Megasquirt Cold Starting Pulse Widths

So, it was a bit chilly here this morning. About 17F. The 20v Rolla didn't want to start so I had to take the Camry into work. Actually, I've been having some cold start issues with it, but it usually would eventually fire. Once started, I was fine. The battery gave out this morning before I had the chance. So, what percentage of bumping up the cranking pulse widths are people using? That is, if 2.5mS works at 160F, what about the buckets colder than that down to 0F? I think my current pulsewidths are just too fat between 0 and 40F or so. Probably too fat above that too. My CLT was about 27F this morning when I was trying to start it, at an ambient temp of 17F. We dropped from 65F last night about 6:30 to 17F this morning. I see that the PnP Miata msq varies from 3.1mS at 160 to 8.8mS at -40 but isn't very linear. So, what has been working for you guys?

A LOT more than you think. Lot.

Lets put it this way, if I were idling at 2.5ms hot, I would probably need somewhere close to 65ms (yes 65) at 20F.

If you are having cold start problems- more fuel. It's not linear, since it's very much based on the fuel properties, but it takes a LOT of fuel to get it fired quickly.

Eric

Ok, some more info...

Lets assume that your hot idle is 1.7ms, for a 70F start, you would need about 2-4x that to crank almost perfectly.

Down to 20F- it's up to 10-20x the idle to crank and fire up.

So kind of ignore what I just said. But not by much.

Can you take data (a/f) when you crank and run up?

I'd be willing to bet that you are not even close to too much fuel. Like a LONG way from it.

Eric

I can't speak for the MS units specifically but Eric is correct in that it isn't linear.

The theory is of course called lost fuel. The fuel needed to actually get into the cylinder is a function of the fuel required for the desired A/f ratio, the fuel temperature and the intake port area temperature. This all assuming you are not direct injection and you are injecting on a closed valve. The later isn't required if you are not injecting directly into the intake port face.

When I was doing the modular engines it was scarey how much fuel they wanted to get a good cold start at -40F. If you shorted the key a few times you could foul a few plugs pretty quickly! The fast burn heads like even more!

A little trick I also used was to shut the ISC motor/valve until after the inital flare of the engine. The peak rpm is dependant on the intake volume to the engine volume. If you look at a few of your rpm traces you can find it pretty quick. At that point you open the Isc to a nominal value for that rpm. It really acts like a choke on a carb and reduces the fuel needed after the initial crank fire. It helps if some one short keys and reduces the chance of a flood condition.

I am sure it is all in the MS, just not sure how to tell you to do it on there! Now if it was an 8065.........

Humm. Well, I can't really take a a/f when I'm cranking. It's a heated narrow band sensor, but it still takes a while to start getting readings. I've looked at a few maps of other people's MSQ's, besides DIYAutoTune's PnP Miata of 3.1 at 160 to 7.3 at 20F. Another map I found that a guy in MI (where I hear it also gets cold once in a while) had good results with was 4.0mS at 160, 18mS at 20F. I don't recall the engine he had; I'm just looking for a pattern.

Looking at one of my logs, my PW at idle fully warmed up is running around 3.5mS, with the A/F "fat" but idle smooth. My 160F bucket was at 2.5, which I guess would be too low, but it had been starting warm just fine. Maybe the priming pulse. What wouldn't start this morning was a 20F bucket of 12 or 15mS, Coolant only, 5MS standard prime, always, twice.

Oh, and the megatune software won't let you enter cranking PW greater than 25.5mS.

I'll put my 10 amp charger on the battery when I get home and start fiddling with it. At 17F the battery didn't have a lot of reserve capacity left. It won't take long at 10 amps to get it warmed up enough to start cranking again. Part of the whole cold start tuning problem is that you only get one shot at it a day. I've been chasing similar issues with the MegaSquirted Sportster.

The 4AGE 20v has the injectors at the port up against the head. I have an ISC valve, but my control of it is sub-optimal. It will fast idle, but I've never really got all the settings down to get it to truely control my idle speed, so I just set it to "close enough" and let the idle vary a little when it's warm. The MS doesn't really have a setting to independantly control it for cranking anyway, but that's an interesting solution.

The Sportster doesn't have an ISC at all. I'm using a Buell TB and the idle is controlled by a throttle stop screw with a cable that comes out so you can adjust it wherever you want.

Well, I got it started. It warmed up to 20F this evening, so that helped. I put it on my charger and as soon as the battery warmed up some and took a surface charge, about 15 minutes, it cranked a lot faster and started up. I also fiddled with fattening up the 20-40F pulsewidths. Of course, once it fires one time, that's it for the day as far as tuning goes. So, I think my problem was:

• Weak battery. The sticker has 2 years punched out, 2002 and 2007. I'm thinking it is probably 2002 as my service database doesn't list me swapping out that battery since I bought it in 06. At that time, I put whatever battery I had laying around in it, so 2002 is probably about right.

• Pulsewidths not fat enough at "WTF are you doing trying to start me?" cold.

It is also possible that the batt voltage was dropping below where the MS would function during cranking.

here's mine. I haven't screwed around with it much but it works in the really really cold canadian winters:

My problem is to keep it running in the first 5 or 10 seconds after getting started but you said yours works ok once it does fire. Can I see your afterstart enrichment settings?

details: 1997 Escort 2.0L SOHC 8-valve. 15lb/h injectors @ stock PSI.

I see your 20F is 3x your 160F. I think mine right now is closer to 6x. That 3x is probably what others use. I'm not sure if my 160F is too lean or my 20F is too fat, but my 160F is 2.5mS and as of right now, my 20F is 14mS. Also, your table is linear, which others say isn't optimum, but hey, if it works for you, whatever it takes.

I don't use ASE at all. I have my ASE set to Cycles, total time 1, Normal Map. If it is real cold, I may have to work the throttle a little for about the first 3-4 seconds, then give it another 5 seconds or so to catch up and it's good after that. With ASE on, I never could get it to do right. WIth it set to 1 cycle, as soon as it starts, it kicks over to the warmup enrichment. I have 3 MS'ed vehicles and I have them all set up that way, including the air cooled motorcycle.

Sweet to know, we were wondering this cause I was trying to fire up the Merkur the other day and it was all of 30F outside and it would not fire, PW was in the 4-8ms range, sounds like we need more like 20, I did read to go linear, but to fatten it up accordingly but now I'll just go with a nice big number to get in the ballpark.

Dr. Hess wrote: I see your 20F is 3x your 160F. I think mine right now is closer to 6x. That 3x is probably what others use. I'm not sure if my 160F is too lean or my 20F is too fat, but my 160F is 2.5mS and as of right now, my 20F is 14mS. Also, your table is linear, which others say isn't optimum, but hey, if it works for you, whatever it takes. I don't use ASE at all. I have my ASE set to Cycles, total time 1, Normal Map. If it is real cold, I may have to work the throttle a little for about the first 3-4 seconds, then give it another 5 seconds or so to catch up and it's good after that. With ASE on, I never could get it to do right. WIth it set to 1 cycle, as soon as it starts, it kicks over to the warmup enrichment. I have 3 MS'ed vehicles and I have them all set up that way, including the air cooled motorcycle.

Your post got me thinking. Can't tell yet if that's good or bad

Like I said I haven't really screwed around with my cranking table much, just got it to work when I need it. My injectors are relatively small for this engine, equivalent to 157cc/min. Each cylinder is ~500cc, make a ratio of that and compare to yours?

I might try disabling the ASE as you have someday to see if that works better. I know my WUE has some pretty high values so maybe ASE is just trying to flood out the engine until a little heat gets into the plugs.

Well, I have 400CC cylinders and the stock injectors, 295CC. It's about a 160HP motor, all stock except for the megasquirt.

Anyway, if it starts tomorrow morning, I'll take it to work and get a new battery at lunch. Otherwise, I'll take the Camry. Those are the only vehicles we have with snow tires. I really like my snow tires. I can go run down Subarus with those.

Dr. Hess wrote: Humm. Well, I can't really take a a/f when I'm cranking. It's a heated narrow band sensor, but it still takes a while to start getting readings. I've looked at a few maps of other people's MSQ's, besides DIYAutoTune's PnP Miata of 3.1 at 160 to 7.3 at 20F. Another map I found that a guy in MI (where I hear it also gets cold once in a while) had good results with was 4.0mS at 160, 18mS at 20F. I don't recall the engine he had; I'm just looking for a pattern.

Can you rig it to be on independantly of the MS?

Also, can you find a WB sensor handy? Those are key to tune starts and open loop operation. (so important that I've spend close to $250k a decade ago for 100 of them here at work, and those were the cheap ones...)

dan- I would suspect that you, too, don't have enough run fuel.

I really think people are stunned how much more fuel a cold engine needs, especially on crank. Not only is there lost fuel issues, one has to look at a fuel distillation curve to see that no fuel will evaporate under about 20F, so all the fuel burnt is either entrained due to motion OR heat of compression.

More fuel, my friends.

Eric

I agree, your cranking pw will be getting big starting around freezing and it's definitely exponentially larger below that.

A few things that are important but sometimes overlooked:

Battery size/condition, a slow cranking engine is much harder to start After start, your cranking might be enough to make it catch but this needs to be right, independent of warmup to keep it there Idle air, sometimes the air you need to idle at 1k cold is about double what you want to get it to initially fire. In other words don't try and pedal it to start ;)

Another thing to remember is that if you think you have too much fuel and and think it's flooding, just crank it with it floored. If it is too much it'll try and start. If it's too lean, it'll just dry crank.

A quick check to see if you are too lean in crank, without an A/F meter, is to add a load or a transient. If you crank and then go to the high cam (ie the rpm spike from the manifold volume) as you are on the down hill slope of that if you introduce a load (ie A/C hit, power steering lock, or a good old load check on the Alternator) if you flame out more then likely you are too lean in crank. Yeah it is cheap and dirty, but it beats dropping 8-10K on an ETAS meter

Again I am not an MS guy, but what is your triggers for crank, and the exit conditons to leave crank conditions?

My cranking RPM is set at 300. That is, if the motor is spinning over 300 RPM then the MS uses the run fuel maps and not the cranking maps. Engine cranking speed with a good battery and warm weather is around 250.

I'm not sure I understand your quick check with the load. You mean I should crank the wheel (power steering) and turn the AC and lights on while cranking? What am I observing to determine a too-lean condition?

It wouldn't start this morning. Last night, after I put the battery on charge it fired up fairly quickly. This morning, it spun about 10x and that was it. It warmed up to 25F this morning from 17 last night. New battery tonight.

The thing that makes me think I've been too rich on cranking is that it tended (when cold) to start easier if I held the throttle open some. Not floored, as that puts it into flood clear.

I just found a setting I haven't noticed before: First Start Enrichement. With it set to On, you turn on the ignition, step on the accelerator to past the floodclear point 3 times, then crank and it will dump extra fuel on cranking at whatever percentage you set. I think I'll turn that on and set it at 25%.

DOC, it sounds to me that the battery doesn't have enough cold cranking amps. Before you start messing with your settings(it starts fine when the temp is warmer), I'd change the battery. My Bimmer has a cold start injector to richen the mix when the temps are colder. Your MS should be fine as is.

Dr. Hess wrote: The thing that makes me think I've been too rich on cranking is that it tended (when cold) to start easier if I held the throttle open some. Not floored, as that puts it into flood clear.

Except most systems will add fuel if you slightly add pedal. May want to check that. I bet it adds more fuel than any air (since you crank at atmospheric).

I still say more fuel. What I've seen people do tends to be way too little.

Our goal here is to fire on the very first injection, not keep adding fuel. It takes a BUNCH of fuel to fire on the first injection.

(BTW, a lot of the data that I posted isn't as extreme as it could be- if you knew what I'm working on right now, you'd know why, but I'm referencing data on a current project- which isn't exactly the fuel system you are using....)

Eric

With a 48-degree temp swing over a relatively short time, I'm wondering if you had a good old fashioned low-tech fuel separation/gas line freeze. A hot battery and some time in the sun helped, especially since it started right up when you hooked up the batt.

The key phrase is, it always started for you eventually but now it didn't. the temp swing is a pretty rare event. A quick freeze will separate the water out since it freezes faster than the other molecules in the gas. That will happen when the temp plummets.

Well, near as I can figger with the MS (and I haven't dug through the source code looking at it), more pedal isn't going to give me more fuel cranking. Maybe just a tad of Acceleration Enrichment if you stomp on the pedal, but I don't think that kicks in during cranking either.

I'll try boosting the PW's up for cold after I put the new battery in and see what happens.

I have a cold start injector on the MS'ed Truck (22R ->22RE), but I never hooked it up. The Truck starts pretty well in the cold. It's that finicky high performance motor that gives me grief. I'm also still using the MAP for engine load, and that's a bit tough with the 20 V and ITB's. Not much vacuum change over a big load change. If I had to start over on my maps, I'd go with straight RPM and throttle position for my fuel maps on that one. That's how I have the Sportster running and it does great.

Damn, you go grab a coffee and you are 4 messages behind on this! Hot tread!

OK first, the load, you are adding load however you can. I am not sure what you are equiped with in the car, but any of the mentioned "hits" a/c clutch engage, a Powersteering bump against the stops, or a heavy alternator load ( I have a load testers I hook up) will give the engine a higher power load then expected and will show if your fuel calibration is stable. If you falter, then more often then not you don't have enough fuel to reject the high load. I am assuming that spark is a constant in crank . (?)

Also if you are just using an rpm crank window, I would suggest going a tad higher. IF you spin with the starter at 250, and are set at 300 moving to 350 is not going to effect anything at 70F air, but will make a huge difference as you are in cold air ( < 0F). I seem to recall it is milli seconds at 70F and more like 1-2 tens of a second in the -10F ambient. Can you change the rpm crank exit with air charge temperature?

Also how does MS transition from crank to run? Might there be a flag or gate between crank and run?

pkrstr8 wrote: Also how does MS transition from crank to run? Might there be a flag or gate between crank and run?

Actually, how does it crank? All at the same time until it fires, or in the two batches until it fires?

If it's a "trickle" amount, you generally don't need as much fuel as what I mention to get it running, say numcyl/my amount. Basically, you want it to fire after numcyl injection events. Much more than that, and it may not fire.

The issue is that you can get into an equalibrium of injection/rejection out of the exhaust- so more is a good thing. You want the fuel to puddle up on the back of the valve until it's the right amount, and then fire.

It's still a LOT of fuel. Again, much more than most people think to get it to run.

Eric

Well, theoretically anything is possible with megasquirt. The problem is that if you go outside the published code, you are totally on your own as far as writting the code in assembler, etc. Assemblers are out there and free. I had to recompile the code for the Sportster because the injector drivers were not inverted on my board and they are inverted on the conventional (store bought) MS'. Bottom line, though, is that there is no setting to modify start RPM based on temp. I also don't see how that will make a difference in cold air. If my crank speed is 250 RPM, I'm set at 300 as the trigger to move over to the VE table instead of the cranking table, how is changing that to 350 going to get the motor to fire that first time at 10F, which is my current problem? I can bump that up to 350 or even 500. Idle is around 1K. Spark is constant in crank. I have 2 Camry coilpacks and run waste spark. I believe there is a flag (variable) somewhere in the program to indicate Running. In MegaTune, a Run indicator comes up when you are running, so I assume it is reading a variable there somewhere which is set depending on the RPM signal. I really don't like to tackle assembler if I don't have to. MS2 code is in C. MS1 Squirt-n-Spark Extra (which I'm running) is in Assembler.

Anyway, so I can add load when cranking with the methods described, especially the AC because my AC clutch is hotwired and doesn't use the built in control circuit which disables it during cranking (long story involving the tach signal, flyback voltages and lack of time to mess with it). If I understand your suggestions, the car has to start with the existing map, but you add the bigger cranking load and it doesn't want to start and that indicates too lean?

Right now it isn't firing at all. In the past when cold, I'd crank it a while, then stop, then try again and it'd fire, if that's what you mean, Eric. Otherwise, I'm going to assume it is in batches and not all 4 injectors at once. My injectors are wired up in 2 banks of 2. The priming pulse probably hits all 4 at once. The motor has to go around once for the MS to figure where it is to determine which coilpack to fire. The timing pickup is off the distributor hanging off the back of the exhaust cam with 2 teeth removed out of 24, creating a 12-1 pulse pattern.

Here's a pic of the Megasquirt derived computer I built for my wife's Sportster. http://www.drhess.net/images/bbms3.jpg The HD collector's cigarette package is shown for size reference only and not to be construed as a smoking endorsement.