accell enrichement problems

BimmerjimÂ® · Post by **BimmerjimÂ®** » Wed Aug 17, 2005 1:52 pm

tim, i am using 1.3 i upgraded a month or so ago. that made the noise a little less. but didnt fix it.

Post by **grippo** » Mon Aug 22, 2005 6:51 am

I have written code to provide averaging for all the ADC channels as much as you want since there will be user inputs for this. However I want to release it as a separate version and before I do I would like to get some data from people. Any ADC sytem will flicker. If you measure down to microvolts you will see constant jitter. MS2 has a 10 bit ADC instead of an 8 bit, giving 4 times the resolution. Just like with RPM - MS1 only measures to the nearest 100 rpm, MS2 to the nearest 1 rpm. So you will see the last 1 or 2 digits of rpm jump all over in MS2 and be rock steady in MS1. This doesn't mean there is more noise in MS2, just more resolution. Also, Bruce told me that if if you put a fixed voltage on an MS2 ADC channel on the bench, it doesn't appear to have any more jitter than MS1, so he doesn't think there is any internal electronic noise coming from the MS board design.

What I would love to see is someone do a side by side on-car (and maybe on-bench also) comparison of MS1 and MS2 tps and coolant. Then take these curves and convert them to common units and plot them in Excel. The reason for this is that it would really be useful to quantify the noise and maybe figure out where it is coming from. In work I see people cover up noise all the time by filtering it, but you should really understand the source before you do it.

Bernard Fife · Post by **Bernard Fife** » Mon Aug 22, 2005 7:30 am

Al,

Just to add to this thread, I have a relatively stable function generator supplying the rpm signal on the bench. I have good quality pots for the TPS, CLT, IAT, and EGO inputs. I have a car battery supplying the power, so it's pretty stable. The following are as reported by MegaTune.

With MS-II, the rpm varies by:

Â±1 rpm, under 3000 rpm,
Â±2 rpm from 3000 to 6000 rpm,
Â±3 at 9000 rpm.

The TPS varies by Â±0.5%. CLT doesn't vary at all (as an integer). MAP varies by Â±0.1 kPa.

With MS-I (and the same test set-up), the rpm varies by:

Â±50 rpm, under 3000 rpm (less if the generator isn't near the break point),
Â±100-150 rpm from 3000 to 6000 rpm,
Â±250 at 9000 rpm.

The TPS doesn't vary at all (though it's being shown in 1.0 steps, rather than the 0.1 of MS-II). CLT doesn't vary at all (as an integer). MAP doesn't vary at all (being shown in 1.0 steps, rather than the 0.1 of MS-II).

Lance.

Post by **grippo** » Mon Aug 22, 2005 8:11 am

Lance,

This is good data and confirms what Bruce said that there is no internal noise problem on the MS2 board. The only question I have is what does a step represent for MS1 - is it also a % like MS2 ?

Bernard Fife · Post by **Bernard Fife** » Mon Aug 22, 2005 9:30 am

Al,

Yes, in the latest version of MT, TPS is in % for MS-I as well as MS-II (I used MT225b596 for both).

I realized I should have made logs of the MS-II and MS-1 under the same inputs - would that help at all?

Lance.

Bernard Fife · Post by **Bernard Fife** » Mon Aug 22, 2005 9:40 am

Here are two datalogs - one with MS1, the other with MS2. The inputs are as close to the same as possible, the only change was the processor.

There isn't a lot in the datalogs that isn't captured above, the only thing is these are captured at about 20 Hertz, so more variation might have been evident than to my eyes, if it had been present.

Lance.

kalvinlk · Post by **kalvinlk** » Fri Oct 14, 2005 2:42 pm

Has there been any more work done on this front? I am finally getting around to smoothing the edges on my MS2 install, which means I need to lower my tpsdot value from 1.7 to something more practical, around 1.0 or so. I cannot put it at 1.0 because of the false accel enrichment triggers. Unfortunately, if the solution was implemented by a code change in the v2 branch, am I out of luck on the v1 code branch? Or would I be able to pretty easily copy the changed written into the v2 branch into the v1 and recompile?

Cullen

TimMoman · Post by **TimMoman** » Fri Oct 14, 2005 6:08 pm

For what my opinion is worth, and my error with the initial feedback, I agree that the sample rate is the issue. THe box is able to do the job a little to well. Smoothing the samples could be the next step. Just holding the value for every other sample might do the trick. Take the last, average with the current. Apply new average and write it into storage for the next sample averaging event. Iam having trouble with the laptop, but will give it a try. My code skills are really bad. So we shall see.

kalvinlk · Post by **kalvinlk** » Sat Nov 05, 2005 1:19 pm

Alright, I am going to play out in the garage today and try different capacitors and attempt to smooth out that TPS flutter, since I do not have the time to learn C. I'll post what I find!

kalvinlk · Post by **kalvinlk** » Sat Dec 31, 2005 11:42 am

I've been away from my car and the megasquirt community for a couple months due to a hard term at school. Has anyone done any work on this sensor averaging? I searched, but could find nothing. None of the changelogs for the code show it has been added there. I am building a new engine for my car, and would like to get my install fine tuned on my current (trashed) engine. Thanks.