Building a Microsquirt Daily Driver

[Piledriver]

Once I remembered to import my old MS2 Lambda delay table my VE table was close to constant after a few extra on ramp "pulls" while commuting to work. (after swicthing to MS3)

Getting it right is the difference between VEAL working "~OK" (maybe a little better than MLVs autotune (with a single delay value, which is obviously wrong >90% of the time) and working dead nuts consistently.

The sample delay is a bit different on MS3, the 0-9 sample delay options are insufficient, making MLV useless other than for viewing logs and manual touchups for MS3 users. (of course viewing sdcard logs at ~300Hz sample rate can be quite useful sorting issues)

Of course if MLV had the ability to use the table from TS//VEAL, you would not really need VEAL: the results would be as accurate if not moreso simply post-processing the logs.

[turbobaja]

Once I remembered to import my old MS2 Lambda delay table my VE table was close to constant after a few extra on ramp "pulls" while commuting to work. (after swicthing to MS3)

Do you mean that VEALS wasn't making anymore changes?

I've been putting miles on it running VEALS and its running great! But I'm still a little lost on where to head with my Lambda delay table #s. The AFR gauge generally shows exactly what I want it to with VEALS running in a given load/rpm range, despite large changes in delay values. There is no "burn" key on that table for me, I'm not sure it's even storing my changes.

When I first ran VEALS it developed a trench in my VE table right at 2,500 rpm. It was a brief test drive so I assumed that trench needed to be smoothed out across the table, so I manually smoothed it before the next test drives. I also re-worked the AFR tables to lean out a little more under low map cruise. It drove pretty good, so I started letting VEALS tune again. It immediately dropped the VE table right at 2,500 rpm again. It seems to reveal that this engine has a major efficiency problem in that range and literally chokes itself off 1,000 rpm later....virtually useless after 4K . However, it makes great power from 2,800-3,600 rpm, and cruises great on a lean tune at any speed. Problem is every up-shift puts it just under 2K rpm and it has to pull up through the VE butt-crack @ 2,500 to make good power again.

Here's my table after about 50 miles of mixed driving w/cell change resistance on Hard. Stock intake trumpet on the air box.

[Piledriver]

With correct Lambda delay and injector dead time, the few changes made by VEAL are tiny once the table is well tuned, basically only "noise".

You want to initially tune on "easy" and work your way up as you close in.

This is even more apparent on MS3, where the VEtable is in tenths (0.1 AFR resolution)

You should have a checkbox for "update controller" and a "send" "burn" button top left of the VEAL screen.
(MS2 might be a little different, has been awhile, but I'm pretty sure that's ~same)

[MarioVelotta]

Karl, does the afr stay consistent through that valley? I can't say I've ever seen something that drastic before.

[Piledriver]

Starting to look about right, I assume you have sample averaging on and the "dot" tracks straight across now?

You may want to scale your reqfuel down ~50% and double the VE table values to increase your effective resolution at the bottom. (you may be able to go >50%/200%)

You should limit the max value in the vetable to ~230 (at least on the older FW)
IIRC the VE table max value may now allow a LOT of headroom with current firmware (could just be MS3, but maybe not)

[MarioVelotta]

You may want to scale your reqfuel down ~50% and double the VE table values to increase your effective resolution at the bottom. (you may be able to go >50%/200%)

Can you please explain this a little more

[Piledriver]

Your lower # VE values are small enough (esp. arond idle) that a single digit change is a significant % change.
Rescaling reqfuel and the vetable can double (or more) your effective resolution.

The easy example is divide reqfuel by two and multiply the whole ve table x2.
Looks like you could divide/multiply x3, assuming MS2s vetable limit is still 255 (want to stay ~230 max)

Here are my current tables FWIW: Stock T4 2L runners on widened plenum, stock TB with only an air filter on it. 1.8L T4, 914 2L heads, web73, non-manifold vac referenced (fixed) fuel pressure. Note that MS3 vetable is already .1 resolution unlike ms2.
(otherwise I would do the scaling previously mentioned for MS2, max vetable value on MS3 is now 16,384, rather than the old 255 IIRC, not sure if the latest ms2 code also changed to match, possible, easy to test if you have one running)

The VE "variation" is quite real and you SHOULD see it: effectively visualizing the actual tuning of your intake/cam/exhaust.
Note the arrangement of the rpm bins, carefully concentrated around pktq and pkhp (the system tune gets "busy")

EDIT---I just realised those are not really "current" I have been trying to track down an intermittent CLIdle bug with the laterst FW, loading old msqs al lot lately, but it's still a perfectly good example, showing a peak at 2100, pktq ~2600, and probably pkhp at ~5200 2X those RPMs. (kinda overlapping) I can add a "trumpet" to the TB and see everything move left (shift to lower RPMs) when running VEAL again...

I recently changed back to the stock 45mm TB as it was in much better shape than the modified 50mm Vanagon unit I was running: This also shifted my current tune enough I probably need to rearrange the rpm bins some.

the really low ve#s in the 30 kpa row are due to fixed fuel pressure, if it was manifold vac referenced the #s would be a lot flatter bottom to top, at the cost of making the accels harder to get right.(IMHO turn accels OFF until your ve table is right, then just turn on EAE...)

The trick to get uploaded pics to display properly here is:
800X600 max, also 125kB max. Max 3 per post.

[MarioVelotta]

I understand the peak torque and peak HP. but it's the scaling I don't understand fully. If you half the req_fuel and double the VE table you would still end up with the same number in the end.

[turbobaja]

Thanks for sharing your tables Piledriver. Very interesting how yours show a similar dip through the rpm range. I understand what you're saying about scaling up the VE #s for better accuracy. When tuning with VEAL the "cell change" value reads into the .XXX range, even though the main table displays only whole numbers.

Mario the AFR stays constant through that 2,500rpm valley. Before VEAL tuned it that way it would run VERY rich AFRs through that rpm range regardless of engine load. Here's a couple scatter plots on the current tune.

[Piledriver]

http://www.mobiloil.com/USA-English/Mot ... _Guide.pdf

I have been running 0-40 M1 and Love it. Very quiet valve train compared to brad pen, better pressures than the Amsoil before that.

I won't even use Rotella-T 15W40 anymore and it has slightly higher zinc/phos numbers than that 0W-40.
Life is a continuous process experiment, I guess, IIRC Web cam (nitrided) what lifters?

Seems actual oil viscosity has a large disconnect from the SAE ratings on the bottle.

I would run the Turbo-diesel Truck 5W-40, if forced to use Exxon-Mobil products. It's also still a true G4/G5 synthetic, not G3 AFAICT.
The 5W-40 Rotella T6 is a G3, but I still prefer it overall, if only due to the TDiclubs interesting long term UOA wear results.

[Piledriver]

I understand the peak torque and peak HP. but it's the scaling I don't understand fully. If you half the req_fuel and double the VE table you would still end up with the same number in the end.

Yes, but if you are (for example) at 20 at idle, a single digit change (smallest change possible for MS1/2) is 5%.
If idle is at 100, +/- 1 digit is 1%. so 5% vs. 1%---
2X-3X Better overall resolution and tuning accuracy is easily possible.
For free.

You will need to retune to take advantage of it, and the accels likely will need tweaked.
(I certainly did not invent this, it was recommended by the devs on the msextra forums ages ago)

I'm not sure 255 is the max VE anymore on MS2 (latest FW) I know the latest alpha MS3 firmware they essentially uncapped it as they had room. (I'm not quite sure what to do with a 16 bit VE table, 1/10ths are sufficient, but you never know)

If MS2 max is still 255, limiting the table to ~220ish avoided headaches in the past.

[Clonebug]

My VE Table is running from 70-144 at this time if that means anything. I only have the kpa areas below 100 tuned to any degree though.

Kpa range is 30-200 at this time. I was starting to think mine were to high and I needed to rescale but reading this I will leave them where they are at and work with them.

[Piledriver]

Once I remembered to import my old MS2 Lambda delay table my VE table was close to constant after a few extra on ramp "pulls" while commuting to work. (after swicthing to MS3)

Do you mean that VEALS wasn't making anymore changes?

Pretty much.

Here's todays "results" run on "hard" (which really isn't all that "hard" on MS3 with the latest beta TS) after some manual smoothing.
extended 90 minute commute today, small town, rural roads and highway/city traffic.

The cell with the info "bubble" was the largest change, and you can almost see exactly where i did the smoothing. It fixed it ~back as it was...

veal-changes-60km-commute.jpg

[turbobaja]

Very cool!

This little torquer has been on the road for the last couple weeks, still waiting to run through it's first 10 gallons of fuel to get an idea of MPG on the current tune. I'm gonna test compression and leakdown this weekend to get an idea of how it's breaking in, haven't pulled a spark plug yet even.

As much as I'd love to re-scale the VE table on this thing, I'm probably going to wait to get that "fine tuned" until I'm tuning on my baja . She's happy with it already, but I know there's room for improvement. It's got a slight bucking problem I want to iron out in the very low map on/off throttle at slow speeds. It pull so much vacuum, it's really responsive to slight throttle input...moves that heavy '74 Super pretty darn good

[Devastator]

It's got a slight bucking problem I want to iron out in the very low map on/off throttle at slow speeds.

Datalog it in this area and look for Accel/Decel events.