounds.

I'll be changing the A.08 to have two regulators, and will be routing the power feeds around a bit. Also I'll be putting a fuse holder in the system.

I don't disagree with excessive fusing, but I only see a need for 2 fuses. That's big power fromt the battery and small power from the battery. Fuses are simply protection against short circuits and blow before the wire catches fire. I think freeEMS should have a big fat wire to power it. preventing voltage issues, so a fat fuse wouldn't hurt here, then the PCB fuse can drop your current requirements to allow for smaller traces on and after the PCB. After that the extra fuses are simply diagnostic tools to help find what popped the fuse.

Perhaps we also want to design a relay / fuse PCB? I don't know if I'd prefer this as in line or on a PCB. I'll have to think a pinch on that.

[Fred]

As long as everything is fused appropriately for the gauge of wire following it there is no safety issue, however, when a fuse dies, it's nice to know what caused it. If you have 50 things hanging off one fuse then it's impossible to diagnose. Hence separating it out.

As for heavy wire, that is always nice, but is a patch. A patch that MS needs to work properly. I'll still be using heavy wires, but given that others won't we should design the system to not care. For that reason we have 4 positive feeds to the box : bat ref, heavy dirty power, cpu supply power, sensor supply power to allow the correct control and noise separation. Grounding wise, we will want 2 separate grounds : cpu/sensor ground and dirty heavy ground. These can consist of multiple physical wires etc.

I'm going to put some good work into the diagram later today so we have a solid visual on what there is, how to hook it up, and how it works.

Fred.

[jharvey]

Perhaps we should design a fuse panel. Just a PCB with a bunch of fuses and solder point on it.

[Fred]

The trouble is that such things are very very application dependent. That translates to a smaller percentage of the user base that will benefit. Which in my mind means that if at all, it should be done much later. I always strongly recommend that my friends and acquaintances do NOT use the MS relay board. I see it as a weak link added to the system. Either way, let me get out a new and improved diagram and we'll see where we go from there.

Fred.

[KW1252]

Looks great

From experience, I could say wiring all the electrics into one relay requires special attention. At least in Mazda 626 wiring, main relay (for powering up sensors and solenoids) is on continuous power while the relay for fuel pump and coil is switched, and simply connecting them to a single pin would create an interesting situation where electricity loops from the continuous power circuit to the switched power circuit. A single diode will cure this, but unless it's taken care of, you might be left wondering why your battery is flat after every driveless weekend...

I like the idea of ECU going to sleep mode on it's own means. My box draws about 150mA and according to the manufacturer, should be connected to a switched power supply (it was not how it was put in wiring instruction sheets though) and that way, it's really easy to lose logged data and wiring a turbo timer would be quite a bit more interesting.

[gearhead]

1) fuses should be separate from the ECU.
2) There needs to be 2 grounds. 18 ga ground for the digital side and a 10 ga (or so) for the other power grounding needs (ignition, injection, PWM valves)
3) There needs to be 2 +12 supplies. 18 ga switched with key or relay to power up ECU. This should also be the vbatt line. The other +12 should be direct to the battery through a fused link. This is the flyback or release pulse line.
4) If we want the car battery to maintain flash, we will need a 3rd +12 unswitched line. The more I think of this, a 3V coin sounds better unless there are some really good reasons to need +12 instead.

Do I have this wrong?

Herr Bosch did it this way and they are dead reliable. Pin 5 is signal ground, Pin 17 is power ground, Pin 35 is switched +12 and pin 4 is constant +12 for flyback reasons as well as continuous power requirements. Note it is only slightly larger than the rest of the wires at 1.0 vs 0.75, though pin 17 is 2.5. Big Ground.

Now, the complicating thing is that we are running ignition as well. If we allow for (in box) high current drivers, we will need another power gnd for ignition. Do we need a snubber diode for each coil as well? If so, then we will need another +12 unswitched which I feel should be separate from the injection side.

KeithG

In my install (as well as Fred's) I do not worry about the ignition noise in my ECU box as my triggers are all low current and not inductive (CDI for me, external igniters for Fred).

[Fred]

4) If we want the car battery to maintain flash, we will need a 3rd +12 unswitched line. The more I think of this, a 3V coin sounds better unless there are some really good reasons to need +12 instead.

Why do you prefer a battery? For a start, I think it has to be 5v battery which could be a pain/impossible to find. Secondly it becomes a maintenance item. What do you have against a 3rd +12v line? I can't see any issues with it. It just requires me to write some code (can be very basic at first) to watch a key switched input and go to sleep and wake up again. It will draw next to no power while asleep anyway and it can come up quickly if need be.

Now, the complicating thing is that we are running ignition as well. If we allow for (in box) high current drivers, we will need another power gnd for ignition. Do we need a snubber diode for each coil as well? If so, then we will need another +12 unswitched which I feel should be separate from the injection side.

Definitely not. The reason you use IGBT there is because the fly backs are very high voltage. Clamping that spike screws your spark up totally so you just want to let it fly where it wants for ignition. I've also thought about the grounding and considered a separate one. My gut feeling is the same, but I have no basis for this. I guess we should just calculate it based on peak current demands for each device or set of devices. Perhaps the best way would be to eat connector pins fast and use one pin in and one pin out PER coil and per injector. That way the current capability of the pins can stay low and any faults only affect one cylinder. That last bit could be good or bad.

In my install (as well as Fred's) I do not worry about the ignition noise in my ECU box as my triggers are all low current and not inductive (CDI for me, external igniters for Fred).

As you know, I'm a big fan of this, but people wanting on board ignition is inevitable so we should probably plan for it up front.

Fred.

[gearhead]

4) If we want the car battery to maintain flash, we will need a 3rd +12 unswitched line. The more I think of this, a 3V coin sounds better unless there are some really good reasons to need +12 instead.

Why do you prefer a battery? For a start, I think it has to be 5v battery which could be a pain/impossible to find. Secondly it becomes a maintenance item. What do you have against a 3rd +12v line? I can't see any issues with it. It just requires me to write some code (can be very basic at first) to watch a key switched input and go to sleep and wake up again. It will draw next to no power while asleep anyway and it can come up quickly if need be.

Fine. I am convinced. A 3rd 12V line for maintaining memory. Can we protect this feed (on board) and use the constant 12V that we are using for flyback? I am sure we could do some testing to see if there is a way to make it work and save a plug pin.

Now, the complicating thing is that we are running ignition as well. If we allow for (in box) high current drivers, we will need another power gnd for ignition. Do we need a snubber diode for each coil as well? If so, then we will need another +12 unswitched which I feel should be separate from the injection side.

Definitely not. The reason you use IGBT there is because the fly backs are very high voltage. Clamping that spike screws your spark up totally so you just want to let it fly where it wants for ignition. I've also thought about the grounding and considered a separate one. My gut feeling is the same, but I have no basis for this. I guess we should just calculate it based on peak current demands for each device or set of devices. Perhaps the best way would be to eat connector pins fast and use one pin in and one pin out PER coil and per injector. That way the current capability of the pins can stay low and any faults only affect one cylinder. That last bit could be good or bad.

With all the noise an IGBT coil driver generates, the coil drivers need to have their own gnd. If someone wants 1 per coil, then they can figure that out. I just do not think we want it on the same ground as injection power. Can someone convince me otherwise? We are trying to have a noise immune setup and ignition is a tremendous amount of noise. If you are only using low current drivers, then jumper ground from power over on teh injection side of the board.

I knew there was a reason we do not want a snubber, but could not remember why. Thx.

In my install (as well as Fred's) I do not worry about the ignition noise in my ECU box as my triggers are all low current and not inductive (CDI for me, external igniters for Fred).

As you know, I'm a big fan of this, but people wanting on board ignition is inevitable so we should probably plan for it up front.

Fred.

I agree with this and see the advantages. that is why I am trying to work out how it should be done, properly.

Gearhead