BBro

Spaceman, thanks for that article. I never even thought to see if there was any published research on this topic, it was an interesting read for sure. I think modeling is awesome but doubt it would really help in my situation. A mode does exist for tuning the bias (feed forward) table and I have done this. I am very confident with the overall shape of my bias table but with changing conditions the overall values can be 20% or more apart from what it was on a previous day. I'm definitely going to try some of your PID tuning techniques because I think a good PID is required to makeup for the bias table being set conservative, (middle of the range of values required in normal conditions). One thing I have noticed when talking to some friends who work in engineering control systems is they are used to seeing P and I values in the 2-5 range meanwhile megasquirt you see stuff up to 200 quite often. The megasquirt manual also advises to use very little if any derivative gain but I have found it necessary to include a good bit in almost all of my PID controls.

HeavyD, I agree but I think the issue is the closed loop control, when at high sensitivity levels, targets max boost for so long and doesn't drop the DC fast enough to allow the system to react and then expel the gas. You can see this in Run 6: the DC is held at max during initial spool-up until the delta of 30kpa is met and it immediately drops to the value in the bias table however it is still not at target boost so it starts immediately increasing the DC and before it can react it has already hit over-boost. This is mainly a problem when dealing with the boost threshold after the lower limit delta has been met because the controller will continue to increase the DC even though it is not possible to hit the target at that given RPM and then once it is possible to hit the target the DC is so high it immediately fly's right past the target and into over-boost.

DNMakinson

iTrader: (3)

Not seeing your target table, but based on your comments, I’d suggest that you match your targets at lower RPM’s to what the turbo can support so that you don’t wind up the controller.

BBro

Target table is just 190kpa across the board, and I agree with you that adjusting it should be the next move. I remember Savington talked about it being important to make your target table an achievable value in every cell, but some people disagreed so I never tried it. Based off my findings I think you do need an obtainable value because having the target at an unobtainable value results in uncontrollable overshoot. To tame the overshoot I end with a small delta and very insensitive closed loop control however the downfall to small delta and insensitive control is that when the bias table is off due to changes in weather the closed loop control lacks the sensitivity to compensate.

BBro

Well, that's all she wrote for me for a while. My whole life I've wanted to move out west and I got an offer interning for a certain arms manufacturer out in Prescott AZ this summer.

It will be some time before I get to really dig into the car again since all summer I am in Arizona and then all year I will be in Alabama and then hopefully back out west the next summer too. Luckily I only have 3 semester left.

For now I'll be enjoying the view.

IMG_20190505_131629

DNMakinson

iTrader: (3)

Arizona is gorgeous and diverse landscape. Enjoy. Do well. The son of a friend will be starting Engr at Bama in the fall. Full scholarship as your situation. I’ll have him look you up.

BBro

Thank you! Definitely tell him to give me a shout once were down there, I can offer him any advice I have.

BBro

I have a big update post that needs to be made soon but lets skip forward a bit. Had a track day yesterday and experienced the cam position sensor failure after about 10 hot laps in a row. I'm assuming my oil temps were huge my oil cooler isn't ducted well and it was like 100 degrees out. Coolant temps were 230 and 235 is where my CLT based rev limiter starts. I had to get towed off track and then the car started fine after I let it cool. Even did some more laps later in the day. Here is the log of the failure happening.

Ordered a new Beck-Arnley sensor
I was reading threads years ago where people were adapting the evo sensor but nothing concrete ever happened. Has anyone come up with a better solution or should I just hope this new sensor doesn't have the issue as well?

Really need to fully duct all my coolers but that will have to wait till I can get a solid week to leave the car on stands at the garage back home but with me having the car down in Bama all school year and working random places over the summer who knows when that will happen.

BBro

Overdue for an update: Some things done a long time ago but never updated: new radio/amp and door speakers. Was never happy with the audio in my last setup and now its all I could ask for out of a Miata and it looks nice and understated. Also flushed coolant, and installed a hallman mbc. For the mbc I replaced the coupler on the coldside IC piping for a couplet with npt ports to use as post intercooler boost reference for the MBC. Heres some beauty shots.

Shameless plug- my Instagram is bribrode and I post a lot more of the car and small updates there.





20191010_085010


20191010_084926


20190902_153207

BBro

Just been enjoying the car. Tune is about perfect now besides some startup dips which I am curing with more cranking duty.

Finally got idle to be about perfect by fixing my fan engagement. Solution was to add way more IAC percentage than actually needed but this allows it to engage smoothly. Previously I had it adding 4% because thats how much extra it wanted with the fan engaged the problem was the engagement was still a huge dip sometimes killing the car. By adding way more IAc % than needed it engages smoothly and my CL pulls it right back down to where it needs to be anyways. I have it set at 10% now.

I also lowered my throttle closed TPS% which helped some stalls when coming to idle after using very little TPS. Super happy with how the car drives now. Feels so much like OEM it is scary.

Enjoy eye ****:

Padlock

abosolutely love how clean this car is

BBro

Thanks padlock! I always enjoy keeping up with your build too. Can't wait to see more k-swap progress.