Dyno Report

#1

Bottom line up front. Got 4hp and identified an ignition problem that I didn’t know existed.

I’ve never dynoed a car before. I would encourage everyone to collect some spare parts and go do it. It was interesting and useful.

With a new air filter I showed up at the dyno this morning with 2 spare 173 ECU’s and a spare AFM. We did two dyno runs for baseline, and did 2 runs on each part I swapped in. We also swapped in some different plugs the guy suggested I try. We also did a re-run of the baseline towards the end, just to make sure that our increases were real.

We did 11 total pulls I think. Dynojet dyno.

Baseline #'s. 145.8hp, 143.9ftlbs.
ECM #3, AFM #1 150hp, 148.3ftlbs.

The curves themselves show a difference more pronounced then the peak numbers imply. The base line runs had serious power drops at 3k and 4300rpms. Those problems only existed with ECU #1.

All pulls got shaky between 4500 and 5500rpms to the tune of ~20hp & ftlbs…which is to say serious power loss. The dyno guy, who did not have a BMW background, told me that since both hp and torque curves were misbehaving the same way, that meant that I had an ignition problem. After looking at my plugs and trying some of his plugs, he suggested a new coil and plug wires.

My Bosh plugs pulled better then his NGK plugs. [file name=SepDynoRunChart.jpg size=110690]http://spece30.com/media/kunena/attachments/legacy/files/SepDynoRunChart.jpg[/file]

#2

I just reread some old dyno posts and found that I should use the SAE corrected numbers. Using those my #'s are reduced to 146.3hp and 143.5ftlbs. That’s kinda lame.

#3

With less power, you went faster. Isn’t that interesting…

#4

I’m curious if anybody has a plausible explanation as to why different ECU’s behave differently? There are some analog signal conditioning components, but aren’t these things mostly digital and hence should be consistent?
thanks,
bruce

#5

This is probably a Jim Levie question. My experience has been that swapping ECUs gives my car a rough idle.

I had heard that there is an adaptive element to these ECU’s, so there are maybe different maps based on what the ECU is seeing, over time it calibrates this better and theoretically runs better. I’m not sure how this would work since the chips are Read Only. But my knowledge of electricity and these early chips is caveman-like.

http://www.opel-scanner.com/files/DME_1.1_1.3.pdf

Of course other things like different ECU types with different versions of motronic 1.1 vs. 1.3, etc. could play a role. I’ve even seen it ‘implied’ that 944’s (which used the same motronic I believe and similar ECU / DME units) had different Siemens chips installed so how quickly those react / process could be part of this.

The AFM, crank sensor, all determine fuel and spark as processed by the ECU so some magic mixture of those elements can all be a determining factor as well.

In other words I have no idea, but the dyno shows a difference and we all know that dynos don’t lie or maybe they do and that is part of this too.

#6

Here is my WAG.

  1. Input - Since the reference voltage and the signal processing of just about all of the engine sensors are in the DME, depending on the age of the filtering circuit, voltage generator, and EM shielding the incoming signal quality may vary from unit to unit.

There is also a fail safe mode within the program such that even if one of the sensor input signal may be bad, the DME will still operate with the internally guessed value.

  1. Adaptive element - Microprocessors do contain small amount of storage memory. It looks like DME does store some key elements of operational variables to speed up the processing time. When swapping DMEs, the internal values would be set for a different engine and may produce either good or bad results. Since idle control is one of the stored variable, that may explain why engine would be rough when changing DMEs.

  2. Output - The ignition amplifier is inside the DME. Depending on the age of the circuit, the output of the amplifier may vary from unit to unit. One may look pretty and on time while another one may be somewhat squashed and delay erratic.

  • I have no idea what I am talking about but it sounded good. *
#7

Chi’s comments are close to the mark.

While the computational portion of the DME is digital, the interface between the compute engine and the car is analog. The analog input portion contains filters, amps, and D/A converters. Even if the analog parts had been precision trimmed at the time of manufacture, they would have drifted by now. So while the compute engine is exact, the input data may be fuzzy. To a lesser degree, the analog output circuits will affect engine operation. That explains why you see a difference in performance among a set of DME’s of a particular model.

Over the production life of the E30’s several different DME implementations were used (153, 173, 525, etc). The internal differences in those DME’s can produce different results for a particular car. And cubed by the aging factor the variation in results can be rather large.

The DME is adaptive below WOT. It will learn fuel trim values based on feed back from the O2 sensor. On these DME’s the adaptation is stored in volatile memory and learned adaptations will be cleared if unswitched power is removed from the DME. That doesn’t affect operation at WOT. so far as I know. But it does explain why an engine may idle rough for a while after a DME swap. At WOT the DME is only using data from the AFM, crank position sensor, engine temperature, and intake air temperature. Fuel delivery is computed from maps stored within the DME based on those inputs.

#8

jlevie wrote:

So, if I pull the battery or throw the kill switch, I’ve cleared the memory in my DME? Is this a good or a bad thing?

#9

Since we are primarily interested in only WOT response, it doesn’t really matter if the adaptation values get cleared.

#10

ah, good point.

#11

Just got back in from dyno effort #2. Replacing lots of ignition stuff cleaned up the curve a little and got me a couple more mid range ponies, but no change in peak. That’s a net win, but I’m a little frustrated that I didn’t end up with a numbers as good as the front runners. I spent the extra money to get a new head and bore .020 over with that expectation. But nope, 150hp (STD correction) is all I’m going to get.

Between dyno effort #1 and today I:
-Replaced plug wires (a colder NGK then our Bosch plugs) and coil.
-Spread the plug wires out so they weren’t bound together.
-Replaced fuel filter.
-Tested TPS switch function and confirmed throttle body fully open at WOT.
-Tested a 3rd AFM today.
-Found a vacuum leak where line to fuel pressure reg hits intake manifold. The screws holding the little plate needed to be tightened a bit.

Lesson learned from today. Engine temp is a huge player. If the engine has been cooling for 10min because you’ve been replacing the AFM, you’ll have to do two dyno runs and take the 2nd one. The first one will show a bogus gain of several hp just because the engine cooled a bit.

#12

leggwork wrote:

[quote]I’m curious if anybody has a plausible explanation as to why different ECU’s behave differently? There are some analog signal conditioning components, but aren’t these things mostly digital and hence should be consistent?
thanks,
bruce[/quote]

Don’t most of the “used” ECU’s come out of junk yards? It is not uncommon for someone looking to gain better performance to have the ECU reflashed before it found an untimely demise and ended up in the junk yard. I recall reflashing a 1990 535i ECU after the dealer warrenty expired for a few extra HP. This issue comes up every time there is a dyno discussion.

#13

Its pretty easy to open up a DME and check if it has a chip or not

#14

Ranger wrote:

That’s why Robinson always waits until the “one to go” signal before he fires it up. He hates heat soak.

Me? I warm the engine at the 3 minute mark so I eliminate some parasitic drag from cool oil. :blink:

Here’s the link to the last thread that devolved into a “yes you can, no you can’t” about flashing the chips or replacing the chips.
http://spece30.com/component/option,com_kunena/Itemid,94/func,view/id,38857/catid,16/

Has anyone experimented with simply re-flowing the solder joints in our 20 year old boxes?

#15

My take-away from Jim’s excellent review of the technical details is:
Drive with WOT all the time !! :woohoo:

#16

I am almost 100% certain that the '90-ish vintage ECU’s could only have the chips replaced, not re-flashed. The EPROMS of that era required higher voltages while in programming mode that the ECU did not have the capability to provide built into them. I’ve replace the EPROM on my '91 M5 a few times.
thanks
bruce

#17

It is highly unlikely that the eprom data has changed if the engine runs.

I don’t know if it correct, but I’ve been told that the AFM signal hits the rail around 4k rpm and if the WOT switch is closed the DME just uses map data above that point. When I get the new car built I intend to log the AFM signal and see if that is a correct assertion. If it is, it might explain the weird bump in the torque curve around that point that we see in dyno data.

If the DME is running in open loop mode at that point, the way to maximize output is to try different DMEs, not AFMs. While the guts of the DME is digital and isn’t affected by component aging or trimming, the analog interface circuits are affected by component aging and trim.

#18

jlevie wrote:

[quote]It is highly unlikely that the eprom data has changed if the engine runs.

I don’t know if it correct, but I’ve been told that the AFM signal hits the rail around 4k rpm and if the WOT switch is closed the DME just uses map data above that point. When I get the new car built I intend to log the AFM signal and see if that is a correct assertion. If it is, it might explain the weird bump in the torque curve around that point that we see in dyno data.

If the DME is running in open loop mode at that point, the way to maximize output is to try different DMEs, not AFMs. While the guts of the DME is digital and isn’t affected by component aging or trimming, the analog interface circuits are affected by component aging and trim.[/quote]

Ya, I figured that swapping out the AFM to get a change of WOT performance was kinda wishful thinking. But I couldn’t help recalling that Robert Patton found several hp with an AFM change so I sucked it up and gave it a try.

It occurs to me that we don’t know much about how ECU’s learn and modify their own behavior. I wonder if we are kidding ourselves when we swap out ECU’s at the dyno shop.

#19

Ranger wrote:

[quote]-Replaced plug wires (a colder NGK then our Bosch plugs) and coil.
[/quote]
Re: colder plugs - Since we can’t modify the engines beyond an 020-over, what advantage is there to a colder plug? Won’t a colder plug foul easier?

#20

Steve D wrote:

[quote]Ranger wrote:

[quote]-Replaced plug wires (a colder NGK then our Bosch plugs) and coil.
[/quote]
Re: colder plugs - Since we can’t modify the engines beyond an 020-over, what advantage is there to a colder plug? Won’t a colder plug foul easier?[/quote]

After the first unspectacular dyno effort, you know, the one before the second unspectacular dyno effort, I asked around on a variety of ignition related issues. I had a particularly good talk with an NGK technical guy. There was consensus that race cars like colder plugs then DD’s. At the time I didn’t know what cold and hot meant in the context of plugs, but for everyone’s edification, a cold plug is designed to absorb less heat from the combustion chamber, and radiate heat better to it’s metal surroundings. Therefore a cold plug reaches a colder steady state temp then a hotter plug.

So I figured that I’d try a one step colder plug then the standard Bosch. Bosch didn’t make 1 step colder so I had to go to NGK.

I don’t know much about how different kinds of spark plugs behave differently. I can read marketing stuff as well as the next guy, but I really don’t know spit about spark plugs. All I can say is that there was consensus on the issue and their reasoning (that race cars tend to operate at higher temps then DD’s) seemed reasonable.