I’ve discovered gradually over the past few years that each MINI Cooper S is dramatically different regarding its engine output. I have measured stock outputs (SAE corrected on all the following) at the wheels from 138 to 156, this huge (12%) deviation is amazing and frankly unacceptable for a series production vehicle. Why the difference?

I’ve been using the calculation that the MCS’s powertrain has a 12% frictional loss. Based on this, a 163HP ’02-’04 MCS should have about 143 HP at the wheels. This seems to be in-line with every dyno pull I’ve seen from both Webb and Helix, just for popular comparison.

Furthermore, each MCS engine has a dramatically different reaction to modification. As a dramatization, I’ve measured vehicles that have the identical modifications (i.e. “200HP package” consisting of 15% pulley, ALTA CAI, Milltek cat-back, and UNIchip Map05) vary 25wHP, from 164 to 191. This is insane!

The MINI's individual reaction to mods, including ECU mapping, means that among all the current products (Shark/Redbox, AmD, GIAC, MTH, UNIchip (and in the past, EVOtech and Powerchip)) will all react differently, even if "configured" for the same mods on the car. Historically, Shark, AmD, and MTH have not had consistent gains. UNIchip, if the mapping is aligned with the vehicle mods, has good gains in both average dyno numbers and drivability. A huge problem in figuring out which ECU is "the best" is the MINI itself. It is very difficult to dyno the MINI consistently, especially on a chassis dyno. I use my Auterra Palm/OBDII-based dyno, which is an interpolation/calculation software dyno. The advantage is real-world airflow. The disadvantages are, it's not eddy-current (LOL!), and you must, MUST, enter into the calculations the correct vehicle and atmospheric datum and do the "pulls" on flat ground in order to get consistent and repeatable results. Even then, the MINI itself is very picky in terms of Adaptive Logic. I've loosely developed a run-in routine after resetting the Logic in order to obtain reasonably consistent results, however it's critical the vehicle[s] be driven identically, always. This gets rough; why can't we go back to carburetors?!

Ok, moving on. A huge advantage of the UNIchip is the custom-mapping capability. This allows you to bring the mapping from "matched to reference" to "tuned for this vehicle exactly". The differences in HP and Lb-Ft will vary from zero to huge, depending on the individual car, of course.

Back to the topic: WHY will the difference vary so much (hypothetically) from car to car? Here is a list of all the items I’ve ruled out, based on my to-date experiments (work with me here):

*Any supercharger pulley size
*Any air-box or cold/hot air intake
*The following ECU systems: EVOtech, Powerchip, UNIchip, Shark. Based on this, can we not assume the other few are exempt too?)
*Any cat-back or header
*Any valid ignition product (wires, plugs, coil-packs)
*Both stock and Schrick camshafts
*Any Throttle body
*Stock or ALTA intercoolers
*Any suspension/brake/wheel-tire combo.
*Any transmission, including clutches and limited-slip differentials

This points to essentially nothing “external” (and by that I mean bolt-on) on the engine. We can safely assume that it must be something “internal” Here is a list of components I haven’t ruled out yet:

*Rotating assembly (crank, bearings, con-rods, pistons, rings)
*Cylinder head
*All Gaskets, from head to intake to supercharger
*Port matching and general alignment of engine pieces (alignment of intake manifold to head, head to header)
*Supercharger, including surrounding plumbing (inlet tube, exit assembly including “horns”)

Here are the general drivability issues:
*Comparatively little torque over the entire RPM band
*Poor, but not unacceptable throttle response
*Very repeatable, including more consistent than average dyno pulls
*No or little reaction to resetting ECU Adaptive Logic

Anyone have any insight on this phenomenon?
Cheers and TIA,
Ryan

 

my rig was the classic woofer; i had tried several exhausts, evo chip, throttle body header, cam, intake, ported snoots, with minimal results on the dyno (153whp)and 1/4 mile. the 15% pulley worked (169hp), but my car was still not up to Andy's, who had a 15% pulley on a stocker; again on the same dyno and even the 1/4.

bolting on the Endyn cylinder head and adding the 19% pulley added about 15hp (189whp) (larger ex valves and custom porting). Adding the Apexi fuel management computer added another 10hp. (198whp) finally, adding larger injectors and re-tuning the Apexi added anther 15hp, total: 212whp, 179ft-lbs.

I have yet to get back to the track, my previous best time was 14.7 in the 15% pulley incarnation.

my suspicion is the head, particularly the ex valve, is the bottleneck. once that is freed up, things can start to work.

check out the funky stock ex port pic and compare the mod.

 

and:

 

The difference in resulting performance mod gains using similar equipment between cars falls in line with the difference in stock performance. There is definitely something inherent to the cars and not necessarily the mods making this difference. A look through NAM will show people with vast differences in fuel mileage (with some having dismally low despite 'sane' driving), varying degrees of the stumble/yo-yo, cold start issues, hesitation, etc. In some cases a trip to the dealer for the latest software works wonders, other times nada, if not worse. For comparison, my car gets decent mileage, practically astounding at a true 31-32 on highway compared to some getting 10mpg less, no stumble, but the yo-yo in varying degrees (made WORSE since a software update). Where is the consistency?

It appears to me that the biggest factors affecting this are the ECU, the blower, and possibly 'sensory input' meaning the quality of the various pressure, temperature and other senders on the engine. The ECU itself, the fuel maps as well as 'base programming' stand to have the most variation judging by the endless revisions to the code and the maps, and the questionable success of reprogramming. There is undoubtedly differences in 'raw' ECUs over (production) time as well, which may operate differently depending on the installed code. Slight changes in the hardware which may affect how the ecu interprets a sensory signal before making it available in the registers for the processor to read and interpret, that sort of thing. At times I wonder exactly how well, if at all, a new 'reflash' of the ECU at the dealer 'takes'.

There is also going to be differences in the overall quality of the blower, how well the rotors seal against the case and each other and the timing gears keeping things consistent. The slightest leak can vary the produced boost by a couple psi, and such an internal leak causes a rapid buildup of heat as the preheated air recirculates and adds to the heat being produced by the compression process. Some people may have slight leaks in the intercooler boots from loose clamps or production tolerances in the rubber.

The sensory/data aquisition part of the puzzle plays a big role too. The quality and repeatability of the O2 sensors, the MAP-IAT sensor, and the engine temp sensors play a huge role in how the ecu sees the engine operate as well as keep tabs on the ecu's operation during closed loop. There is no doubt that most, if not all MINIs run quite rich, and on the rich side of stoic it does not take a big change in fuel to make a big change in output (and mileage). It seems to me that the ECU is programmed to 'safe' the engine at the slightest hint of detonation or excessive IAT and keep it there much longer than necessary.

Fuel quality plays a big role as well, and in this day of high gas prices and excessive profiteering of all stages in the oil industry, the chance of getting a tankful of 'regular' out of a 'premium' pump is very likely. Drive the car anywhere close to hard, and the engine will be running in 'safe mode' quite often.

I dont know if there are different camshafts for each version of the MINI engine, though it makes sense that there should be. A normally aspirated engine likes to see more overlap to aid in cylinder scavenging, where a blown engine would want less overlap so as to not essentially blow it's boost through the exhaust. It would not surprise me if some engines may have escaped with the wrong cams installed at the factory.

Being that today's CNC milled engines are fairly consistent in their assembly tolerances, I dont see fitment issues like port alignment being much of a concern except for those trying to extract the last bit of power through porting and matching. Timing is fixed by the crank timing disk and not subject to big variations there, and timing itself is set by the ECU. Cam timing can see the biggest variation via a stackup of tolerances with sprockets, keyways and chain length, but since it is a chain system it is inherently more consistent that one using a belt. Fuel pressures and injectors can show up differences as well, especially in open loop (WOT).

I think that once the huge variations in stock MINIs are explained, we can fix the modded ones by proxy.

 

This is sorta along the same lines...People are always saying that the ECU 'learns' your driving habits, especially during your run-in period. Is this in anway contributed to variances in current performance? Could it possibly lend to the differences atleast?

 

Greatbear brings up an excellent point regarding potentially broad tolerances in sensor outputs and subsequent variation of ECU inputs. Also, supercharger rotor sealing variances seem quite probable, as the tolerances must be very very tight to maintain repeatability.

jlm's concept of the cylinder head breathing being the culprit, while possible, doesn't seem likely based on Greatbears' comments on production CNC consistency, which seems plausible, but of course can only be speculation at this point.

--->RallyMINI; while the Adaptive Logic in the MINI's Siemens ECU is very reactive to driving style, it's been generally ruled out as an overall issue. Its' ability to be reset and then re-taught will get the most out of the car from the ECU's perspective.

Thanks for all your comments thus far!

Something to potentially build on regarding Greatbears' concept of sensor variances. I was chatting with a representative of Auto-Meter today regarding their mechanical boost gauge tolerance and accuracy. Their production tolerance is approximately +/-1/4 PSI. When I'm doing 2nd gear dyno-pulls, my Auto-Meter boost gauge reads 16 PSI at 7000 RPM with a 15% pulley, however my MAP sensor, read via the OBDII port, reads 58 inHg, or assuming an average atmospheric of 29.9 inHg, is 28.1 inHg of boost, or about 13.8 PSI. Why the disparity of 2.2 PSI? (which is clearly beyond the tolerance of the boost gauge)

 

as far as CNC machining the head, if you look into the stock ports, you will see that the machining leaves a huge portion of the port as cast, and as cast, is not only begging for flow inprovement but is subject to major part to part variation.

The blower is very finely machined where it needs to be for rotor clearance, but once again, the machining is not extended to the blower intake and output ports, which benefit to a greater or lesser degree depending on the luck of the draw.

What I found was that when the flow restrictions were removed, my rig responded more typically. I didn't think to change any sensors on my rig to get improvements, but that is an interesting idea as well.
My knee jerk reation was to attempt to blueprint the motor, starting with the head (the block is coming next); that is, methodically inspect and optimize whatever mechanical details I (or Endyn) could find. For decades that has been the tried and true hot rod method to deal with manufacturing inconsistencies and poor finish in Detroit iron. The Mini is not a bad platform, the block is strong as is the crank. The head is a joke compared to a Vtech Honda, but getting 250hp out of 1.6 liters is not too bad.

 

Interesting thread. To build on the thoughts here and jlm's thoughts in particular. It's interesting to see that the JCW head has exhaust port work done, and I believe that is all they do to the head. While I still think this is an excuse for MINI to charge what they do for the JCW. Has anyone noticed JCW's being more consistent from car to car than other MINIs?

Paul

 

paulmon - certainly not! I've done baseline dyno runs on JCW's with results ranging from 158 to 176 wHP peak, which calculates to 180 to 200 crank HP (strangely round numbers, actually).

---->jlm; don't all the MINI heads come from the same casting? How much material difference can there be between castings? I agree that blueprinting the long-block is a sure-fire way to get a solid, consistent foundation, and ideally we'd all be doing it.

I think based on the JCW inconsistency, even with the machined head, can imply that the headwork isn't the major source of the evil here.

 

Ryephile,

all your baseline dynos come from the same dyno with the same conditions or from the Auterra Palm/OBDII-based dyno?

BTW...I will go so far to say that all aftermarket chip/piggyback computers have inconsistent gains. Saying that it is only AMD,Sharky and MTH is far fetched.
Some websites sell other ECU tuning even advertise with that fact....like saying will give you an 10hp increase but we saw more on other cars.

If you had a 25whp difference you must have felt that the car was slower than the other. Was that the case?

How about wheel weight, tire diameter, weight of the car (sunroof or not/fuel amount) since the Auterra Palm/OBDII-based dyno requires you to drive, etc? This might not make a huge difference but a difference it will make.

I do not believe the head is a major culprit either....if you have a full blown done head that makes 15hp with cam.....smaller cast imperfections in a stock head will not make a huge difference if at all.


Alex

 

it occurred to me that major improvements were realized from the Apexi, over-riding the ECU determined injector pulse requirements.


The ECU takes some of its data from the O2 sensor mounted just ahead of the CAT, using that and its own calcs to judge A/F.

When we dynoed the Apexi, we attached a wideband O2 sensor to a separate plug in my header, also just before the cat, and used that to set realworld A/F.

different sensor, different results.

 

Ryephile,

How many JCW's have you dynoed? The 158 whp seems like a fluke. I know of two people who have experienced low numbers on their JCW and found out the dealer had overwritten their ECU with the stock software and didn't know that they then had to load the JCW software on top of the stock software again.

Can you post all your dyno numbers. If you have seen ranges of 150-170, but only one of them was at the 150, then that may be an outlier. I would be interested in seeing how the distribution looks.


Paulmon,
The JCW heads have work done on the intake and exhaust ports. I would think the JCW heads are more consistent than the stock heads, but that doesn't mean the kit itself is more consistent.

 

I had my first opportunity to use a Dyno on Friday.

This is the second dyno shop that has told me that the wieght of the stock wheels makes a BIG difference. He told me that he had one MCS go up by 17 hp with a wheel change. If you figure that in use the back wheels are also rotating, there is another 17 HP, for a total of 34 hp to the road from a wheel change.

I figure his 17 hp is optomistic. They probably had a HOT car, that cooled off while changing the wheels, but even if it was only 10 HP might this account for the different values being reported by different MCS owners?

I never worried about the unsprung wieght, but I may be changing wheels to reduce the wieght now. I am currently running the 17" Mini Lites, which are HEAVY.

I got three runs on his dyno.

The first run and second run were essentially identical, in peak numbers, and the traces nearly overlayed. This makes me think his dyno is at least consistent.

For the third run I removed the stock air filter and left the box open. (It sounded like an ambulance siren by the time it hit red line.) Peak HP and Torque on the run were with in 1 HP / pound of the two base runs, but the shape of the torque curve was different. It as flatter from low to high speed, but lower between 4 and 6 than with the stock air box and filter.

I have nothing to compare this Dyno to, but is sure was generous. It showed 154 WHP. I specifically asked if it was wheel or implied crank and he said it was Wheel.

The engine was HOT and we did NOT cool the intercooler for any of the three runs. The hood was open for all three runs.

Based on the car being hot, with an intercooler warm enough that I did not want to hold my hand on it, I suspect his dyno is a bit generous.

Still this makes me wonder about the variablity between cars.

 

for kicks and a good hotrod read, check out this engine building link. Larry is getting in the tens, naturally apirated, with this bugger, a civic.


http://www.theoldone.com/articles/badtothebone/

 

All my figures come from AVERAGES of several runs of each car. The Auterra software corrects to SAE, of course, assuming you input the exact atmospheric and vehicle datum. Naturally I am very careful in doing so.

I agree aftermarket ECU maps are inconsistent. Of course, this is reactive to the individual vehicle and its' own ECU and sensors.

From behind the wheel, 10wHP is noticable, 20wHP is "night and day"

I agree rotating mass weight will technically change results; however on a rolling test it is less significant.

--->jlm, thank you for your insight. So far it is looking like the ECU and its' sensory input are the biggest variables in the whole equation.

 

Ryephile,
I'd be very sceptical of readings from an on-board dyno, IMHO. Having designed and built on-board data acquisition systems for top-fuel dragsters, as well as less 'performance-oriented' cars, I know there is a lot that even well-informed car-guys don't, or can't, take into account. Determining real, meaningfull HP from acceleration & mass is difficult, if not impossible - the data is just too variable to give consistent or accurate results.

At one point in the development of the previously mentioned data system, I tried to incorporate an HP calculation based on acceleration (using an accelerometer, as well as the timing numbers) and the well-known mass (top-fuel cars are weighed on every run). I found that just the vibration of the car caused a variability of 500-1000 HP - for an engine that makes somewhere on the order of 3000 - 5000 HP, that's 10-20%.

Granted, a Mini doesn't vibrate anywhere close to a top-fuel car, but variability on the order of 10-20% doesn't surprise me in the least. Bottom line - I wouldn't get too concerned about the numbers until you can get a bunch of Mini motors on the same engine dyno, then worry about the variability.

 

You're reading way too much into that. This is actually a clever marketing "trick" so to speak. And look - they got you! The vendor usually will provide some kind of "evidence" backing their original claim (10hp). But then they'll add "we saw more on other cars" or a common one - "we saw as high as 20hp on other runs". This lets them LEGALLY put things in your head. So you're thinking, "wow! up to 20hp" yet they admitted that 10 is what they can back up. It's like those weight loss commercials that show everyone who lost 100lbs - they don't show you the 5-10lbs losses or mention the people who it didn't work for. But they have to put it in somewhere that these results are not typical. An other example is an attorney asking a question that is off-topic or can't be proved and the opposing attorney will object. The question/answer is disregarded yet you know the jury heard it and it will be in their head.

 

Yeah but even 170 is only 191hp at the wheels (11% loss) - still 5% below specs. And that's the BEST dyno.

 

--->grnmene; so I suppose doing chassis dyno and Palm dyno pulls simultaneously, getting the same results are just coincidences? How about the benefit of a doubt on the numbers themselves, as that is NOT the thesis of the thread.

By all means, please enlighten us! What are we not taking into account so far?

--->people freaking out over JCW figures: Of the handfull I've dyno'd, they have been a fairly even spread over the bookends I noted above, so the probability of an "odd one out" in my experience is not likely. Again, the numbers themselves aren't the point here, but WHY they are different.

Thank you.

 

Ryan,


I think the numbers do matter. It plays into why they are different. I come from the world of nuclear chemistry, and part of the job of investigating results that don't appear consistent is to look at distribution of results and total number of results. This isn't the only thing to look at, but it helps. I would still be curious looking at all the peak numbers that you have received. It would also be interesting seeing graphs. Do the graphs look similiar but just different numbers, or are you seeing differences in the graphs as well?

 

Ryan,

Another thing is that dyno's are not absolute measurements. They are based off of readings that are affected by various conditions as well as the fact that they have compensation values built into them. As an absolute scientific instrument the dyno is a very weak instrument on it's own. Instruments that are susceptible by a lot of variables are quite common, and we use them in the scientific community quite often (The kind you see on CSI a lot). One way that a scientist removes variability is to run control samples along with the sample run.

I would think to remove more variability on a dyno is to do the following:

• Take a car (possibly your car) and run it across a few different dyno's (the kind that reflects what you are using), across a couple of different days. Take the uncorrected results and find an average. This could be considered a control car. This is not an absolute control, but probably would make a half decent one.
• Then when you are ready to run your sample cars, try to keep your control car as near of a condition (mostly temp wise) as the sample car you are testing. Run it once before the sample car, again with uncorrected results. Then run the sample car, with uncorrected results.
• Determine the difference your car is to your control number and use that as a calculation against the sample car.

This will remove some variability. I would also think of not using the corrected results. It is not perfect, but somewhat better than just running cars all different days with no control. Granted there are most likely other issues affecting the cars as noted above, but at least you could take out some variability based on conditions.

 

Just to put some context into this thread, an '04 MCS w/ JCW kit got ~164HP (max) on a Dynojet. Immediately afterward, my '04 MCS with just a -15% pulley got 172, 170, and 169 HP at the wheels - in that order.

I think the whole point of this thread is to determine the cause of the variance, not the accuracy of the testing equipment. There's way too much variance for it to be the testing equipment.

 

dotbob,

You want to determine the cause of the variance, but not the accuracy of the testing equipment? That doesn't make any sense. If the equipment is not accurate, then that will introduce a lot of variance.

Without a control, you have no idea what the difference is between the two cars. Maybe yours is just a stronger car. There was one Mini on here that beat your number with no mods. He had multiple dyno graphs to prove it. How you run the dyno and your ability to remove variance has a lot to do with the accuracy of the results. I have seen some dynos that are all over the place, and other ones that are dead on, run after run.

We all know that dyno's are not good at absolute measurements. Wouldn't it make sense to remove that variable (since that is what is reading the results) and then you can see some true numbers, and start focusing on other areas, and see what the true differences are.

What kind of dyno are you using Ryan?

 

You're putting words into my mouth. You must take the performance figures within their context. My numbers were comparing two MINIs within minutes of each other on the same dyno. You can't compare HP numbers across the board in North America without realizing there will be differences in accuracy.

That's the point. We all know there are variances; now what's the cause? It's not just testing equipment and methodology.

I think it's a given that an absolute measurement would be perfect. I don't think we can prove that everyone gets the most accurate measurement possible, but that's not the point. The point is, even with the inaccuracy and multiple variables, what is the underlying reason MINIs vary in horsepower?

 

Ryan, DotBob, et al:

The numbers are the thesis of this thread! You are questioning the variability between MCS's based on the HP numbers you have measured. I claim that you can't do that with the test descriptions and results that you have posted so far, because the HP measurements you have done are not necessarily accurate.

You were the one that said putting a Mini on a (chassis) dyno was difficult. If it is, then there is going to be a lot of variability in the results, simply due to the difficulty of getting the car set up on the dyno the same every time. This is especially true with a Dynojet. I've spent a whole bunch of time using a Dynojet and found that it is only useful as a relative instrument for the same vehicle - different vehicle, different results - different dyno, different results - different tires, different results. The only really accurate way to measure HP is a water or Eddy current brake engine dyno, and even then getting results that are within 5% from pull to pull requires extremely careful work and preparation.

As for enlightening you - you'd have to re-live with me the last 20 years of my career as an electronic engineer. No matter how good the instrumentation, there is always variability in physical measurements. In some cases, you may know the source of the variablity - vibration, electronic noise, etc. - and to a certain degree, with very careful work, you can minimize the variability from the known sources. But in many cases, you don't know the source of the variability, and hence any attempt to minimize the errors may, in fact, corrupt the measurement.

Yes, averaging a number of results together does improve the accuracy, but only by the square root of N, where N is the number of samples averaged. To get a factor of 10 improvement in accuracy, you have to average 100 samples that are all taken with the same instrumentation, under exactly the same conditions, executing exactly the same test. If you can manage to do that in a car, then you are some sort of magician.

As for getting similar results with your Palm dyno and a chassis dyno - did you get the same results all of the time, under all conditions, doing the same tests? If not, then you can't really trust either one of them to be accurate (which one was right?). Agreement between the two at one point (ie peak HP) doesn't mean a thing. Think of something like a volt meter - I can connect two meters to the same battery at the same time. If both meters measure the same voltage, does that mean they are accurate? NO! It only means that they agree at that point. If I then hook them up to a known calibrated voltage source and neither one measures what I know to be correct, what have I learned? Nothing, other than I can't trust either meter for ANY measurements.

You wanted to keep this on topic - and I claim that I am. The topic was the variablity between different MCS's. The point that I have been trying to make is that the numbers that you have measured are subject to too many variables - test conditions, instrumentation errors, dyno calibration, etc., etc., etc. - to be taken at face value. Unless, and until, you can start to eliminate these sources of error, then your results will not be consistent or believable, and you will continue to be perplexed by the observed variability between similar cars.