Cosworth on the rollers

[mike-p 116]

I took the Cossie to a Pistonheads RR day at Clive Atthowe's in Norwich this morning, having just collected it from the bodyshop yesterday.

It made 281.5bhp, which is almost exactly what Cosworth reckon a Stage 2.3 car should do.

Most pleased.

[mike-p 116]

Here are the details from yesterday's run (Run 5) compared with the figures from a year ago (Run 2).

The engine is down on torque at the low end but up on bhp at the top end.

GT86_Cosworth_RR.pdf

[Kodename47 446]

The blue run is overall the better run, the red power run is only above for about 900RPM out of the 5000 on the chart...

[mike-p 116]

But the car now feels better on the road, smoother, pickup more eager, it only goes to show that RR figures don't give the whole picture.

[will300 812]

Any idea what has happened to the torque Mike? That just looks weird.

[Kodename47 446]

3 hours ago, will300 said:

Any idea what has happened to the torque Mike? That just looks weird.

What's "weird" with the torque? The AFRs look much better on the recent run though, looks lean on the initial dyno. Would explain why it currently feels smoother.

[Ade 517]

AFRs still look a tad leaner than mine which are at about ~11.8-12 through the rev band > 4krpm

[Kodename47 446]

Stupid website, can't edit the post.... taken from what Ade, the inaccurate OEM sensor? Don't forget the dyno will read tail pipe mixtures. Inherently leaner plus add in a cat too. The cylinder mixtures will be richer than the graph, possibly 0.5 or more.

[Ade 517]

ah I didnt see your previous post.

Fair enough. I didnt realise the lambda was that bad.

Yes im talking about ecutek logs.

[Kodename47 446]

6 minutes ago, Ade said:

ah I didnt see your previous post.

Fair enough. I didnt realise the lambda was that bad.

Yes im talking about ecutek logs.

It's as good as the sensor is scaled to but the sensors still seems to vary from car to car. So it's a ballpark figure especially once below 13:1 as it loses accuracy the further away from stoich.

[Ade 517]

Thats surprising given how rich these cars run.

Well if you are correct about 0.5 richer, those afr values look very similar to mine taken from ecutek logs. I guess it shouldn't be surprising since they are both Cosworth Matt Calibrations.

So do tuners use the measured afr from the exhaust on the dyno to fine tune the target afr values or is it not that critical?

[Kodename47 446]

2 minutes ago, Ade said:

Thats surprising given how rich these cars run.

Well if you are correct about 0.5 richer, those afr values look very similar to mine taken from ecutek logs. I guess it shouldn't be surprising since they are both Cosworth Matt Calibrations.

So do tuners use the measured afr from the exhaust on the dyno to fine tune the target afr values or is it not that critical?

On the stock calibration they have only set the sensors to read as low as 12.1, if it reads any lower then the calibration has been changed. Seeing as the sensor only has an effect on closed loop (~14.7-14.0 as standard) then there's no need for it to be accurate any lower.

Depending on the setup you could use a range of techniques:

• A wideband Lambda in each manifold runner - time consuming and expensive but most accurate and get each cylinder's AFR for optimum tuning.
• A wideband Lambda in the manifold similar to OEM position (you can't remove the OEM sensor) - more common and good level of accuracy if pre-cat
• A wideband in the tail pipes is the least accurate but you can manually compensate for the leaner readings.

Yes, you would ideally dyno (or any form of accurate WBO2) tune the AFRs as it's one of those things that can vary from vehicle to vehicle. With ECUtek you can re-purpose the rear O2 input to be able to put a Wideband in.

[will300 812]

*ignore*

[Sonic 4]

from what I've seen the stock semi-wideband is fine down to about 11.2:1 AFR

BUT

you have to apply a re-curve to the sensor to increase its accuracy as knightryder is correct in saying as stock with the stock calibration its accuracy isn't great away from stoich

and the issue I have with dyno / roller widebands is often they're poked into a tailpipe which is a long way from the engine so reading a cold exhaust mix and also its very difficult to get sufficient flow over the sensor for an accurate reading as most tailpipes are too large and diffuse a flow 

 

if the roller dyno sensor is mounted into a small pipe and that pipe is poked a long way into the exhaust then this is a lot better but still suffers from a much colder gas mix so changing the accuracy of the reading 

[Kodename47 446]

1 minute ago, Sonic said:

from what I've seen the stock semi-wideband is fine down to about 11.2:1 AFR

BUT

you have to apply a re-curve to the sensor to increase its accuracy as knightryder is correct in saying as stock with the stock calibration its accuracy isn't great away from stoich

You can re-calibrate the OEM sensor to read as low as you want, however it loses accuracy the further from stoich. How would you even know that any calibration is near accurate without confirming on a wideband?

[Sonic 4]

9 minutes ago, knightryder said:

You can re-calibrate the OEM sensor to read as low as you want, however it loses accuracy the further from stoich. How would you even know that any calibration is near accurate without confirming on a wideband?

you're correct, you do have to have a standalone wideband running in paraell ( mounted a bit further down the pipe)  to achieve the calibration

thats what I did anyway