Speedo reads low when ambient temp is cold, accurate when hot

[usafdarkhorse]

Primers,

First, details: 1990 Honda NSX that reads speed in kilometers per hour.

I'm running into a head scratcher. I noticed a few months ago that I felt like I was going pretty fast down a stretch of road I normally take. Too fast.
I used a GPS app to confirm my suspicion that I was going faster than my speedometer was reading, sometimes up to 20mph faster.
After I left work, when the sun was up and it got fairly warm out (>80F), my speedo was reading right on the money.
I've just been living with it here for months, maybe longer!

The most I've been able to research and read here on Prime, it sounds like the vehicle speed sensor (VSS) is on its way out. So, being a pretty inexpensive part I changed it out.

No change. Am I looking at a speedometer cable? Any NSX Master Techs want to chime in and help? Thanks!

 

[docjohn]

what size tires are you running

 

[usafdarkhorse]

Close to OE 94-01 specs except 215/45ZR16 Front at 32 psi and 255/40ZR17 Rear at 33psi.

 

[Old Guy]

No speedo cable on an NSX.

Need some clarity around the speedo that reads in km/hr. Did you import a Canadian spec NSX or did you import a metric cluster and transplant it into your 1990 (1990???). If you did the latter, then you might have a calibration issue. Has the problem always existed or did it materialize recently?

If its a recent change, the fact that it is temperature sensitive makes me think that the electronics on the circuit board for the speedo might be having issues. I have never pulled the instrument cluster; but, I expect that there are capacitors in the speedo drive circuit and a lot of owners have become aware of the ageing electrolytic capacitor issue. Electrolytic capacitors and resistors are slightly temperature sensitive by nature. Ageing capacitors could be demonstrating increased temperature sensitivity. However, that said I think your error is going in the wrong direction as I would expect the tolerance drift to increase with temperature making the error greater at higher temperatures. That is something I would have to check out.

You might not want to 'just live with this' as I recall a post sometime ago discussing an NSX loss due to a fire that originated in the instrument cluster. I seem to recall some discussion that the fire was attributed to capacitor problems; but, it was a while ago and the details are kind of fuzzy. You might want to contact Brian K at NSX Erepair. He may be able to clue you in on cluster issues and provide a repair if required.

 

[MotorMouth93]

A couple months ago I was seeing similar inconsistent speedo readings. Sometimes perfect within 1-2 of what the GPS said and other times I was going 20mph faster. The issue ended up being capacitors in the cluster.

 

[Spottwd]

I, too, have been questioning the accuracy of my speedometer. When driving at or above the speed limit, in town at "city" speeds or when on the highway at posted speeds of 70-75mph, there sure are a whole bunch of folks just zooming by when I'm reading 75-80. So I think that my speedometer is reading high.

Back in the day there were places you could take your car and get it on a dynamometer or some other sort of machine and have it run in order to check speedometer reading again wheel rotation speed. I can't find anyone that does that anymore. The best suggestions I get are to time the travel between mile markers on the highway. Haven't done that yet, and I have no GPS.

Does anyone know what the tachometer vs. speedometer correlations are for gears 1-5? No, no, I don't want reverse!!!

1991, OEM wheels and tire sizes.

 

[The King]

Around here there is no problem checking your speedo, as there are many speed check radar signs in school zones. Too bad drivers don't know the rules for school zones, as the lower speeds are only for when children are actually present.

 

[Honcho]

I, too, have been questioning the accuracy of my speedometer. When driving at or above the speed limit, in town at "city" speeds or when on the highway at posted speeds of 70-75mph, there sure are a whole bunch of folks just zooming by when I'm reading 75-80. So I think that my speedometer is reading high.

Back in the day there were places you could take your car and get it on a dynamometer or some other sort of machine and have it run in order to check speedometer reading again wheel rotation speed. I can't find anyone that does that anymore. The best suggestions I get are to time the travel between mile markers on the highway. Haven't done that yet, and I have no GPS.

Does anyone know what the tachometer vs. speedometer correlations are for gears 1-5? No, no, I don't want reverse!!!

1991, OEM wheels and tire sizes.

Assuming you have the US market gear stack, at 8,000 rpm, you have the following speeds (using the 94-01 OEM tire size 245/40/17 - 24.7 inches):

1st --> 47 mph
2nd --> 84 mph
3rd --> 118 mph
4th --> 150 mph
5th --> 188 mph

Obviously tire pressure and brand can change this slightly. I use the Conti Extreme Sport tire as my reference, since that is what I run.

 

[Old Guy]

Be aware about separating the issue of a general drift in speedometer accuracy versus the temperature sensitive accuracy problem. It would be pretty normal for a non digital electronic speedometer to have a permanent accuracy drift with component ageing. I expect that all first gen NSXs are probably suffering from this problem to some degree. Electrolytic capacitors age whether they are turned on of off and the rate of ageing is a function of the capacitors long term temperature. NSXs living north of the 49th may go for decades without capacitor issues. NSXs that get parked in the sun in Phoenix - maybe not so long. I recall that Kaz raised the issue of drift in one of his posts. There are calibration pots on the cluster that should allow you to recalibrate both the speedo and the tach (easy for me to say when I have never had t do it!).

The variation of speedo accuracy with ambient temperature is a slightly different problem

 

[Spottwd]

Thanks for the info. Yes to US market.

Does anyone have data for 1991 OEM wheels and tires (205x50R15 front, 225x50R16 rear). And, maybe for less adventuresome me, at lower rpms.

 

[Kaz-kzukNA1]

Variable speedo offset under different temperature and humidity is the typical phenomenon with the leaky capacitor and failing pot (variable resistor) on the gauge circuit board.

Have you ever seen the false 'BARKE LAMP' or 'Door ajar' light inside the warning indicator at the centre bottom of the gauge in the past?
Those are the typical first sign of this leaky cap issue.


If you have electronics background or know someone, just remove the gauge and bench test it.

In order to eliminate the different chassis setup, tyre/wheel size factors, all Japanese car manufactures must follow the JIS definition for the VSS output and that is VSS 637rpm = 60KPH.
As you can see, it's based on each rotation of VSS and not the tyre/wheel so all JDM cars would follow the same rule.
The number of pulses per rotation is also defined with several options but for our NSX, it's 4 pulses per rotation.

f [Hz] = V [kph] x 4 [pulse] x 637 [rpm] /60 [sec] / 60 [kph]

So,
30kph = 21.23 Hz
60kph = 42.47Hz
100kph = 70.78Hz
180kph = 127.4Hz

Just set your function generator as square wave, 0 - 5Vpp, 50% duty.
If you want preventing the odometer from accumulating the mileage while on the bench, just disconnect the odometer connector from the circuit board.

If your gauge unit was never refurbished in the past, just send it to BrianK before it gets worse.

All NSX will suffer from this.


Kaz

 

[docjohn]

haha next time
I get a speed infraction I will hand the magistrate a copy of this post:biggrin:

 

[Old Guy]

haha next time
I get a speed infraction I will hand the magistrate a copy of this post:biggrin:

Probably not. You have left a digital finger print in the thread which means that you are aware of the problem. I expect that awareness of the problem negates the problem as a potential mitigating factor in your defense. In fact, float the idea as part of your defense and you might irritate the Court. Best to suck it up and pay.

I spent a fair amount of time with lawyers during my career and it appears to have altered my thinking!

 

[usafdarkhorse]

No speedo cable on an NSX.

Need some clarity around the speedo that reads in km/hr. Did you import a Canadian spec NSX or did you import a metric cluster and transplant it into your 1990 (1990???). If you did the latter, then you might have a calibration issue. Has the problem always existed or did it materialize recently?

If its a recent change, the fact that it is temperature sensitive makes me think that the electronics on the circuit board for the speedo might be having issues. I have never pulled the instrument cluster; but, I expect that there are capacitors in the speedo drive circuit and a lot of owners have become aware of the ageing electrolytic capacitor issue. Electrolytic capacitors and resistors are slightly temperature sensitive by nature. Ageing capacitors could be demonstrating increased temperature sensitivity. However, that said I think your error is going in the wrong direction as I would expect the tolerance drift to increase with temperature making the error greater at higher temperatures. That is something I would have to check out.

It's a Japanese import and their cars follow manufacture year. I bought the car when I lived in Japan and brought it home with me to the US. Most North American 1991 NSXs were actually built in 1990. I can see how that might be confusing as I've been looking for a Canadian speedo to get the wider speed band and ODO in kms while also having a reference to mph.

Thanks for the capacitance tip. The issue probably didn't materialize recently, I've just noticed it recently. Honcho's article in NSX Driver came at a good time.

A couple months ago I was seeing similar inconsistent speedo readings. Sometimes perfect within 1-2 of what the GPS said and other times I was going 20mph faster. The issue ended up being capacitors in the cluster.

Thanks for the tip. I just read Honcho's article in NSX Driver.

I use the Conti Extreme Sport tire as my reference, since that is what I run.

PM'd you about your article.

Variable speedo offset under different temperature and humidity is the typical phenomenon with the leaky capacitor and failing pot (variable resistor) on the gauge circuit board.

Have you ever seen the false 'BARKE LAMP' or 'Door ajar' light inside the warning indicator at the centre bottom of the gauge in the past?
Those are the typical first sign of this leaky cap issue.


If you have electronics background or know someone, just remove the gauge and bench test it.

In order to eliminate the different chassis setup, tyre/wheel size factors, all Japanese car manufactures must follow the JIS definition for the VSS output and that is VSS 637rpm = 60KPH.
As you can see, it's based on each rotation of VSS and not the tyre/wheel so all JDM cars would follow the same rule.
The number of pulses per rotation is also defined with several options but for our NSX, it's 4 pulses per rotation.

f [Hz] = V [kph] x 4 [pulse] x 637 [rpm] /60 [sec] / 60 [kph]

So,
30kph = 21.23 Hz
60kph = 42.47Hz
100kph = 70.78Hz
180kph = 127.4Hz

If your gauge unit was never refurbished in the past, just send it to BrianK before it gets worse.

All NSX will suffer from this.


Kaz

Kaz, thanks for weighing in. I haven't seen the Brake Lamp indicator or anything else on the gauge cluster. I have around 98K km on the car, so this may just be the first indications of the capacitors failing. I will see about removing the gauge cluster and sending to BrianK.

Thanks for the replies everyone.

 

[drew]

Kaz

I wish to replicate your bench test rig. What are you using as a signal function generator?

Actually I want to make something like the T3TEC meter test board, but I'll start with the tach first.

 

[Old Guy]

Kaz

I wish to replicate your bench test rig. What are you using as a signal function generator?

Actually I want to make something like the T3TEC meter test board, but I'll start with the tach first.

If you need a scope and waveform generator (and digital analyzer on some of the options), check out Analog Discovery and The Waveforms software.

https://analogdiscovery.com/

Its now the go-to tool for electrical engineering students around here. If you search, you can find less expensive scope only versions for less than $100; but, some of the software is a little flakey.

If all you need is something to generate a pulse for calibration of the tach or speedo (a pulse being a square wave with 100% offset), an Arduino or Trinket or whatever equivalent should do the trick for around $10. Given your history, I would think you would have no problem coding it; but, if you are interested I think I still have a very clumsy piece of code that I was running on an Arduino to generate synchronized cam and tach pulses for testing a one-off coil on plug installation on another car. It used external pots to adjust the pulse frequency (engine speed). The only problem might be that the pulse duty cycle was written in code to replicate the cam sensor and it is not 50%. If the NSX speedo needs a 50% duty cycle you would need to modify the code and naturally enough, my code is not commented. At that point might be easier to write your own code. All you need to do is add an output transistor in pull-up configuration as the Arduino may not have enough current drive to drive the speedo.

 

[drew]

begin: thread_hijack

Old Guy, the device you refer me to is US$280+s/h. While nice, it is overkill and a bit pricey in comparison to the alternatives.

The DSO138 scope that Kaz shows is US$13 inclusive from AliExpress, which is an easy ship to NZ. The deluxe version is a killer US$19!

I am hoping Kaz will spill the beans on the function generator so I can do some click shopping today.

There is a possibility of creating a decent and complete calibration and test board for less than US$100. It might look like a Tony Stark creation out of a cave in Afghanistan, but it should do the job. I even have a spec list and a set of cluster connectors in anticipation of this project.

 

[Old Guy]

Yes; but! I believe the DSO138 is just a single channel scope. You were looking for a function generator which the Analog Discovery does in addition to being a scope. It also has good FFT software should you ever feel the need and a digital logic analyzer option.

Really all you need to calibrate the speedo and the tach is a pulse train generator. At the pulse rates that Kaz identified you could probably do it with a Trinket. You only need the scope if you want to verify the output of your pulse generator.

 

[Spottwd]

Nice discussion, but.....

Can anyone just tell me the speed (mph) in 5th gear at 3000 rpms 1991 OEM 15 and 16" wheels 205 and 225 ties is supposed to be.

 

[drew]

Suspend: Thread_Hijack

A quick google shows:

https://www.nsxprime.com/wiki/Gear_Ratios

8000 RPM = 186mph, therefore (186/8) * 3 = 69.75MPH.

But this is academic, as you have no idea if your tach is even correct. If you doubt your speedo then your tach readings will be suspect too. Mostly pointless really.

The BEST thing to do is use a multi-billion dollar calibration tool that integrates atomic clocks and the Theory of Relativity to verify your actual speed. And this calibration tool is available to YOU by virtue of the American taxpayer.

IOW: get your smartphone out and compare the relative speed as stated by GPS and compare it to your speedo reading.

Resume: Thread_Hijack

[MENTION=26435]Old Guy[/MENTION], I am a software puke, but I treat technology like seafood, only buy what you are going to use. (Unless I am on a government project and then I try to add on every bit of junk I can justify as it will give me future capabilities, stuff to trade with others and account for internal "reacquisition" losses from the supply chain. I consider the implementation of 30% of the original spec to be an extreme success on government projects.* )

Here is my inspiration (top photo courtesy of Kaz)




I'm shooting more for a "built-in a cave" look without the silkscreened cover and compact electronics.

* Usually, I have to buy a lot of expensive stuff to meet the minimum contract amount. It is easier to get $10M than it is to get $10K. A while back I needed a $7K item, but I had to spend $150K to get it, I bought all kinds of stuff that I thought I could use over the next five years. I bought stuff for everybody as well to get their sign off. It took two years to get my $7K item and by that time I had "found" (an internal reacquisition that nobody was guarding at the time) a suitable replacement and I didn't need it anymore. However, the extra stuff I bought went to super good use, I was able to extract a 5x increase in productivity with half the people...it was awesome.

 

[furinax]

Wow that looks like the holy grail of dash repairs!!!

I can't imagine bad capacitor in a speedo could cause an entire car to go on fire.

Electrical problems suck. I wonder what the behavior is like as the capacitors start to dissipate. My guess is that it will slightly drift over time as opposed to failing very suddenly?

You could get one of the escort radar detectors which I believe use GPS to determine your speed. It's very handy in case the VSS ever fails (or if you run larger tire sizes).

 

[drew]

[MENTION=31854]furinax[/MENTION]

nsx cluster fires (courtsey of Kaz)
http://www.nsxcb.co.uk/showthread.php?11827-Speedo-problem&p=115142#post115142

[MENTION=29139]Spottwd[/MENTION]
Google Maps will show your speed too. Just go into navigation mode by picking a destination and clicking Start.

 

[usafdarkhorse]

You could get one of the escort radar detectors which I believe use GPS to determine your speed. It's very handy in case the VSS ever fails (or if you run larger tire sizes).

I just used an app that utilizes GPS to confirm. Civil GPS isn't mil spec accurate, but it can get you within a couple knots at 500+ knots groundspeed, so fairly accurate.

RPM in gear is a good way to confirm too and it's how I had been driving the car when it's colder out until I could figure out the fix.

I'm pulling out the gauge cluster this weekend. No fires for me thank you.

 

[Old Guy]

Electrical problems suck. I wonder what the behavior is like as the capacitors start to dissipate. My guess is that it will slightly drift over time as opposed to failing very suddenly?

Depends on the nature of the failure. Gradual ageing of the components will result in drift. However, one of the common failure modes of electrolytic capacitors is swelling where the rubber plug on the end of the capacitor is dislodged allowing the electrolyte to leak out. Depending on the capacitor the electrolyte can be more or less corrosive leading to the damage on the covering on the circuit traces shown in the photo from Kaz. The damage may be sufficient to destroy the trace - some of the photos of CCU and Bose door amps show severed traces requiring a 'green wire' repair. In such case you might have a gradual drift in accuracy as the capacitor changes and then an abrupt failure if you have a severed trace. Green wire repair is more difficult than simple board clean up and capacitor replacement.

The specific mechanism of instrument cluster failure leading to a fire is hard to verify because the evidence tends to get destroyed. However, if a capacitor leaks out electrolyte, it is possible that this may lead to the development of an internal short in the capacitor. Electrolytic capacitors have rupture discs for such events; but, that does not guarantee removal of the faulted component. Leaking capacitors don't guarantee that you will have a vehicle fire; but, it is a possibility with an un quantified risk.

It becomes a bit of a risk assessment like going to the dentist. Do I go early and perhaps catch the problem while it is a small surface cavity or wait until its a root canal and crown repair.

 

[usafdarkhorse]

Pulled out the gauge cluster yesterday morning. Wasn't too tough to get out with the removable steering wheel, took maybe 30min.

We'll see what BrianK has to say about the boards.