Hypergound wires?!??

[LightningGSX]

Well ADC resolution is meaningless here.8 bit resolution is about 0.019 volts,deviations in the 5 volt ref or ground ref can easily be greater than 0.019 volts which would the shift the lookup table.Which is not neccessarily the case here.With a quick test ,with a snap-On Vantage waveform viewer and a 10 guage wire with aligator clips,I can see electrical system noise is probably the issue here which(you being an ecm designer would know) can cause all sorts of problems.Now I'm sure your gonna say, the ECM is filtered for noise which is true.But I doubt the OEM's add any more filtering than they needed at the time and by adding less suppressive plug wires or by the overall age of vehicle,engine,alternator electrical system noise will increase.This is still pretty much speculation, but I'll stick by my previous posts and I will say again I'm NOT making a case for buying these pre-fab wires.By the way who do you design for? firmware or electrical?

 

[brads]

Originally posted by LightningGSX
Now I'm sure your gonna say, the ECM is filtered for noise which is true.But I doubt the OEM's add any more filtering than they needed at the time and by adding less suppressive plug wires or by the overall age of vehicle,engine,alternator electrical system noise will increase.This is still pretty much speculation, but I'll stick by my previous posts

But you have to look at where the noise is coming in. The sensors are not grounded to the engine, the only connection between them and ground(or power) is at the ECU. So if there is noise transmitted on the ground plane that affects the sensor signals, it will affect the ECU ground the same way. As long as the ground for both of them is at the same potential its not really a problem. The worst noise is the ignition, and the ECU is isolated from this somewhat due to the external coil driver setup. Plus most ECU designers seem to be smart enough to not try to do A/D conversions as they are firing the coils, which is a pretty simple way to eliminate noise issues in practice. At least the ones I have seen so far have avoided that.

Originally posted by LightningGSX
By the way who do you design for? firmware or electrical?

Motorcycle aftermarket, a few companies. I do firmware, some hardware, programming software, a little of everything. Motorcycle and other small engine setups don't have the luxury of separate coil driver packaging, so the noise of firing a coil driver(or firing a capacitor in a CDI style ignition) within a few inches of your main ignition microprocessor is a pretty big issue.

Brad

 

[LightningGSX]

Looking at a ECU out of my 90 Talon TSI I see

1.No APPARENT isolation between digital and analog planes
2.What looks to be a 12 Mhz xtal(on the part# at least) and if the instructions are executed every fourth clock cycle(which I assume they are) and given the charge times of ADC caps,sampling the karman sensor,ignition charge delays etc. it's possible conversions are happening during fire.
3.Even though sensors and the ECU may be at the same potential, each sensor has a transfer function(programed into a lookup table) ,deviation in the positive voltage reference would still cause error.

Besides, the noise doesn't neccessarily come from the ignition
Anyways, I'm not saying your wrong and skepticism is definately a good thing ,but I believe there is some truth to the claim's and who knows I could very well be wrong.And noise could interfere with the counting of the karman signal, but I guess that's a whole other discussion.

 

[LightningGSX]

And it's pretty cool this didn't turn out to be a "my dick is bigger than yours" type thread

 

[LightningGSX]

And as I think about it this deviation would be huge in resistive type sensors that use a voltage divider setup to infer the readings(IAT,BARO?,CTS,TPS, etc).These sensors use a pre-calculated, pre-programed transfer function(And I seriously doubt any software correction is implemented).Also the benefit would be more noticable in MPG and throttle response than in HP anyway.

 

[brads]

Originally posted by LightningGSX

2.What looks to be a 12 Mhz xtal(on the part# at least) and if the instructions are executed every fourth clock cycle(which I assume they are) and given the charge times of ADC caps,sampling the karman sensor,ignition charge delays etc. it's possible conversions are happening during fire.

It's possible sure, but its very easy to toss out the conversions done at the time the coils are firing. The other noise isn't bad enough to cause errors, at least not in my experience. Of course that is dependent on how the PCB is set up.

Originally posted by LightningGSX

3.Even though sensors and the ECU may be at the same potential, each sensor has a transfer function(programed into a lookup table) ,deviation in the positive voltage reference would still cause error.

The positive voltage reference comes from the ECU, and is referenced off the same ground signal as the sensor, and the rest of the ECU.

Originally posted by LightningGSX

Anyways, I'm not saying your wrong and skepticism is definately a good thing ,but I believe there is some truth to the claim's and who knows I could very well be wrong.

I see too many people screwing up dyno results just to sell products, to give these guys the benefit of the doubt based just off a few dyno runs, when everything else tells me its not legit.

Originally posted by LightningGSX

And noise could interfere with the counting of the karman signal, but I guess that's a whole other discussion.

It could, but it'd be even less likely. A very little filtering on the input line would help with that.

Originally posted by LightningGSX

And as I think about it this deviation would be huge in resistive type sensors that use a voltage divider setup to infer the readings(IAT,BARO?,CTS,TPS, etc).These sensors use a pre-calculated, pre-programed transfer function(And I seriously doubt any software correction is implemented).

I have yet to see any noise spikes on these using DSMLink sampling them at 20 Hz. If the ECU was reading huge noise spikes on any of these sensors signals, I'd be able to see it pretty easily with the logger. If the noise you are thinking of is very small, then the whole AD resolution discussion becomes an issue.


Brad

 

[out there]

i think it would be a good idea for one of you that purchased the wiring kit to take a picture of the engine, and label the "best" places to ground everything. that way, those of us that aren't made of money can go buy 8g wire and ground everything better with it

 

[LightningGSX]

Originally posted by brads

I have yet to see any noise spikes on these using DSMLink sampling them at 20 Hz. If the ECU was reading huge noise spikes on any of these sensors signals, I'd be able to see it pretty easily with the logger. If the noise you are thinking of is very small, then the whole AD resolution discussion becomes an issue.

At a slow 20 Hz,very little noise would be visible,you could miss huge spikes .The noise is there and I've seen it and with your profession you should have tools available to see it.At any rate you didn't design the ecu or any other aspect of these vehicles(and neither did I) and you can't say any of my points are impossible.So only time and testing will tell.

 

[GaRn]

well we had someone test, the dude with the NA car, which imo woudl be a better baseline car than a turbo car for dyno comparisons for obvious reasons.. yes he bought a new o2 sensor and took out his crap splitfires(which should only yield 1-2 hp gain if any, if he needed new plugs and o2 anyways), along with his ground kit and netted no gain.. so it might idle better or whatnot but i doubt it adds any noticeable hp..

 

[brads]

Originally posted by LightningGSX

At a slow 20 Hz,very little noise would be visible,you could miss huge spikes.

Sure you could miss them once. But even a broken clock reads right twice a day. And in the same manner, if there was this horrible noise, I'd bet that at SOME point in one of my logs, I'd have sampled during the noise occurence. You are saying the ground would affect the AD readings, correct? So even if it were sub nanosecond noise, we don't care. What is more important at is the setup time of the AD converter. How long does it take to get its reading? I'd bet its somewhere around 10-100 microseconds. Given a 20 Hz, or 50 millisecond sample rate on the logger, and the huge number of minutes of logging I have done, I'd still bet I would have seen one of those enormous noise spikes by now.

Originally posted by LightningGSX

The noise is there and I've seen it and with your profession you should have tools available to see it.At any rate you didn't design the ecu or any other aspect of these vehicles(and neither did I) and you can't say any of my points are impossible.So only time and testing will tell.

Are you sure the noise isn't an artifact of your test equipment? Even with Mhz frequency scopes, you can have issues with try to read ground noise, as to whether its an actual bounce, or an error with the scope or its probes. Anyhow, let me know how the testing turns out.

Brad

 

[LightningGSX]

Actually they are quite small ,missing huge spikes at 20Hz was just an example of possibility.These spikes,while invisible to your logger, could be enough to shift lookup tables PERIOD.Anyway it's pretty common knowledge automobiles are a fairly noisy environment,no one doubts this but you.If the noise lied in my test equipment it would still exist when the engine was not running.

 

[GaRn]

Ok.. If they worked at all, they would only add a few hp or lb ft of improvement!!! its nothin to argue over.. sheesh.. if you believe it so strongly, do it.. if you did it to your car an care so much about what other people think, you have problems.. i wish this would END!

 

[brads]

Originally posted by LightningGSX
Actually they are quite small ,missing huge spikes at 20Hz was just an example of possibility.These spikes,while invisible to your logger, could be enough to shift lookup tables PERIOD.

If it is invisible on the logger, how far is it going to shift a lookup table? How much of a lookup table shift do you get when there is a shift of less than 1 bit on an AD output?

Originally posted by LightningGSX
Anyway it's pretty common knowledge automobiles are a fairly noisy environment,no one doubts this but you.

I never said that at all, you must not be reading my posts very well. I said that the DSM wiring seems to be done properly so that noise doesn't affect the ECU operation.

Originally posted by LightningGSX
If the noise lied in my test equipment it would still exist when the engine was not running.

Wrong. Oscilloscopes aren't perfect. Just because they CAN measure however many million samples per second, that doesn't make them accurate at that speed for all purposes. If you are measuring with your Snap on tool, you need to find how well it handles common mode signals at high frequency. If you exceed the scopes maximum common mode slew rate, you can get severe distortions in the scope output, that aren't actually there in the signal. The Vantage looks more like its geared towards testing lower frequencies, and probably wasn't designed with good common mode rejection at high frequencies.

Brad

 

[LightningGSX]

1-We already went over the lookup thing.
2-I could never say the dsm wiring is proper, maybe adequate for a stock relatively new production vehicle.Not older modded vehicles.
3-How I could see spikes with the engine running and they disappear when the engine is off and blame it on error is the less likely thing so far.the vantage is good to 1khz,not fast, but considerably faster than your logger.

garn
This is hardly an argument, this is a pretty civil discussion, for dsmtuners.com at least(most threads look like an AOL shit talking room).But this is getting boring, since no proof exists as of now.So this is my last post

 

[brads]

The DSM wiring is done properly, using a method that has become pretty common over the years, as design engineers have determined what works, and copied and built upon successful designs. Modded or not isn't an issue, unless they have done some really wierd mods to the wiring.

The way you see a spike with the engine running, then have it go away when the engine shut off, is part of that common mode rejection I was talking about. The higher the frequency, the harder it is to reject common mode signals. I looked up the vantage specs, they don't even mention things like CMRR, Max CM slew rate, etc. With a 1kHz sample rate, I bet its not that great for that. I'm not knocking the tool, it looks very useful for other things, just not for this.

You are saying the noise affects the ECU lookup tables. But my logger is looking at the results of the AD conversion in the ECU. So if it affects the ECU, surely my logger would show it. The ECU isn't sampling those sensors at some super high frequency. So its not like there is a 0.0000001 chance of me catching the noise with my logger, if the ECU caught it. If the ECU sampled at 1kHz like your Vantage, I'd have a 2% chance of finding the noise with the logger on any occurence of it, each engine revolution. The odds of me seeing the noise multiple times in even a 1 minute log of 1000 RPM idle would be pretty high. Thats if the ECU samples at 1kHz, which is way higher than I'd expect.

Looking at the ECU reading seems like a much more valid way to see if the ECU is reading noise on the sensors, than trying to check it with another seperate tool.

Brad

 

[LightningGSX]

Originally posted by brads
The DSM wiring is done properly, using a method that has become pretty common over the years, as design engineers have determined what works, and copied and built upon successful designs. Modded or not isn't an issue, unless they have done some really wierd mods to the wiring.

That is just an unintelligent thing to say.Bordering on stupid.

The way you see a spike with the engine running, then have it go away when the engine shut off, is part of that common mode rejection I was talking about. The higher the frequency, the harder it is to reject common mode signals. I looked up the vantage specs, they don't even mention things like CMRR, Max CM slew rate, etc. With a 1kHz sample rate, I bet its not that great for that. I'm not knocking the tool, it looks very useful for other things, just not for this.

This doesn't make much sense.If it is harder to reject common mode signals at higher frequencies,how would that effect my vantage working at a relatively low frequency.Anyway ,BY DEFINITION common mode rejection would not apply here.Maybe you should recheck the definition.

You are saying the noise affects the ECU lookup tables. But my logger is looking at the results of the AD conversion in the ECU. So if it affects the ECU, surely my logger would show it. The ECU isn't sampling those sensors at some super high frequency. So its not like there is a 0.0000001 chance of me catching the noise with my logger, if the ECU caught it. If the ECU sampled at 1kHz like your Vantage, I'd have a 2% chance of finding the noise with the logger on any occurence of it, each engine revolution. The odds of me seeing the noise multiple times in even a 1 minute log of 1000 RPM idle would be pretty high. Thats if the ECU samples at 1kHz, which is way higher than I'd expect.

This also doesn't make sense.Your logger only reads the ECU's interpretation of the sensor input's.So short of you measuring outputs of all the sensors and then comparing them to your logs, you would NOT see this.Your logger is buffered by the ECU, so I don't see how your logger would show any noise at all(20Hz or 1Khz).

Looking at the ECU reading seems like a much more valid way to see if the ECU is reading noise on the sensors, than trying to check it with another seperate tool.

False.Seperate tools are isolated from the vehicles power supply.

 

[MNGSX]

Originally posted by LightningGSX
That is just an unintelligent thing to say.Bordering on stupid.

Some one would have to border on stupid with a mod that the wire kit fixes. Like swap a clutch and forget to reinstall a ground strap.


I cannot find any differences enough to matter between grounding on a DSM and any other modern car. From Kia's to Ferraris. From frame and powertrain component grounding to sensors. ECU's analog signal conditioning and conversion to digital data are nearly identical. When it comes down to signal lines and conditioning on ECU's everybody is doing pretty much the same thing. In fact DSM's have one up over some with the frequency based MAF sensor. The frequency type signal is much more tolerant of interference and noise. It does'nt sense voltage, capacitance, resistance etc. Just the frequency. The frequency range was chosed to best reject common mode interference. Any frequency based interference outside this range does'nt matter because it is'nt looking at it at all. Since it is looking for the "5" and "0" volt states and how fast they come even relitively large noise caused voltage fluctions dont matter. 1/10th of a volt above or below 5v or 0v is a 1 or a 0 frequency pulse just the same.

Originally posted by LightningGSX
This also doesn't make sense.Your logger only reads the ECU's interpretation of the sensor input's.So short of you measuring outputs of all the sensors and then comparing them to your logs, you would NOT see this.Your logger is buffered by the ECU, so I don't see how your logger would show any noise at all(20Hz or 1Khz).

That is like checking the frequency response of tweeters at 40kz and doing nothing else to the audio system. Nobody can hear it or make out any differences. If the ECU can't see it. It wont log it so it wont change anything. Bottom line the minor interference you are seeing has no effect on things.

Want to cut interference on sensors? Add a drain sheild or even a driven sheild (google). It is a seperate circuit from everything else. Lower resistance or higher gauge grounds do nothing for this at all.

Originally posted by LightningGSX
Seperate tools are isolated from the vehicles power supply

So look for problems in the power supply then try to figure out where they are coming from. Probably not the ground wires gauge, metalurgy or construction that is for sure.

Put a filter on the alternator charge circuit. Make a more robust noise supressor on the ignition drive circuit than just a itty bitty cap. Switch to COP or DECOP. (No plug wires near sensors or at all) . Here is an experiment. OBD or other datalog the ECU. Track down the interference at the source. Eliminate it. Datalog again recreating the previous conditions of the first log as scientifically as possible. Did what you did even matter?

Honestly, DSM's are 100x more hindered by the cam/crank sensor setup than anything else. It just counts time between 0-5v pules to determine degrees. That is a sensor resolution problem. The ECU's time calculations are just resolution enhancement thru signal processing. As far as what tells the ECU when TDC for the next slug is coming it is just four teeth.

A multi tooth system with 60 teeth (100 tooth triggers exist). Will make more power with the same mods than a stock crank/cam angle sensor. Now add the signal processing to increased resolution. The stock ECU can't do it. It has a more accurate method of sensing engine rotation and matching the required events with it.

Want to figure out the position of something with a 4 pixel map and a ruler or a 60 pixel map and a ruler? Simplified but that is the idea.

Since the crank and hence pistons accelerate at different rates depending on factors like engine acceleration, deceleration, load, throttle and RPM.... Dont you think counting from one crank angle degree triger to another across a much narrower angle then counting helps.

 

[95GST]

I looked through the manual almost all the sensors have their own ground wire going back to the ECU.

Most people who install new ground wires disconnected their battery and reset their ECU. Then they claim that it runs smoother, faster.. it might be the car felt "different" because the ECU had just been reset.

The only thing I can see is that your accessories like head lights and fans might run a little more stable with less resistance putting slightly less load on the alternator, which could explain the slight HP increase?

But the whole grounding concept could be ideal to prevent rotten stock grounds from creating problems in the future.

Anway, you can do it yourself for about $20 and have it fit alot better too. No need to pay $100 for pre-made kits.

 

[brads]

Originally posted by LightningGSX
That is just an unintelligent thing to say.Bordering on stupid.

Surely you can come up with a more intelligent reply than "you're stupid"? What mods are going to affect the grounds? turbos? exhausts? oh ohh plug wires? spark plugs? injectors? What part of that statement was stupid? Please enlighten me? Not ALL of us are so fortunate as to have a 1 kHz scope that apparently is good enough to read accurately read noise in the MHz range Some of us don't have such quality equipment, and have to resort to using scopes that measure in the MHz range.

Originally posted by LightningGSX

This doesn't make much sense.If it is harder to reject common mode signals at higher frequencies,how would that effect my vantage working at a relatively low frequency.Anyway ,BY DEFINITION common mode rejection would not apply here.Maybe you should recheck the definition.

The noise you are looking for occurs at higher frequencies, thats how it would affect you. The vantage is less likely to be able to reject high frequency noise, and more likely to present erroneous output when high frequency common mode signals are present. Why would common mode rejection not apply? Are you not looking at the sensor power, ground, and signal wires? I thought that was what you said the noise was affecting? Common mode rejection should apply here.

Originally posted by LightningGSX
This also doesn't make sense.Your logger only reads the ECU's interpretation of the sensor input's.So short of you measuring outputs of all the sensors and then comparing them to your logs, you would NOT see this.Your logger is buffered by the ECU, so I don't see how your logger would show any noise at all(20Hz or 1Khz).

The logger is spitting out what the ECU is reading on its analog channels. If the logger is spitting out a smooth data stream, that must be what the microprocessor in the ECU is seeing. Which means that either they did their job right when they did the horrible stock ground design(which you still haven't explained why it is bad, just that it is), or they did a lot of buffering/filtering in the ECU to account for that(whhich would mean your serious doubts of the software filtering are wrong). Odd how that works.

Originally posted by LightningGSX
False.Seperate tools are isolated from the vehicles power supply.

The ECU isn't though. And isn't the whole purpose of this exercise to clean up the signals that the ECU sees? So when is it good enough? When the ECU doesn't register the noise, thats when. If you had a scope that was good enough to work in the frequencies to accurately measure the noise, I'd say you had an argument. But a 1kHz scope isn't even close to good enough for that. At that point, you are just better off looking at what the ECU is reading.

Brad

 

[LightningGSX]

1- I did not call you stupid.I said that was a stupid thing to say, and I'll stand by that.The electrical systems are designed by corporate engineers, under corporate pressures,It's usually "get it working and get it into producton".Non-oem and aging oem alternators,plugwires,blowermotors, etc can produce more noise than new oem parts.I doubt the designers of the ecu account for aftermarket stuff.The more I look into to this I am also starting to believe the so called "phantom knock" stems from noise.

2-I never in any way said our grounds were horrible(I don't know where you got that from).I said it is possible that some vehicles(with older or non oem parts) could benefit from different/better grounding points.

3-The more I look through books and app notes, the less I see how common mode signals are making my readings erroneous.If common mode signals were such a problem with my vantage they would also be a problem witn the ecu.Besides the common mode voltage present with the engine running, would be the same common mode voltage present with the engine off ,therefore not effecting my readings.Maybe you could explain this to me.

4-How you think your logger would show noise or error still continues to not make sense to me.Like I said before, unless your comparing analog results from more than one ADC.

5-I also never said this noise was huge and high frequency.Whether it's its small or large high or low frequency it will have different effects on different ecu's(vehicle to vehicle and oem to oem).

 

[VRMAN]

Anyone interested in a set of grounding wires that doesn't want to pay HKS $100 pricing here is a link for the same damn kit for $20:

http://cgi.ebay.com/ebaymotors/ws/eBayISAPI.dll?ViewItem&category=33574&item=2433518846

 

[LightningGSX]

MNGSX, The MAF frequency was not "chosen" for that reason at all.That frequency correspond's to the frequency of the von karman vortices inside the sensor.And as for the rest of what you said , you obviously didn't read or understand prior posts.

 

[shayne54]

Im not sure if anyone else read the Import Tuner Magazine artilce or not, but they even stated that there had been an engine fire in this vehicles history and that it hadnt been driven for quite a while. Im not 100% sure, but wouldnt that make a huge difference? In my opinion, i think that the ground wires were a fix to thier car and i wouldnt count them out for helping anyone elses car that might have a less than perfect ground. for only $20 (DIY) i know that it cant hurt to try.

 

[MNGSX]

Originally posted by LightningGSX
MNGSX, The MAF frequency was not "chosen" for that reason at all.That frequency correspond's to the frequency of the von karman vortices inside the sensor.And as for the rest of what you said , you obviously didn't read or understand prior posts.

Simple.

You they used a frequency to voltage converter at the ECU for the MAF channel. Read some books on sensors and data acquisition systems. A frequency signal is very resistant to inerference. Think AM radio static vs FM. A frequency data line from a sensor rejects noise that much better than say a 0-5v signal.



The karman vortices can are a factor of sensor housing design (volume vs diameter ETC). So yes I'm pretty sure they actually meant the sensor to have the frequency output it has.


I highly doubt this..... Even if these frequncies at a given airflow are "just what happened" when they designed the sensor the descision to not convert it to a different type of signal (like 0-5v) at the sensor was done because the frequency output is good at rejecting interference.

Anyway. If my terminals and cables are all clean and tight I'm not doing sh&t to my ground wires.

If you are looking for power via electronics upgrades bump up your ignition voltage to 16v and get COP ignition.

If you have aftermarket EMS, Look for alot better timing sensors that are compatable.

 

[cait sith]

Originally posted by VRMAN
Anyone interested in a set of grounding wires that doesn't want to pay HKS $100 pricing here is a link for the same damn kit for $20:

http://cgi.ebay.com/ebaymotors/ws/eBayISAPI.dll?ViewItem&category=33574&item=2433518846

While HKS has made parts of dubious value/performance in the past, these super "phat" Hyperground wires are not made by them.

Just a FYI.