My failsafe mechanism.
 

Water injection is great- right up until it stops working. Then, depending on how carried away you got during tuning, it may be time to start shopping for a new engine. I decided I wanted a failsafe system to put the engine in a “safe” condition should the WI fail.

My original thought was just to put a pressure switch on the water line between pump and nozzle and have it trigger a retarded ignition map on the EMU. Unfortunately this sort of external map-switching isn’t supported by the software. So after some tinkering, this is Plan B.

What I’ve done is to construct a bypass system around the boost controller (MBC) such that should the WI system fail, the system will only produce “stock” boost- around 5PSI for my configuration. Further, the system is implemented such that should the failsafe itself fail, the system will also revert to the safe mode.

It’s actually pretty simple- only two major components. First is a 6-30PSI pressure switch, currently on sale for $20 at Devil’s Own: http://www.alcohol-injection.com/adj...witch-p-1.html and the second is a boost control solenoid, $50 from Flyin’ Miata: http://www.flyinmiata.com/index.php?...0%20%201990-97

The solenoid is a bit tricky to figure out at first- it has three ports, labeled IN, EXH(haust), and OUT. If you’re electronically inclined you can think of this thing as a SPDT air relay, where OUT is the common terminal, EXH is the N.C. terminal, and IN is the N.O. terminal. So when de-energized, air can flow between EXH and OUT. When energized, air can flow between IN and OUT.

So, referring to the attached schematic. I plugged the IN port with a 1/8” NPT pipe plug, so that I now have a simple on/off valve that passes air when de-energized and blocks air when energized. I then screwed some brass fittings onto the remaining two ports, such that the solenoid is installed essentially in parallel across the MBC. When the solenoid is open, air flows freely through, bypassing the MBC and limiting the system to stock boost. When the solenoid closes, the MBC is the only path to the wastegate, so the system functions normally.

The switch was pretty easy- just put a tee in the water line right before the nozzle (after the check valve) and use the normally open contacts.

Theory of operation is simple- under normal conditions (WI off) the solenoid is open and the WG is exposed to normal manifold conditions. This isn’t a problem since if the WI is off, we’re probably not trying to make boost anyway. As soon as boost starts to come on, the WI pump pressurizes the water line, closing the switch, closing the solenoid, and shutting the bypass. Should the WI pressure ever drop (empty tank, pump failure, broken hose) the line pressure drops, the solenoid opens, and boost drops. When boost suddenly dumps to 5PSI (from 13) it’s also an obvious and unmistakable indicator that “Something is Wrong. You must stop and investigate NOW!”

In other words, the failsafe isn’t something that comes into play only when something is wrong to limit boost- rather it operates only when everything is normal to permit higher boost. Any industrial process engineers in the audience will appreciate this fine distinction.

So my plan is simply to tune the EMU maps assuming that no water is present below 6PSI, and only get aggressive with the timing above 6PSI (which, if I’m really in a hurry is where I’m gonna be anyway)

I’ve had the system in for about two weeks now, and it’s working splendidly. There has been no negative effect on spoolup, no weird oscillations, nothing. Under normal circumstances it behaves normally, and if I turn off the water it drops right down to 5PSI. This system has also had an unintended but interesting side-effect: it gives me an easy way to limit boost just by pulling one wire- handy if it’s raining heavily or for AutoX on street tires where 13PSI is merely a one-way ticket to coneville. (Yes, I am running SM2 on street rubber. Go ahead and laugh. Get it out of your system.)

The rest of the pictures show the brass valves on the solenoid, the hot side of the engine compartment showing the solenoid installed and plumbed (there was no good way to photograph this one) and the cold side showing the pressure switch hanging from one of the radiator bolts and plumed into the water line just below the nozzle.

wow nice :bigtu:

I bet you could also use the GM boost solenoid, they're from a Typhoon and cost around $30 shipped IIRC.

Nice job Joe - as usual. :bigtu:

off topic....damnit, i wondered why it never worked for me...i have one sitting in front of me right now, i had it hooked to a EBC with IN/OUT, not EXH/OUT, makes sense since I would only see 7psi no matter what the EBC was configured for. Looks like I have a project for the MS for boost control.

back on topic....nice job!

Hey, I know all about SM2 on street tires with >200whp. Hoosiers didn't help much. :(

but I thought people were saying EMU could do EBC... It's a shame you live in Cali, you would love MS, and could make tons of great writeups.

Good plan Joe. Just a couple of random comments:

You may want to have an indicator lamp come on when the solenoid opens. Just for a visual indicator of trouble.

You may wish to replace the MBC with a second selenoid and control it from the EMU.

This would certainly help to prevent damage occuring if the pump fails or tank empties, but would not prevent failure from a clogged nozzle--like what happened with MagnaMX5. Perhaps a second "over pressure" circuit could be added as a failsafe against the other extreme?

I do hope you're joking. I've been waiting to see where people go with the new 275/35-15 A6s once they start hitting the dealers this month. Thinking maybe if I'm lucky I can pick up some used 13x8 wheels on the cheap towards the end of the season and run 'em next year.

Actually, if I end up keeping this car for any length of time I'll probably end up doing a Megasquirt eventually. The EMU was a nice way to get up and running quickly and with minimal fuss, but I'm really ithcin' to try a full standalone at some point. It's not technically legal, but then neither is the EMU.

Thought about it, though it'd actually be a little harder than that, since the solenoid is normally open all the time, except when the water pump is running. I think merely not being able to make boost is in itself a pretty good indicator.

No EBC on the EMU. I'd have to go with a standalone EBC. Don't really wanna spend the money on that right now, since the TurboXS MBC is working pretty well.

This is true. I think what I'd really need is a flow sensor- if water is not flowing then there's a problem (dead electronics or clogged nozzle) and if water is flowing and yet there's no pressure then that's also a problem (open line). For now I'll just have to keep debris out of the system. Which reminds me, it's probably about time I pull the nozzle and inspect the screen. Schedule that for this weekend.

True that it doesn't run closed loop boost control like a standalone EBC can, but it can still control the dc of a solenoid.


Heh flow sensor. See that's why you do what you do and I do what I do. I didn't even think of such a thing.

nope, the 225 hoosiers really can only put down power in a perfectly straight line. The other local turbo NB runs on 15" koseis with V710s, and has just as much trouble. you just can't fit enough tire under a miata it seems. I've been considering making an EBC map to make the turbo hit more softly. you may have more open courses out there though, it may not be a big problem for you.

that's cool about the MS, you'll be doing writeups for wild add-ons in no time.

This wouldn't help with a clogged nozzle would it? With the nozzle close you'd still have a pressurized line no?

Hmmm.... Maybe I'll be flaring the fenders once those 275s come out after all...

In San Diego, we're fortunate to have a pretty big space to run in. Our typical courses can run 70-90 seconds, and there's usually one or two big, wide turns exiting onto straights where it'd be nice to be on the power without having to worry about whether I'm about to go +20 and kill a corner worker. Here's a picture of the space we typically run in. You can guess where the hairpin that always bites me in the ass is located. :D

Good point... you might need to get a range switch... so it is only active when 25<X<60 (or whatever you pump pressure is normally). I've seen adjustable ones at one point... but I don't remember where.

I have been planning the same thing, but want to use a temp switch instead. It drops the boost if the IAT passes a threshold. That way it can account for a clogged nozzle or simply a very hot day.

I'm afraid all this business of overpressure switches to detect nozzle clog just isn't going to fly.

The Shurflo pumps that most of us are using have built-in pressure limiters. When they reach a certain head pressure, the motor is interrupted. Basically a primitive pressure regulator. As far as I know, this limit point is what the adjustment screw adjusts. Remember, these pumps are just a modified version of those originally designed to pressurize the potable water systems of boats and RVs- not a very high precision application.

So because, in normal operation, the pump is operating against the pressure limiting switch anyway, a clogged nozzle isn’t going to cause the line pressure to increase- it will only decrease the operating duty cycle of the pump.

What you'd need for accurate clog detection is a very sensitive flow switch- one capable of accurately reading the very low rates of flow we're dealing with. I'm sure somebody makes one, though the only flow switches I typically come across in my work are generally sized in the gallons per minute range, rather than gallons per hour. They'd never even detect 2 or 3 GPH- like trying to measure crank endplay with a yardstick.

I do kinda like the temp sensor idea. I wonder how much the temperature actually drops between the throttle body (where my nozzle is) and the back of the manifold? Despite “common knowledge” to the contrary, I've always imagined (with no particular supporting evidence) that the majority of the charge cooling took place inside the cylinder during the compression cycle.

If you position the temp sensor such that it the nozzle is in the direct flow of water/meth I'd think that in the absence of water the temp will jump.

Probably the best thing to do is wrong a knock sensor and be conservative with the tune :)

putting a temp sensor in the direct path of the water/meth would introduce a convenient location for puddling to take place.

BTW, the OP should be a sticky. great design and mod

for the super rich out there.... aquamist sells a flow sensor with a little spinny wheel in it that detects flow. sounds like something I'd stay away from as far as reliability (too fiddly), but it's interesting none-the-less.

http://www.aquamist.co.uk/sl/plist/p...nstruction.PNG

Snow Performance has a flow meter solution as well:
http://www.snowperformance.net/product.php?pk=13