Gauge Cluster Tachometer Driving Puls Specification

[Heineken]

Since a few days my instrument cluster is out of the car, primarily for replacing the capacitors to prevent the popular cluster fire issue.



To avoid any issues before installing it into the car a small test bench is built to check lamps, indicators and the two main meters.
Drew helped with the pin-out of the gauge which is almost matching this 1997 JDM cluster except some small differences (seat belt, catalytic converter and high beam). Thanks!

The speedometer seems to be a simple circuit with 5V supplied from the cluster at pin B22 (with a ~10 k resistor in series). Note that the needle starts moving when touching B22 to ground fast and repeatedly.
About five lines of code on an Arduino are sufficient to create a timed interrupt switching 5V at pin 12. Using an S8050 transistor to pull down B22 is sufficient to get the cluster (including the odometers) to run.





Now the tachometer is of a different type of input, it seems. Power is supplied to the circuit via pins A13 and A27 (indicated by the brake light warning light coming on). The tacho pulses are coming in via pin A28.
There is no voltage coming out of the pin. Connecting the Arduino 5 V output directly to A28 distorts the signal but no needle movement. Manually contacting A26 to 12 V does not yield any movement either.


Has anybody been successful in driving the tachometer of the NSX cluster so far?
When looking at T3TEC's (and other) calibration rigs they don't look like voltages larger than 12 V are being used but I'm out of ideas

 

[Big McLargeHuge]

Have you read through this (http://www.nsxprime.com/forum/showthread.php/196322-Tachometer-Calibration) thread? Lots of good discussion and some secondary resources.

I've not done this before, but I think the output to the tach needs to be a square wave, 300 pulse/min = 100 RPM, not sure if you were doing this already with the 5V Arduino output. I've read a Kaz blog post (http://www.nsxcb.co.uk/entry.php?2471-Cooling-Sys-Brake-OH-etc-28 and http://www.nsxcb.co.uk/entry.php?2471-Cooling-Sys-Brake-OH-etc-28&goto=next) showing him using an oscilloscope (DSO138) or similar to drive the speedo and tach. It looks like the Arduino needs some ancillary components to generate a square wave, maybe you could buy one of the cheap oscilloscopes instead.

 

[Heineken]

Thanks for the information on the setup from Kaz, I wasn't aware of that post but the small DSO is a good idea to check the frequency of the speedometer signal generator and those are cheap (~30 Euro per piece).
The threads on nsxprime I covered, it seems. Someone talked about 100 V, which sounds highly dubious to me but I performed no measurement of my own so I can't tell.

After talking to a friend who works a lot on 90s Hondas he suggested to look into the manual for information on the ECM and there it was:


C8 of the ECM is the same as A28 on the gauge cluster and it's a 12 V Open-Collector kind of output with a transistor.
Taking note of that, I used a 10 V 1kHz calibration signal from an old Tektronix oscilloscope and after touching A28 with it the needle moved (way beyond a reasonable value as 1 kHz is roughly 20,000 RPM) :smile: but that comfirms that it's a 12 V square wave driving the tachometer.

The next step is going to design a matching driver circuit for the Arduino and incorporate everything into a test setup.

 

[Big McLargeHuge]

Good to know, I'd offer more advice but that's as much as I know .

By the way, if the needle moved way past the 9k RPM mark (or pegged itself 360 degrees on the backside of the "0" pin), I'd check to make sure the needle wasn't mechanically moved too far from its factory "range" from around 0-9(250?) RPM. When I was messing with my tach, you can move the needle past redline and feel a very slight resistance when it hits the internal stepper motor magnet, which I think changes the calibration somewhat, so if all of a sudden you're giving it 300 pulses and you're not getting 100 RPM then that might be the issue. Just an FYI, and certainly don't remove the needle from the gauge or you'll break everything :frown:

 

[Heineken]

The needle moved quite far, I'm afraid but as it was all driven by the internal coil mechanism I hope nothing was shifted .. so on the one hand good to have a test rack as I'm going to need it, it seems, on the other hand there's potential that something is already broken .. *sigh*. Not planning to touch any of the meters and needles as I heard they are super delicate (which you confirmed).

Just have to ensure to check the readings before changing the capacitors to ensure everything is recorded as-is and then re-check after the replacement

 

[Big McLargeHuge]

That would be a good idea. I didn't bother calibrating my cluster after I replaced the caps, mainly because my tach was already nearly non-functional due to some previous owner's work and I couldn't fix it. The speedo was still accurate at lower speeds, but I haven't tested much at higher speeds so some pot adjustment might be in order.

I would only go so far as removing the cluster lens and using a freshly gloved finger to gently push the needle back and forth, and only if absolutely necessary if the test rig doesn't output the results you expect and you've tried pot adjustment already. Like you said, better to leave that as a last resort.

I doubt you've broken anything, but once you go to re-assemble the cluster after you've replaced the caps, just be very careful tightening the screws between the gauges and PCB or you'll break a gauge coil wire and have to re-solder it, which is quite difficult with the gauge face in the way. Maybe you've read about this on Prime already, but it would be good to be careful since it's necessary for cap replacement and there have been a few Primers who've made this mistake and had to disassemble the cluster again to fix it (including my car's previous owner :redface.

 

[Old Guy]

I had drafted a rather lengthy response this morning on the difference in the tach and speedo circuits; but, then the Prime server did its 'too busy due to overwhelming interest in the new NSX' thing (as if!) so I sat on the post for a few hours. In the intervening time it looks like you have already discovered the internal pull up configuration on the C8 tach output pin in the ECU.

In that second link that @Big McLargeHuge provided where Kaz is testing the tachometer it looks like the Vmax on the oscilloscope display is 11 volts, so I am guessing that the tach is configured for a 12 v pull up.

Assuming that the tach circuit requires an external 12volt pull up, you could do this by connecting a 10,000 ohm resistor between +12v and the collector of the 58050 transistor that is being driven by the Arduino. Using your Arduino, this should give you a pulse that switches between 0.5 and roughly 12 volts. In Kaz's tests of the cluster he uses a 50% duty cycle for his pulse. If you configure your Arduino for a 50% duty cycle and measure the voltage on the collector using a DC averaging voltmeter you should measure something around 7 volts (the voltage never drops down to 0 volts). That is a quick check to confirm that it is working correctly.

You will note that on the internal schematic that you posted for pin C8 of the ECU indicates that in addition to the internal pull up resistor there is a resistor between the collector of the transistor and the C8 output pin. The function of this resistor is not clear to me. I suggest that you try something like a 1000 ohm resistor between the collector of the transistor and pin A28 when testing the tach.

If you look at Kaz's second post with the tach test, it looks like he is using a 150 hz signal to generate a tach reading of 3000 RPM. The service manual indicates 100 RPM per 300 pulses per minute of the igniter unit which is 100 RPM for a 5 hz signal which is equivalent to 150 hz for 3000 RPM - what Kaz is using.
​

 

[Heineken]

Thanks a lot for the information - why I was wondering about the relative silence regarding my post you were already busy writing up detailed information - next time I'm going to be more patient
Good point to look at the wave forms measured in Kaz's DSO to get a reference about the voltage, I missed that one.

Been a little busy and showed my driving circuit ideas to a friend and co-worker for inspiration. The results are incorporated into the following diagram:



The replacement caps for the gauge cluster arrived yesterday so that's set up, too.

A similar small DSO like Kaz's has been ordered and is expected to arrive in a few days.
Playing with ideas for a small universal test box and thinking about various designs.

 

[drew]

>that something is already broken .. *sigh*. Not planning to touch any of the meters and needles as I heard they are super delicate

FYI: The NSX clusters can be repaired by robbing 1991-1995 Legend clusters. They use the same exact stepper motor as the NSX.

I buy Legend clusters on ebay for US$40 shipped and just take the stepper motors. The needles are very close in appearance too in case those get buggered.

You can practice taking apart other Honda/Nippon Seiki clusters as they are common and cheap, but nothing on them that I found can be migrated over.

 

[Big McLargeHuge]

>that something is already broken .. *sigh*. Not planning to touch any of the meters and needles as I heard they are super delicate

FYI: The NSX clusters can be repaired by robbing 1991-1995 Legend clusters. They use the same exact stepper motor as the NSX.

I buy Legend clusters on ebay for US$40 shipped and just take the stepper motors. The needles are very close in appearance too in case those get buggered.


I bought several Legend clusters ove the COVID Holiday as junkyards were using their free time to liquidate old inventory. I believe the Vigor cluster will also work as well, but this needs to be verified.

You can practice taking apart other Honda/Nippon Seiki clusters as they are common and cheap, but nothing on them that I found can be migrated over.

Not trying to hijack the thread, but that is good info and if they're interchangeable I might get a Legend cluster and see if I can get the stepper motor to work for my broken tach. I would only have 2 chances to remove the Legend needles without breaking the motor again to re-assemble the motor onto the NSX gauge face, but for $20 a try it would be worth it to see if I could save my cluster.

Is there any special procedure for removing needles without breaking the internal stepper magnet out of its tracks? Even a speedo shop locally said they are not always successful removing needles from motors so they'd rather not try it. Two forks remove the needle easily enough, but I'm hoping there's some trick to keeping the motor intact besides luck. Maybe I could break the needles away from the post and leave the post inside the motor.

If all this works, Heineken's posts here might help me too if I decide to try and recalibrate my speedo & tach :biggrin:.

EDIT: This was the best picture I could find of a Legend stepper motor. Looks pretty much the same at a glance to me, so I might give this a try.

 

[drew]

My special procedure:
1. grip spindle of stepper motor with grippy needle nose pliers
2. move needle left and right, not up, to break glue and friction
3. keep moving needle left and right while pulling slowly until the needle comes off
4. to re install. Power up cluster, each side as a power and ground. The stepper motor goes to zero and reinstall the needle.

What I don't understand is how JP companies Miner and Spoon can offer a "gauge face change as an exchange service" without an extremely reliable method of needle extraction and reinsertion. One botched job and the profit from 15-20 units evaporates.

The gauge maker Nippon Seiki is opaque and doesn't seem to offer anything other than OEM Factory support.

 

[Heineken]

Maybe they just had enough Legend clusters to play with :biggrin:?

I don't have any special procedure other "The Two Forks" method.

What I don't understand is how JP companies Miner and Spoon can offer an "gauge face change as an exchange service" without an extremely reliable method of needle extraction and reinsertion. One botched job and the profit from 15-20 units evaporates.

The gauge maker Nippon Seiki is opaque and doesn't seem to offer anything other than OEM Factory support.

Regarding the current status of the test bench, a front plate design has been created:


The casing should be big enough to provide space for the Arduino and the other components:


Today's test of the tachometer driver circuit was successful:



100 Hz equals 2000 RPM so nothing is broken *phew* ..
The driver needs a little more punch with a smaller collector resistor, it seems. 1.33 kOhm is providing a much sharper wave form than with 10 k (below):



Did not have a third hand to take a picture of the improved waveform.

Hopefully the parts list is completed soon and the order can be sent out.

 

[Big McLargeHuge]

Maybe they just had enough Legend clusters to play with :biggrin:?

Maybe they have proprietary Spoon needle extractor machine of their own design. I read on a past thread (http://www.nsxprime.com/forum/showthread.php/165296-Cluster-needle-removal post #25 ) on that topic that Sudesh had his own way of 100% removal success, so there must be something out there no one is willing to share specifics on. IMO not particularly benevolent since the % of people with access to their services over the years is very small and then nonexistent when they go dormant. Anyways, enough about that here :wink:.

Very cool, thank you for sharing all this! I have an Arduino Mega lying around that I was planning on using to drive addressable LED lights, but I might use it in the future to do as you are and get my cluster back in order. Once you finish, any info you have on your final setup would be greatly appreciated since I have never used an oscilloscope or designed any circuits, and I doubt anyone near me would care to calibrate the cluster for me. Once I get that set up and assuming the Legend stepper motor works, I wouldn't mind posting my own cluster calibration experience here if it adds to the conversation.

Here's my little spreadsheet of the caps I replaced. There are quite a few MLCC's/tantalum ones I replaced just because I could, but it was probably a waste of time.
No guarantees they are the same as your cluster especially since mine is from a 1990, but hopefully it helps a bit.
I ordered ones with the highest lifespan I could find (5000-10000+ hours) and a temp rating of at least 105°C, most ended up being Nichicon which I think were mostly the same as factory.
The two largest caps (C1 on both boards) needs to be 20mm and NOT 25mm tall as I ordered the first time, the extra 5mm makes the circuit boards not fit back into the case properly since the top of the caps will interfere. Just FYI. Also, C17 and C18 on Board A are both bipolar/non-polarized caps which tripped me up too.

 

[nsxmugen]

Not trying to hijack the thread, but that is good info and if they're interchangeable I might get a Legend cluster and see if I can get the stepper motor to work for my broken tach. I would only have 2 chances to remove the Legend needles without breaking the motor again to re-assemble the motor onto the NSX gauge face, but for $20 a try it would be worth it to see if I could save my cluster.

Is there any special procedure for removing needles without breaking the internal stepper magnet out of its tracks? Even a speedo shop locally said they are not always successful removing needles from motors so they'd rather not try it. Two forks remove the needle easily enough, but I'm hoping there's some trick to keeping the motor intact besides luck. Maybe I could break the needles away from the post and leave the post inside the motor.

If all this works, Heineken's posts here might help me too if I decide to try and recalibrate my speedo & tach :biggrin:.

EDIT: This was the best picture I could find of a Legend stepper motor. Looks pretty much the same at a glance to me, so I might give this a try.

I just found this interesting thread, it’s something I been working on too.

I took apart few honda/Acura clusters and the motors look the same. On my table I have Acura CL 2003 and Honda Civic 97. Yes removing with fork works just put some leather under it so does not indent the cluster. I notice one fast woosh is better then slowly applying the force but each it’s own. The needle can easily brake inside the motor. I left tech on my table when my son thought it was a toy and broke it of easily.

Here are few pics for reference. Great work btw!!

Sent from my iPhone using Tapatalk Pro

 

[Heineken]

Thanks for the plentiful information. My list is available in this thread: http://www.nsxprime.com/forum/showt...-Light-Bulbs?p=2014095&viewfull=1#post2014095 (needs Desktop view to see the complete table)

 

[Big McLargeHuge]

Thanks for the extra info @nsxmugen :smile:. I just ordered a cheap '93 Legend cluster to cannibalize for the motors, I can try to "rip/whoosh" the needle straight off to see if it works okay but I think I'll try breaking the needle away from the post first to circumvent the issue. If it works, it would be a great alternative to the purchase of a $700+ NSX single gauge face & motor, which I find absurd.
It looks like those fuel/temp/etc. gauges have some extra components that probably makes them incompatible with the MPH/tach gauges, like the NSX cluster. I'll do some experimenting once my S2000 cluster conversion finally arrives (thanks USPS) and report back.

Thanks for the plentiful information. My list is available in this thread: http://www.nsxprime.com/forum/showt...-Light-Bulbs?p=2014095&viewfull=1#post2014095 (needs Desktop view to see the complete table)

Ah I saw that a few days ago, you have more info than me :biggrin:. Well, this should be helpful for someone looking for the differences between the 90/97 clusters.

 

[nsxmugen]

Thanks for the extra info @nsxmugen :smile:. I just ordered a cheap '93 Legend cluster to cannibalize for the motors, I can try to "rip/whoosh" the needle straight off to see if it works okay but I think I'll try breaking the needle away from the post first to circumvent the issue. If it works, it would be a great alternative to the purchase of a $700+ NSX single gauge face & motor, which I find absurd.
It looks like those fuel/temp/etc. gauges have some extra components that probably makes them incompatible with the MPH/tach gauges, like the NSX cluster. I'll do some experimenting once my S2000 cluster conversion finally arrives (thanks USPS) and report back.



Ah I saw that a few days ago, you have more info than me :biggrin:. Well, this should be helpful for someone looking for the differences between the 90/97 clusters.

Correct, the fuel, coolant and oil motors have extra resistors on them. This is the same for acura cl and civic cluster that I have, those motors also have extra resistors on them. MPH and RPM do not have any resistors on them and this is also the same for acura cl and civic. It might be the same for all honda cars until they were forced to go canbus starting with 2007/2008 model lineup, I don't know. Yes, the first time I remove the needle from the motor it was difficult and those things feel like they are crazy glued. But the next time you will know what to expect. I removed many on vw/audi and those are easy. I was also suggested to keep the cluster in freezer overnight and try it, but I did not notice any difference, they are still stuck on there pretty good.

 

[Heineken]

Parts for the tester have been ordered. I'm looking forward to assembly and test

 

[drew]

>Parts for the tester have been ordered. I'm looking forward to assembly and test

When are you taking orders? Looking forward to being a supporting first adopter...

 

[Heineken]

Once the first one is assembled and working we are going to see what's next.
In general I'm OK with creating multiple copies if my costs are covered and people are OK with waiting until it's ready

 

[drew]

While Heineken is performing magic,

I wanted to see what a cheap bastard hack like me could do with a <$2 signal generator from AliExpress.

The NSX tach is driven by a 12V square wave doing 50% duty at 50hz per 1000rpm. Or 1hz = 20 rpm. *



Above cluster is showing 143hz for 3,000rpm on the tach side. Ths cluster has yet to have it caps replaced.


The NSX speedo is also driven by 5V square wave doing 50% duty at 42.47hz for 60kph. Or 1hz = 1.41 kph (NOT 12V)

[photo in the future, this cluster will have to be repaired from previous issues...which is why I have it]

I used a cell phone 5v usb power supply using a "USB Male to 5.5 mm/2.1 mm 5.5*2.1 DC Barrel Jack Power Cable AC plug Transfer Connector Charger interface converter" This wire can also power the signal generator which makes it output 5v.



My goal was to make an inexpensive test rig with 2x signal generators, some analog switches with resistors, and others for lights. It would look like something built while in a prison cave located an Afghanistan. Then Heineken comes along and rains on my parade, and shows everybody who the real master is . He also graciously helped me with my math too. I'll be supporting his work because he is doing it right.


*Other Honda vehicles (Accord, Civic) that use the smaller stepper motor are driving at 33HZ per 1000rpm. Accord and Legend appear to use 12v for both speedo and tach....not the NSX.

Get your own cheap signal generator here "1Hz-160Khz DC 4V-30V 5mA-30mA 2 Channel Dual Signal Generator PWM Pulse Frequency Duty Cycle Adjustable Module LCD Display"

Amazon also has a similar signal generator (at 4x the price: $8 vs $2
https://www.amazon.com/Frequency-1H...ctangular-Generator/dp/B08793P3XJ/ref=sr_1_20

 

[Honcho]

While Heinekin is performing magic,

I wanted to see what a cheap bastard hack like me could do with a <$2 signal generator from AliExpress.

The NSX tach is driven by a 12V square wave doing 50% duty at 50hz per 1000rpm. Or 1hz = 20 rpm. *

View attachment 164858

Above cluster is showing 143hz for 3,000rpm on the tach side. Ths cluster has yet to have it caps replaced.


The NSX speedo is also driven by 12V square wave doing 50% duty at 42.47hz for 60kph. Or 1hz = 1.41 kph

[photo in a bit]

My goal was to make an inexpensive test rig with 2x signal generators, some analog switches with resistors and others for lights. It would look like something built while in a prison cave located an Afghanistan. Then Henekin comes along and rains on my parade, and shows everybody who the real master is . He also graciously helped me with my math too. I'll be supporting his work because he is doing it right.


*Other Honda vehicles (Accord, Civic) that use the smaller stepper motor are driving at 33HZ per 1000rpm.

Get your own cheap signal generator here "1Hz-160Khz DC 4V-30V 5mA-30mA 2 Channel Dual Signal Generator PWM Pulse Frequency Duty Cycle Adjustable Module LCD Display"

I'm buying one now- I'm gonna test and calibrate my cluster while it is out. Thanks for this link!

 

[drew]

>I'm buying one now- I'm gonna test and calibrate my cluster while it is out. Thanks for this link!

I need to start a Shopify Affiliate or something...

For $20 dollars more you can purchase an o-scope to verify the signal generator.
https://www.aliexpress.com/item/32826369650.html

1990-1994 NSX (I'll have to verify the others)

Tach side pinout:
P2 Ground
P3 12V
P4 Tach signal 12v

Speedo side pinout:
P8 Ground
P9 12V
P24 VSS Speedo signal, 5v

Orientation
Hold cluster as you would normally view it while driving
Put cluster on the table on its back.
Numbering is Left to Right, with the locating gap on top.

 

[Heineken]

Important

Before someone (potentially) kills his cluster

The speedo input signal from the VSS (P24 on 1990-1994) is 5 V (not 12 V)
To create this signal, the signal generator from Drew needs to be powered by 5 V

Everything else is fine

PS: Thanks Drew for fixing it in your post!

 

[Heineken]

Update

Since Drew cut some corners with a very simple but working test setup it time to give an update on the sophisticated version.

After the first set of parts arrived I managed to botch it up and ruined two set of cases and had to re-order. On the positive side, a huge number of issues showed up that are now resolved and several items that have been improved:

• Relay pin-out and circuit (fix)
• Switch-sensing circuit (fix)
• Case layout (fix)
• PCB layout (simplification)

As requested by Drew a larger case is available, too for those that would like to extend the rig with cables and switches check lamps and other functionalities (beside the gauges).

The design is now upright since there is more space needed to fit the back of the switches:


The circuit has been fixed several times but should be OK now:



The PCB layout was changed accordingly (it's not going to be etched, just for prototype wire routing)





For those interested, the work was done using the freely available KiCad. Complicated but powerful (consult a training manual before use - even I needed one).

The final PCB is going to an Arduino Shield, easy to separate and attach:





Note the famous Arduino Uno Layout-bug with a half 2.54-grid offset on the second connector row - luckily there no additional pins to connect.