Virus

iTrader: (3)

Good to know. Thank you. Any particular gear you do this in?

Virus

iTrader: (3)

...see? I didn't know that there could be "noise" in the mix. I just figured that there was a spike and it didn't look normal and therefore I should be worried. I didn't know what real knock "looked" like.

Any one have a datalog of real knock?

m2cupcar

iTrader: (5)

actually real knock looks very similar to what is seen in your log- it's just that your spike is so small. My logs are loaded with noise- but it's more or less a linear increase. Then at 7k I get a huge spike- but running the car in 3rd, out of boost up to 7300 I got the same spike. So it's something that only rattles at 7k. I've also got to find a better mounting spot for my sensor. Using the adapter plates for the engine swap has eliminated the recommended location. I have learned that the further away from the front the better. Apparently all those pulleys make a lot of noise which the sensor picks up. Moving the sensor back cut the noise in half. The next thing are my lifters- which are in bad in shape. Such is the case with most JDM engines- sounds like a few of them never pump up. And that seems to get the sensor excited to. All this said, your knock line looks very clean. And like I said, I don't see that knock occuring in your other logs. So it's most like some kind of noise. Repeatability is also key in detecting knock... though not good for the engine. But if you make at least two runs in your log, you can then reference them to eachother, so if you do see anything suspicious you have something to go back on to reference and back it up- or not. - rob

hustler

iTrader: (6)

there is a really good tuning shop in the DFW area, and for $350 max, it will make more power safely, and with a near perfect cold start in about 2 hours (cold start may take longer). I'm just going to drop the cash, and be done with it, and know its safe.

jayc72

iTrader: (14)

I'd rather learn to tune the car to 90% of it's potential than have someone else tune it to 99%. What happens when you need to tweak the tune? Spend more money on having someone else tune it? Take that $350 an invest in dyno time learning to tune it yourself. The original poster is doing exactly the right thing in trying to figure it out for himself rather than relying on FM to do it for him.

This is a great thread guys! I especially like the idea about seperating actual knock from ambient engine noise.

Jay

m2cupcar

iTrader: (5)

That's exactly what I'm doing. I was intimidated at first, but like anything, the more time you spend on it the better you get. And with all the right sensors/gauges/data it's hard to make any stupid tuning mistakes. I've gotten my fuel map fairly well tuned and have now started tweaking the IGN map. Another nice thing about DLL is the power graph. It's good for measuring the before and after of tuning changes. I've also got to get the EBC up and running- but that appears to be a wiring issue. Then I'll head to the dyno and let my expert tuner pal take over.

y8s

iTrader: (8)

Havent boosted high enough to see knock yet, sorry...

but a knock sensor is basically a microphone for engine noises. it responds to high frequencies of knock but since it's a wideband mic, it'll hear whatever noises the engine makes. a rattle here, a vibration there... etc.

m2cupcar

iTrader: (5)

If I had new miatas like you guys, I probably would have so many rattles.

Virus

iTrader: (3)

May I get you to look at this fuel map screenshot and give me ideas on what to change? When I'm in boost, I feel hesitation at points to redline.

y8s

iTrader: (8)

it's not so simple. you have to look at your datalogs and see where AFR is off-target and then make the changes.

Virus

iTrader: (3)

Sorry. I meant may I get you to look at the spreadsheet to give me an idea of what you would change to "smooth" it out? I just thought that doing this may make my car feel smoother...but I found out that my IC piping had come apart after installing new engine mounts.

y8s

iTrader: (8)

it might make it a little smoother. Just try to eliminate the steep transitions. better to err on adding a little than subtracting a little. if you have to go up or down more than 5%, try adjusting the adjacent cells.

best way to find out how close it is is to do a datalog with autotune off and see how it reads.

mcsinc

great info guys, keep it up. I'm learning to tune a TEC3, but the basic info is the same.