Page 20 of 21

Re: "Noisy" tpsDOT

Posted: Tue Apr 24, 2012 2:21 am
by jsmcortina
robs wrote:I started with Peter's "quick" check: unplug the MS and check that the ground terminals of each of the engine room sensors -- TPS, MAT, CLT -- have no path to the battery -ve. The only path to ground should be via the MS -- in at the sensor ground pin (pin 19 in my case) and out by the other connected ground pins. So the sensors should be completely isolated if the MS is unplugged. If there is a path to ground, something's wrong.

(snip)

[*]I think the installation manual could be improved by including Peter's simple check when the harness is about to be plugged in.
I'd added the note to the manual last week
http://www.msextra.com/doc/general/grounding.html

Where do you think this page should this be linked from to help direct most users there? (I've added a few links to that new page, but not that many yet.)

James

Re: "Noisy" tpsDOT

Posted: Tue Apr 24, 2012 5:30 pm
by robs
gslender wrote:Rob, do you have a before / after log to show how the changes to GNDing all sensors via MS made a difference? Is it just a case of "do it" and you will be amazed?
Not amazed and, with luck, you've already got it covered. As an example, here's a before/after of CLT:
clt.jpg
I guess it should have been obvious that a jitter of .7 degrees or so was way too much. Like I said, it was my first installation and, besides, that amount of change wasn't going to affect the PW much. Still good to see the smoother graphs.

I would have posted TPS except I haven't updated the .ini file to give tenths yet. Integers are a bit chunky.

Have fun,

Rob.

Re: "Noisy" tpsDOT

Posted: Tue Apr 24, 2012 6:07 pm
by robs
jsmcortina wrote: Where do you think this page should this be linked from to help direct most users there? (I've added a few links to that new page, but not that many yet.)
Hard to say. When I did my installation I followed the B&G manual instructions so have never worked through the process in MS2/Extra. As I recall, B&G had the wiring diagram at the end of the Megasquirt assembly instructions. That seemed logical to me -- having built and tested it, get it installed -- so a link in the corresponding place might be an idea. But then the assembly page won't be visited by people who've bought their Megasquirts pre-built, so having this page accessible under a general "getting it wired up" heading would also seem appropriate.

For a beginner faced with Megasquirt, though, there is a lot to take in and quite a breadth of knowledge needed to make sense of it all. Perhaps a "top 10 installation pitfalls" page would be a worthwhile addition to the documentation -- a list of common mistakes and links to the parts of the existing doco describing the right approach. I think most new users would be happy to work through such a list and it might help structure their reading through what seems to be a great wall of documentation. But I suspect making such an "easy start" page would not be as easy to write as it seems on the surface.

Have fun,

Rob.

Re: "Noisy" tpsDOT

Posted: Tue Apr 24, 2012 6:44 pm
by Peter Florance
robs wrote:
jsmcortina wrote: Where do you think this page should this be linked from to help direct most users there? (I've added a few links to that new page, but not that many yet.)
Hard to say. When I did my installation I followed the B&G manual instructions so have never worked through the process in MS2/Extra. As I recall, B&G had the wiring diagram at the end of the Megasquirt assembly instructions. That seemed logical to me -- having built and tested it, get it installed -- so a link in the corresponding place might be an idea. But then the assembly page won't be visited by people who've bought their Megasquirts pre-built, so having this page accessible under a general "getting it wired up" heading would also seem appropriate.

For a beginner faced with Megasquirt, though, there is a lot to take in and quite a breadth of knowledge needed to make sense of it all. Perhaps a "top 10 installation pitfalls" page would be a worthwhile addition to the documentation -- a list of common mistakes and links to the parts of the existing doco describing the right approach. I think most new users would be happy to work through such a list and it might help structure their reading through what seems to be a great wall of documentation. But I suspect making such an "easy start" page would not be as easy to write as it seems on the surface.

Have fun,

Rob.
I could add to my MegaSquirt and Getting Started Powerpoint

Re: "Noisy" tpsDOT

Posted: Wed Apr 25, 2012 11:04 am
by bubba2533
Rob,

Not sure if you will be able to answer this because I'm not sure exactly which sensors were correctly grounded and such.

But I have all of my sensors grounded at the MS except for the IAT sensor which is going to the body. Do you think that will effect my other sensors such as my TPS? Or will the sensor that is not grounded correctly be the one with the noise and the other ones relatively ok?

I will be grounding it to the MS when I have time, I just wanted to see what your situation was like.

Thanks,
Branden

Re: "Noisy" tpsDOT

Posted: Wed Apr 25, 2012 11:56 am
by jsmcortina
Of all the sensors that are important to ground correctly, the IAT is at the top of the list. False changes in IAT will directly impact fuelling.

James

Re: "Noisy" tpsDOT

Posted: Wed Apr 25, 2012 9:50 pm
by robs
bubba2533 wrote:Not sure if you will be able to answer this because I'm not sure exactly which sensors were correctly grounded and such.

But I have all of my sensors grounded at the MS except for the IAT sensor which is going to the body. Do you think that will effect my other sensors such as my TPS? Or will the sensor that is not grounded correctly be the one with the noise and the other ones relatively ok?
Being an electronics newb I'm sure I'm not the best person to answer this, but I don't think the incorrect grounding of one sensor will affect the others. Does depend on how you wired it though. I followed the original Megamanual wiring which had all sensor grounds coming to one pin. If this pin also had a path to a "wrong" ground, that'd mess it up for all sensors. Looking at the MS2/Extra manual's wiring, each sensor's ground comes back on its own pin, so that'd be safer.

As James says, IAT/MAT is an important variable. If you have a NA engine, IAT doesn't change all that quickly, and it would be reasonable to set a low lag value on it until you can get the grounding straightened out.

Have fun,

Rob.

Re: "Noisy" tpsDOT

Posted: Thu Apr 26, 2012 9:46 am
by bubba2533
I have the ability to swap the stock ECU in, cause I haven't had much time to tune and I need to have a reliable DD.

Re: "Noisy" tpsDOT

Posted: Sat May 12, 2012 6:14 pm
by Hahns5.2
So far my biggest breakthough with TPS noise has been moving the TPS ground to the MS ground (I have piggyback install with the stock PCM). With the gslender 3.2.1 2.3c firmware I can run a TPS lag factor of 100 with a sliding window of 3 and run a TPS threshold of 15 without false triggers, before the sliding window had to be 7-8 and the lag factor had to be 95 and I could only get the threshold down to 30.

Re: "Noisy" tpsDOT

Posted: Sat May 12, 2012 6:24 pm
by gslender
Hahns5.2 wrote:With the gslender 3.2.1 2.3c firmware I can run a TPS lag factor of 100 with a sliding window of 3 and run a TPS threshold of 15 without false triggers, before the sliding window had to be 7-8 and the lag factor had to be 95 and I could only get the threshold down to 30.
Do try the later fw mods, as the later versions have a better system of TPS filters (over sampling) that produces the same outcome, but does so without any response penalty. Definitely worth a look.

G

Re: "Noisy" tpsDOT

Posted: Sun May 13, 2012 6:16 am
by Peter Florance
robs wrote:
bubba2533 wrote:Not sure if you will be able to answer this because I'm not sure exactly which sensors were correctly grounded and such.

But I have all of my sensors grounded at the MS except for the IAT sensor which is going to the body. Do you think that will effect my other sensors such as my TPS? Or will the sensor that is not grounded correctly be the one with the noise and the other ones relatively ok?
Being an electronics newb I'm sure I'm not the best person to answer this, but I don't think the incorrect grounding of one sensor will affect the others. Does depend on how you wired it though. I followed the original Megamanual wiring which had all sensor grounds coming to one pin. If this pin also had a path to a "wrong" ground, that'd mess it up for all sensors. Looking at the MS2/Extra manual's wiring, each sensor's ground comes back on its own pin, so that'd be safer.

As James says, IAT/MAT is an important variable. If you have a NA engine, IAT doesn't change all that quickly, and it would be reasonable to set a low lag value on it until you can get the grounding straightened out.

Have fun,

Rob.
If you only ground a sensor to the block or car and not MS, you will mostly likely only effect that sensor. If you ground it to both MS and car/block, you can effect all sensors.

Re: "Noisy" tpsDOT

Posted: Wed Jul 04, 2012 10:14 pm
by robs
A few things have stopped me participating here for a while. Most relevant to this forum is that my Megasquirted car was run into. No injuries, and the damage isn't all that serious, but because of the state's write-off laws, I may not be able to keep the car. The "good" news is that her insurance company look like they aren't going to pay up, so either I chase them in the courts, or repair it myself at my own expense -- but at least I get to keep this car I've had for >20 years. In the meantime I'm left without a Megasquirted car to play with.

But I think I had undertaken to post to this thread a comparison of my TPS ADC before and after fixing up the sensor ground. Here is an extract from a couple of log files:
tpsground.jpg
So the raw AD value jiggles by just one unit now where it used to be much noisier, around 10 AD units.

My basic config is likely to be as noise free as they come. Injectors are high ohm, and ignition is just a logic level, with the sparks being generated by a Bosch module at the coil. So it may be that other setups will be noisier than this. Still, it gives a picture of what the electronics are (at least sometimes) capable of.

I have a second MS unit, and may see about fitting it to the elderly Peugeot 604 I'm driving these days. It has the same odd fire PRV V6 that the DeLorean uses. Other Megasquirters have blazed the trail, so I should be able to follow. May take me a while though. There are some other substantial draws on my time these days.

Have fun,

Rob.

Re: "Noisy" tpsDOT

Posted: Wed Jul 04, 2012 10:24 pm
by gslender
robs wrote:A few things have stopped me participating here for a while. Most relevant to this forum is that my Megasquirted car was run into. No injuries, and the damage isn't all that serious, but because of the state's write-off laws, I may not be able to keep the car. The "good" news is that her insurance company look like they aren't going to pay up, so either I chase them in the courts, or repair it myself at my own expense -- but at least I get to keep this car I've had for >20 years. In the meantime I'm left without a Megasquirted car to play with.
Well that sux - what was the reason the insurance company gave for not paying up - hopefully not because of modifications to the car (ie ECU) !!

Re: "Noisy" tpsDOT

Posted: Thu Jul 05, 2012 2:29 am
by Peter Florance
Rob, terrible news. Glad you are all right.

AFA worst case noise, recall my datalog with Low Z injectors. I don't have any of those mods with 12V return all hacked up.

Hope you are back in an old, cool ride soon.

Re: "Noisy" tpsDOT

Posted: Thu Jul 05, 2012 5:43 pm
by robs
Hadn't realised that yours were low Z injectors Peter. OK, so that evidently doesn't introduce much noise. Discharging the coil through the MS is the only remaining suspect on my list -- but I don't think that's all that popular around here (with all the wasted spark and COP users). So the apparently common problem with TPS noise is just down to people not getting their wiring sorted? Mine was at any rate.

As for the insurance company, nothing to do with illegal mods, I think it's just their way. The other driver had comprehensive, I had property damage insurance, so as long as her insurance company doesn't chase me, my insurance company won't get involved leaving me as the small fry vs. the 500 lb gorilla. It's an impossible case for me to prove (disputed traffic light), but the balance of probabilities is definitely on my side. They're probably expecting that I have other things I'd rather be doing than chasing them in the courts for a couple of grand. They're right about that.

Have fun,

Rob.

Re: "Noisy" tpsDOT

Posted: Sat Sep 29, 2012 12:10 pm
by JoseMiguel
Back from the dead with this thread

One of my cars had a terrible TPS noise, I had to use something absurd like 15 lag and 80 TPSthreshold trigger to make it work relatively well, and I still got triggers sometimes... Gslender 2.2 and the sliding window partially solved this problem, but I had to use also a very high number (10) to make it work smoothly. CLT and MAT sensors were also very noisy.

I thought I would never get this problem solved. Car was wired by an experimented MS so I did not know how sensors and stuff were wired. I read all of this thread and went to the garage. Wired the sensors CORRECTLY as in this diagram

http://msextra.com/doc/general/grounding.html

and voila! car has never been this smooth. Im running no lag in the TPS (100) and I can use a TPSthreshold of 30 without false triggers. I get instant throttle response like never before in the megasquirted life of this car (more than 5 years)...

To anyone having issues with TPSdot noise, I highly reccomend checking sensor grounding. I thought that was ovbious and didnt check it for more than 5 years!!!!

Re: "Noisy" tpsDOT

Posted: Thu Feb 14, 2013 10:16 am
by JasonC SBB
Another thread resuscitation post.

Someone had mentioned "synchronous" sampling earlier in this thread, where the analog signals are sampled at some quiet point, e.g. just before any injectors and coils are turned on or off.

I work for an IC company whose IC's often have pins that are sensitive to uA and mV, a few mm from pins that switch 600V and 10A in 50 ns... otherwise known as integrated high voltage power supply IC's. Needless to say, noise mitigation is a HUGE part of our development time.

I recently completed a study spanning several months on how to make the low voltage pins ignore the noise, without a bypass capacitor (this is difficult). Here is a summary:

- synchronous sampling is *the* way to go. It's far better than continuous-time averaging - i.e. in the MS "continuous sampling" would be 8x or 32x oversampling (8-32x between spark events) and averaging them all. Synchronous sampling would be to take a sample just before a coil fires or an injector opens or closes, whichever has the longest time since the last "noisy event". Continuous time averaging includes the obviously noisy portions in the averaging.
- a big part of the process is making sure the sampling point is indeed quiet - i.e. you stick a scope on the pin and cycle the ECU through different duty cycles and spark advance
- it's best to sample just *before* a switching edge, rather than x microseconds after a switching edge.
- it's also best that the sample occurs while the nearby high current FETs are not conducting - in the MS this means you kinda don't want to sample just before a solenoid or coil is about to turn off
- you need to scope the input signal and check that any noise that jiggles the signal has long "settled out" before the sampling point
- to ensure rapid settling the source impedance must be low enough that it settles any bypass capacitance, in the time between samples
- obviously things still get better with a bypass capacitor - in the MS, it means an analog filter at the ADC pin
- note that MAP values show a repetitive waveform that varies between each intake valve opening event. A "true" reading of MAP would be some kind of average between valve events, or perhaps at the point that the intake valve is open widest.
- the above does not eliminate the need to do some kind of averaging or smoothing over a few engine cycles or spark events.

Re: "Noisy" tpsDOT

Posted: Thu Feb 14, 2013 10:33 am
by jsmcortina
JasonC SBB wrote:- note that MAP values show a repetitive waveform that varies between each intake valve opening event. A "true" reading of MAP would be some kind of average between valve events, or perhaps at the point that the intake valve is open widest
From data in Heywood it isn't as simple as this. This may be true at very low rpms, but at higher rpms, the actual MAP is a mix of many overlapping pressure pulses.
Hence our current advice to use "event average" on firmwares that support it.

James

Re: "Noisy" tpsDOT

Posted: Thu Feb 14, 2013 10:53 am
by JasonC SBB
So say, 8 points averaged between valve opening events should give an accurate facsimile of "true" MAP.

IRTB's would be a special case - it may need a MAP sensor for each cylinder, and the min reading per engine cycle may be the most accurate.

Re: "Noisy" tpsDOT

Posted: Thu Feb 14, 2013 11:21 am
by elaw
Not to drag this discussion too far off-topic but here's a log that illustrates the problem well. It's a zipped .msl file, viewable in MegaLogViewer. The fields of interest are RPM, MAP, and MAPwindow which shows when MAP is being sampled. The file comes from the high-speed MAP logging feature that was added to the MS3 firmware a while back, and is a log of a sweep I did from idle to high RPM on my Audi 5-cylinder engine.

In this file, you can see at very low speed (below 850 RPM) the MAP trace has a nice repetitive waveform and the sample window is well positioned to read the lowest MAP during the cycle. At 1044 RPM the waveform gets more distorted and it's clear the sample window is not at quite the right time. At 1561 RPM things start to get ugly and it just gets worse as the speed increases.

After I saw this I switched my MAP sampling to "cycle average" and it cleaned up amazingly.