ScoobySnack

Spent this am testing various ECU's on my MY00. Cars only mod is a full catless exhaust system and was putting out 243bhp and 239lb/ft torque.

The aim was to change ECU's and plug in a select monitor to see what came out after some runs from 1500rpm up to 6500rpm using 3rd gear and trying to keep all other variables the same.

First some simple impressions about driving my car with ECU's:

Original AE802 (Green) ECU - Car seemed to have little torque over 5000RPM and felt very snatchy in its power delivery through the RPM range, like constant small changes were occuring to its boost delivery. However I'm use to this and felt this must be the norm for the car over the last year.

AE800(Black)ECU from MY99 - Car seemed smoother on boost than before, and did feel to push slighty more past 5000rpm in a smoother way. This ECU gave the impression of the car being more torquey. I remember noticing afterwards from the data that boost on this ECU was slighty higher by about 0.3PSI

Delta 1(Prototype ECU from Stephen Done)- Car picked up much quicker and pushed harder through the RPM range right the way up to 6000RPM. Impression from this was like a completely different car was being tested.

I'll leave everyone to draw their own conclusions.

My thanks for this go out to Stephen Done for his time and effort in the am and for taking the time out to explain in detail why and what the ECU does to my car. Hopefully he can add some more of the recorded details from the test to show what was happening to my car e.g boost figures etc.

What have I discovered ........
Resetting an ECU can have a difference.
ECU from MY99 (AE800) seems IMHO to be better than MY00 (AE802)
I need a new ECU from P.E when it comes out, now I've tasted I must have.......


J

ScoobySnack

Some interesting numbers......

Boost PSI at 3000rpm 3500rpm 5000rpm 6000rpm

AE800 ECU 14.5 13.9 13.8 12.2

AE802 ECU 14.2 13.8 13.8 12


Ignition Timing

AE800 ECU 20 19 25 29

AE802 ECU 19 18 19 23


J

john banks

6 degrees difference at 5000 and 6000 rpm is quite a lot plus a little extra boost here and there and not so rich - no wonder it is better. Very interesting.

No results shown for the Delta ECU I see

[Edited by john banks - 1/27/2002 4:25:00 PM]

ScoobySnack

Perhaps someone would be able to add them to the list if possible??

J

EvilBevel

Steve, if you have a data log like on the other thread, I would be *very* interested to see the comparison.

john banks

More data for all three ECUs for us to get our teeth into would be SUPERB, but if you need to keep the delta stuff under wraps a bit we understand (for now )

Steve, if you told me what I needed to send to the Select Port at what baud and how to read the data I could have a go at plotting my ignition advance for a boosted PPP ECU - 256 bytes of EEPROM could record a bit of data once I get a new micro and read the RPM signal or I could borrow a laptop. Mail me off line if you prefer.

[Edited by john banks - 1/27/2002 4:36:49 PM]

[Edited by john banks - 1/27/2002 4:38:42 PM]

EvilBevel

Sheesh, that is a *lot* of difference ... ultimate proof really that the MY00 "complainers" were spot on with their observations ?

6 degrees less advance, and richer, yup, that sounds a lot like "hesitating bag of ****"

I can't understand *why* they did this though ... surely it will not help emissions ?

StephenDone

I'd say the boost is the same on the two - what boost the ECU is trying to achieve, and what it manages are two different things.
Ignition timing is very different at the top end.
I'll get all the versions dyno'd at some point and publish the results.

> No results shown for the Delta ECU I see
>
Not just yet John! My map is rough and ready at the moment, and people on here would tear it apart :-) I'm waiting for Mervin at PE to get back to do the proper mapping. I've done the software & electronics for reprogramming - I'll leave the mapping to the experts !

All I've done so far is up the boost, stretch it out a little, and make sure there's ample fuel and no det. Currently about 295BHP and 290lbft torque with std intercooler and turbo. 211BHP at the wheels. More mapping and learning to be done.

Cheers

Steve

EvilBevel

I wondered if with the same boost solenoid duty cycle, boost wouldn't be a tad lower just because of the ignition/fuelling ?

john banks

Sounds plausible doesn't it? Having said that the errors are quite small and I am not sure how many bits the ADC is... 97.5mV/PSI with a +-1LSB error at 8 bits over 5V would be +-19.5mV - ie 0.2PSI. If it is 16 bit then it is still within variation from one run to another?

[Edited by john banks - 1/27/2002 4:45:43 PM]

EvilBevel

One more thought (sorry) ... how far apart where the different runs ? You know better than I do that overcooking the inlet temps could induce the retardation.

In what order were the ECU's tested ?

john banks

Did they have a good drive to settle them in?

Questions, questions, you are a good sport Stephen!

StephenDone

Standard boost control on the Impreza is closed loop, believe it or not.

The ECU has two 8x8 maps. One called boost aim, and one called boost achieve. Aim contains duty cycles. Achieve contains boost pressures. RPM on the y axis, load on the x.

If you choose a throttle position - load (ish!) and a particular RPM, the ecu chooses a duty cycle. That is its best guess of the duty cycle required to achieve the boost it wants. After it has applied that duty cycle, it then adjusts the duty to match the boost value in the other map.

So, in summary, fuelling and ignition would make differences to how the turbo responds. However, since the boost is under closed loop control, it will adjust the aimed boost to match what it wants to achieve.

This control method explains why you get overboost when you do mods such as downpipes. The turbo will spool up quicker than the ecu expects. Because of this, its boost aim is too high and it overshoots. The ecu eventually catches up with this and reduces the duty cycle to match the required boost. Sorting out overshoot on a modded car should just be a case of modifying the aim map, leaving the achieve alone. More research required on my part.

The more I learn about the standard ecus, the more impressed I am with what it can do. Multiple 256 zone fuelling and ignition maps, 64 zone boost control, all fully interpolated topped off with a 32bit risc processor for good measure !

Steve

john banks

In practice it seems to only boost AIM after spool up, with lame static duty cycles giving a wonderfully slow spool up! Does this fit with what you have learned? I suppose for a standard car this is OK. Shame they didn't do self adjusting PID during spool up and it would save me a lot of work!

I have a MAX232 @ 4800 baud and know the data comes back in 3 byte groups with an identifier then HI LO. What do I send to it to get ignition advance figures back?

[Edited by john banks - 1/27/2002 5:04:03 PM]

[Edited by john banks - 1/27/2002 5:06:06 PM]

StephenDone

>One more thought (sorry) ... how far apart where the different
>runs ?
>In what order were the ECU's tested ?
>
ae802 (green)
ae801 (black)
delta1 (black, but not really!)

The 802 was then tested again last, after an ecu reset.
The constant knock correction then went away.
(I'd never seen a car before that was retarding all the way from 1000 to 5000RPM !!!)

>You know better than I do that overcooking the inlet temps
>could induce the retardation.
>
Agreed. But the coolest run was on the worst ecu.

>Did they have a good drive to settle them in?
>
About 5 minutes a piece !
It wasn't that scientific.
However, the car was fully warmed up before all runs.
'Orrible rainy day, so charge temps probably lower than normal.

I'll work out some more stats and post in a minute.

Cheers

Steve

john banks

Is it request 0E and byte 1 is (b-128)*0.5 as Ian Gratton posted?

Is the table he posted correct?

EvilBevel

>>ae802 (green)
>>ae801 (black)

OK, so *if* any negative influences came from a potentially hot intercooler, the difference would have been even bigger ...


That's enough proof for me thanks

>>delta1 (black, but not really!)

You should put a pink sticker on it

Thanks for the info, appreciate it.

Theo

StephenDone

Here are some in gear acceleration times 3000 - 5500RPM.
Sorry I can't do speeds, but the ecu doesn't update it's mph value as often as its rpm value, so it's less accurate.

Jason, was it 3rd gear we used ?
Told you my memory was bad :-)

ae802 (green) 5.41 seconds
ae800 (black) 5.25 seconds
delta 4.67 seconds

I'll try to get a graph out of Excel...

Steve

P.S. John. Lets talk offline :-) btw, Ian's data was OBD-II and is only useful for MY01 onwards.

john banks

Pink would look like PPP!

StephenDone

How about purple ?

john banks

Purple sounds cool. How about PurplE for PE initials? Tacky? Cool? Who knows Or "The *P*urple *E*cu - better than black or brown, much better than green" oh dear I've lost it. But very exciting results!

StephenDone

Keep prescribing yourself the drugs John ;-)

Graph accessible here

http://www.steve.ukmail.org/car/temp/ecus.htm

john banks

Nice graphs. Funny the 802 wins VERY slightly to start with, but the 800 soon gets back that miniscule lead.

I've been having trouble with emails. Did you receive email from me Steve?

StephenDone

I got your mail

I think the 802 won the start because that run was done starting on a motorway slip road, so it was slightly down hill !!!
Like I say...not very scientific. Dynos _do_ have their place John. I know you swear by them.
Or is it at them :-)

Steve

babber

So I wasn't imagining it then ? Not a lot in it though!!

Would be interesting if to see what it was like after the ECUs had been in the car for a few weeks. I understand that it wouldn't be possible.

Or what about getting three cars that were almost the same power and torque, and running them for a few weeks with the different ECUs then testing them on the rollers ?

Cheers Phill C

PS Cheers to Floyd for suggesting the MY99 ECU, still VERY happy

ScoobySnack

Hi Steve,

Yep it was 3rd gear....

And you were right about the intial decline on the start of the AE802 (green)ECU. Possible slight incline for the AE800 (Black) ECU aswell.

Seems like the Green had another slight advantage there......

Purples a cool colour, or perhaps a Cheetah background it went like one

J

babber

A picture of Kelly Brookes bangers

I'm sure Salsa man (??) can sort it out

Cheers Phill C

StephenDone

Sorry John,

Wasn't ignoring you here...

=========
Sounds plausible doesn't it? Having said that the errors are quite small and I am not sure how many bits the ADC is... 97.5mV/PSI with a +-1LSB error at 8 bits over 5V would be +-19.5mV - ie 0.2PSI. If it is 16 bit then it is still within variation from one run to another?
=========

ADC is 10 bit. Though there'll be a lot more than 1 LSB of noise on the line. Maps are all 8 bit. Remember pressure is absolute too, which means the pressure range is double what you might think too - 0 to 35 psi (ish). Rather than 0 to 20 (ish).

Cheers

Steve

john banks

Yeah I use 10bits and throw away the last two for this very reason, it also keeps us poor guys who have to work with 8 bit RISC chips a lot happier, so I have about 0.1PSI error from "my end of things" on a 3 bar MAP sensor which is "adequate" - probably more still from noise as you say. However, my whole routine could run 33 times each duty cycle and I only need to run it once so no processing speed problems yet.

GTJ

Nice info,

But what does AE781 means on my ECU?
My Scooby is from 1999.

Are the codes really MY related?

Oh yes, the colour is more or less brown (dark red) while the colour in my car is purple PPP


GTJ

[Edited by GTJ - 1/27/2002 10:16:06 PM]