PWM Closed loop idle
 

Anyone try PWM CL yet?
I managed to get a stable idle on PWM warmup, but temps have been between 25 and 38°C overhere so I don't know what to expect when it starts getting colder.
IOW, should I start playing with CL or is Warmup fine?

I had assumed that pretty much all of us were using closed-loop PWM idle. It works pretty well for me. Had a problem when my TPS sensor started sticking and it wasn't falling below the threshold, but apart from that I've had no complaints. My idle is stable at around 1300 RPM cold, 900 RPM warm, from 40°F to 100°F (roughly 4.5°C to 38°C).

I spent a lot of time trying to get closed loop idle working with MS2 and never got it working better than PWM warmup, which is rough but effective.

On a lark I ordered this PWM Converter board a few days ago; when it arrives I'm going to see if it helps. The main problem seems to be the tiny effective range of duty cycles. Maybe it's worse in MS2/e...

Anyway, to answer the specific question, I tuned the PWM warmup back in February in -5C weather and it's still working OK in 30C weather. It just uses less of the table since it starts warmer. So you might need to tweak it a bit but it won't be wasted effort.

Was thinking of ordering Jean's board as well, hoping to get rid of the iac hum. Let us know how it works out.
To be honest, I'm a bit overwelmed by all the settings for closed loop, so if it isn't significantly better than warmup, I'm not going to bother with it for now.

We just hacked the INI file to allow for a larger multiplier. Worked OK but not perfectly. We've got one of Jean's PWM converter boards on the way to test it out too.

hahaah - wow, I didn't think of that. I'd assumed it was a hardware thing. I never cease getting caught off gaurd by... interesting... issues in the .ini.

I'd really love to see if it helps. I've seen stuff the code does which DOES NOT MAKE SENSE.

But the folks on the MSextra forums literally don't want to hear it. It's been on my list to get that board, but, proportional-only-control not being proportional seems to be sort of a wrench in the works. Obviously, it can be made to work, and I shouldn't be so upset that what they call PID isn't PID, even the second time around... Still, it just doesn't sit well with me.


Joe: MS-I PWM/CL works well from all reports. No one I have ever talked to has MS-II idle working well except the people who wrote the code. :-) Actually, other cars seem to get along ok.

I've used warm up and found it to be awesome. I never have ANY issues with it. The only downside is I idle at 1,050 rpm to give myself buffer for the lights on, stereo bumping, power windows, fans, heater all running at once thing.

The INI had that limit because closed loop doesn't work all that well at high frequencies. It works sort of, but not spectacularly well.

Here's what Ken said about the why of it:


I'm going to allow higher numbers in 2.1.1, but keep in mind that the higher the number the lower the likelihood that the closed-loop algorithm will work due to reduced resolution as you increase the frequency.

This is both a hardware and software limitation:

500 Hz would require a multiplier as high as 17. With the multiplier this high, your accuracy will be 6.8%. This means that the valve will only change duties in 6.8% increments.

This is because we don't have a hardware PWM pin available to run the idle valve, and the software algorithm that we had to use instead gets accuracy of .4% at 30Hz. Each time you increase the frequency you decrease the resolution by .4%.

I knew the accuracy went down, but I'd assumed that was the limit of the MS-II's ability to flip the state of a non-PWM pin. I wanted to play with it to see if I could tell the difference. Anyway. As is, if you set it for P-only control (valve position proportional to the difference in RPM ACTUAL and RPM REQUESTED), you can raise and lower the RPM and not see the valve move. And I don't mean physically, I mean, MegaTune reports that the valve is happy to sit where it is.

Moving the target (with the idle at, say, 1200) from 700 to 800 to 1100 to 1400 to 1600 makes no difference. All the way from 0 to 5000, the function seems to only move the valve during the times the RPM is changing. It's complicated, but, if I put the brake on, drag the clutch, the idle valve's position will move ONLY during the time period the RPM's are dropping. Once they are stable, it sits still. This seems like a D-only control, but a very insensitive one since small changes (uneven idle) has no effect.

With things like this left unexplained, I'm reminded of the times I redid my entire wiring on their suggestion only to find out my problem was a software setting. On and on.

In short, I would love to see someone try this, but I won't be running out to do so, the problems I'm seeing are MUCH more basic than the accuracy of the valve. Maybe it's my own fault, but there's no way I'll find that out since other than telling me to go buy more hardware there's little explanation.

At some point, I'll read the code again, but I've been busy messing with things that work.