ECU Light Stays On

[G3.Spiffy]

I’m finishing up the wiring on my spfi conversion and had absolutely everything working, but now my CEL won’t come on with the key in IG and the light on the ecu is on constantly when the car is running.

The bulb is good, it lights up when grounded properly. There is power on the IG side of the bulb, and for some reason the other side when disconnected is getting 5 point something volts from the ECU. I don’t think the ECU is giving it the proper signal and don’t know what needs to happen to fix this issue.

Any input is appreciated!

[DaveT]

the light on the ECU - do you mean the red LED in the hole?  That should be blinking - either trouble codes, or the all is well code.  I know of no time it should be on steady.

[G3.Spiffy]

Yep, that’s the one. Blinks five fast ones when key is in IG position and the engine is off, but is steady once it’s running. Any idea what might be going on?

[DaveT]

Wild guess.   Make sure the 2 diagnostic test connectors are disconnected.   The single pole green, and white, near the wiper motor.

[G3.Spiffy]

I’ll make sure of that today, although I’m pretty sure they’re disconnected. Any other thoughts? Could I have mixed up two wires in the process of splicing in the donor harness which could be sending something to somewhere it shouldn’t on the ECU?

[naru2]

If the O2 sensor input to the ECU is constantly rich,the LED will be on constantly.

[G3.Spiffy]

So could it just be due to the engine being cold when I checked it while running? It wasn’t on long enough to heat up

[naru2]

Hmmm...,not too sure about that.

If it is on constant immediately after a cold start,it must an electrical problem as the sensor takes a minute or two to warm up and generate  a rich signal.

[G3.Spiffy]

16 hours ago, DaveT said:

Wild guess.   Make sure the 2 diagnostic test connectors are disconnected.   The single pole green, and white, near the wiper motor.

Test connectors are unplugged

4 hours ago, naru2 said:

Hmmm...,not too sure about that.

If it is on constant immediately after a cold start,it must an electrical problem as the sensor takes a minute or two to warm up and generate  a rich signal.

Yea looks like you’re right. It seems like I have something funky going on electrically. I was poking around under the dash today double checking things and suddenly my fuel pump primed and was pushing fuel back into the tank. It only stopped when I jostled things around a bit so it looks like I have exposed metal making connections it shouldn’t somewhere... or maybe managed to damage my ecu somehow? The car still runs fine though so I doubt that.

I’m leaning towards bad connection or I made an error when getting power to things. I had everything working just before this happened and have everything else working still, except the CEL, weird fuel pump issue, and light on the ecu.

[DaveT]

Yeah, go looking for that first.  Generally, I've rarely seen anyone on here with a dead ECU.  I've never had a dead one.  I've been running and maintaining EA82 powered wagons since 1988.

[G3.Spiffy]

On 5/21/2020 at 9:50 PM, DaveT said:

Yeah, go looking for that first.  Generally, I've rarely seen anyone on here with a dead ECU.  I've never had a dead one.  I've been running and maintaining EA82 powered wagons since 1988.

After some extensive poking around I’ve determined that the only thing not working as it should is the CEL. The fuel pump “issue” was just test mode probing the fuel pump. The light on the ECU acts as it should in each test mode or read memory mode configuration, but I can’t get the CEL to come on for the life of me. If I  take the bulb out and test the voltage on the circuit board, there is ~12v on the IG power side, and 5.something volts on the other side coming from the ECU. Same thing if I probe the proper pin at the ECU. I feel like this 5v signal from the ECU should function as a ground to turn on the CEL seeing as how it has power on the other side. Should my ECU be sending 5v out to the CEL? If not, what could cause this and what can I check to try and fix it? I’ve checked all my power going into the ECU and cleaned all the grounds which are associated. 

[DaveT]

What is the voltage on the bulb, both sides, with a known good bulb in the socket? 

With it open circuit, bulb removed, the voltages can be misleading. 

[G3.Spiffy]

17 minutes ago, DaveT said:

What is the voltage on the bulb, both sides, with a known good bulb in the socket? 

With it open circuit, bulb removed, the voltages can be misleading. 

With a known good bulb in and it closed-circuit, there is still ~12v on the IG side but more like 10.5 on the other side. Is this normal? I’ll throw a different bulb in just in case, but did make the current one light up by opening the circuit and applying ground to the non 12v side of the circuit 

[G3.Spiffy]

I tested the other bulbs the same way, and all of them get the same reading on the IG side as they should, but two of them have ~2.5v on the other side and one has ~145mv. Those are the ones that work, so it’s weird that they don’t all read the same on the other side as well. That being said, the one that only has ~145mv is brighter than the other two working ones. Don’t know what to think of that

[G3.Spiffy]

Could I have fried the IC chip related to the CEL?

[DaveT]

Re. This- " With a known good bulb in and it closed-circuit, there is still ~12v on the IG side but more like 10.5 on the other side. Is this normal? "

Ok, this [combined with the info from your open circuit test] indicates that the bulb is lit by the circuit grounding the 10.5V side of the bulb.  I have not had to trace or trouble shoot one of these to this level before, but from general electronic design experience, I figure that the ECU pin for the lamp is an open collector type drive.  That means it is open circuit until the ECU signal makes it tie to ground, maybe a few hundred millivolts would be on it in that case, and the light would be on solidly.  It is possible to blow out the transistor - if something were wired wrong, like it was tied to +12V and then the ECU tried to light the CEL.

The other bulbs - I'm thinking you are referring to the other red lights on the dash, for alternator, brakes, etc?
Those are not controlled by the ECU, they have their own circuits.  Some are switches, so they will read very low V on the GND side.  The alternator one, for example, is wired into part of the regulator circuit, and likely not driven as hard, so a couple volts on it is not unexpected.

 

[G3.Spiffy]

27 minutes ago, DaveT said:

Re. This- " With a known good bulb in and it closed-circuit, there is still ~12v on the IG side but more like 10.5 on the other side. Is this normal? "

Ok, this [combined with the info from your open circuit test] indicates that the bulb is lit by the circuit grounding the 10.5V side of the bulb.  I have not had to trace or trouble shoot one of these to this level before, but from general electronic design experience, I figure that the ECU pin for the lamp is an open collector type drive.  That means it is open circuit until the ECU signal makes it tie to ground, maybe a few hundred millivolts would be on it in that case, and the light would be on solidly.  It is possible to blow out the transistor - if something were wired wrong, like it was tied to +12V and then the ECU tried to light the CEL.

The other bulbs - I'm thinking you are referring to the other red lights on the dash, for alternator, brakes, etc?
Those are not controlled by the ECU, they have their own circuits.  Some are switches, so they will read very low V on the GND side.  The alternator one, for example, is wired into part of the regulator circuit, and likely not driven as hard, so a couple volts on it is not unexpected.

 

Yes, those were the other lights I was referring to.

That makes a lot of sense, and is sort of what I got to as well. For whatever reason the ecu isn’t giving the signal to ground the connection. Based on this I uninstalled and opened up the ecu to inspect for burnt components or corrosion, which I found none of. The ecu is spotless inside. 

Is there any course of action from here? Can I track things farther somehow and make a repair if needed or do I have to pay some professional who probably doesn’t know what they’re doing to repair the ecu?

I haven’t looked at the part number yet but I may have a spare ecu that I could swap in to see if that fixes things but if there’s some power issue still present in the car I’m concerned about damaging the second unit as well.

[Rampage]

9 hours ago, G3.Spiffy said:

Should my ECU be sending 5v out to the CEL?

Possible. But not sure. I can not find internal info on the ECU. In electronic circuits, they sometimes use a pull up resistor on the collector of a NPN transistor and it will show voltage if the transistor in not switched on and not connected to a load circuit (bulb). A lot of the circuitry inside the ECU runs on 5 volts from voltage regulators built into the unit. When the collector is connected to a circuit like the CEL light you would normally see about the same voltage on both sides of the bulb. If they are using the 5 volts on the pullup resistor instead of 12 volts then you would see 5 volts with an open circuit.

Newer ECU's monitor that voltage and if it is not higher than 5 volts it will say hey you've got a burned out CEL bulb.

In case you don't know, a simple transistor has 3 terminals, Emitter, Base and Collector. With an NPN transistor the Emitter is grounded. When a little voltage like 3 tenths of a volt is applied to the Base, the Collector will pass current to the Emitter, effectively grounding the Collector. Like turning on a switch.

[G3.Spiffy]

3 minutes ago, Rampage said:

Possible. But not sure. I can not find internal info on the ECU. In electronic circuits, they sometimes use a pull up resistor on the collector of a NPN transistor and it will show voltage if the transistor in not switched on and not connected to a load circuit (bulb). A lot of the circuitry inside the ECU runs on 5 volts from voltage regulators built into the unit. When the collector is connected to a circuit like the CEL light you would normally see about the same voltage on both sides of the bulb. If they are using the 5 volts on the pullup resistor instead of 12 volts then you would see 5 volts with an open circuit.

Newer ECU's monitor that voltage and if it is not higher than 5 volts it will say hey you've got a burned out CEL bulb.

In case you don't know, a simple transistor has 3 terminals, Emitter, Base and Collector. With an NPN transistor the Emitter is grounded. When a little voltage like 3 tenths of a volt is applied to the Base, the Collector will pass current to the Emitter, effectively grounding the Collector. Like turning on a switch.

So if I understand what you’re saying (correct me if I’m wrong), it’s possibly normal for me to get the 5v reading mentioned earlier when the bulb is out (open circuit) depending on how the inside of the ecu is configured.

If that’s the case can I visually observe this inside the ecu? Or would that be a part of an IC chip? 

Speaking in terms of the function of the CEL, the wire that exits the ECU and goes to the signal side of the CEL should be connected to ground whenever this transistor flips its switch, correct? 

If I understand what’s going on, there is something preventing the ECU from flipping that switch. This could be:

•The switch is damaged and can’t function.

•The ECU doesn’t know to turn on the CEL to test function with the key in IG and the engine off. This could be because of some other damaged internal ECU component or due to bad connections upstream. 

I’ve checked all connections outside of the ECU pretty extensively and am pretty confident my issue lies somewhere inside the ECU. 

What I need now is to diagnose what actually failed and figure out a course of action to fix it or replace the unit if necessary. I’m located in CA so I need the CEL to function to pass emissions unfortunately. 

Thank you for the input!

[Rampage]

3 hours ago, DaveT said:

It is possible to blow out the transistor - if something were wired wrong, like it was tied to +12V and then the ECU tried to light the CEL.

100 % correct. Just a momentary direct contact with 12 volts would burn out the transistor.

[G3.Spiffy]

6 minutes ago, Rampage said:

100 % correct. Just a momentary direct contact with 12 volts would burn out the transistor.

It seems highly likely this is what I did. Any advice on how to replace the transistor/IC chip? Or is the ECU no longer viable? 

If so, I have two other ECUs which are from a 90 and a 92 Loyale spfi manual trans. The systems that they run seem pretty close to identical but the sticker on the housings all have different numbers.

Any info on spfi manual trans ECU interchangeably?

Edited May 24, 2020 by G3.Spiffy

[Rampage]

What I've run into with a lot of electronic units, not just automotive, over the years is, any OE manufacturer will have parts (chips and transistors) labeled with "house" part numbers on them, so there is no public cross reference for those part numbers. In the past, if I had a schematic of the unit, or could draw the circuit on paper I could come up with a replacement part that would work. But, that was like 15-20 years ago.

Probably your only options are to try your ECU in another original car that it came from and see if it behaves like normal, or recheck your wiring (again) and try your other ECU.

[Rampage]

30 minutes ago, G3.Spiffy said:

Any info on spfi manual trans ECU interchangeably?

I know someone else can answer that question for sure, but I think they are different.

[DaveT]

Those years are pretty close, they probably will work.  All of them have  a pin that tells the ECU if it is is connected to an Automatic or a standard shift.   If you have both schematics, you can find it.

I will see if I can find some time to look at one of my spare ECUs.  A few years ago, I started to reverse engineer one.   Many of the ICs were normal parts I could find data sheets for.  Transistors, I can figure out, marked or not, I just have to trace out which one is connected to the CEL pin.

 

[G3.Spiffy]

Just now, DaveT said:

Those years are pretty close, they probably will work.  All of them have  a pin that tells the ECU if it is is connected to an Automatic or a standard shift.   If you have both schematics, you can find it.

I will see if I can find some time to look at one of my spare ECUs.  A few years ago, I started to reverse engineer one.   Many of the ICs were normal parts I could find data sheets for.  Transistors, I can figure out, marked or not, I just have to trace out which one is connected to the CEL pin.

 

All of the cars are manual so theoretically I wouldn’t have to rewire anything, it would just be plug and play right?

What are the chances that I could damage something if I throw the problematic ECU into the 92?

This would allow me to test for function and interchangeability before installing a good ECU into a car that could still have something shorting out that circuit and see if it runs everything but the CEL in the 92.

If the ECUs are the same, and I know for sure I’ve fixed whatever damaged the original ECU, I’ll probably just use the one with the good CEL just for my emissions test, then switch back because the 92 is all original and I’d like to keep all of its parts together until that’s no longer possible.