Tuning for Altitude/Knock Issues
 

I had a weird experience yesterday when we went over to Ellensburg for a local autocross event. The car is tuned at essentially sea level (~50-100 ft most of the time), usually around 50 degrees ambient (welcome to the scary PNW).

At the Ellensburg event, my dad took one run, and the car felt a little down on power from the passenger seat, but he tends to be ginger with the throttle anyway, so it's hard to tell. I took my first run and it was obvious that the car had barely more power than a stock Miata. We tried pulling the hood off to cool it down, and my dad took another run and it was obvious he had no power too. At this point, we pulled out of grid and spent a half an hour diagnosing.

The AIM was being finnicky, so I couldn't pull useful longer term data, but just playing with the Hydra, free revving, I noticed it was way down on timing from where it should have been. After messing around for another while, I noticed it seemed to be using the timing values from the backup spark map, for who knows what reason. I put in a call to my tuner friend, but he didn't call back for a bit, so for the time being, I copied the normal timing values into the backup spark map, and that seemed to bring all the timing back. I took a run, and the car seemed to have most/all of its power again. We ran our last few runs without any issues.

Later on, my tuner friend called back and said to check the check engine light (which turned out to be on -- it's covered by the AIM dash, so hard to see without looking carefully for it), which meant that one of the diagnostic codes had been triggered, which is why it was on the backup spark map. We key cycled the car, so we don't know which code it was, but it never came back.

Today, a day later, I had time to beat the AIM into submission and get the data out of it, which let me look at logs of a ton of parameters. My dad's first run, it looks like he had all the power on tap, but decided not to use it (highest TPS is ~65% that run). However, just as he was first starting, turning the first corner, there was a big knock spike up to 1.1V (I've never seen knock over 0.1V before on datalogs, even when dynoing hot), with some following knock at 0.5-0.8V for a couple seconds before going away. My entire first run, the knock basically didn't go BELOW 0.5V for the first 2/3s of the run. The knock retard was pegged at 12 degrees for almost the entirety of my run.

So, obviously, knocking, somehow. The car is tuned for 91 octane, and we never saw a lick of knock on the dyno, even when we pushed the timing a few degrees past where we ended up. The temperature was very hot out (85 ambient according to TWC, but probably way hotter in the sun on asphalt). Looking more carefully at the datalogs, it we were way down on boost (this is a supercharged car, so it's not boost controlled by anything other than PV=nRT). My datalogs from driving around the house here show that I tend to be around 210-230kPa for 4-8k RPMs. Datalogs from yesterday showed that we were doing 170-190kPa in the same range, quite a drop.

Doing the math on the air density for the temperature/altitude shift, it works out about right (85% less air than we're used to). However, this was putting us well out of the range that we had tuned in, which was both extra rich (was seeing 10.2-10.8:1 on most of the WOT sections), and also had higher timing, since we'd just interpolated up from the boost range to what we found at 100 kPa. Thinking about that, I noticed a bunch of the knock is starting at part throttle. We only tuned WOT timing, and interpolated from WOT levels back to the 100kPa (N/A WOT) values. This means that, for example, at 140kPa, at 6000 RPMs, it's running 25 degrees of timing, whereas at WOT (210kPa) it's at 13 degrees, which is a big bump in timing for what's still a nontrivial amount of pressure.

So, with that really long intro, a couple questions for anyone who has any clue:
1. Do you find that, at altitude, you need the same timing values you needed at your original pressures, even though you're making less pressure now?
2. How quickly should I be ramping down my timing from the N/A/DD-ish regime into the boost regime? Seems like I need to be pulling a bunch of my timing out of the boost area, since that seems to be most likely to be causing the knock.

What altitude are you at? I live over 7k feet (77kpa ambient) and I've been using standard timing values on my car with no knock. From what I've read altitude is almost a knock deterrent in itself (lower compression ratio) so if anything you should be able to get away with the same or more timing.

Of course you are going to be making more heat to create the same absolute pressure you would at sea level, so I'm not sure. I used a mspnp base map and pulled another 2 degrees to be safe (91 octane).

Oh yeah, and post your timing map.

Starting at sea level, going up to 1700 ft for that event. With the temp/humidity differences, it's actually a 15% drop in air density from what the car was originally tuned at. It was running quite rich, but I'm more worried about the timing...

My full Hydra map is here:
Index of /pics/miata/20130614

It sounds like the car just needs to be tuned more. Did you only have the dyno for like 20 minutes or something that you didnt have time to actually tune the rest of the map. I think just tuning the rest of it in should do ya.

Well, it's actually kind of hard to tune slightly below max boost on a supercharged car. I can't just change the boost knob and then re-adjust like a turbo car can.

For now I'm just tempted to copy my higher boost timing values down a few cells, then interpolate slower up to 100kPa values.

Unlike a stock horse power miata, a boosted miata throttle pedal has more than 2 positions.

Yes... But to get anywhere near consistent values for the pull you'll need to start at like 50% TPS at low RPMs, then slowly increase up to like 80% TPS at WOT. While I admit that yes, that's possible, I sincerely doubt my ability to reliably back-to-back tune ignition timing that way...

How are you tuning the car? What base timing map did you start with?

Edit: I have no idea how to view or read a hydra map. Post a screenshot of the timing map for us?

Get on a better dyno. I'm driving 75 miles each way to go to a dynapac rather than driving 3 miles to the land air and sea next to my house or 20 miles to the dynojet.

Added timing map to dir:

http://www.akilla.net/pics/miata/20130614/timingmap.gif

I started with the timing map that Jeremy sent with my Hydra, then tweaked the WOT timing on the dyno after doing a bunch of fuel with load control.

Better dyno? He's tuned on a DynoDynamics.

Then I dont know what the problem is. Have dyno hold wheel speed, vary throttle to hit cells. Do full passes through all cells at various throttle positions to confirm when done.

I'd put my money on an untuned fuel map at 170-190kpa.

It's definitely a bit richer there. However, fueling in the mid to high 10s shouldn't be causing crazy knock, I'd think? I pulled some fuel out of that range last night anyway, though.

I'd actually expect it to be running lean there. Hmm.

Do you trust your knock sensor?

Yeah, before I had the fuel areas that I hadn't verified yet as a little fat, in case I went into them by accident. Definitely not lean -- the datalogs from the working runs show dipping down into the mid to high 10s once in a while, but that knock-filled run, I was getting the max 20% hydra enrichment on knock retard, so I was seeing closer to mid 9s.

It's a brand new knock sensor from ~6 months ago. So, I'd hope so, but who knows for sure. On the dyno, I never saw any knock above 0.12V or so, but the small noise curve on it always went up with RPMs.

Could always just be some combination of the altitude and the untuned parts of the map to be running overrich, combined with the much higher timing in that part of the map, to make actual knock.

Tonight, I'm going to pull the spark plugs to look for any signs of det. I didn't hear anything while running, but it's hard to notice that sort of thing.