If the D term gets sufficiently negative it can.... I think?

I wish the MS3 would log the internal computed PID terms.

--Ian

 

do it, make sure all is fine and no glitchy-ness, and come on over

 

D is meant to keep spikes from happening. I don't see it creating those.

 

So I am running ms2 3.3.1 and I know it is different, but after struggling with closed loop ebc this is what helped. I noticed when set to basic slider mode the pid was set to all 100 IIRC. Everything I was able to find suggested starting from 100 0 0, which I had zero luck with.
My point is that I think we should start with pid values at whatever the simple algorithm sets them at for the firmware we are using.
I left my slider at 0 or 1 I think, and my solenoid at 19hz. I know people will disagree but I also had better luck with 30ms interval than 50ms and now I am actually running 15ms interval.
I am very happy with my closed loop ebc now, it reaches my target as early as possible and holds it to 6500. Granted I am on a completely different setup.

 

30ms did nothing. The valve can't react that fast.

 

D allows more P without instability and spiking so you get faster response.

 

P and D should be on all the time, with I being turned off when greater than some distance to target.

 

I realize the valve can only react at 51ms or so, but the way I see it the algorithm can run 3 times per valve cycle at 15ms. I could be wrong but it seems to oscillate less at a shorter interval.
My main point is that I think you are correct about starting from the basic pid settings.

 

Yes, but that's not the way the implementation in the ms3 works.

--Ian

 

1.4.0 is much better. The plunging DC% issue is gone. This is Basic mode, slider at 360. Bias duty is 33% across the board. I almost want to go higher with the slider or add more P with advanced mode.

 

Glad you finally upgraded and it is working much better.

 

Make sure you test different conditions. Heat soak. Cold. Etc. 1.4 is so easy to setup because the bias table is always in use. So PID doesn't have to do much in the conditions you tuned for. If conditions change that's when you actually need to tune PID.

Something I did was set the bias table off by a few percent in different directions along the rev range. Forcing PID to do some work. Then you can fine tune it.

With your baller wastegate you won't see the improvements in 1.4 as much. My hanky wastegate needs a lot more duty (10% more) randomly at 5200rpm. Then needs to go back down. So in 1.3.4 I had a nasty dip. With 1.4 I can tune it out with the bias table.

As sonofthehill stated. I'm not sure if the code changed. But starting at 100,100,100 for PID made a world of difference.

 

A quick change of direction, and a peek into the future. One of the many things we'll use this chassis to develop is a T5 swap kit, and this is the beginning of that development. The box is a -238 code from a 1996 3.8L Mustang, which is desirable for a few reasons:

• It's a World-Class box
• It has a 3.35 1st gear which means that you can swap the input shaft from a 3.35 V8 box into it with ease.
• The 0.75 overdrive is a big step down from the .843 in the car now, but not as deep as the .63 ODs in all the V8 boxes
• It has an electronic speedo sender, and with a Dakota box and a little work, it should talk nicely to the NB speedometer and MS3
• It was $130 at the junkyard, not $800 like all the V8 boxes are

I snagged this after confirming with G-Force that it's a suitable candidate to build later on. They actually said it's more desirable than an older V8 box as the basis for a built gearbox, since the newer V6 boxes are typically in much better shape.

For people who want a 100% bolt-in solution, you'll have to shell out $1100 for a new T5z box, but converting the V6 box to V8 spec is easy. You need the input shaft, bearing, bearing cover plate, and a shim kit, all of which can be had for ~$140 or so, which means you can put the gearbox itself together for well under $300 without any special one-time deals or hard to find parts.

The clutch disc was $110 and it's a performance organic disc with a 10-spline T5 center. It will get swapped for a 26-spline piece when I do a G-Force gearset next year. All the G-Force boxes are set up around 7.18" input shafts (V8), not the ~7.8" shafts that came in the V6s, so in order to accomodate built gearboxes, we'll build around the 7.18" length.

The bellhousing is cut off a blown 5-speed (everyone considering this swap probably has a blown 5-speed in their garage) and it will be milled down to the correct length and welded to an adapter plate which will be done in a few weeks. Using a chopped-up OEM bellhousing means I can retain the stock TOB, stock clutch fork, and stock hydraulics, so from the front, it will look like a Miata transmission with a 10-spline input shaft sticking through, and there are no issues with reliability or fitment on any of those critical items.

So that's the trans>engine attachment all sorted out. Next big puzzle is the shifter extension. The PPF bracket should be a little easier.

 

Excellent Andrew. Much happier to see you doing this than BEGI which won't ever really do it.

also interested in seeing what you come up with on the shifter extension...my biggest issue. But if you're requiring/expecting customers to get into the guts of their boxes then supplying a longer shift rod would be a simple supported solution.

And yeah, stay away from the .63 5th. I dislike mine a lot.

 

This is fukin awesome. Are you getting the adapter plate made? Or is there a t5 input plate readily available?

Do you see this as a future TSE project (should I go buy a t5 right now? )?

What are those t5's rated to stock? Are you doing a G-force gearset because of ratios or strength reason.

Will you be able to run the v6 gearbox even though it is the 7.8" shaft?

 

You should be able to run this straight to the MS3 and then run the speedo off the MS3.

 

Not necessarily expecting people to do that, since you can just buy any V8-specific T5 and it will work here, but I am anticipating that a lot of people will opt to buy the V6 box and swap the input shaft since you save several hundred dollars in the process. In fact, you can buy the box from LKQ (~$250), a Tremec rebuild kit ($330), plus the input shaft and requisite parts ($50), and probably pay someone else to assemble it ($300ish), and you'll still have a new V6 box with the better OD for less money than the V8 box.

I know you opted to do a shifter adapter and cut the tunnel to make it fit. What do you think about removing that freeze plug? I don't think it can be left open, but I think it could be sealed back up fairly easily.

 

I wasn't able to find one so I'm having an adapter plate machined.

It is something we will offer. It's clearly a needed product, and I will clearly need one for this car, so it will definitely see production. Whether we decide to do hybrid 5-speed bellhousings or cast our own remains to be seen.

My -238 box is rated to 265ft.lbs, same as the WC V8 boxes. I'll do a G-Force gearset for the ratios.

I am swapping a 7.18" shaft into my V6 box. See last post for details/cost

 

Saw that, the first post confused me about the input shaft lengths.

Excited to see this progress. Makes me want to go get my blown 5 speed and cut the bellhousing off. I did just get a TIG...hmmm. I need to find twin disc clutch discs for the t5 too I guess.

Either way this is going to be cool. I like that price. And maybe its time to take a stab at rebuilding a transmission.

Custom driveshaft? Or chop and weld the mustang and miata.

 

Custom. Hot rods guys get custom ones done all the time so they are readily available and relatively cheap.