Mistakes were made, my R34 Story

[GTSBoy]

1 hour ago, MBS206 said:

Put in excel, graph it, fit a trend line, display formula.

I totally am not reinstalling MATLAB. Hopefully you have it installed if you want to use it 😛

But if HPTuners can only fit a linear equation, no need for any of the above with some simple mafs...

I said "Matlab clone". I could throw one onto one of my linux VMs. Probably only take 10 or 20 hours to sort out the dependencies, relearn how to use it, etc etc.

[soviet_merlin]

21 hours ago, Kinkstaah said:

The other wire I thought I'd use oil temp. This is where I've gone crazy.

HPTuners requires you to implement a formula so you know how much volts = how much temp. This seems relatively simple to me. However I cannot find the scale for this anywhere on the internet, nor decipher how to figure it out without removing the sensor from the car.

All I know is that voltage actually goes up as temperature goes down. I am using the actual gauge, so I can see what the temp is. The signal wire has been branched off into the MPVI3.

 

Just as an aside, my mechanic advised against branching off from temperature sensors. I don't fully understand why, but from what I remember because it is all about resistances, and messing with the wiring will mess with the resistances and throw off the results. Something like this. He said to run a dedicated temperature sensor per input.

Pressure sensors on the other hand are fine to be branched off and passed onwards.

[GTSBoy]

5 minutes ago, soviet_merlin said:

 

Just as an aside, my mechanic advised against branching off from temperature sensors. I don't fully understand why, but from what I remember because it is all about resistances, and messing with the wiring will mess with the resistances and throw off the results. Something like this. He said to run a dedicated temperature sensor per input.

Pressure sensors on the other hand are fine to be branched off and passed onwards.

That's true for things that are treating the input as a "temperature" input, ie that are interested in measuring the resistance of the sensor. But in the case of Greg's MPVI3, he's just measuring a voltage. Voltages are measured with a high impedance input, so no significant current flows, so they have almost no effect on the circuit being measured. It is exactly the same as probing the sensor's terminal with a multimeter set to DC volts.

[MBS206]

2 hours ago, GTSBoy said:

That's true for things that are treating the input as a "temperature" input, ie that are interested in measuring the resistance of the sensor. But in the case of Greg's MPVI3, he's just measuring a voltage. Voltages are measured with a high impedance input, so no significant current flows, so they have almost no effect on the circuit being measured. It is exactly the same as probing the sensor's terminal with a multimeter set to DC volts.

That would totally depends on how the circuit in the GM ECU is actually made as to if it could make an impact.

The big thing to check for would be to use a multimeter to measure the voltage of the sensor with it on the gauge only, and then what it is when the sensor is plugged into the GM ECU too. If it varies at all, then I would disconnect it.

But yeah, totally depends on how the circuit is taking that voltage measurement and the componentry selected.

[GTSBoy]

1 minute ago, MBS206 said:

The big thing to check for would be to use a multimeter to measure the voltage of the sensor with it on the gauge only, and then what it is when the sensor is plugged into the GM ECU too. If it varies at all, then I would disconnect it.

I don't think he's got it on a gauge and on the ECU. I think he's got it on the gauge and on the HPTuners DAq thingo. Remember, we're talking about oil temp here, not something that the ECU is actually interested in for its own sake.

[Kinkstaah]

Correct. The ECU cannot read oil temp. (Well, I think it probably can in some situations. I did have the thought of potentially repinning the ECU when I was doing oil pressure).

I am using this into the MPVI dongle, so that the MPVI dongle can read oil temperature. It is attached to a VDO gauge which is obviously calibrated to whatever curve the sender actually is using.

This would be easy if I could setup a table of voltage to temperature like many sensors, but it appears I cannot do this and can only setup the transform rule which appears to be Input (voltage) x Multiplier, and add an offset.

This to me means it MUST be linear. So it may be a complete waste of time wiring this into the ECU. The idea was that the MPVI3 has standalone logging.

I wanted to use this instead of a laptop with serial cable (for wideband) for long datalogs. Given the wideband also has electric interference, I may never trust this either in a world where the serial wideband and the analog output wideband do not agree.

Last time I did a trace I could see the two wideband traces follow each other, but one was a little leaner than the other. I plan on playing with voltage offsets and actually driving the thing to see how close they correlate.

If they never correlate... then, well, maybe I'll never use either. Ideally I'd like to have the Analog wideband read ever so slightly leaner than the serial one, because the serial one is 'correct'. Tuning the car to be ever so slightly too-rich would be the aim.

Not needing to have a laptop flying around in the footwell connected with cables is... an advantage. About the only one from the forced upgrade to MPVI3.

[MBS206]

1 minute ago, GTSBoy said:

I don't think he's got it on a gauge and on the ECU. I think he's got it on the gauge and on the HPTuners DAq thingo. Remember, we're talking about oil temp here, not something that the ECU is actually interested in for its own sake.

Ah, I thought he'd wired it to one of the spare ECU inputs! Too long ago since I read that post, ha ha. I've been arguing with radiators, harmonic balancers, alternators and rust since reading it.

[Kinkstaah]

Welp, I have been driving it around (a bit) and messing with idle and return to idle characteristics. This is all well and good until I discovered my starter motor... is nearly dead.

This is unfortunately after I booked my exhaust and tune in.

This is the new section of my exhaust, after losing my mind with vbands and clamps.
Yes I now have 4 mufflers. The two middle ones are 5in body 3in pipe round straight through mufflers or incorrectly named 'resonators' whatever you want to call.

How does it sound? Muffled.

 

Car runs richer now in low load. Is it restriction? Or lack of exhaust leaks.
I also figured out a way to make the Analog Readout match the serial WB Readout.

All of this requires a car to you know.. *start* to tune the maths, which was not ideal when I needed it to get to the exhaust shop and back. It started when cold.

30 mins after the car got home, it was time to get stuck into it.

Unfortunately this thing has no identifying marks on it at all. I contacted Mal Wood and asked if he could send me a replacement and he asked me to remove it because they've changed starters with their kits a few times over the years. I asked other companies with LH LS starter conversions and they all use different starters unique to their kits. I did find CAE performance supply Mal Wood - But they are out of stock for this starter for the next 3 weeks.

The tune is booked for Friday 30th.

Here's hoping the local starter motor repair place (which interestingly has a 5 star review from Trent @ Chequered) can get this rebuilt/sorted out so I can get it on and put the car on the dyno.

Here's hoping!

[MBS206]

I see you've never had to push start your own car... You could save some weight right now...

[Kinkstaah]

I don't think there's any way someone is push starting this car.. I honestly can barely move it, and moving it and steering it is just flat out not possible.

I'm sure it is, but needs a bigger man than me.

I have a refurbished starter now.
The starter man was quite clear and consise showing me how nothing inside a starter really should contribute to slow cranking, and turned out that for the most part... my starter was entirely fine.

Still, some of the wear items were replaced and luckily it didn't show any signs of getting too hot, being unfit for use, etc. Which is 'good'. I also noticed the starter definitely sounded different, which is a bit odd considering nothing should have really changed there....

Removed and refit, and we'll pretend one of the manifold bolts didn't fully tighten up and is only "pretty" tight. GM only wants 18ft/lb anyway.

I also found a way to properly get my analog wideband reading very slightly leaner than the serial wideband. There's Greg related reasons for this. The serial output is the absolute source of truth, but it is a total asshole to actually stay connected and needs a laptop. The analog input does not, and works with standalone datalogging.

Previously the analog input read slightly richer, but if I am aiming at 12.7 I do not want one of the widebands to be saying 12.7 when the source of truth is 13.0.

Now the source of truth will be 12.65 and the Analog Wideband will read 12.7. So when I tune to 12.7 it'll be ever so slightly safer.

While messing with all of this and idling extensively I can confirm my car seems to restart better while hot now. I did add an old Skyline battery cable between the head and the body though, though now I really realise I should have chosen the frame. Maybe that's a future job. The internet would have you believe that this is caused by bad grounds. In finding out where my grounds actually were I found out the engine bay battery post actually goes to the engine, as well as a seperate one (from the post) to the body of the car.

So now there's a third one making the Grounding Triangle which is now a thing.
I also from extensive idling have this graph.

Temperature (°C)	Voltage (V)
85	1.59
80	1.74
75	1.94
70	2.1
65	2.33
60	2.56
55	2.78
50	2.98
45	3.23
40	3.51
35	3.75
30	4.00

 

Plotted it looks like this. Which is actually... pretty linear?

I have not actually put the formula into HPTuners. I will have to re-engage brain and/or re-engage the people who wanted more data to magically do it for me.

Tune should be good for the 30th!

[MBS206]

Temp = -21.052 X Voltage + 114

 

That should get you pretty close. Calcs based on two points I could do easy calcs on (30 and 70 degrees).

 

It also says your sensor should only read as low as 9 degrees when it maxes out at 5V, and should hit a peak of 114 degrees at 0V...

Just as a heads up if you were going really cold places, or wanting to be aware when temps really go up with it.

[Kinkstaah]

Yeah - Half the problem is I know this sensor actually goes to 150C.. I'm pretty certain it is min of 11C.

So still more data required I suppose. It's really quite hard to get the oil temp to 100+ then immediately pull over and take a reading before the temp drops. Annoyingly I suppose the range I really 'want' is likely 80 -> 120C. TBH the ecu can't really *do* anything with it, and the gauge itself is very visible...

...but you know how it is.

[GTSBoy]

3 hours ago, Kinkstaah said:

So still more data required I suppose. It's really quite hard to get the oil temp to 100+ then immediately pull over and take a reading before the temp drops. Annoyingly I suppose the range I really 'want' is likely 80 -> 120C. TBH the ecu can't really *do* anything with it, and the gauge itself is very visible...

You just need a datalogger of some sort. A handheld oscilloscope could do it, because it will make the trace visible on screen, so you can look at the peak, or whatever you need to look at. And there are cheap USB voltage loggers available too. You could get a 2 channel one and press a button to feed voltage to the second channel at points that you want to check the sensor voltage, when you knew what the guage was saying, for example.

[MBS206]

Or just wire a multimeter in, sit it up like it's a gauge, go for a drive, read temp gauge, read multimeter, speak to phone and tell it to take notes.

[Kinkstaah]

5 hours ago, GTSBoy said:

You just need a datalogger of some sort. A handheld oscilloscope could do it, because it will make the trace visible on screen, so you can look at the peak, or whatever you need to look at. And there are cheap USB voltage loggers available too. You could get a 2 channel one and press a button to feed voltage to the second channel at points that you want to check the sensor voltage, when you knew what the guage was saying, for example.

I already have the ability to datalog the voltage... what I can't do is correlate it to what the gauge is showing when I'm looking at the data at a later date.

4 hours ago, MBS206 said:

Or just wire a multimeter in, sit it up like it's a gauge, go for a drive, read temp gauge, read multimeter, speak to phone and tell it to take notes.

Or something like this. Except getting the oil to 125C+ involves me going 200+kmh at the track so you can imagine this is a little tricky unless I do something like note what the voltage is when I am at the top of 5th gear so I have a point to reference on.

Drove car today. O2 still reading richer because hopefully not a leak instead of massive restriction I've introduced. Car really doesn't sound like a LS. It's quite strange.

Pulled a bit of fuel out up top. This does potentially explain why my spark plugs looked like the car was running pretty over-rich. If I was tuning it to a ~12.7 which in reality was far richer than that in the past.

It will be handy to have the dyno WB to compare my WB to to compare notes that's for sure. At this point I'm happy to just sit in the passenger seat and tune the thing given I've got about 7 million hours on the platform lol.

[TurboTapin]

I didn't go through every post in detail and this is probably already known but a VDO 320.021 as per the VDO site is scaled 323-18 ohms/50-150°C. Linear thermistors do exist. 

Edited May 27 by TurboTapin

[Kinkstaah]

That could very well be the thermistor, but the ECU only sees Volts. VDO don't seem to provide a 0-5 volt curve, only the resistance curve.... (or line).

[GTSBoy]

4 minutes ago, Kinkstaah said:

VDO don't seem to provide a 0-5 volt curve

Probably because they couldn't, because the use of the variable resistor to create a "signal" in the ECU is managed by the ECU's circuitry. The only way that VDO could do it would be if they made a "smart" sensor that directly created the 0-5V signal itself.

And that takes us back to the beginning.

On 25/05/2025 at 9:27 PM, Kinkstaah said:

what I can't do is correlate it to what the gauge is showing when I'm looking at the data at a later date.

Well, in that case, you could do the crude digital (ie, binary, on or off) input that I mentioned before, to at least put a marker on the trace. If you pressed the button only at a series of known integer temperatures, say every 2°C from the start of your range of interest up to whatever you can manage, and you know what temperature the first press was at, then you'd have the voltage marked for all of those temperatures.

And you can have more than one shot at it too. You can set the car up to get the oil hot (bypass oil coolers, mask off the air flow to oil coolers, and/or the radiator, to get the whole engine a bit hotter, then give it a bit of curry to get some measurements up near the top of the range.

 

On the subject of the formula for the data you provided, I did something different to Matt's approach, and got a slightly different linear formula, being Temp = -22.45*V + 118.32. Just a curve fit from Excel using all the points, instead of just throwing it through 2 points. A little more accurate, but not drastically different. Rsquared is only 0.9955 though, which is good but not great.

If you could use higher order polynomials in the thingo, then a quadratic fit gives an excellent Rsquared of 0.9994. Temp = 2.1059*V^2 - 34.13*V + 133.27. The funny thing is, though, that I'd probably trust the linear fit more for extrapolation beyond the provided data. The quadratic might get a bit squirrely. Hang on, I'll use the formulae to extend the plots....

It's really big so you can see all the lines. I might have to say that I think I really still prefer the quadratic fit. It looks like the linear fit overstates the temperature in the middle of the input range, and would pretty solidly understate what the likely shape of the real curve would say at both ends.

[Kinkstaah]

Well, I'm tired.

I'm tired because about 4PM yesterday, before today's appointment someone immediately bought my bumper. They couldn't get it any other day as they're on the way back to NSW.

So I had to do that big GTR conversion I had been planning. Unfortunately, the information on SAU about what you need and how this is done is incomplete.

So what should be a simple bolt on affair, yeah, it's not. Did you know if you use all GTR items the bonnet won't close? This little manuever sent me into about 1am the night before trying to dodge a way to get it closed. I will have to revisit this in the next few days  - or maybe not, I may let a body shop figure it out. It all needs to come up and my motivation to pull the bumper off is low. It also seems to hit things in the bay where the GTT bonnet didn't.

Yes I used 100% new OEM GTR items.

Today, I had the joy of driving to the dyno looking like this:

Given I had roughed in the fuel and given sensible but pretty conservative timing, I didn't really bet on having the car drive out any real difference than when it drove in. Sadly due to a miscommunication and laptop fun and games (and almost bricking the dongle, prayers and firmware updates indeed), I ended up using HP Tuner credits to licence the car that was already licenced.

So in the end my laptop was used. It turns out my butt dyno is still well calibrated after all this time.

The 325kw was on 74% Ethanol, the 313kw line was on 98. The other line is the 'before' line which was 281kw. While the numbers are pretty low, they're pretty in line with what you'd expect. Even if US dynos bump the whole result up about 50KW, gaining 10-15% is similar gains. 

The curve of the cam is pretty much spot on with what was discussed as well. All this said, it still feels bad to not see the number you secretly want to see. Even if the car drove great beforehand, and I knew pretty confidently the car would drive out much the same way it drove in due to the nature of a wellish dialled in LS1 not gaining much if anything at all from being tuned from where it was.

As expected, the car isn't particularly sensitive to running it at anywhere between 12.0 and 13.0 - And the initial timing at 20deg and 12.0 made 308KW. So 3 degrees of timing, and leaning it out to 12.7 for 5kw, anything above stopped giving any benefit until E85 (which has an additional 2 deg as before).

Car itself behaved entirely fine. I found out that 100C = 1.15V!
IAT at about 7pm was 19C.

I might mess with the bonnet mounting.. but given the REO NEEDS TO BE CHOPPED TO FIT A GTR BAR this is possibly something I may leave gathering (more) dust until it returns to paint jail.

[Kinkstaah]

Had a chat to Tony Mamo - he advised that looking at the graph's shape (regardless of number translation) it shouldn't be levelling off like that up top. He said it looks like the car is running a restrictor plate or something of that nature.

Luckily the last dyno pull on E85 was saved to the PC - So I did a quick trace of Barometric KPA vs Intake manifold pressure, and lo and behold I see a copy of the 'waves' I was seeing on the dyno:

I'm not entirely sure how to go about addressing this. It may be worth sacrificing IAT to run a more direct intake for the motor. Though packaging this could be flat out impossible given the constraints of the car. I should have removed the intake pipe to see if this made a difference but looking at this graph I now have something to potentially test in the future and think about.

This one is from an old track day log in the OLD setup going through gears on track.