Improving sensor inputs / stability of 5V+12V, Ms2 V3

[Dennis_Zx7r]

I'm currently trying to improve some sensor readings, or more specifically the noise.

These sensors are 2 barometers, one connected to JS4 and the other to JS5.
I built the circuits like described in the manual. First I used .22µF capacitors and 1k resistors. I wasn't satisfied with the stability of the sensor signal, so I tried increasing the capacitors size and the resistance. Biggest values used were 100µF (yes, 450x the recommended size) and 10k.
I then tried capacitors between .22 and 2200µF (yes, 2200) on the voltage supply of the sensors, in trying to stabilize the +5V.
It did not make ANY difference regarding noise. The signal just got very slow, as expected.
I learned yesterday that the type of capacitor (ceramic for example) might make a difference. I used polarized ECaps as the bigger ones, if this could be an issue.

This has led me to the conclusing that the noise is entering the system after JS4/5. From datalogs, I discovered that the signal is way better with the engine off. This picture shows this representatively.

Sensorrauschen.png

You can clearly see where the engine is being started.

The displayed baro is mounted in the proto area. I only have logic outputs for the ignition (the FET variant as in the manual), and the injectors are 14Ohm each. There is nothing drawing power from the MS besides this but the sensors (TPS, CLT, MAT, MAP, Baro1+2) and a bluetooth module. I already tried this with every other sensor and the bluetooth module disconnected, no improvement.

---

1) What I haven't tried yet is stabilizing +12V, which I think the processor uses as supply. I know the capacitor in boot seems to kind of filter out noise already, but I guess enlarging it would also smooth the battery voltage signal even further. Could this be beneficial? In this case, which mod be most effective?
Or are there hardware limitations on the Processor (Ms2, V3 Board) which make all of this futile?
Or may I have another issue I haven't thought of?

2) Especially for the constant baro, a smoother signal would be beneficial. I can get an nice signal with a stronger lag factor, but this unfortunately also affects MAT and BattVoltage, therefore this is no option. As an alternative, is it possible for me to modify the code so that only the baro is smoothed with a high factor? Smoothing it in MLV works fine (green vs. red graph), but that's not an option when I want to use the signal for correction in live operation.

BaroSmoothed.png

[DaveEFI]

Large electrolytic are not designed to remove all noise on a line. They will, however, smooth out any voltage changes. But can't supply a rock steady voltage for ever. For noise, you'd generally add smaller caps - something like an 0.1 uF ceramic is parallel with it. But both of these will be most effective as close to the load as possible.

MS itself doesn't produce a regulated 12v supply. Very difficult thing to do from a 12v source. Hence the important one being 5v.

[Dennis_Zx7r]

Large electrolytic are not designed to remove all noise on a line. They will, however, smooth out any voltage changes. But can't supply a rock steady voltage for ever. For noise, you'd generally add smaller caps - something like an 0.1 uF ceramic is parallel with it. But both of these will be most effective as close to the load as possible.

Alright, I'll try one of those soldered onto the sensor directly then for the supply voltage.
Should I try that for the input pin of the processor, too? Not sure if this counts as load.

[whittlebeast]

Change your baro scale to say min of 0 KPA and max of 105 KPA to put the noise into perspective.

[Dennis_Zx7r]

As requested

BaroSmoothed2.png

I know it's not that bad, but it could be way better if I could somehow smooth it out. Especially because this is a value which is expected to change only very slowly.

[ol boy]

My baro sensor is all over the place too. I did the same thing. Keep adding bigger and bigger caps to both the 5v input and signal output to have no affect. I'm wondering if noise is introduced closers to the processor. I'm on ms2 though 3.0V board.

Sent from my SM-G920V using Tapatalk

[ol boy]

What idle control type are you using? I wonder if my 2 wire fidle valve is causing some weirdness. The frequency that the fidle runs at is now 3x to 6x higher than what was available a few years back on 3.0.x to 3.2.5 firmware.

Sent from my SM-G920V using Tapatalk

[whittlebeast]

Post your latest MSQ and matching log please.

Andy

[Dennis_Zx7r]

What idle control type are you using? I wonder if my 2 wire fidle...

None.

@whittlebeast, in case you meant me:

Old Msq, but matching the log
https://dl.dropboxusercontent.com/u/106 ... 0_1030.msq
Last available log
https://dl.dropboxusercontent.com/u/106 ... -21-08.msl

[ol boy]

Got a quick log this morning with the msq.

baro.msq

2016-09-07_06.49.01.msl

I have to have the lag factor down around 40 for normal driving, would like to get it up to 90 but baro swings to wildly at that level.