Sport Auto still doesn't trust dyno measurements

[greenberet]

The German car magazine Sport Auto just stated that they will not put the cars they test on a dyno. They measure cars' lap times on the Nürburgring and their actual lift and drag in a wind tunnel. Measuring horsepower would be child's play by comparison and their readers would like to see the results.

However, on page 79 of the July 2011 issue they stated "After another try at running dyno tests that deliver objective results, we have again decided not to carry them out. During intensive discussions with acknowledged experts it became clear that in the higher horsepower classes, repeatable results cannot be achieved due to a range of factors that cannot be controlled for."

I'd imagine that if you always took the same baseline car along as a benchmark and measured the difference between that and the car currently being tested, you could get a pretty good objective measure of rear wheel horsepower. Sport Auto, however, don’t think that chassis dyno results are reliable enough to print. Oh well.

To make this NSX-related: Sport Auto tested a 3.2 liter NSX in 1997 and a 3.2 liter NSX-R in 2002. If they could have carried out reliable dyno measurements, it would have been interesting to see what the difference in horsepower was.

 

[Ponyboy]

I don't get it. Considering all of the obvious variables associated with 'Ring lap times or potentially miscalibrated wind tunnels et al, a dyno is probably the most repeatable measurement they can do.

Seems silly beyond measure (pun intended).

 

[greenberet]

Yeah, it would have been interesting to hear what variables are beyond their control during dyno measurements to understand their reasoning better.

They use the same driver for all of their Nürburgring lap times so at least that variable is controlled for during the track test. But his driving style surely suits some cars more than others and since they’ve been using him for about 15 years now, his reaction times are probably slowing down. And they do carry out the tests at different air and track temperatures. Their lap times for different cars are probably as comparable as they practically can be but there are still uncontrolled variables impacting the times.

The Mercedes fixed-floor wind tunnel they always rent has four scales to measure the weight at each wheel and the force with which the car is pushed back at different air speeds. Those scales can probably be calibrated pretty accurately so the drag figures should be as comparable as you can get. But since it is a fixed-floor tunnel, the lift figures are screwed up if the car uses ground effects. The modern Ferraris they test, for example, always show less downforce than Ferrari claim.

I agree - there are obvious variables impacting the Nürburgring lap times and the wind tunnel measurements. And then Sport Auto states that uncontrolled variables impact dyno measurements just too much for them to publish those figures. Readers regularly write in asking them to spend the couple of Euros it would cost to carry out dyno tests. It would have been nice to hear more of their reasoning why they are sticking to their guns and don’t trust the results enough to carry them out.

For what it’s worth, here’s the original text from page 79 of the July 2011 issue of Sport Auto: Auch nach erneutem Anlauf, der Leistungsmessung auf dem Rollenprüfstand die notwendige Objektivität abzugewinnen, haben wir uns entschlossen, weiterhin auf diesen Test zu verzichten. Nach intensiver Befragung ausgewiesener Spezialisten kristallisierte sich erneut heraus, dass es in den hohen Leistungsklassen zu keinen Ergebnissen kommt, die einer wiederholten Überprüfung standhalten. Die Abweichungen sind aufgrund diverser, nicht beeinflussbarer Kriterien zu hoch.

 

[mlambert890]

All of these tests are hyper variable. Same dyno, different days, will absolutely deliver different results based on various factors. ANd of course it goes without saying that different dynos has insane variance (esp from different manufacturers)

Dyno's are really only effective for baseline testing. As in... Test car stock. Put on exhaust. Test car again on same dyno same day. At least here you can trust the variance.

But in terms of establishing an actual absolute measurement, *especially* if comparing two different vehicles, possibly years apart, its pretty much worthless from an absolute science standpoint.

That said, I do still think it would be interesting for mags to do it.

*I* think they dont because of manufacturer pressure honestly, and I think that may be the issue here as well.

Manufacturers often use ringer cars. A dyno would be a great way to weed THOSE out "hey, why did this NSX dyno at 300rwhp when it should be 290 at the crank!?" That kind of thing

 

[OneRedNSX]

Also there are some higher end cars that just wont perform on a dyno without the right circumstances. I seem to remember some test somebody did on a BMW where it would cut power if it didn't sense the correct amount of airflow coming through the cars cooling system for the speed it was going. Trying to deal with all that junk would be a nightmare. Wouldn't be that big of a deal if you were testing sport compact cars... The last thing you want to do is have the new Aston Martin come up short on power on your dyno because it isn't supposed to act right on one... That would probably be the beginning of the end of for your "testing".

 

[goldNSX]

You've asked me about the dyno:

Out of interest, what kind of dyno did the facility use and how many horsepower of driveline loss were measured at 200 km/h?

Bosch FLA-203s measured my NSX as having 27 HP driveline loss at 200 km/h, Enzo’s as having 31 HP, and Austrian Type-R’s as having 36 HP or 39 HP depending on the gear used. Maha dynos measured Procar Special’s NSX as having 58 HP and m666’s as having a whopping 90 HP of driveline loss at 200 km/h. Unfortunately, all those tests were done at different facilities.

If your facility was especially accurate, it would be interesting how much drag the engine, gearbox, and tires really create when the throttle is closed and the car slows down from high speeds.

How fast did it go? And how did the dyno know how much resistance to use?

It was a MAHA LPS 2000, upgraded to 3000 (Software).

In 4th driveline loss was about 62 hp. I had 02+ wheel/RE050A 255/40/17 in the rear. To be concrete. The driveline loss is not measured with the throttle closed, it's not depending on the throttle as the clutch pedal is pressed after max rev have been reached. So, it's the tires and gearbox alone.

260 km/h was max in 5th.

 

[greenberet]

In 4th driveline loss was about 62 hp.

Was that at 200 km/h or at the power peak? At what rpm did you get your peak rear wheel horsepower?

The driveline loss is not measured with the throttle closed, it's not depending on the throttle as the clutch pedal is pressed after max rev have been reached. So, it's the tires and gearbox alone.

My mistake. You’re right. Those coast down-measurements only measure the power it takes to spin the gearbox and differential and the rolling resistance of the tires. They don't measure any resistance in the engine itself.

260 km/h was max in 5th.

Do you know how the dyno knew how much resistance to apply to the rollers? Did it estimate the correct resistance based how much driveline loss it saw at various speeds and then the assumption that an NA1 with 270 horsepower at the crank (274 PS) can go 270 km/h?

 

[goldNSX]

Yes, 62 hp at 200 km/h in 4th. At peak it was 69 hp (214 km/h). Max. power, wheel and flywheel was at 7880 rpm, 20 rpm below max. reached rpm.

I have no idea how the dyno calculated and set things. We were sitting in the car and the car simply stopped accelerating in 5th at around 260 km/h. Due to the hot IAT the timing was so much retarded that it hit a wall. We didn't measure hp then.

 

[greenberet]

goldNSX: first of all, thank you for the information. You and rsevo6 are the first people I know who have tested the top speed of a Comptech supercharged NSX and I find the results fascinating.

Secondly, maybe your CTSC NSX develops more than 351 crank hp before heat soak sets in. Did the dyno measure engine rpm from the green tachometer connector in the engine compartment or did it estimate engine rpm based on the roller speed? If you kept your foot on the gas until the rev limiter kicked in but the dyno thinks you only got up to 7900 rpm, maybe the dyno underestimated your actual engine speed and therefore your horsepower. If you actually got up to 8300 rpm (the OEM fuel cutoff according to the PGM-FI System Description in the Service Manual), the dyno should have come up with 351*(8280/7880) = 369 hp.

That happened a few years ago at an NSX dyno day. Engine rpm weren’t measured from the tachometer connector but only estimated from the roller speed. And the same gearing and tire sizes were assumed for all NSXs present. Since some NSXs had five speed transmissions, some had six speeds, and a variety of tire sizes were mounted, the assumed rpm and therefore the calculated horsepower were often wrong. So even though the measurements were carried out on the same dyno on the same day, the results still weren’t comparable.

 

[goldNSX]

Secondly, maybe your CTSC NSX develops more than 351 crank hp before heat soak sets in. Did the dyno measure engine rpm from the green tachometer connector in the engine compartment or did it estimate engine rpm based on the roller speed?

I agree. In the US dynos take much less seconds than here on a MAHA. I'm pretty sure if I give it a go in 2nd there are much more than 350 ponnies behind me as the IAT can't climb that fast. But I'm no poser and stated the 351 hp. In the first run we had 357 hp.

The rpm was taken from the lash (spelling, in German: Lasche) in the engine compartement. It's the directest rpm measure an OBD I car allows. I don't think OEM rev limiter is at 8300 rpm. Mine is clearly at 8k and I went back to the stock chip.

I still have my AP-22 and nearly worn tires (and a clutch on it's way ). It would be interesting to see where 0-100 are now. But need some time to do so.

 

[MvM]

Perhaps there is also another explanation.

If the dyno used is always the same and the results are always adjusted for atmospheric differences I think the results would be VERY MUCH comparable between different cars and years. Or AT LEAST as much comparable as the Sport Auto lap times can which every body is always talking about.

However, I wonder if it might be that constantly testing high-end sports car would reveal that many cars don't actually live up to the claims manufacturers make. NOT doing such tests would be a nice way to stay friends with car makers and keep the magazine itself in business.

I think we pretty much know that the NSX usually will be very close to the claimed HP. I think this has to do with the old Japanes 280-HP rule, which was rather easy to reach by any Japanese car maker.
But, IMHO, I still think that claiming the Ferrari F355 as having 380HP (engine that is) was always overstating actual facts since I have hardly seen any F355 dyno break much above 300 RWHP.
And perhaps there are many more such cars that would have difficulty reaching the output the brochures will claim in the sales department.

 

[goldNSX]

However, I wonder if it might be that constantly testing high-end sports car would reveal that many cars don't actually live up to the claims manufacturers make. NOT doing such tests would be a nice way to stay friends with car makers and keep the magazine itself in business.

Magazines live from the adds of the manufacturers who lie to the readers/buyers. Don't you smile at your wife's beauty magazines with all their false promisses and wedding-stories? Well, she does think the same of you and your car magazines. :tongue: But in the end, both end in the same boat, so it's still a happy end!
Now, Autosport just needs some picky reasons to make the readers understand that it's good for them not being told the truth or better: it's good for them being lied to.

 

[greenberet]

If the dyno used is always the same and the results are always adjusted for atmospheric differences I think the results would be VERY MUCH comparable between different cars and years.

I thought so too until I read about LoveFab’s experience in this thread. They developed a new exhaust system for the NSX in 2009 and measured the horsepower gain it provided. As a baseline, they first measured the horsepower of the car in stock form and came up with 229.0 rwhp. Then they built their new exhaust system, installed it on the same car, and went back to the same dyno. Their results this time: 274.5 rwhp, or a gain of 45.5 hp at the wheels.

To confirm the results, they decided to carry out before and after measurements on the same dyno on the same day. This time they only measured a gain of 22.3 hp at the wheels. The "stock" measurement was higher than before and the "modified" measurement was lower even though all of the measurements were carried out on the same dyno.

LoveFab’s lesson learned was “The best comparison results can only be gathered by testing components as close to back-to-back as possible.”

However, I wonder if it might be that constantly testing high-end sports car would reveal that many cars don't actually live up to the claims manufacturers make.

I agree but then they shouldn’t really carry out wind tunnel tests that contradict manufacturer claims (like for the Nissan GT-R) or measure Nürburgring lap times, either. Maybe they just REALLY don’t trust dyno measurements.

 

[goldNSX]

LoveFab’s lesson learned was “The best comparison results can only be gathered by testing components as close to back-to-back as possible.”

Well, mmmh, I would state it differently: "by doing it correctly" This was a clear and obvious error to be seen in the plots. :wink:

 

[greenberet]

It’s astounding that various dynos measure between 27 hp and 90 hp of driveline loss at 200 km/h when an NSX’s clutch is depressed and there’s no load on the transmission. Unless someone’s transmission is filled with sand, there’s no way that one NSX’s transmission is going to swallow over three times as many horsepower as another.

If the driveline loss in fourth gear is 69 hp at the power peak, as was measured for yours, then stock NA1 5-speeds should only dyno 201 rwhp and just about every dyno quoted here on NSX Prime is way too high.

What driveline loss at 200 km/h was measured during your first dyno test before the CTSC was installed?

 

[goldNSX]

What driveline loss at 200 km/h was measured during your first dyno test before the CTSC was installed?

w/o CTSC, 4th gear, 200 km/h, 55 hp, but this was with 245/40/17 tires. I've mounted 255/40/17 to cope with the added torque and we still had to put 120 kg in the trunk and heat up the tires to avoid slipping.

The conclusion is very simple: The BOSCH is simply underestimating or the MAHA is overestimating the true values. Those 90 hp are maybe an exception, a dot very right in the normal distribution curve.
But does it care? I guess that our tests here in Europe are more serious than the 5 seconds dynos we see in the US videos sometimes. :wink:

 

[greenberet]

That’s a 13% increase in driveline losses due 120 kilos in the trunk and warmed-up 255 instead of 245 tires. I don’t know, man.

I’m sure 90 hp of driveline losses at 200 km/h are on the normal distribution curve. Looking at the data points we have, it wouldn’t even have to be many standard deviations from the mean because the variance of these dyno measurements is so huge.