[rusEfi] Miata NA 1990, 1.6L stock #21

[stefanst]

Good points/ questions.

I assumed that, just like with my NB1 and NB2 Miatas, the TPS is 5V referenced. However, it seems that the '90 Miata doesn't even have a stock TPS sensor, so I'm at a loss here. @? any data on your TPS sensor?

Looking at Toms calibration, 0% TPS is 102ADC and 100% TPS is 793 ADC. So 0% throttle is approx 0.33V at the input pin and 15% is 1.0V (if my math is correct, but somebody else better check!)

Some of the short-term TPS noise pattern seems to be following the same pattern as the Vbatt and the AFR noise, so it appears that some if it is systemic and not related to actual throttle movement.

In other news, TPS pots usually have fairly low resistance, maybe 1kOhm of so, precisely because we often have to deal with noise. That's all I got. Tom?

[tomiata]

I guess I should start at the real world, by inquiring if @tomiata drinks coffee, energy drinks, etc. Noisy TPS could be the real world, loose cable, or other things like that. I doubt it, but should always cover the things I can't see personally.

As a matter of fact, I am a coffee addict. Drip, pour over, french press, espresso, cappuccino, latte, whatever....

But, my TPS is a generic OEM sensor with a potentiometer and I hacked it up to fit in place of the original Miata TPS/switch. I ran new wiring for ground, +5v, and sensor direct from the ECU. The voltage range I remember seeing was about .1v to 2.5v. I'll remeasure that and report back on it.

CLT and IAT are stock sensors and stock wiring. Stock AFM is still in place and IAT in the AFM.

[tomiata]

Here is a trace of the TPS signal at a connection just outside the ECU with the engine running. This is at 0% throttle. Scope input is set to AC mode and 20mV per division. Look like roughly 20mv of noise, but there is an occasional spike of about 50mV like you see here, if I'm reading this correctly. The larger spike seems random and occurs every few seconds. And the noise is greater at higher throttle. Just realized I didn't capture a shot at higher throttle.

tps-noise.jpg (26.38 KiB) Viewed 17476 times

Measuring voltage with a digital volt meter at the same point shows .48v at 0% throttle and 4.41v at 100% throttle.

I'll try to get more measurements of noise for CLT and IAT. Which of the signals are more critical to affect for fuel calculations?

@stefanst thanks a TON for the VE tune error report. I applied updates for the cells that were off by more that 1%. I'll try that first and see how it goes.

[kb1gtt]

Can you measure the original ECU's resistance from the signal pin to GND? That's measured with the ECU on the desk, with nothing connected, just DMM with ohms between GND and that signal input pin.

[stefanst]

Unfortunately the stock ECU doesn't have a connection for a potentiometer style TPS. Stock '90 Miata is a simple switch. I'll measure my '99 Miata ECU which has a resistive TPS.

EDIT: Just measured it- resistance from TPS pin to GND is 11kOhm. Resistance to +5V is 6kOhm. Any idea how we get to the input impedance from there?

[stefanst]

Sorry, but I just don't like the noise in your signals. Most critical are of course TPS and MAP which should get their reference GND driectly at the ECU. @: Is that how you wired them, or are you grounding them somewhere in the engine bay? They need a dedicated GND wire coming directly from the ECU. This wire should not be used to power anything- just supply GND to the sensors.
IAT and CLT are often grounded right off the engine block, so if the GND on the ECU floats up or down in comparison to that, we need to support the ECU with bigger ground wires with lower resistance.
Not sure how you wired your WBO2 gauge, but it should also get GND directly at the ECU.

Good luck hunting down grounding issues- a perennial favorite of all car-guys

[AndreyB]

Unfortunately the stock ECU doesn't have a connection for a potentiometer style TPS.

Stock automatic wiring and ECU has a potentiometer TPS and that includes my hunchback. Next time you see the hunchback we should put your osc on the TPS line - as you remember I have the 2nd connector added into my harness for the data logger, so we can nicely connect with a plug.

NHMS racing weekend data logging of OEM TPS pot: https://svn.code.sf.net/p/rusefi/code/misc/logs/1991_miata_hunchback/2016_oct_nhms_race_weekend.7z

[stefanst]

Right - forgot about the automatic!

Just looked at a random log from your last post: Your TPS signal looks great, but MAP is pretty poor. Also Vbatt and CLT are not that great. IAT and TPS I'd say are the best looking signals in that log, followed by AFR.

<threadjack>
Two things I noticed, even though they belong in the hunchback thread:
- Your AFRs at WOT are waaaay too lean to make all the power the little 1.6l can.
- Whoever drove the car at log 091339 got damn close to a money-shift at 2640.5s
- That track in a Miata is supposedly 90% full throttle and 10% braking - you guys need more full throttle

</threadjack>....

So it looks like maybe the stock ECU wiring for NA Miatas doesn't really supply very good GND connection.

[tomiata]

Sorry, but I just don't like the noise in your signals. Most critical are of course TPS and MAP which should get their reference GND driectly at the ECU. @: Is that how you wired them, or are you grounding them somewhere in the engine bay? They need a dedicated GND wire coming directly from the ECU. This wire should not be used to power anything- just supply GND to the sensors.
IAT and CLT are often grounded right off the engine block, so if the GND on the ECU floats up or down in comparison to that, we need to support the ECU with bigger ground wires with lower resistance.
Not sure how you wired your WBO2 gauge, but it should also get GND directly at the ECU.

Good luck hunting down grounding issues- a perennial favorite of all car-guys

My MAP is a INSIDE the Frankenso enclosure. Vacuum hose running from spare nipple at throttle body through firewall to TPS ground comes direct from Frankenso. I have Frankeso case grounded to same ground as used nearby to my nearby AFR gauge. I'll try grounding that point to chassis with a big wire.

@, russian and stefanst already commented, but the stock TPS is just a switch and stock ECU has nothing useful to measure there. I think the generic TPS pot I have is 10k ohms, but I need to measure again to make sure.


And, here is another driving log after updating the > 1% VE tuning error cells that stefanst called out from his analysis.

[stefanst]

So how does it feel? I still notice a handful of times where you went pretty lean when enrichment should have taken care of it. And of course some occasions where it went too rich for a perfect world. But overall the log looks pretty solid. Your VE tune of all cruise cells seems to be near perfect.

With the MAP sensor being directly on the board, I'm with you that external reasons for the noise can be ignored for now. But the el_delta signal does still look a bit wonky- especially at part-throttle where there really shouldn't be any el_delta. The MAP signal actually looks pretty good, but yet the el_selta is moving about without apparent cause. Is engine load just MAP, or is there something else thrown in?

There's also considerable noise in TPS delta. Any ideas? Could be a bad TPS sensor....

I'll run th elog through my DB tomorrow.....

[kb1gtt]

I'll assume you used analog IN3 for TPS. How would you feel about changing R231 for an 11k ohm resistor? You could simply put a new resistor on top of the existing resistor. This would be an SMT thing, so probably less than desirable. Alternatively you could install a 11k ohm thru hole resistor on the via from W53 to GND. If you already have a wire installed in that via, you could lay the resistor flat on the back side, then solder the thru hole lead to the bit of wire that pokes beyond the PCB.

Could you install a 6kohm resistor for W49. If you have a wire for that which is changing the pin out, you could replace that wire with a thru hole resistor. If you have the SMT 0R resistor, you could remove it and add a thru hole resistor.

Once the new resistor(s) have been installed, you will need to re-calibrate your TPS.

Another tip, twisting your TPS wires together helps reduce noise coupled in via RF sources. I commonly twist wires by cutting the wires to length or a bit longer, then putting all three in a vise on one end. and in a hand drill chuck on the other end. Spin the drill until you get a nice bundle of twisted wires. Then I go add connectors and install as a harness.

Is MAP noise an issue? This most recent log appears pretty clean to me.

[stefanst]

[...]
Is MAP noise an issue? This most recent log appears pretty clean to me.

I agree that it looks pretty clean in the log. What has me baffled is that the el_delta value for accel enrichment is often showing significant positive values, where there isn't much of anything going on. The only explanation for this that I can think of is short-term order noise that doesn't show up in the log itself, but makes its' presence know in el_delta.
Even more weirdly el_delta shows positive values where the MAP goes down! Also tps_delta is only ever positive. No clue why we never see negative values for tps_delta and rarely ever for el_delta. Again: I think I'm not understanding the code properly here....

[kb1gtt]

Hmmm, I wonder if the long-ish small-ish hose could account for some of this. I know the MAP sensors typically have a reaction time of 1mS, as well 6ft of hose would account for at least another 5mS of signal delay. That's 6ft / 1126ft/S = 0.0053S. As well don't forget there will be some amount of "water hammer" but to a much lesser extent as this is light weight air, not heavy non-compressible water. The MAP noise might simply be real world mechanical issues.

https://en.wikipedia.org/wiki/Speed_of_sound

I put in a reminder to myself to remember and look at this a closer this evening. I don't have log viewer right now.

[kb1gtt]

Can we get an ohms reading from the wiper to GND wire in the idle position and the WOT position on the Miata TPS?

Here's a QUCS simulation that shows the difference of the 11kohm vs the 500kohm. This one shows ECU's voltage is very linear, while it's also conducting about 0.004 mA.

Capture.PNG (31.39 KiB) Viewed 17426 times

Then if I turn the 11kohm back on, we get the below. Note we are now 0.1mA, and slightly non-linear. I'm looking to learn the range of the TPS, as we might be in the linear region, or we might need to compensate in software.

Capture2.PNG (30.69 KiB) Viewed 17426 times

QUCS file attached.

TPS.7z

(891 Bytes) Downloaded 310 times

[stefanst]

I ran the latest MSL through my database again and it looks like improvements over last time are marginal, if any.
On the other hand, you're off by a max of 3% or so in the driving cells- that's probably better than stock!

Your tune for the extreme cells can most likely still be improved, I notice you don't like to drive at high rpms and full throttle for example. You need to fix that
We're going to need that data for a perfect tune. But your cruise-cells look pretty close to perfect.

At this point I'd say your tune is definitely good enough to get a bit more aggressive with the target AFRs. I like to lean out the cruise cells a bit to improve fuel efficiency and then I run fairly rich at higher MAPs to achieve max power. You should probably aim for 12.5 AFR at 100kpa and a taper from there to 14.7 at 70kpa or so. I have the same x- and y-axis scaling for AFR target and VE tables in order for me to have to think less. But in the end AFR table scaling is not very important.
Also: my target for idle is a bit richer than I would recommend for you- my setup is a bit special and that creates some trouble in maintaining idle AFRs. So 14.7 at idle for your car should be good.

Your VE tune seems to be about as good as we can get it under the circumstances and a fair bit better than most of the cars driving around out there. If you want to, I'll share my -fairly aggressive and tuned for 93 Octane- ignition table with you now. But burning or blowing up a piston with timing that's too advanced is definitely possible. Det-cans would be perfect to make sure you don't knock...

[stefanst]

The TPS response not being perfectly linear doesn't make much of a difference in anything. Accel-enrichment based on TPS change is always fairly heavy-handed anyway. And the non-linearities could be tuned out with adjustments in the TPS/TPS table for enrichment. Noise is much more critical, so whatever we can do to reduce noise, like increase the current is OK by me, even at the expense of some non-linearity.

[tomiata]

kb1gtt wrote:

I'll assume you used analog IN3 for TPS. How would you feel about changing R231 for an 11k ohm resistor? You could simply put a new resistor on top of the existing resistor. This would be an SMT thing, so probably less than desirable.

No, I'm using IN5, and I see that would be R251 which is also 500k, right? I'm not too keen on soldering SMT, I'll look to see what I can do to put in parallel to R251. Probably won't have time to try that before the weekend. And I should have thought of twisting the wires, I'll try that too and see what the noise does.

Can we get an ohms reading from the wiper to GND wire in the idle position and the WOT position on the Miata TPS?

TPS wiper to GND at WOT = 3.82k ohms

Thanks again for the help!

[tomiata]

Trying to clean up TPS, I twisted the wires together, cut off about 6" of extra length, and added a heavier ground wire to the ECU case and to TPS ground. It looks cleaner on little dso scope. See if the log looks any better.

stefanst wrote:

I notice you don't like to drive at high rpms and full throttle for example. You need to fix that
We're going to need that data for a perfect tune. But your cruise-cells look pretty close to perfect.

At this point I'd say your tune is definitely good enough to get a bit more aggressive with the target AFRs.

This log has a bit of WOT at high rpm. I need a place to do that more
Do you mean update target AFR table and manually tweak the tune rather than using auto-tune?

Thanks!

[kb1gtt]

TPS looks significantly better, there is still some room for improvement, installing the thru hole resistors would probably help with that last little bit of noise.

How about your Vbat, how does the scope look compared to the log?

I believe you are using the OEM harness for Vbat, and CLT, is that correct? Would it be hard to make those into twisted pair?

Is your CLT using the block as it's second wire, or does it have 2 wires on it? If it's GNDed via the block, perhaps find the straps that connect the block to the frame, and take it apart clean it up and reinstall the bonding wire. I'm a fan of No-Ox Id, it helps prevent corrosion issues and helps make a good low impedance electrical connection. Keep in mind the RF energy flows in the surface of the wire, not in the center of the wire. So if you have corrosion on a wire, or corrosion on a connector, the higher frequency electricity will flow through the corrosion, not the wire. For 10kHz the skin depth is about 0.039 inches. If you measured with a DMM, you would be measuring with DC, which would go through the center of the wire and show low ohms. However higher frequency stuff would see much higher ohms. AKA, clean connectors and a dab of No-Ox goes a long way.

http://circuitcalculator.com/wordpress/2007/06/18/skin-effect-calculator/

https://www.amazon.com/NO-OX-ID-Compound-Electrically-Conductive-Grease/dp/B00HSW341A

[tomiata]

TPS looks significantly better, there is still some room for improvement, installing the thru hole resistors would probably help with that last little bit of noise.

How about your Vbat, how does the scope look compared to the log?

I believe you are using the OEM harness for Vbat, and CLT, is that correct? Would it be hard to make those into twisted pair?

Is your CLT using the block as it's second wire, or does it have 2 wires on it?

vbat 5v output to CLT/IAT/TPS looks pretty noisy on the scope, I put a 470uf cap on it and it helped. I'll get some scope shots on it when I have more time. CLT is using original wiring and has two wires running to it, replacing the wiring is an option, but I think vbat is the main problem for CLT and IAT. Two of the engine grounds that I found look corroded. I have some permatex dielectric grease. Is No-ox about the same? Is it worth paying that much extra?

EDIT: my mistake, I didn't put scope on vbat yet.

[kb1gtt]

Dielectric grease is an electrical insulator. No-Ox is petroleum jelly with some copper or other metals in it which increase conductivity. The jelly provides an air barrier, as well the metals that are in the jelly tend to sacrifice themselves kind of like how galvanized metal works.

I forget was your alternator controlled by the Frankenso board, or by an OEM system? Perhaps we should do a bit more tuning on the alternator control.

[AndreyB]

small #299 progress - CUSTOM_OBD_SKIPPED_SPARK = 6045 code should be indicating a spark miss in the log. Works best with warningPeriod = 0

Can you please try driving a bit to confirm if the code would end up in the log?

[stefanst]

I forget was your alternator controlled by the Frankenso board, or by an OEM system? Perhaps we should do a bit more tuning on the alternator control.

Miata NA alternator is standalone and AFAIK quite nicely stable.

[kb1gtt]

If this is stock alternator control, than the Vbat noise is probably a bad wire, or early stages of a failing alternator. I would say first try fixing those engine ground straps, as well check any GND wires from the ECU for corrosion or bad connection issues. Also check your Vbat wire going to the ECU. Corrosion on relay contacts or any of the various connections between the battery and the ECU could be causing a higher frequency issue. Perhaps even add an additional GND wire direct from the frankenso, to the battery negative terminal. Frame GND's are commonly bad for a variety of reasons. Or perhaps temporarily add a wire direct from battery positive terminal direct to the Frankenso. Remember to fuse it very close to the battery.

IAT and CLT should be fairly low impedance which by itself should be fairly immune to noise issues. Are the OEM wires twisted? If not adding a twist and running them separately might help.

[tomiata]

Alternator is standalone.

Not sure if stock CLT wiring is twisted, it's in a sheath and in a tough place to get to on the back side of the engine.

I'll clean up the nasty looking ground I found next.
Thanks for the tips.

[stefanst]

None of the wiring on the Miatas I worked on was ever twisted, so I doubt that's the case for his '90.
Clean up the grounds and see what happens!

[tomiata]

small #299 progress - CUSTOM_OBD_SKIPPED_SPARK = 6045 code should be indicating a spark miss in the log. Works best with warningPeriod = 0

Can you please try driving a bit to confirm if the code would end up in the log?

Yes, I see the 6045 warnings. Log attached.

I cleaned 3 of the engine bay ground points and added another ground wire on the ECU enclosure. I don't see any noticeable improvements so far, will keep trying.

before:

12v-vbat-before-cleanup.BMP (38 KiB) Viewed 17366 times

after

12-after-gnd-cleaning.BMP (38 KiB) Viewed 17366 times

[kb1gtt]

Also check the noise at the battery. I expect the battery will be clean, and that you are getting your noise over the wire with it's various voltage drops and such. Just to make sure it's not the alternator bumping the voltage around, check it at the battery. It should be reasonably ripple free there. If it's clean at the battery and dirty at the ECU then you know the problem is with the wire in between.

[AndreyB]

Fresh log with #299 spark progress - seems like fuel is still an issue, will try same approach for fuel schedule.

[stefanst]

Interesting. The situation around 267s looks pretty bad....
Same around 288s and again 303s.
Every time we can clearly see oscillations of the RPMs as well.
If I understand this correctly, it's a problem with scheduling of fuel when the amount requested passes a certain threshold.
If you look in the log, we get a bit of chatter in the commanded PW, which may be based on noisy sensor inputs. In order to eliminate as much of this as possible, I suggest zeroing out the intake air fuel multiplier. Just set all the multipliers to 1.00

I would also like to see a bit more MAP signal smoothing, but I'm not sure where and how etc...