Jim Morton

As a small contribution to this discussion... I used to have several throttle stop blocks I would use while using the Sharktuner tool to make changes in the maps while "closed" loop. The largest was slightly adjustable and would be set to stop the throttle pedal just before the WOT switch would fire on the car being tuned. I used this block to remove variability during tuning sessions with the idea that is was not realistic for the driver to get to the same throttle position for several successive logging runs.

I had two smaller blocks, each to serve for the same data logging stability used during logging runs at lesser throttle settings. Again, the idea was to get consistent logged data for making best managed and repeatable changes to the maps.

This idea for using the blocks was hard to sell as most of the "approved" tuning gurus "poo-poo"-ed the idea as rubbish and instead told folks to simply use "auto tune" or consult an approved guru. If you really want to better your tune, do consider the basic engine inputs, how to hold them stable while you collect data to decide what is needed. All in all, YMMV.

Best regards

Jim

Hilton

The bit in bold above is the bit I thought was wrong and is what lead me to write my last post.

The point I was trying (badly) to make was that I think when tuning a 928 in open-loop mode with the sharktuner/sharkplotter, we need to consider where the adaptation parameter will tend towards, when the car is run in closed-loop mode.

Yes - this is exactly what I as getting at, and will correct the concern I have about the adaptation parameter impacting the open-loop parts of the map.

Can you please post up a pic of the sharkplotter target AFR map which shows exactly which cells should be tuned to 14.7AFR when tuning a car which will later run with an O2 sensor?
(edit: just bolded this bit, as this is the solution I'm looking for. just before this post I almost posted a longer worked example with questions - but instead of discussing the problem further, I'd like to just get the solution)

My obtuse point in that prior post was that those of us with sharktuners need a new target AFR map with 14.7 in the cells where the LH defaults to closed-loop,
if we're going to tune a car which will run with the O2 loop enabled once its done being tuned. I just wasn't aware that you'd established which cells it is (I only saw your plot with the artificial 165 load limit).

Yeah sorry, I just picked random numbers to illustrate a principle.

jcorenman

In the context of tuning, and begin able to adjust the O2-loop limits as we've been discussing (i.e. the next version of ST), I think that is the right way to go about it. Don't tune around the O2-loop but rather fit the O2-loop to the tune. .

Currently, if we want to use adaptation, then you are correct that we need to consider that when establishing the fuel map. But not much for a NA car, the factory wasn't stupid. It is also not a case that what we've been doing is wrong, we've just figured out how to improve the tools and do a little better job.

At this point there are options. Yes, I can revise SP's target map to consider adaption-- but it won't be a big change for NA cars, maybe extend 14.7 down to the 180 row. That's only a half an AFR unit change from what you posted above. Let me look at some plots and post something tomorrow. But you can also do all that better than I can, it was never meant as a "one size fits all" target. Drive the car, log data, and see what it looks like. Ares where it runs closed-loop should be set to 14.7.

The second option is let us finish these revisions to ST, then we'll have the flexibility to move the thresholds where we want. I've only done one test run so far and then found a loose TB so our S4 is down for a bit. I can do some work on the GT and then switch it back to a MAF if needed. We also need to catch up on documentation on the new new idle controller and some other new stuff.

I hope this helps.

Cheers, Jim

Hilton

Yep, I wasn't suggesting it was wrong - I'm just a fan of improving the details, hence my current quest to measure injector opening time accurately we already discussed (which reminds me I still need to borrow a loose knock sensor off someone local - will call him tonight).

Ah cool.. I wasn't aware that the sharkplotter can distinguish open loop from closed - I'll have to go check what I could be logging and see. I've had the ST installed in my 87 for a while, but keep forgetting to stick the laptop in the car so I can log to/from work etc.

I think its cool that there's further development going on, and that it gets discussed/tested by the community. This improvement to managing the O2 loop, plus the other changes you mentioned via PM should make for some good improvements to the utility of the tuner, and the portability of it too.

jon928se

Don't call now about to go to bed LOL !

They're very loose at the moment - like in a box somewhere. If you can remember which box they're in your welcome.

ptuomov

Hey Jim Morton --

Good to see you here.

I would have to agree on the importance of two things.

First, one has to understand what happens inside the internal combustion engine. I foolishly first thought that one could just optimize it as a black box. With hindsight, I should have immediately understood that a single "function evaluation" on a dyno is way too expensive to make any progress that way. Instead, one needs to bring any and all outside information about how the internal combustion engine works into the analysis to have a hope of optimizing the tune with an acceptable number of dyno runs.

Second, when tuning the engine, it is important to be able to hold as many variables as possible constant. When holding those variables constant isn't possible, one at minimum has to be able to incorporate them in the analysis. Otherwise the whole thing is pretty much hopeless. (For example, I wasted a ton of time and money before I realized by fuel pressure and temperature were not stable between runs.)

Some people may have disagreed with you on the utility of those throttle stop blocks, but I like that idea. It seems like a very good way to repeatedly test the same medium load conditions. We're doing a similar thing with John, but instead of using the throttle stops we're using the boost controller to give us varying levels of boost, repeatedly, at WOT.

One area where we're finding difficulties is the intake air temperature. The direct effect of ambient air temperature and pressure is compensated for by the dyno correction factors, such as SAE correction. This works well for normally aspirated cars. However, what they can't compensate for is that most pump gas turbo cars are severely knock limited. Colder ambient air is going to always allow a turbo car to make more power, even after SAE correction. We can run much more boost and/or timing in 40F weather than in 90F weather. Intercooler heat soak acts similarly. We've tried to run tap water thru the intercoolers to stabilize the environment. I've also once tried to run the SAE correction factor with the intake manifold temperature instead of the ambient temperature -- not to produce high dyno numbers (although it definitely would accomplish that) but to get something repeatable out of the dyno for tuning purposes. Even then, it's the case that the knock constraint shifting will not make the constrained optimal tune repeatable. Any ideas?

Other things that I personally consider good practice:
- Disable o2 loop, adaptation, and reset adaptation from ST2 before attempting to tune the maps.
- "Neuter" the knock retard on ST ignition page.
- Set the WOT fuel map initially to zero in the beginning of the tuning project, and only bring it on later.
- Make sure that the engine is at operating temperature.
- Monitor fuel pressure.
- Turn the A/C off during the runs and run the A/C for a short period between runs. This will help making the runs more repeatable and to cool the fuel between.
- Start with as slow passes as possible on the dyno, within reason, to minimize the impact of transient tuning parameters.

I don't know the above are good things to do, but I believe they are. I am not an authority on any of these topics, so if they don't make sense someone please say something!

Jim Morton

Tuomo

Thanks for considering the advice. One of my main rationales for the need to have repeatable test data is that when you consider the amount of change to maps being proposed, how do you tell noise from real data ???

When first working with the Sharktuner tool, I found the interface wonderful, but the data acquisition awfully slow as would be needed to capture data in "real" time or "on the fly". It was not easy to align logged Sharktuner data with data logged from other sources such as WBO2 loggers, dynos, etc.. After studying the time alignment of logged Sharktuner data, it was clear that more "steady state" measurements were needed as well as multiple runs for logging data to resolve the data in meaningful ways to make intelligent changes to fuel / ignition maps. Again, this my opinion (full disclaimer) and I know the above "flies in the face" of many folks frequenting this board. I keep "contributing here as I do like to promote good tuning practices as well as reinforce the value of the Sharktuner tool... that is if used properly.

As relates to your efforts, I am looking forward to hearing more about your upcoming tunes. With my no longer owning a 928, I frequent the board mostly to stay in touch with folks and seeing new developments. Your project is a great inspiration for many to follow.

On my front, I am currently tuning my E36 racer using a Link G4. It's a nice ECU but using it as the ECU, I need to integrate it with my Aim MXL dash / logger. As I learn more about the value of the MXL as driver / chassis development tool, I may "upgrade" to the MoTec M1 platform and truly integrate the engine and chassis systems. It's a lot to learn, but like your twin turbo program, it keeps my out of more serious trouble !

Enjoy

worf928

When I have one I'll post it. Andrew has posted his at least once or twice in the past years.

Again, I really do appreciate the can-do attitude. I may be able to get to the dyno in the coming months. I would be highly interested in experimenting with a feature for switching off 02 trim and adaptation learning. If I do make it to the dyno I would be quite willing to use a beta version and provide private feedback if the feature isn't yet released by then.

RKD in OKC

Jim Morton, those block things sound like a great idea. Especially in my experience tuning a Turbo car...makes a big difference in not only when you try to get back into the throttle after letting off for a downshift, but also effects the power when going full throttle.a

Once when I got my 944 Turbo back from a professional tuning it ran great and logged good numbers but seemed a little sluggish. Had a 350lb friend go with me to help with data logging. With both of us in the car a 3000rpm to 7000rpm 3rd gear WOT run took 13 seconds. The numbers looked good. The friend started the logging early once and we noticed the numbers before we went full throttle were not consistent. After many runs smoothing out the area below 3000rpm where we went WOT to Red line run we got it down to just 6 seconds AND the power came on so much smoother it actually felt slower. If we had known what needed tuning before we started the runs it would have been a lot easier to tune those areas if there where some way to limit going into WOT.

When I talked to the professional tuners about the significant difference in power and that I was tuning below the actual WOT area they called that area the tip in voltage and said it took a lot of tuning time to get it just right to make the most power.

Jim Morton

RKD:

I agree with the advice you got. Tuning these cars through the transition points of the OEM engine management takes lots of discipline combined with making sure you can rely on the data that's logged. When first working with Dennis Kao's engine, there were many anomalies when reviewing the logged data in terms of having planned changes lead to predictable results. As we got the basic tune "pretty good" in terms of how the car drove and observed improvement, we were still no where near a "good" result in terms of comparing the result to other cars... why ?

After doing lots of tests, getting consistent data was but the first step to having planned moves result in predictable results. Without some basic helpers such as the throttle stop blocks, I do not think we would ever have been able to make real gains in the torque "dip" as the "flappy" valve opened. Bottom line, repeatable test apparatus is a key ingredient, often overlooked in the Internet age.

Stay tuned !

Brett Jenkins

I looked over about half the pages in this thread to try to find a fix for my problem, no luck.

I am trying to tune my 88 to use 36# injectors. I have the ST2 connected to the diag connector, it lights up yellow and then yellow and green when the key is on. But, when I connect it to the laptop via USB, windows 7 pops up an error about an unrecognized device and won't ever load a driver or make the device available. I've tried several reboots, removed USB devices, etc, no luck. Am I doing something wrong or did I miss something?

polecat702

Brett, send Jim Corenman, a PM. Jim knows the Shark Tuner better than any one. He helped develop it. Jim's a great guy, and he'll help ya out!

jcorenman

Brett, that is not an error that I have seen. Have you tried a different USB port on the computer, or a different USB cable?

I've sent a query to John and Niklas, they are the brains of the outfit.

Brett Jenkins

I will re-install the software tonight and try it with a different USB cable on all ports.

AO

I have an old laptop and the STer is only recognized on one of the USB ports. I had to mark it with a sharpie to make sure I always use that port. Didn't look into why, but my guess is that the STer requires a USB 2.0 port? Just a guess.