28-Jan-00

I extracted the timing advance curve of my engine. Some people have said one reason the 90hp engine has lower output than even the U.S. Auto-tranny 1.6, which have the same cams and compression ratio, rated at 105hp, is that the ECU restricts topend power. Incidentally our Miatas are sold here with a 95hp rating, not 90hp. This may be because of the lack of a catalytic converter. But still, there's a 10hp shortfall.

 I still doubt that these engines are really only 90hp. The bone-stock dyno measurements show 72 rwhp,, which goes up to 78 rwhp if the timing is set to 16 degrees. A U.S. 116hp Miata measured with the same dyno yielded 86 rwhp at 14 degrees timing. If you add the +9% S.A.E. correction factor for temperature and humidity, that's 85 rwhp for the "90hp" engine, and 94 rwhp for the 116hp engine, which jives with Dynojet figures in the U.S. That's a 9hp shortfall, so if the US spec engine is 116hp then the version we have is 107hp, quite close to the 'matic-tranny version engine of 105hp.

(incidentally the same US Miata I dynoed showed 97rwhp after the JR CAI/headers/exhaust, which is 106 rwhp after the +9% correction - again this jives with the Dynojet figures of this setup.

Anyway, how could an ECU restrict power at the top-end? By making the mixture either too rich or too lean from optimum, or retarding the spark advance from optimum. Too lean can yield detonation, too rich could cause carbon buildup. Retarding the spark will yield high exhaust gas temperatures which could overheat the valve seats. None seem to be a proper way of reducing power with the ECU. If indeed the power was reduced from 105hp to 90hp, the "proper" way to do it would be to make the cams even milder, which would at least increase low-end torque, or to place an intake restrictor.

Another question I had is why does the stock engine's power drop so abruptly from 6400 to 6750 rpm? It's almost as if it's engineered in to make you want to upshift before the revlimit.

 So I logged my ignition advance on WOT to compare with the curve for the 116hp engine in Norm Garrett's book, as follows:

I used my trusty MiniDisc recorder, connected one channel to the Crank Angle Sensor (CAS) output, (after my MSD) and the other channel to the ign terminal in the diagnostic connector. I checked the schematic in the book, the CAS signal goes to the ECU, then an ECU signal(s) goes to the ignitors, which has a wire that goes to the diagnostic connector. So the MSD is taken out of the equation.

I used Matlab to extract the info in the resulting pulse train:

I took 3 readings:

-In neutral, slowly stepping on the gas until redline, trying to keep vacuum above 10 inches at all times.

-In neutral, WOT.

-Going up a fairly steep hill in 2nd gear, WOT.

Here is the uphill run result:

The jagginess is due to "sampling jitter". Just filter it out visually (draw an imaginary smooth line through the thickest portions). Proper filtering can be time-consuming to program so since this is a one-time operation I left it at that.

Take note that the AFM reading would have an effect on the timing, and because of the turbo, it would probably max out at a much lower RPM than stock. The above run was taken with a 6 psi setting on the wastegate. It was uphill so I probably hit 6 psi by 4000 rpm. Also, assuming that the AFM maxes out near max power of a stock engine (~80 rwhp), and since I get 80rwhp by 4000 rpm with the turbo, it's safe to assume that the AFM in the above run was definitely maxed out by 4000rpm. Note that my revlimit is 6750 rpm and not 7200. Probably useless to rev it that high with the mild cams we have. I hope to get around to doing this extraction on a stock Miata.

Some of the features of the above curve is that the ignition is retarded below 1800 rpm. If you add in the 10 deg factory static timing, it's just barely 0 degrees at 1000 rpm. The bump in the curve from 600 to 1000 rpm is probably where I stepped on the gas. The timing curve is fairly flat with a small kink from 3000 rpm to 500 rpm, then it drops sharply to 5500 rpm then rises a little bit.

I then got curious as to what the timing was in units of time, and not degrees, here it is:

Note the Y axis is in milliseconds. There's something a bit suspicious about the timing curve from 5500 to revlimit. Seems almost like a simplification.

But, it's there even in the no load curve. The no-load curve is roughly the same shape but shifted up 15 deg.

How does it compare to the '90-'93 (116hp) Miatas? I don't know yet, until I get the graph from Norm's book.

On the stock cars bumping up the timing from 10 deg to 14 degrees improves the whole rpm range a lot, from 14 to 16 helps the low-end the most (but a small improvement only), the top-end slightly, and the midrange hardly changes. Going from 16 to 18 helps again the low-end a bit, the top-end slightly, and the midrange begins to drop. Now we can see how retarded the bottom-end is, no wonder it likes the static timing set so advanced.

The U.S. Miatas are supposed to have best top-end at 12 deg, and best low-end at 18 deg, and would only show a +2hp change going from 10 deg to 14 deg, but the measurements I took showed +5ft-lbs in the low and midrange, and +5hp up top.