bliffle

Put a freshly charged battery in and tried it again without luck, whether the AFM was connected or not. Not a hint of firing, now. Maybe that means that any stored fuel in the FI system has been exhausted.

I get a very faint whiff of gas, but I think that proves nothing.

Found my old fuel pressure gauge but it only goes up to about 10psi. I'll see if I can rent one. Then I have to connect it, and it looks like the proper tube is the outermost on those two that are paired up on the drivers side of the firewall. the manual shows the gauge attached to the metal pipe on the firewall, but that's so crowded and cramped looking I'd rather attach to the neoprene tube up on top where it connects to the pipe that ducks under the solenoids on the drivers side. Even if I have to make a short length of tube to connect two hoses.

I couldn't find what the flow should be from the pump, but I'll look at the FSM again.

I've gotta re-check coil primary voltage, just in case I've regressed something. Maybe secondary too. Yeah, why not, I've got a timing light.

bliffle

I found a fuel pressure gauge at Kragens for $50 complete with appropriate adapters, etc., hooked it in and got nominal readings per the FSM (60 psi actual vs. 50-70 nominal with gauge right on fuel pipe, 42 psi vs 30-50 nominal teed into fuel line). So it looks like the pump is OK. It filled a quarter of a pickle jar in just a few seconds.

Also, checked coil voltage, getting 12.5 v. at each coil in ON, about 10.5 while cranking.

Doesn't fire. So tomorrow I'll check spark secondary with timing light.

Fuel, compression, ignition. If timing light is good tomorrow, that just leaves compression. Maybe the walls are soaked with gas. I'll pull plugs and check for gas. Also, put in some good Autolite 2626s!

Siraniko

iTrader: (13)

Im too lazy to read the entire thread. before you even start the engine, connect a wire from the borrower to both trailing and leading coils + terminal. This is to ensure that you will have spark provided that the igniters are in good shape.

bliffle

"Borrower"?

I think I'm getting good voltage at the coils, altho it drops to 10.5 when cranking.

Siraniko

iTrader: (13)

I was working @ home while postwhoring. that should be battery, not borrower. I guess everyone now knows I work in the finance industry.

bliffle

I'll try that, I've got plenty of test leads around here, but I wouldn't expect a change since it appears that I already have sufficient voltage at the coils.

I'm going to pull the plugs and crank it over to see if I get the right kind of puff-puff-puff from the lobes. And to see if it ejects a lot of surplus fuel. Even with the plugs out my electronic timing light should show if timing is correct (I have no good reason to expect it to be wrong because it had run in short spurts before, and I can't account for what would have changed the timing).

I'll probably spray some MMO in the plug holes to see if that changes anything.

I'm going to hypothesize that the fuel circuit is OK up to the engine compartment, so the fuel pump and filter are exculpated (for the moment). That leaves the fuel system in the FI system and ignition. I have no supporting info for thinking compression is kaput, but I'll try to verify that while the plugs are out.

I've got to wash the poor thing today with a bucket and hose. It's accumulating a lot of dust from sitting around, and must look rather tatty to the neighbors.

I've been assuming (and I may be wrong in this, so lemme know if you know better) that when the ECU decides, in it's wisdom, to kill the engine that it shuts off the ignition, not the fuel. Fuel is only cut by that gizmo that detects upside-down condition.

I'm also going to cycle in some different ignitors because i can imagine that an ignitor could pass our primitive DC test under static conditions but fail when things are changing in dynamic operation.

Could be coils, too. I believe that if the leading coil failed that would account for this. I can probably just do a substitution test with a coil from another car, and/or swap leading and trailing. I infer that if the trailing coil failed the car would just run a little poorly and the tachometer would fail (if the fault is in the primary). As I recall, there's some kind of primitive coil ohmmeter test so I'll measure that, too.

But first and easiest today, I'll check ignition secondary with the timing light.

It would be nice to have a diagnostic computer for this stuff. I see that "Pico Technology" has such a thing for about $2000, which is a lot since I doubt that it does much for a 23 yr. old car. But with the proper probes connected to a USB a DIY guy could readout static values and even to a 'scope display for such things as coil voltage, spark voltage, injector voltage, etc. I would think that a couple hundred bucks of cannibalized stuff from cheap Harborfreight timing lights and some acupuncture probes, etc would do it for the hardware, and I bet there's some freeware or open software for the PC side. I think I'll post a query in the tech section (would apply to all generations of rotary) to see what develpos. Then a guy could hook that up to his carputer and have 24/7 diagnostic readout! It may be that would require an aftermarket or hacked ECU, though, which is apparently the way that FD guys do it.

gsl-se addict

iTrader: (3)

When you shut off the car, the primary kill to the engine is shutting off power to the coils, etc. When you turn off the key, you kill power to the main relay, that in turn kills power to the coils, ECU, injectors, etc. The SE ECU does not have the ability to decide to shut down the engine on its own. As long as it gets power and the signal from the tach/trailing ignition, it will try to fire the injectors. It might squirt too much or too little if there is a problem with one of the sensors, but will still try.

If the main relay was acting up, that could intermittantly kill power to the coils/injectors/ECU. The main relay is a dual pole setup (really two relays). It is possible that when the car dies, you could be getting power to the coils, etc. but the other half of the main relay is cutting out. The 2nd side powers part of the ECU, AFM, fail check connector (for testing ECU codes), etc. You can see the main relay on pg. 50-16 of the FSM. Maybe you could try to hook your multimeter to the various outoputs of the main relay and see if the power drops out when the car dies?

It is really strange. There are at least 3 SEs on here all having this same problem (or at least a very similar problem). I'm going down to help cjf in north carolina with his SE maybe next weekend. We'll see if we can figure anything out.

Kent

Rx-7Doctor

iTrader: (13)

If the connectors at the coolant temp sensor are not tight and making a good connection the engine will not start. The CTS is the only sensor on the engine that if not connected the engine will not start and you will just have a flood situation.

I just don't understand it. Because of your description of the symptoms with the engine firing then running for about 10 seconds then dying only leads to a couple of things.

1. The fuel shutting off or not enough pressure/volume, which you have performed the test and verified that it is within specs.

2. Bad trailing ignitor. Which you have swapped them out with no change.

3. Unless the flap door on the AFM was bent from a backfire situation it would not give your symptoms. Like I stated you can unplug the AFM and the engine will still start and run without shutting off.,

4. The coolant temp sensor if bad generally goes to the rich side and may just cause hard starts but would not start then shut the engine off it was bad. If the connection was lost as stated above from the connector to the CTS then the engine would not start and you would have very wet plugs.

5. The ECU generally if gets damaged from an electrical spike or water for the most part will just not allow the engine to start and you would be able to perform the pin test and find faults there.

6. I believe at this point you probably have her flooded due to the low voltage of the battery. You are going to have to de-flood her and go back to testing.

I recommend that after you get her de-flooded that you hook up your fuel pressure gauge and install either a inline spark tested (pick up one cheap at harbor freight) then with the assistance of someone, crank the engine and see if she fires. If she fires then dies you can see if you are losing spark (would put the checker on the trailing side) because if you lose the trailing side the fuel pump cuts off. If you lose that signal then your issue is there in either the ignitor circuit, coil wiring or maybe main relay area. If you don't lose the signal from the trailing side and it dies then it is possible that the main relay is the culprit.

When you open the hood and are standing against the driver side fender and looking straight down on the main relay. There are 2 connectors to the main relay. The one on the right as looking down if there is a problem in those connections then the engine will not start. The connector on the left (as staring down) will only cut off a portion of the fuel injection system but the engine will still start and run.

I recommend that when you are de-flooding that you connect a jumper vehicle with the engine running to get the most amount of cranking amps to spin the engine so that it pumps all of the fuel out. I take a white piece of paper and hold it in front of the plug openings and crank until there are no more spots forming on the paper.

bliffle

Aha! With Wackys help I found a Thermo connector for $7 from NAPA. It's not easy to find in their books, but the NAPA number is 2-17427, maker = Echlin, and there are 14 more in fresno.

Rx-7Doctor

iTrader: (13)

Wacky's help. It was I that supplied you with that information.

bliffle

You're right. Sorry to get the attribution wrong.

I started a diagnostic thread on the General Rotary Tech forum, here:

https://www.rx7club.com/showthread.p...05#post8240405

Thanks.

bliffle

Oh crap!

I just went out with my multimeter and tested the voltages on the ECU with the key at "ON" (with a cold engine) , as instructed and found a hatful of anomalies.

Here are the exceptions to the nominal values on p. 4B-70 in the online FSM:

terminal_____nominal value_____actual value_____Connection to
-------------------------------------------------------------------------------------------------------
C____________1 - 2 v.____________5,5 v.__________water thermo sensor
E____________4 - 6 v.____________0 v.____________Air flow meter
L____________0 - 12 v.___________4 v.____________Variable resistor
Q____________2 v._______________.02 v.__________Air flow meter
R____________7.5 v._____________0 v.____________Air flow meter
U____________12 v. _____________0 v.____________ignition coil (T) - terminal
V____________12 v. _____________0 - 4 v._________main relay

h____________< 1.5 v.___________12 v.___________vent solenoid valve

Some of those zeros may have been because it was difficult to get the reading out of the back of the connector.

Looks like several readouts are wrong. Problem is, I don't know what it means

gsl-se addict

iTrader: (3)

Now we are getting somewhere.

C- thermosensor - don't worry about that one. The car is not up to temperature, so that one won't read as expected.

h - vent solenoid - this just means that the TPS is not adjusted properly. It may also be off since the engine is not warm. It should be reading 0v if the engine is warm and TPS adjusted correctly.

L - Variable resistor - that looks fine. That is for the adjustment of idle mixture. It is in about the middle of the range, so that is probably good. Wouldn't cause this problem if not adjusted properly.

V - main relay - that is that 2nd section of the main relay that I was talking about before. If that loses power, that will also throw off those readings to the AFM

U- trailing coil (-) - go back and double check this reading. Also check voltage at the trailing coil (-) directly. If the coil has power, but the ECU terminal doesn't, there is a yellow connector uner the windshield washer bottle. That carries power for the injectors and the trailing coil (-) signal to connect between the engine bay harness and the ECU. However, if this was unplugged, then you wouldn't have power at terminals A, H, or F. So, double check this one.

If you could, go back and check terminal O. That should be the same as V. If O is getting power, but V is not, something is a bit weird as they are supposed to be connected electrically in the harness.

Take a look at the main relay. It is on the drivers side of the engine bay, near the rest of the relays. It is the one with two round parts if I remember correctly. Make sure it is plugged in, check the voltages at the terminals, replace the main relay if you can get another. You should have the following readings at the main relay (key turned to 'ON').

B - 0v (ground) - there should be two wires with this color
BY- 12V - this powers the coils, injectors, terminal A of ECU
RW - 12V - this leads to V and O on the ECU .. might be a problem here
W - 12V - this powers part of the relay (powers BY wire when ON)
WL - 12V - this powers another part of the relay (powers RW wire when relay ON)
BW - 12V - should be 2 wires with this color ... powers both portions of the main relay

Let us know what you find out.

Kent

bliffle

OK. I re-measured and re-tested as you suggested. here's the latest new/revised data:


I re-checked this and after some pushing and shoving of the probe into the back of the connector I got 12 v. at both "U" and "V".

I also checked the actual coil voltage and it's 12 v.

Since both "O" and "V" are reading 12 v. I guess that is answered.

I undid the phillips screw mounting the relay (and two others) to the sheet metal, attached alligator test leads to re-establish any interrupted grounds, and inverted the Main Relay to test the leads underneath. They all read nominal (ignition is "ON").

So now it seems that all the voltages read nominal, as expected.

I don't know what to make of it. Unless there is an intermittent that changes when I push on the connectors on the ECU.

I didn't attempt a start. I just turned the ignition switch to "ON" 3 times to do the voltage tests.

I also attached my LED test light to the test socket next to the ECU and again checked that I only got the initial "ON" light and no further codes.

bliffle

Now I'm worrying about that CTS connector. All that's there now is a hacked connection fashioned from the broken parts of the connector from when I swapped in a new CTS. Now I have a couple of new connectors from NAPA, but the keys and keyways don't match (there are actually two each) so I'm going to have to grind a key off to install it properly. I'm planning to grind off the largest key on the outside of the CTS itself so that the small internal key will pass into the keyway in the new connector.

I think the CTS is just a thermistor and thus not polarized, nor is either lead grounded. So it doesn't matter which key I grind off.

gsl-se addict

iTrader: (3)

Correct. The CTS is a thermistor, so polarity is not an issue.

j9fd3s

iTrader: (3)

the ecu in the gsl-se has no control over ignition. the gsl-se is an evolution from the carburated 12a cars. the ignition is basically the same as the 12a cars, its self contained, ignitors are on the dizzy.

anyway the way the gsl-se works, it gets its rpm signal from the trailing ignition, the ecu triggers the fuel pump also. so if the trailing ignition doesnt work, the ecu sees no rpm signal, and doesnt output anything.

the fuel pump also has a switch in the airflow meter, if the airflow meter is closed all the way, it opens the fuel pump circuit. when the engine is running the afm is open a little. the main fueling computation is done with rpm x airflow meter voltage x coolant temp.

the coolant temp sensor basically changes the mixture based on coolant temp, when the engine is cold, the ecu sees that and out puts more fuel. the gsl-se is primitive, theres no failsafe, like there is with the newer ones, if the sensor is unplgged, the ecu outputs a TON of fuel = instant flooding.

bliffle

So, I conclude, the ECU does NOT cutoff the ignition. It only examines the trailing for a sense of RPM.

The ECU, I conclude, controls the fuel pump and the injectors only.

So, if the ECU is broken, or misled by a bad voltage read, it can cause the engine to stop only by cutting the pump or injectors. So the question is: was my original engine cessation because the ECU turned off the pump or the injectors? Since I could subsequently turn off the ignition then restart the engine briefly (presumably resetting the ECU in the process), what can I conclude?

If the ECU had caused a cutoff, why couldn't I get a code? Does the lack of code suggest the ECU itself failed, not one of it's inputs? Or could it be that something the ECU turns on failed to turn on?

j9fd3s

iTrader: (3)

the gsl-se ecu is faithful, but not very bright. it does NOT have any learning features (which is a plus!) it simply looks at its inputs, does its computation, and outputs. garbage in garbage out

it will only set a code if theres an open or short in the circuit, its not smart enough to figure out if its out of range or not.

try it, unplug the afm, and see if it sets the code.

bliffle

Ha!

I got it running. Running smooth and troublefree. Just like new.

Reviewing the data I figured it wasn't the fuel feed, which left compression and ignition. So I removed the leading (lower) NGK BR8EQ14 plugs, which had a faint smell of fuel but looked almost like new (only a couple hundred miles on them), and put paper over the holes about an inch away. Disconnected all the spark wires. Cranked it over long enough to hear a dozen good strong uniform puffs. The paper had faint discoloration and faint smell of fuel. Nothing excessive.

So, compression is off the hook. You could only wish to have such good compression. And it isn't flooded.

I replaced the two NGK plugs with a pair of Autolite 2626s that I had just pulled out a few weeks ago. Why? Because I LIKE those 2626s! They work when the NGKs fail, sometimes, especially with flooded engines.

I also decided to replace the ignitors (again), one by one. I replaced the ignitor at the top, nearest the alternator, gave it a try and it started immediately. On the first rev.

Bingo!

The ignitor I pulled out was one that tested OK with our primitive little static test. I always wondered if that static test was good enough, and now I believe it is not enough. I think we need a better test. Or else, just keep a supply of J109s so a guy can try one after the other when there is a failure. I've always carried at least 2 in every car because I had my suspicions (being an old electronics guy, including designing and building specialized military receivers) about static tests. Especially static tests of non-linear gizmos like transistors in circuits that may have energy storage like an inductor coil or capacitor, etc. The static condition may bear no relation to the dynamic condition EXCEPT that if static fails then dynamic is almost sure to fail.

Passing a static test at nominal voltages proves nothing about dynamic operation.

But it is possible to monitor something like the ignitor because it is involved in the electrics of the coil primary, so one could at least track normal operation and compare against normal profile for a pathology. In fact, with proper tracking of coil primary voltages (secondary would help, too) this problem would have been kinda easy to shoot.

In the meantime, try another J109, and another, until you're really convinced that both are OK.

Over on the General Technical Forum I queried what other folks are doing for dynamic debugging, and got sortof poopooed by folks.

https://www.rx7club.com/general-rotary-tech-support-11/what-hardware-software-available-pc-diagnosis-760284/

But I think they're wrong. I think I can come up with a rather simple monitoring system that a guy can install a few simple adapters, and then when there's a problem just connect up a laptop with a couple of simple A/D converters and shoot these mystery bugs dead!

I think an ignitor problem can be shot by examining the coil primary voltage history, for example.

What I need is an intercept connector for each of the harness leads that come into the ECU. That's just a male-female pair between the harness and the ECU that brings a set of leads (perhaps a ribbon cable) out to the edge of the carpet where it can be connected to an adapter for the laptop, or even run directly into your Carputer!

With the proper software, the computer can pinpoint this and light up a FAILED IGNITOR display.

It would be good to add in a Fuel Pressure sensor (perhaps introduce a tee with a sensor in the line fom the pipe to the FI) and some kind of Injector display, which I haven't figured out yet, but since the injectors are electrical there's got to be a way.

Compression is still such a gross phenomena that it's still difficult to monitor but easy to detect.

gsl-se addict

iTrader: (3)

Nice! Glad to see that you got it figured out. Turned out to be something pretty simple afterall.

It wouldn't be hard to monitor some of these things. You could even use a Megasquirt to measure/log some parameters and just not use it to control fuel. Or rig up some ADC and interface with a laptop.

Fuel pressure could be measured by teeing into the fuel line with a pressure transducer and an ADC. The injector duty cycle, frequency, etc. could be measured fairly easily too.

Anyway, this shows that going through and doing logical testing is the way to go. Also, it shows that even if you are sure something is okay, it is best to go back and double/triple check when other possibilities are eliminated. It seems very common for people to say that they know a part is good because it is new, worked before, etc. and eliminate it as being the problem. Then to find out there is a problem with the part.

Kent

j9fd3s

iTrader: (3)

i think they show it better in the fd book, but the ecu tester mazda had is a lot of led's that turn red or green with various outputs.

it would be really simple to make for the -se

Rx-7Doctor

iTrader: (13)

This is why it is so frustrating sometimes doing diagnosis over the internet. :-)

By putting an inline voltage tester like I suggested and verifying spark to the trailing ignition this would of been found long ago. :-)

Good to hear that it is on the road again and that it was not some weird anomaly that I had not encountered before.

bliffle

I'd like a time-base readout instead of a simple go/nogo green/red test. That way I can catch nuances and tune a profile.

It ALWAYS pays to redo tests and go over your assumptions. That's why I write stuff down. that's why engineers and scientists keep lab notebooks. That's why they try to teach you that in lab courses. We all commit oversights and misremember history. It's amazing what we learn from regarding every detail.

Egoless. That's what one must be to solve tough problems. Otherwise you are lost.

But I'm not finished with this problem just because the engine is running. I got a couple followup issues:

1-better test for ignitors. Probably in situ, by putting them in live. That means I've got about 10 spare ignitors to test next week by this laborious process.

2-need heatsink compound to cool off ignitors. Maybe Radio Shack.

3-need a more concise and direct debug process to zero in on this common failure.

BTW, it pays to have the little $6 ratchet with a phillips bit in the glove compartment as well as a 12mm and 14mm wrench to move the alternator out of the way. At least I had those ready in the car!

I spent 2 weeks elapsed downtime and $300 unnecessarily (as it turns out) on this problem which could have been solved in 10 minutes at no cost if I had been better prepared with nothing more than knowledge (I had the tools and extra ignitor in the car all the time).

It could as well have been a coil, ECU or fuel filter or pump. So I wonder if I have the tools to detect and solve those problems on the road. Nothing worse than getting stuck in the boondocks with a brokendown car. Well, I have the hardware, now I have to improve the software.

gsl-se addict

iTrader: (3)

It sounds like the trailing was failing and perhaps working okay for the static test and low RPM (cranking until just after start). Once the rpm climbed to idle or so, the trailing cut out and you lost fuel pump, trailing spark, and signal for injectors all at once. Some people will say that ignitor work or they don't (so static test is good enough), but this shows that this isn't the case and they can be intermittent.

It would be fairly easy to build a dynamic tester. It would be like the static test (shown in FSM and Haynes), but replace the switch with a transistor and drive it with a function generator or a simple 555 timer circuit. You could sweep the RPM and even duty cycle if you wanted. The simple setup would be to use a lamp like in the Haynes/FSM, but to give it a more real world test, you could mount a spare coil, coil wire, and sparkplug to your tester. You could then sweep from say below idle to above 7000RPM and make sure the ignitor works correctly over the entire range. It would be simple to do and you could probably build a tester for about $20...possibly much less if you already have a case to build it in. It would be a simple setup. You could hook up an o-scope too if you wanted to look at what the ignitor was doing as well as what the coil (-) primary voltage was doing if you were so inclined.

heatsink compound - you will want this. You can grab a little tube at RS for about $3.

http://www.radioshack.com/product/in...ductId=2102858

My little tube has handled many ignitor swaps between my car and others' cars that I have helped to work on. Still have lots left. You only need a tiny amount to cover the back.