Winter Upgrade - Megasquirt

[bobcowan]

Update:

I finally got the wiring harness completed. The ECU uses a DB37 connector for the main harness, and a DB15 for the ignition and a couple of extra features. If you're not familier with them, it's very similar to the serial connector in the back of your computer. It's a very small connector, designed for 20-22g wires.

I started out with a solder connection. But, apparently, my soldering skills were not good enough for that (which is why I didn't build the ECU). So 4 of the injectors (bank 1) were wide open all the time. That was a bad thing.

I bought a new connector and a crimping tool, and replaced the DB37 with a crimp type connector. That worked out really well. But those connectors are really tiny! I had to wear magnifying glasses to get that job done. But that worked out a lot better.

Spent the last month struggling with a sync problem. The computer had trouble reading the crank trigger wheel. Since it can't tell engine position, it wouldn't run well. I tried everything I could think of. I bench tested every component, and it all tested out fine. The darned thing still wouldn't run. I was so frustrated I was ready to pul the whole thing out and go back to my Haltech.

Out of desperation, I replaced the hall sensor - $57. That was the (main) problem! A brand new Hall sensor was not working right, and feeding a bad signal to the ECU. Unusual to have a bad component right out of the box, but it happens.

That cut my sync problem down significantly. But it was still there. After another week or two of messing around, I opened up the ECU and made a couple of potentiometer adjustments on the main board. Sync problem seems to be completely resolved.

Last couple of evenings have been spent doing some fuel map tuning. Still on the rack, hasn't left the garage yet. It really is pretty much self tuning, and that feature seems to be working well. Tonight I plan on driving up and down the street a few times to see what happens.

Currently, it's fuel only. I have not installed the ignition yet. If tuning goes well today, I'll add ignition this week end.

Once I get those maps pretty well done, then I'll start on the E85 map.

It's been a bit of a struggle, and has taken a lot longer than I expected. I'll let you know how the tuning and ignition install goes.

[Luce]

Good to hear it's working. I thought the hall effect pick ups side stepped all of the VR problems. In fact, read through the docs about using the hall sensors. You may need to cut and reroute a couple of jumpers. I don't think the hall is supposed to use the VR conditioner circuits the pots are associated with. I believe the hall sensor should be hooked in where the VR conditioner circuits hook to on the board. You may be amplifying an already amplified signal and causing sync errors. Did the assembled ecu docs say?

And oh yea. Packing 25 or so 16ga wires into a head shell designed for 22ga to 28ga wire takes a little carefully applied force. I'm still tempted (and if any gremlins arise, i will) to buy one of the DB37 pigtails diyautotune offers.

I'm working on getting the rest of my car drivable before I proceed on the tuning. Glad to hear the selftune part works well. I boned up and bought the upgrades for the tuning software, but haven't yet hooked in the O2 sensor. I still had exhaust leaks, so it wouldn't be accurate yet anyway.

Edit. In the megamanual MegaSquirt - Electronic Fuel Injection Computer by Bowling and Grippo ©2010 Assembly guide (I'm only familiar with V3 that I'm using) for the V3 item #52 you have to select VR OR hall. It tells you what jumpers need to be connected. The hall sensor connect directly to the opto isolator. The hall sensor shouldn't be ran through the VR conditioner where the pots are. The megamanual is a lot to slog through, but check this out.

[bobcowan]

I had the guys at DIY build the ECU for me. I didn't think my skills were up to that job. When I got it, their documentation said they installed a VR circuit, and they sent me a Hall sensor. I asked them about it, and they have found that the Hall works better when used with a VR circuit. I searched through the boards, and sure enough, I found that was the general concensis.

I'v been thinking about this sync loss issue since last night. My plan for today was more on road tuning, see if it was a heat issue. But we had a surprise snow storm last night, so that will have to wait. Today I'll decrease the air gap from about 1.25mm to about 0.75mm. Since I can't drive it today, I'll take the dash apart again and adjust the R56 pot to 2.5 turns.

If that doesn't work, not sure what I'll do. Probably keep the standard ignition system and run the ECU off the negative side of the coil.

I took a close look at the commercially available wiring harnesses. If you're using it with the MSII or MSIII, it works well. All the labels and pins matched up properly. But I'm using the MSII/Extra, with an added DB15 on the other end of the case for ignition control. I would have to modify the plugs and the harness, and then none of the labels would match.

We have a surplus electronic store here in town that's a real treasure trove. I got all the wire I needed for about $10. I spent another $50 for a crimping tool and the DB37 connecters, hoods, and pins. I now have extras that I'll probably sell on e-bay. My only regret is that I should have used more 20g and 22g wires than I did. It wouldn't work any better, but it would have made the job easier.

One of the big problems with the Megasquirt is the documentation. If you thought the FFR build manual was tough to work with, you're in for a big surprise.

[bobcowan]

Well, no joy in mudville. I thought maybe it was a heat issue, as the last two times it occurred after the engine had been running for a while. Today it started within 10 minutes. Air temp was about 25-30*, coolant temp didn't go over 100*. So, it's not a heat issue.

Then I noticed that as I sat there with the key on and engine off, I was still getting a trigger signal. The rpm's would jump up to 20, 80, 210, 30. Sometimes the trigger would go on long enough to trigger the fuel pump relay. With the engine off.

When I unplug the hall sensor, it all quit. If I pull the sensor away from the whel, it also quits. I'v come to the conclusion that these expensive hall sensors are cr ap. Not sure what I'll do now.

Just to get the car going, I think I'll switch to a neg coil trigger, and continue to use the mallory for ignition. That means I'll have to take the dash apart and reconfigure the ECU. Again.

I didn't expect this to be a turn key project. But I did expect the individual components to actually work as advertised.

[Caprimaniac]

Is the hall- sensor shielded/ grounded as it's supposed to? Is cable routing up to standards? Check everything, because these sensors usually "work" and are accurate... Unless there's some kind of noice/ interference, which will mess up things.

There's something wrong somewhere, I bet.

Good thing your car is running already! I've done the cable & ECU installation/ routing. I will have to get the engine together & into the bay before I do the last adjustments and eventually fire it up....

[bobcowan]

I'm using shielded cable supplied by DIY. I followed their instructions on routing and wiring. The voltage seems to be pretty steady at 4.93v as supplied by the ECU; minimum is 4.75v. Air gap is 0.75mm. After mulling it over, here's my plan:

- check the harness, and make it didn't get damaged during install, or is lying on top of the starter motor, etc.
- play with the air gap on the hall sensor. I'll keep increasing it by 0.25mm until the noise goes away, and then see if I lose any signal.
- wire the sensor directly to 12v switched power.

Yes, I'm stubborn. Before I dump it completely, I want to assure myself that I'v tried everything I can think of.

[Luce]

Before you go too far into changing things, read up on adjusting the VR pots. One is for hysteresis and one I think is sensitivity. If it's getting extra signals, you may need to tweek up the hysteresis. They also have a few suggestions on pull up or down resistors and such to tame noises. You may also consider changing the jumpers to bypass the VR circuits altogether.

The VR is an AC signal and the hall should be 0 or some voltage. I think the VR circuit is seeing a very slight dip to neg output on the hall and makes it trigger when it fluctuates between -.01 volts and .01 volts.

If I was in your shoes, and decided it was time to try something different, the first thing I change would be the jumpers in the ECU to bypass the VR circuit. At least your not opening up the harness.

Hope this helps

[Tony Ripepi]

Hi Guys,

Been skulking around this thread and one of the local guys has been installing the MS system on a 347, he found and error in the harness that the MS guys have acknowledged, it makes the link to the ECU ineffective. Before you go much further in blaming your process or installation, reach out to the DIY/MS folks.

TR

[bobcowan]

I didn't buy their harnesss, I made my own. But I followed their instructions. I got a message from Matt at DIY this morning. And he told me not to follow the MS instructions, but told me to wire the Hall sensor a little differantly.

Previously, the instructions said to feed the sensor through the ECU, to get a 5v signal. Today Matt tells me the sensor should get a 12v switched feed. Fortunatly, I had a spare wire in the engine harness in case I wanted to do something like this in the future - something I did when I wired the car 7-8 years ago.

So, tonight I spent about an hour or so redoing the harness in the front of the engine. It really wasn't that big of a deal for me to do it. I could have done it in about 10-15 minutes, but I decided to take the whole thing apart, check all connections, re-wrap, and re-tie everything. Probably not necessary, but it made me feel better. By comparison, opening up the ECU and putting it all back together again takes about 90 minutes.

A VR sensor is simply a magnetic "switch". As the ferrous object moves in and out of range, it trips the "switch" and sends an analog signal. The analog signal goes in to the ECU, and through an additional circuit the signal is then modified to a clean digital signal.

A Hall sensor is a VR sensor with a microchip attached. The microchip uses power to modify the analog signal to a clean digital signal. In basic terms (that even I can understand), the additional clean up circuit is part of the Hall sensor instead of the ECU.

According to the people who built the ECU (and the people who designed it) experience shows them that the best results come with a Hall sensor on a VR circuit, without a pull up. So, that's what I'm using.

I did some more reading today, and have a couple of ideas. The sensors (VR and Hall) are quite sensative to the air gap. The instruction manual says it should be 0.75-1mm. Other sources say 0.5mm. DIY recommends 1.5mm. Again, some conflicting instructions. I'm going to try moving the gap up and down until I get a clean signal. I'll leave it where the signal is best, regardless of what the actual number is.

It was too late to run the engine tonight, but I'll try again tomorrow.

[Luce]

Actually, the difference is a little more fundamental between hall and VR. A VR is simply a coil of wire with a magnet and toothed wheel, the output is an asymmetric A/C signal that varies frequency and voltage with RPM (like a generator). The Hall is a transistor that is sensitive to magnetic fields and it turns on and off with the trigger wheel. It's output is more of a square wave that toggles from high to low. It never goes negative like the VR.

None of that means a hall sensor can't toggle the VR input, but it has to be adjusted to look for the right signal inflections.

I'm using the VR in a MDS dizzy. I had similar problems at first. In order for MS to see the tach signal at a slow enough RPM, I had to turn down the trigger point on the VR circuits to the point I was getting extra trigger events too. My solution was to close down the air gap (increases the output voltage) and turn up the trigger point so it was enough above the noise level to not give false triggers.

For your case, if the hall can take the full 12V (I would expect it could) that will raise the voltage of the high signal. Another problem I had was I installed one of my pots backwards, so when they say turn clockwise, I have to turn counter clockwise. hat wasted about 2 hours figuring out. Unfortunately, I'm guessing there's no easy way for you to run the trigger wheel without cranking the engine. I had it easy. I hooked the dizzy to my cordless drill and could run it from 0 to 3600 RPM (would be 7200 crank RPM) with the squeeze of a trigger.

So, if you up the supply voltage to the hall, run it's gap on the tight side, that will maximize your trigger signal. Doing that will allow you to adjust the VR conditioner to trigger at points high enough to filter the noise signals that are now triggering your MS. Check this. Adjusting the V3 Main Board VR Input Circuit Potentiometers

[bobcowan]

I'v done everything I can think of, and anything anyone else can think of. Instead of getting better, it's only gotten worse.

After rewiring the hall for 12v, changing air gaps, adjusting pots, checking wiring, chanting prayers, and sprinkling it with holy water, I gave up on the Hall sensor. I'm never going to get that to work.

So I reconfigured the board, tossed the Hall, and connected the trigger to the neg side of the coil. Nothing. now there's no trigger at all. Not even a little. Won't run at all.

I'v wasted 5 months and over $1K on this POS. I didn't expect this to be a bolt on project. But I did expect it to actually work. I built a car, this should have been possible. It sure looks pretty, though. Maybe I can sell it as modern art.

Unless I get a better Idea in the next 12 hours or so, I'm going to rip the whole thing out, and put the "old" Haltech back in. I can probably have it up and running again in a week or so.

Maybe next winter I'll take another look at something like BigStuff3

[Tony Ripepi]

HI Bob,

Just starting into a project to retrofit an old Tec3 EMS with the MS system and a TWM intake manifold. One of my friends has the MS installed on a new 347 with TWM, it is not running but the initial up front bread boarding is complete. My project will be on a windsor 410 with a new TWM.

Almost ready to install on the 347 engine stand and fire for the first time. The process seems to be simple starting with all new parts, I will update here as we move forward. I am surprised by your conclusion to abandon the MS and looking at the details of moving forward very concerned.

As opposed to abandoning the MS, have you considered any other alternatives, you seem to be very close?

Tony R.

[Luce]

Bob, are you running some sort of CD ignition box? If so, you can't hook up to the coil neg. CD boxes output up to 500V to teh coil. Hook to the CD box tack output instead.

Don't let such a simple (yet understandably frustrating) error totally put you off. All you're missing is some way to trigger MS accurately. Unless you've smoked the VR circuit with the CD box, all I think you're missing is some pot tweeking. You may need to seek out someone with an o-scope to get it straight.

Are you using a missing tooth wheel with the hall? Also, when going from a missing tooth to a tack signal, you have to go into the trigger menus and select the correct one. If MS is looking for a missing tooth, it'll never find it on the coil or tack sig, so it'll never sync up to the motor and start injecting and firing.

[bobcowan]

There's no CDI box in the system - I removed that when I installed the MS. There must be soemthing wrong with this ECU. I actually had it running for a couple of hours. I did a base line tune in the garage, and got things kinda close. Then I drove it around the neighborhood for about 45 minutes, using the computer to tune the VE map. That was going pretty well, and the below 3500 rpm maps were feeling pretty good.

All of a sudden, a sync error showed up and I almost didn't get the car into the garage. Since then, the sync error has gotten worse. Now there's no trigger signal at all, and (of course) the engine won't even fire.

The problem is, I'm out of time. I'v been working on this project since October. The first race of the season is only 7 weeks away. I'm going to pull the system out, and send it back to DIY for testing. It will be two to three weeks before I get it back. Then, if it still doesn't work I'm left with 2-3 weeks to get the car ready for the track.

Maybe I'll try again next winter. The BigStuff3 is looking better and better - although I still don't like the price.

[Luce]

What type ignition are you running now? Points? If your trigger wheel was a missing tooth and you went to neg coil connection, you will have to change the trigger in the setup menu of the tuning software. I don't have any of the tuning software installed on my work computer, so I can't walk through it and see exactly what the options are, but I remember there are a lot of options on the software side. If you're going to trigger off the dizzy anyway, I would suggest pulling the dizzy and running it with a drill until you get the trigger issues cleared up instead of wearing out the ring gear on your flywheel.

I'm far from any expert, but I have been wanting an excuse to install one of these for 3 years, and have studied this thing a LOT hoping to set everything up the best possible way for my purposes and side step as many problems as possible. So take everything I say with that understanding. Also, my car isn't ready to drive, so all I have any experience with is setup and starting the engine and getting a decent idle. My initial problems with getting the VR to trigger reliably had me considering a points and condenser dizzy and using it and a pull up resistor as a trigger. I would have to assume the hall sensor is the best considering the whole automotive industry went points-VR-hall (plus a few people trying optical somewhere in the middle).

[bobcowan]

Right now the ignition is still a mallory distributor and MSD coil. I didn't want to touch the ignition system until I get the fuel dialed in fairly close.

For a trigger, I'm using the supplied Hall sensor and a VR conditioner circuit. It did not come with a pull up resister, and the shop I bought it from did not recommend one. But, on advice from people on the MS board, I installed one anyway. It's cheap ($1.38) and easy (15 minutes), so I figured Why Not?

The engine started, and seemed to idle OK. But it was late, so I didn't run it more than a minute or so. The issue I'v been having is that I can get the engine to start and idle, and I can get it tuned to about 200-2300 rpm's. Any more than that and the trigger noise overwhelms the circuits and I get a sync loss.

I'll try again tomorrow

[vector1]

might be noise off your alternator, you have it hooked directly to the battery and all other precautions taken with the wiring?

[Luce]

My understanding is the hall puts out enough of a signal to directly drive the optoisolator without the VR conditioner circuit. The VR conditioner greatly amplifies the signal, and the induced noise as well, to the point that MS thinks the noise is a trigger. But it's where MS knows there shouldn't be a signal, so it loses sync. That's why when the engine was sitting still, you were getting little blips of tach readings.

I don't know where the idea to run the hall through the VR circuits came from other than someone realized it would work (sometimes) and made it easier to market the MS as a one size fits all kit. I see no other reason for it. Since you opted for the assembled and tested ecu, I'm assuming you're not comfortable with changing the jumpers, but that would be my first attempt at a fix. If you dig into this, make sure the optoisolator, U3 is present as well as D1, D2, R12 R13, C12 and C30 are present. Also make sure you have all 3 jumpers in place.
To try this approach-
TACHSELECT needs to be disconnected from VRIN and connected to OPTOIN.
XG1 and XG2 need to be connected.
TSELECT needs to be disconnected from VROUT and connected to OPTOOUT.
Be sure you can locate all 3 of these sets of jumpers and that you do have a quality low wattage soldering iron and the skills to use it. This is a fairly densely populated multi layered board, and you can ruin the whole thing.

If you still have noise issues, disconnect XG1 and XG2 and use a vacant DB37 pin and run a wire from XG1 to that vacant connector pin in MS box, and run the wire to the ground on the hall sensor. If possible twist the hall signal and ground wire together. This eliminates most of the induced noise.

In my opinion (again, take it for what it's worth...the musings of just another internet troll) using the VR circuits for anything other than a VR pickup is like hooking the speaker output on a home stereo to the microphone input on a tape deck to copy your buddies albums to tape. Yes, if you turn the stereo volume down to the absolute minimum while recording, you may be able to make out what song you're hearing, but it will be far short of enjoyable music.

If someone can present an argument for the merits of using the VR circuit with a hall sensor, I would love to hear it.

[bobcowan]

Originally Posted by vector1

might be noise off your alternator, you have it hooked directly to the battery and all other precautions taken with the wiring?

Good thought. 2 weeks ago I removed the belt to stop the alternator from spinning, and ran off battery only. Didn't help.

Originally Posted by Luce

My understanding is the hall puts out enough of a signal to directly drive the optoisolator without the VR conditioner circuit. The VR conditioner greatly amplifies the signal, and the induced noise as well, to the point that MS thinks the noise is a trigger. But it's where MS knows there shouldn't be a signal, so it loses sync. That's why when the engine was sitting still, you were getting little blips of tach readings.

I don't know where the idea to run the hall through the VR circuits came from other than someone realized it would work (sometimes) and made it easier to market the MS as a one size fits all kit. I see no other reason for it. Since you opted for the assembled and tested ecu, I'm assuming you're not comfortable with changing the jumpers, but that would be my first attempt at a fix. If you dig into this, make sure the optoisolator, U3 is present as well as D1, D2, R12 R13, C12 and C30 are present. Also make sure you have all 3 jumpers in place.
To try this approach-
TACHSELECT needs to be disconnected from VRIN and connected to OPTOIN.
XG1 and XG2 need to be connected.
TSELECT needs to be disconnected from VROUT and connected to OPTOOUT.
Be sure you can locate all 3 of these sets of jumpers and that you do have a quality low wattage soldering iron and the skills to use it. This is a fairly densely populated multi layered board, and you can ruin the whole thing.

If you still have noise issues, disconnect XG1 and XG2 and use a vacant DB37 pin and run a wire from XG1 to that vacant connector pin in MS box, and run the wire to the ground on the hall sensor. If possible twist the hall signal and ground wire together. This eliminates most of the induced noise.

In my opinion (again, take it for what it's worth...the musings of just another internet troll) using the VR circuits for anything other than a VR pickup is like hooking the speaker output on a home stereo to the microphone input on a tape deck to copy your buddies albums to tape. Yes, if you turn the stereo volume down to the absolute minimum while recording, you may be able to make out what song you're hearing, but it will be far short of enjoyable music.

If someone can present an argument for the merits of using the VR circuit with a hall sensor, I would love to hear it.

More than one person recommended the VR circuit and the hall sensor. There are a lot of ECU's out there set up that way, and they seem to be working fine. I only know enough to be dangerous, so I take their word for it.

Speaking of dangerous, me getting anywhere near that board with a hot iron would be dangerous. I can move pins around, solder wires together, etc. But my skills are not near good enough to come near this SMT board. That's why I had them build it.

The hall sensor should not need a pull up resister. But, out of desperation, I put one in anyway. More than one person recommended it, including an exclusive builder. It seemed to make it better. We'll see this evening.

[Luce]

Another option would be try using a dizzy with a VR sensor. A stock one should work, and let MS run the coil from the single coil driver that should be installed. Don't forget to set the options in the setup menu so MS isn't looking for the missing tooth.

You need something to drive the oil pump anyways, right? Unless you're dry sumped.

I'm running an MDS distributor that has a VR triggering the MS. I use the coil driver to drive a 6AL box, and I have a bypass plug so I can make the coil driver directly drive the coil if the 6AL crapps out. So I have the advance curve of the dizzy, but I can add or subtract a few degrees with the MS. But my setup can not run individual coils because MS has no idea what plug to fire.

Another set up I've read people successfully use it to run a full factory edis system, and let MS control the advance through the SAW and PIP signals. I think this is as good as the MS 4 coil setup I think you were after performance wise.