STD ECU Capabilities.

[Tricky-Ricky]

Thought it was time to clear this up once and for all,

can the std ECU self learn? hoping for a definitive answer from RyanG really.

It is mentioned on here with regularity, and seems a bit of an urban myth, that you need to reset your ECU to cope with BPU, different ron petrol etc etc.

IMO it is not buy any means self learning, now i am no expert but expecting an ECU designed in the early nineties, when wide-bands didn't exist AFAIK, but in any case certainly wouldn't be used on production cars.

Anyway i digress, AFAIK most ECUs are designed to read and adjust/switch to a different part of a set map, according to the sensors attached, IE primary being, airflow/pressure, lambda, RPM and temperature, and secondary being knock sensors, exhaust temp etc, now the ROM chips that the ECU consists of, not RAM which AFAIK they don't have, until pretty recently that is.

 

The ECU contains pre written maps, primary being fuel, ignition timing, secondary being warm up, which is linked to temp sensor feedback , and knock control by knock sensor feedback, and accessory loading, IE lighting, air-con etc.

 

Now as far as I'm aware, there are not lots of different maps for any condition, just the basics for all that i have mentioned, and sub maps that deal with knock retard, overheat, or lambda or other sensor failure,

the ECU cannot tell what type of fuel is in the tank for the simple reason that a normal narrow band lambda sensor is only effectively a switch that tells the ECU that the mixture is rich or lean in the closed loop area of the fuel map, so in turn the ECU is programed to keep AFRs to stoic = 14.7 AFR and anything over 4.000RPM (in the case of the Supra) or + boost is just dealt with in the same but different portion of the map.

The only way the ECU can determine if the petrol you have put in, is by knock sensor feedback, IE if it knocks the ECU will switch to a default timing map, if no more knock is detected IE better petrol, then the ECU switches back to the original timing map, so although most eventualities are covered, it is by no means intelligent or self learning.

 

So i will now wait for Ryan to come along and either correct the bits i have wrong or totally shoot me down in flames, but in any case it would be good to have this all sorted and maybe put in the reference section when all corrected etc:)

[MarkR]

[bolarbag]

I think your pretty much spot on, i.e they dont self learn and they dont hold values depending on certain parameters or previously recorded values

 

As you probably know there are other ecu's out at the time of release that do though, hence the urban myth

[TrickTT]

The ECU does have RAM capability, it uses it to store fault codes for one. Also the ECU has a permanent power supply to maintain that memory. When you "reset" the ecu it switches to a base area of the map which is safe and allows the car to start. Once its running it takes input from the various engine sensors and uses that data to switch, if necessary, to other areas of the map to adjust, for example to fuel octane rating.

You are basically right in saying the ECU can't self learn as that would imply that it can constantly re-write its own map according to conditions, but Toyota programmed a very broad spectrum map with enough scope to allow for things like BPU tuning. If the ECU sees these conditions every time it is switched on, it will "remember" that and go directly to that area of the map in future.

It is also mapped to allow for variations in sensor input (eg my temp sensor might give a slightly different voltage at 75deg than yours) The ecu can adjust for this and again, "remember" the input its getting. That is why the engine runs slightly rough after an ECU reset as it is re-adjusting from its initial state and compensating for the inputs it sees.

So basically it cannot self-learn, but it can compensate, within Toyotas set of pre-programmed parameters, and remember that next time the engine is running.

Edited August 20, 2008 by TrickTT (see edit history)

[bolarbag]

The ECU does have RAM capability, it uses it to store fault codes for one.

 

good point

[monkey3]

 

 

The ECU does have RAM capability, it uses it to store fault codes for one.

 

And I was about to say that.

[Tricky-Ricky]

My point being precisely that, it is just pre programmed and can only deal with a preset amount of input parameters from sensors, it cannot learn what fuel it has or that the car is BPU it simply has default areas of the main maps, and just switches to them if fuel octane is low enough to generate knock/det, and then revert to the more aggressive map once all knock has subsided or not as the case may be.

OK RAM for fault codes, but thats not what i meant, can you tell me which wire is a permanent live? not come across this.

[bolarbag]

it cannot learn what fuel it has or that the car is BPU it simply has default areas of the main maps

 

What ECU can tell you that?

[TrickTT]

The 2 main ECU fuses under the bonnet are permanent live to the ECU, there is a 3rd ecu fuse in the footwell that is ignition switched.

[Wez]

My understanding is that the stock ECU has base fuel and ignition maps, the base ignition map is pretty aggressive and based on its brilliant knock control it is able to pull back the timing to prevent det, this is happening all the time.

 

Where it stores this information I have no idea but this also explains why people say the car feels faster after a ECU reset.

 

If for example you dumped some race fuel in the tank and reset the ECU, the more aggressive timing will unlikely be pulled and therefore more torque/power.

 

This is prob more evident on the jspecs as the jspec ECU may well have more aggressive maps to suite the higher octane fuel available at the time of release.

 

 

All my opinion of course

[Thorin]

My understanding is that the stock ECU has base fuel and ignition maps, the base ignition map is pretty aggressive and based on its brilliant knock control it is able to pull back the timing to prevent det, this is happening all the time.

 

Where it stores this information I have no idea but this also explains why people say the car feels faster after a ECU reset.

 

If for example you dumped some race fuel in the tank and reset the ECU, the more aggressive timing will unlikely be pulled and therefore more torque/power.

 

This is prob more evident on the jspecs as the jspec ECU may well have more aggressive maps to suite the higher octane fuel available at the time of release.

 

 

All my opinion of course

 

 

That's my understanding too Wez. It's quick to pull the ignition timing when you reset the ECU, but if you stick some race fuel in then advancing the ignition timing again happens much slower.

Edited August 20, 2008 by Thorin (see edit history)

[gaz1]

The 2 main ECU fuses under the bonnet are permanent live to the ECU, there is a 3rd ecu fuse in the footwell that is ignition switched.

 

yes the one under the foot well caught me out on the motorway:rolleyes: its a 10amp fuse and when it goes you stop:Pling: it knocks out speedo/rev counter temp gauge and a few other things it also knocks the alternator off so the battery drains

[Tricky-Ricky]

The 2 main ECU fuses under the bonnet are permanent live to the ECU, there is a 3rd ecu fuse in the footwell that is ignition switched.

 

Sorry i meant which wire on the ECU plug is a permanent live, 31 E10 by any chance?

 

Wez, interesting point what you're saying about knock control, as i have monitored std knock sensor feedback and stand alone monitor, and i hardly ever see a spike from the std knock sensors, but even with the sensitivity turned down on the stand alone unit, i still see knock activity, which is puzzling unless either the std sensors are not that good, past there sell by, or they don't do such a great job, and the thing that clinches it for me is that on at least two occasions i have been using det cans that verified the reading on the stand alone unit, but i didn't get a register from the lodged std sensors, which to me is a little worrying:(

[Wez]

What are you using to log the std knock sensor output?

 

The AEM uses the stock knock sensors and they pick up loads of noise and det is very easy to see amongst normal noise, if you increase the rate of capture the knock sensor output is a fuzzy line of data

[Hermit]

Urban myth I reckon. And the idea that the ECU is so aggressive that it causes knock and then backs off just reeks of the sort of thing Mycroft would come out with. No offence meant to Wez and Thorin as this kind of myth (if it is) gains credence simply by being repeated often enough.

 

My (limited and entirely theoretical) understanding is that if the ECU detects knock, it will retard timing to a static 10deg BTDC, throw error 52 or 55, and go into limp home mode until the ignition is switched off.

 

For sure the ECU learns some things after a reset - the range of throttle position sensor voltages is an obvious one. But for many sensors (e.g. water temp) it has no reference to recalibrate against, so it would have to use a static lookup table. I'd say it appears to learn more than just TPS range, but it's not clear what else it learns, or how long it goes on learning for.

[Matt Harwood]

Urban myth I reckon. And the idea that the ECU is so aggressive that it causes knock and then backs off just reeks of the sort of thing Mycroft would come out with. No offence meant to Wez and Thorin as this kind of myth (if it is) gains credence simply by being repeated often enough.

 

 

I'm with Wez on this one. Except I believe the ECU will always attempt it's maximum ignition map, and retard timing if it detects knock. I don't think it stores the fact that it had to retard the timing on its last full throttle cycle, it will simply advance as much as possible, and retart 'x' amount on knock detection.

My reason for this belief is because I have had my car on a dyno whilst tuning for normal road fuel with an E-manage. Once we had finished the road fuel map, we started to map for higher boost using race fuel. - I emptied the tank and re-filled with Klotz 118, and ran the dyno again, with no alteration to anything except the fuel, and the dyno was shwoing about 45hp higher than previously.

The E-manage cannot take advantage of better fuel. The only thing that could have altered is the stock ECU and maybe the engine had cooled slightly in the 5 minutes it took to re-fuel. I doubt 5 minutes cooling would have added 45hp by itself, and we didnt notice any significant drops in output as we continued, and besides, we had two damned great big fans inches from the front of the car!...

[Tricky-Ricky]

Hermit, while I'm inclined to agree with you're comments about timing, apart from retaining the fact that its retarded the ignition until shutdown, the second part i don't as i have said before it only has a set amount of actions it can take, IE, sense knock retard ignition via secondary map, same for most other variables, it can by no means be described as learning.

 

Wez my EMU monitors knock, and the settings are completely configurable so are set to the correct resonance and KHz same as stand alone knock sensor, whether its as comprehensive or as sensitive as AEM i have no idea, although i suspect not, what setting is the AEM knock configuration able to be set to and what are the parameters?

 

 

 

Another thing that comes to mind while discussing this, is that if the ECU was able to comprehensively adjust by big or numerous maps, then surly adding piggyback ECUs would cause havoc, as a lot of the inputs are adjusted before being fed to the ECU and a few of these are also adjusted after the ECU, so if it was so capable, it would be continuously trying to make adjustments/map points, and from my experience this is not the case, accept perhaps for closed loop lambda feedback, but there are ways around that.

[Tricky-Ricky]

I'm with Wez on this one. Except I believe the ECU will always attempt it's maximum ignition map, and retard timing if it detects knock. I don't think it stores the fact that it had to retard the timing on its last full throttle cycle, it will simply advance as much as possible, and retart 'x' amount on knock detection.

My reason for this belief is because I have had my car on a dyno whilst tuning for normal road fuel with an E-manage. Once we had finished the road fuel map, we started to map for higher boost using race fuel. - I emptied the tank and re-filled with Klotz 118, and ran the dyno again, with no alteration to anything except the fuel, and the dyno was shwoing about 45hp higher than previously.

The E-manage cannot take advantage of better fuel. The only thing that could have altered is the stock ECU and maybe the engine had cooled slightly in the 5 minutes it took to re-fuel. I doubt 5 minutes cooling would have added 45hp by itself, and we didnt notice any significant drops in output as we continued, and besides, we had two damned great big fans inches from the front of the car!...

 

Matt, wouldn't the fact that the fuel would have a far better bang for buck so to speak also account for power increase? but I'm with you on the ignition side;)

[Hermit]

Matt, were you logging timing during those runs? Could it not be explained by the different burn rate characteristics of the 118 fuel?

And, out of interest, how much extra power did you get out of that fuel after mapping?

 

Rick, I was thinking the same regarding piggybacks, if the ECU was so clever there'd be no point modifying timing with a piggyback.

Once knock has triggered an error code, fail-safe mode is active until the ignition is switched off (that's straight from the workshop manual, thankfully I've never experienced it!). I suppose what's under discussion here is whether it reacts to knock that doesn't last long enough to trigger an error code (i.e. less than 4 turns of the crank).

As for learning, if all it does is 'learn' the range of the throttle sensor and maybe others, then 'learn' isn't really the right word, but it'll do

[Matt Harwood]

Matt, were you logging timing during those runs? Could it not be explained by the different burn rate characteristics of the 118 fuel?

And, out of interest, how much extra power did you get out of that fuel after mapping?

 

 

Couldn't log the timing. It was the E-manage blue.

I'd be suprised if burn rate alone could amount to 45 hp with no other adjustments. The guy operating the dyno said he'd seen similar things before with aftermarket ECU's that were running knock control.

Power wise, it ran 469 rwhp @ 1.4 bar and 564 rwhp @1.8 bar and racefuel. - That was on a Dyno Dynamics roller.

[Tricky-Ricky]

'learn' isn't really the right word, but it'll do

 

 

 

LOL! Noooooo! thats the whole point of the thread:p:d

 

It makes for an interesting discussion though;)

[TurboBrett]

[Ryan.G]

Have to say i dont unfortunatly know an exact answer to this question but i have worked with a guy who developed a method to be able to remap the stock ecu on the Mr2's which was very cool.

 

More info on that found here but make sure you are sitting down Ignore the first 3 pages though just GB stuff

 

http://www.imoc.co.uk/forums/viewtopic.php?t=75023&postdays=0&postorder=asc&highlight=jeremy&start=60

 

Im hoping when he gets time he will crack the supra ecu

 

Now when mapping this ecu and as well as Emanages that i have done on supra's i have found the timing to be pretty stable across the board and dont really see any changes between runs unless knock has been detected.

 

Now when it has detected knock i saw it pull the timing and then after a decay time, puts it back in... But it also adds abit more fuel aswell as pulling the timing. It certainly doesn't pull the timing and then not put it back in to an ecu reset.

 

IMO on an ecu reset nothing changes and its all in the mind.

 

Ryan

[Ryan.G]

I'm with Wez on this one. Except I believe the ECU will always attempt it's maximum ignition map, and retard timing if it detects knock. I don't think it stores the fact that it had to retard the timing on its last full throttle cycle, it will simply advance as much as possible, and retart 'x' amount on knock detection.

My reason for this belief is because I have had my car on a dyno whilst tuning for normal road fuel with an E-manage. Once we had finished the road fuel map, we started to map for higher boost using race fuel. - I emptied the tank and re-filled with Klotz 118, and ran the dyno again, with no alteration to anything except the fuel, and the dyno was shwoing about 45hp higher than previously.

The E-manage cannot take advantage of better fuel. The only thing that could have altered is the stock ECU and maybe the engine had cooled slightly in the 5 minutes it took to re-fuel. I doubt 5 minutes cooling would have added 45hp by itself, and we didnt notice any significant drops in output as we continued, and besides, we had two damned great big fans inches from the front of the car!...

 

Not sure i would put the 45bhp down to timing Matey. I would more go with fuel being oxygenated and that the tyres and everything had cooled down after mapping the road fuel.

 

On some of the Big Hp cars that i do, i take the car off the dyno and let it cool down after the mapping and then when i have finished with the other stuff we pop it back on and normally find quite abit more HP has appeared. Mainly its due to cooking the tyres when mapping and it grips lots better when its cooled down

 

Ryan

[Matt Harwood]

Now YOU, are one person that I will not question.