• Protip: Profile posts are public! Use Conversations to message other members privately. Everyone can see the content of a profile post.

Gauge Cluster Calibrator

Joined
5 August 2014
Messages
208
Location
Germany
Even though working-at-home, home-schooling and home-kindergarten doesn't leave much spare time it's important to have a bit of a hobby for a change. Since I'm working through the ECUs related to capacitor replacement it's now the gauge cluster's turn. Replacing its capacitors pro-actively is especially important as leaks can cause shorts which have already destroyed whole clusters and even burned down cars (according to rumours).

IMAG3554s.jpg

The gauge cluster is fully electrified (no mechanical links) but not a (CAN-) bus system yet. As a result, the signals driving the tacho- and speedometer are rectangle pulses with a defined frequency. These pulses are converted to voltages by an analogue circuit which drives the needles. This circuit needs calibration after a replacement of the capacitors, a job for the two variable resistors on each half of the cluster PCBs.

Unless there's access to a dyno these input signals need to be simulated to be able to calibrate the gauges. Calibrating them by means of a test drive on a closed circuit is possible, too but certainly not advisable on public roads. Jacking the car up might be possible as well but in case of my 1997 AT it caused the traction control system to kick in and needs matching gear/RPM ratio values as well as a separate RPM gauge ..

The first issue to come across is the pins on the gauge. These changed several times across build years and differ based on region, too. Luckily the tachometer/speedometer pins seem to differ on build year only so information from the repair manual as well as the Electrical Troubleshooting manual can be used.

IMAG3645~2.jpg

IMAG3823fs.jpg

To drive the tachometer on a 1997-2002 cluster only these three pins on the left green connector (when looking a the dials) are required:



  • Power Supply – A13
  • Ground – A27
  • Tachometer Signal – A28


The Electric Troubleshooting manual does not offer further information on the type of signal that is provided from the engine ECU to the cluster but the repair contains the missing link at the iPGM-FI section:

Screenshot from 2020-04-26 16-54-43.png

The cluster is supplied with an open collector style setup controlled by a rather simple transistor circuit creating a near 12 V rectangle signal. The frequency of the signal can be calculated as such: RPM = Hz * 20 (e.g. 40 Hz for 800 RPM) - a common standard.

Looking at the speedometer, it's the right green connector's turn with these pins:



  • Power Supply – B2
  • Ground – B7
  • Speed Signal – B22


The repair manual mentions that it's a 5 V rectangle signal. The frequency is according to the Japanese standard implemented at Honda: km/h = Hz * 1.41.
A somewhat important difference from the tachometer is that the 5V are supplied from the cluster and pulled-down to ~0 V by the Vehicle Speed Sensor (VSS).

Creating a tester for these signals is not exactly complicated and considering a minimalistic approach not even necessary (it's sufficient to buy a cheap 5-12 V digital PWM generator from AliExpress) but my target was to provide an easy to use version that helps to speed-up the calibration process as well as to re-activate my somewhat rusty electronics know-how :smile:

The amateur's choice of microcontroller for such a task is typically an Arduino Uno as it's quick to set up, easy to program, has good support for rectangle signals as well as for all the additionally required elements such as LEDs, buttons and the like. Before reaching that point, though we need to check what's necessary to actually drive the cluster and create a working prototype of these circuits.

Starting with the speedometer we need to drive a transistor by means of the Arduino to pull down the 5 V from the cluster at the required frequency. As we are not dealing with high frequencies an S8050 is sufficient for this task:

IMAG3592~3.jpg

Using a few lines of code (utilizing the TimerOne Arduino library), jumper cables and crocodile clips the dial shows the expected ~100 km/h at 71 Hz :smile:

IMAG3579s.jpg

The tachometer needs a 12 V rectangle input and therefore the circuit is a little different:

IMAG3588~3s.jpg

A successful test confirms a working prototype (100 Hz equals 2000 RPM):

IMAG3602s.jpg

During the actual calibration it's required to quickly switch between low and high calibration values. The change should therefore be quick and easy, preferably by means of a button press. There is a need to switch between speedo- and tachometer and their respective calibration values. If possible we could add a few switches to test the small gauges (oil pressure, temperature and fuel), too.

After the decision for an ABS plastic housing with the dimensions of 115 x 90 x 55 mm a quick check on how much space is actually available for buttons, LEDs and switches (due to wall thickness, switch/button size and other factors) was done. The front plate design was drawn in Inkscape as it creates 1:1 scaled PDF prints which allow to draw everything using their real dimensions.

screenshot.png

A front plate design on its own is not sufficient of course as there is the need for the driver circuits, connecting to the Arduino, etc. To accomplish that task another free program is used: KiCad. It's a little complicated at first (a tutorial is highly recommended) but quite powerful, too. Work starts by creating a circuit diagram and later moving to a PCB layout:

Schaltplan.png

A 5 V relay switches the Arduino's output based on the top switch's position to the corresponding speedo- or tachometer driver circuit. The Arduino reads back the relay state to provide the matching calibration frequency which is selected by means of the push button and displayed on six individual LEDs.

The small gauges (oil, temperature and fuel) require specific resistance values towards ground, nothing more. Unfortunately, these values are rather strange (resulting in 17 pieces overall) and at least the oil and temperature ones require a wattage of >1 W which makes them quite large, too. We are going to have a more detailed look at them later.

<continued on="" next="" thread="" due="" to="" picture="" limit<continued="" reply="" limit="">.. continued on next page due to picture limit</continued>
 
Last edited:
KiCad's next step is the PCB designer. It does not feature an auto-router but manual routing is sufficient with such a relatively low complexity circuit. A raster size of 2.54 mm is chosen (to match the prototype PCB boards used at a later stage). Resolving the rat nests was rather successful resulting in a nice one-layer PCB design:

screenshot_1.png

KiCad offers a nice 3D render feature of the populated circuit board. It's quite motivating to see the potential final result without actually building it up:

screenshot_2.jpg

The two-step-process of creating a circuit diagram that is turned into a layout works out well. A circuit diagram is easy to understand and the corresponding layout easy to implement.
KiCad or similar tools ensure no errors are introduced in the process which reduces the risk compared to a more traditional pen-and-paper approach.

Etching or ordering a board right away would be too risky considering my failure rate so it was decided to realize it by means of a universal prototype PCB.
Not a difficult task, just a lot of counting, wiring and a bit of solder work:

IMAG3736s.jpg

IMAG3735s.jpg

After connecting to an Arduino the programming could be completed (frequency control, button logic, de-bouncing, etc.). A first prototype testing on the uncalibrated cluster looked like this:


Nevertheless there was still significant work to do. Namely drilling the case, installing the LEDs, buttons and switches as well as wiring everything up.
In addition, it was figured out that the previously used Arduino Uno runs on 12 V but it's linear voltage regulator runs way too hot when enabling the relay.
That issue was solved by switching to a Freaduino. It's fully compatible but utilizes a switching power supply that runs much cooler.

Let's continue with the casing. The front plate design is printed on paper, taped to the housing cover and used as a drilling template.

IMAG3792s.jpg

It's printed a second time on transparent Laser printing foil (mirrored), cut-out and taped onto the cover in reverse - protecting the letters from scratches.
After all the parts are installed the foil shows almost no movement which hopefully gives it a long life.

IMAG3796s.jpg

The wiring was done using jumper cables and shrink tubing. It attaches to the pin headers on the board and can be easily separated if required.

IMAG3799s.jpg

Now the user interface can be tested:


The calibration steps would be as such:



  • Selecting the highest frequency
  • Adjust the variable resistor for a match of the dial
  • Select the lowest frequency
  • Adjust the second variable resistor
  • Re-check the highest frequency and potentially adapt
  • Check all intermediate values


Continuing with the resistors for the small gauges. Supplying the cluster with 13.8 V (the dials are somewhat sensitive to supply voltage) they seem to require rather uncommon values:



  • Temperature - L: 130 Ω, M: 29 Ω, H: 20 Ω
  • Oil pressure - 0: 76 Ω, 4: 33 Ω, 8: 13 Ω
  • Fuel - 0: 90 Ω, ½: 30 Ω, 1: 7.5 Ω


Realizing these resulted in overall 17 resistors to be connected. The oil and temperature gauges load the resistors with ~1 W and therefore require larger resistors, in addition.
As there was no obvious way to fit them into the housing it was decided to put them along the cables towards the cluster. It's not beautiful but it works ..

IMAG3813s.jpg

To make the tester's wire harness more flexible a supply voltage and a ground wire was added to each side. These can be used to test other parts of the circuit (like warning lamps).

IMAG3817s.jpg

Unfortunately all of this did not went as smooth as it's written here. It took more than a month to realize it.
The front plate design had to be changed completely after it turned out that the switches couldn't be installed the way it was intended (two cases had to be scrapped).
Numerous versions of the circuit and layout were made fixing various issues and even one that would have killed the Arduino right away.
Three orders were required until all the parts were finally on board.

Investing all this effort for a single cluster is a bit over the top but it was fun nevertheless :smile: .. at least I'm able to replace the capacitors now.
 
Last edited:
Made to Order

As there was so much effort involved in creating the tester I would like to let others benefit from it as well.
I'm open to providing identical or modified copies (e.g. different frequencies, deletion of mini-gauge support, larger case, kmph vs. mph, etc.) to those interested.

Note that it's just a hobby project but I'm nevertheless looking forward to any requests or ideas
smile.png


Bill Of Materials (BOM) - not heavily optimized

QuantityNameDescriptionItem NumberCostOverall Cost
1
Case
ABS plastic housing 115x90x55 mm
4603113.85 €3.85 €
1LED Clips
5 mm, 10 pieces
4425730.55 €0.55 €
1Pin Header
1x 40 pins, gold plated, 10 pieces
4515514.95 €4.95 €
1Relay
5 V print double contact print relay HONGFA HFD3/0053407611.54 €1.54 €
1Resistor SetE12 Resistor set 0,25 W, 100 pieces2214647.25 €7.25 €
1Jumper CablesMale/male jumper cables, 40 pins5111592.95 €2.95 €
1LED5 mm, red, 10 pieces1200230.35 €0.35 €
1Switch TopON/ON4201790.85 €0.85 €
3Switch BottomON/OFF/ON4201800.85 €2.55 €
1PushbuttonPush-to-close4202070.60 €0.60 €
1BNC ConnectorFor panel installation4504230.36 €0.36 €
1Power Connector5.5/2.1 (out of production – needs replacement)4505900.55 €0.55 €
1Prototype PCB160 x 200 mm RM 2.54 mm4402602.05 €2.05 €
1Power socket4 mm, gold plated, black4500450.75 €0.75 €
1Power socket4 mm, gold plated, red4500440.75 €0.75 €
1Resistor47 Ω, 5 W, 5 %2213290.19 €0.19 €
1Resistor180 Ω, 5 W, 5 %2213350.19 €0.19 €
2Resistor120 Ω, 5 W, 5 %2213330.19 €0.38 €
3Resistor10 Ω, 5 W, 5 %2213210.19 €0.57 €
1Resistor39 Ω, 5 W, 5 %2213280.19 €0.19 €
2Resistor68 Ω, 5 W, 5 %2213300.19 €0.38 €
2Resistor8.2 Ω, 5 W, 5 %2213200.19 €0.38 €
1Resistor5.6 Ω, 5 W, 5 %2213180.19 €0.19 €
1Resistor33 Ω, 5 W, 5 %2213270.19 €0.19 €
1Resistor12 Ω, 5 W, 5 %2213220.19 €0.19 €
1Resistor22 Ω, 5 W, 5 %2213250.19 €0.19 €
1Resistor15 Ω, 5 W, 5 %2213230.19 €0.19 €
1Universal Cable0.25 mm², 25 m, white5600482.50 €2.50 €
1ShippingPollin.de for above itemsn/a5.95 €5.95 €
1FreaduinoEF0100115.40 €15.40 €
1ShippingKomputer.den/a3.50 €3.50 €
1TransistorsS8050 5 pieces via ebay.devarious1.20 €1.20 €
1Shippingebay.den/a2.00 €2.00 €
1Velcro TapeSelf adhesive for Arduinovarious5.99 €5.99 €
1Shippingamazon.devarious3.99 €3.99 €
1Zip ties100 x 2.5 mm, black, 100 pieces via MaKaShop24.de10101000025010.49 €0.49 €
1ShippingMaKaShop24.den/a3.90 €3.90 €
Sum:78.05 €

<tbody>
</tbody>


Available as a download from my server:


 
Fantastic post, thanks for adding to the already excellent value of NSX Prime :cool:
 
Are your item numbers for a specific vendor such as Digikey or Mouser or ….? I know that with Digikey, you can create a BOM and make it available to other which simplifies things for anybody who wants to replicate your idea.
 
I'm ordering one: I'm popping for the super extra-wide case so I can homebrew additional functionality. I would like to fit in a power supply inside and explore testing the LED indicators.

Amazing how fast the BOM adds up to real money.

(btw: always ctrl-a [select all], ctrl-c [copy] your response text before posting...that way if 'prime goes bananas you can just ctrl-v [paste] and resubmit)
 
Are your item numbers for a specific vendor such as Digikey or Mouser or ….? I know that with Digikey, you can create a BOM and make it available to other which simplifies things for anybody who wants to replicate your idea.

Digikey' shareable BOM is a super nice feature which I normally use extensively - unfortunately, digikey is also super expensive.
The cost for the tester would probably be several times (4 x or more) expensive if the parts were ordered from them (haven't checked it though).
Due to that, I opted for the German shop www.pollin.de which isn't top quality but very reasonably priced.
 
I'm ordering one: I'm popping for the super extra-wide case so I can homebrew additional functionality. I would like to fit in a power supply inside and explore testing the LED indicators.

Amazing how fast the BOM adds up to real money.

(btw: always ctrl-a [select all], ctrl-c [copy] your response text before posting...that way if 'prime goes bananas you can just ctrl-v [paste] and resubmit)

It's a little scary how the costs starts to rise, indeed ..

The real PCBs have been ordered and are expected to arrive within 1-2 weeks. This should speed up the assembly, costs mostly stay the same.
Pre-manufactured jumper cables need to be replaced with self-crimped ones as they potentially have issues with handling the current for an extended period in one row.

As there was another request via DM: Do you have the 1991 pin-out of the gauge cluster available? (you should, if I remember correctly).
 
Nice job.

If you ever will need an extra rpm meter for auto, let me know.
Still have one in stock and no further need.

I only do the NSX-R gauge cluster conversion, I was not able to find a proper solution to integrate the auto gear display into that rpm meter.

Do you know this tool?
NSX Tacho Messgerät.JPG
 
>Do you know this tool?

Yes, this is a tool used by T3Tech.jp to service NSX clusters. They have gone through a few evolutions of this tool with each one getting more sophisticated and inclusive. The silk-screened face and connectors are just too cool and showing off their quality workmanship. I've emailed them regarding purchasing such a tool but received no response. I believe this tool to have been created by or exclusively for T3Tec. (I do see repair services advertised on auctions.yahoo.co.jp using this tool but my Japanese is through Google Translate.)


I was working on a quick/cheap tool that would work well enough and was collecting all the test specs to build such a device.

Recently [MENTION=30613]Heineken[/MENTION] requested my notes and he was able to implement an elegant solution in a month.



Johan from https://www.sjoebergdesign.com/nsx2s2k was also a great help in getting test specs. Johan is the one that created the S2000 conversion units for the NSX.

[MENTION=5576]Procar Specials[/MENTION] are you selling NSX-R faces?
 
Digikey' shareable BOM is a super nice feature which I normally use extensively - unfortunately, digikey is also super expensive.
The cost for the tester would probably be several times (4 x or more) expensive if the parts were ordered from them (haven't checked it though).
Due to that, I opted for the German shop www.pollin.de which isn't top quality but very reasonably priced.

That is interesting. I have used DigiKey, Mouser, Allied Electronics and BE Electronics (Canadian) and on a mixed order of parts Digikey has always been the lowest cost vendor. Plus, where I am located north of the US border relative to their facility I can order the parts at 4:00 pm and they land on my doorstep the next day for $8 (or free if I binge on $100 of parts).
 
Nice job.

If you ever will need an extra rpm meter for auto, let me know.
Still have one in stock and no further need.

I only do the NSX-R gauge cluster conversion, I was not able to find a proper solution to integrate the auto gear display into that rpm meter.

Do you know this tool?
View attachment 165445


When searching across the internet there seem to be three NSX gauge testers showing up:

Regarding the red one, it's unclear who it belongs to but it's probably not T3TEC as it does not show up on any pages related to T3TEC - seems to be an independent repair shop.

PS: My AT is a 1997 Tiptronic so the dial face is again different from ATs before 1997 ;)
 
Last edited:
That is interesting. I have used DigiKey, Mouser, Allied Electronics and BE Electronics (Canadian) and on a mixed order of parts Digikey has always been the lowest cost vendor. Plus, where I am located north of the US border relative to their facility I can order the parts at 4:00 pm and they land on my doorstep the next day for $8 (or free if I binge on $100 of parts).

Probably related to their storage being in the USA. It made a few spot checks, just to be sure:



Here in Europe it's know to be the shop with most parts - by far - but also quite expensive. They offer free UPS air shipping (all things included) starting at 50 € - that shipping alone is pretty expensive so I'm expecting their margins to be significant when dealing with overseas.
 
When searching across the internet there seem to be three NSX gauge testers showing up:
T3TEC Gauge Simulator from 2017: http://www.nsxcb.co.uk/entry.php?166...ulator&bt=2316
T3TEC Gauge Calibrator from <2017: http://www.nsxcb.co.uk/showthread.ph...142#post115142
Kudo Engineering (?) Gauge Simulator: https://www.kudo-engineering.com/items/21133098


Regarding the red one, it's unclear who it belongs to but it's probably not T3TEC as it does not show up on any pages related to T3TEC - seems to be an independent repair shop.

PS: My AT is a 1997 Tiptronic so the dial face is again different from ATs before 1997


OK, if your AT is 1997 Tiptronic, it wont fit. Lucky man that you have such a young car :smile:

BTW, the tool which my electronic shop uses is kudo engineering.
 
Wow! That is excellent work. Phenomenal write-up! I had recently replaced all the caps in my gauge cluster and wondered how it would turn out. Your outlined device would have helped me for I could have bench tested the cluster before installing it in the car. Fortunately, everything worked out very well though had to validate each function manually and separately which took a fair amount of time. I believe your project will prove to be quite useful for others for the caps will need to be replaced on virtually all NSX gauge clusters in due time.
 
>Fortunately, everything worked out very well

How did you verify the accuracy of the tach and speedo?

Did you use discrete test criteria for the other gauges?

I'm looking to compare to my list and I'm looking for the trigger point for the low fuel LED...
 
drew,

1) For the speedo, I used GPS which is super easy with modern cell phones.
2) Voltmeter...a voltmeter connected to the battery. Before the repair, the dash gauge was reading 10V when the battery was really at 13-14V when the engine was running.
3) For the tach, I borrowed a timing light from an old muscle car mechanic.
4) For the fuel gauge, it was the most challenging for I had to do numerous trips to the gas station to test the level. I know it isn't exact but it worked out and hasn't been off. For the low fuel light, I just let the tank run to near empty and when the light came on, I filled the tank up and noted that the activation level was equivalent to what was written in the manual (can't remember off the top of my head what that level was/is).
 
appreciate the response: We're just trying to nail down all the test criteria.

>I filled the tank up and noted that the activation level was equivalent to what was written in the manual

Excellent, that will work.
 
This is awesome! Replacing the GC caps has been on my todo list for a while, this tool would make it significantly easier.

I may modify the hardware to just take a signal generator input since I've got access to one. It's a less clean solution, but for doing one cluster it simplifies things. I'll post the hardware files if I do so, I'll likely get PCBs made through OSHPark. I've done breadboard work before and it's well worth the $20 to get PCBs made, especially if the board is already designed.
 
Back
Top