Dashboard instrument cluster repairs

[casey]

HELP! I have a weird issue with a customer's cluster that I cannot understand. I wonder if anyone on here can explain why this might be happening.

Customer has a 1992 FD and sent me a Knightsport cluster with a reported intermittent tacho. That problem was fixed by me and a few other minor issues I spotted too. I'm unsure whether the cluster came out of the customer's car, or was sourced by him as an upgrade to fit whilst his car is with a mechanic. My final test photo, with all gauges set ~halfway:

Returned to customer, and his mechanic refitted the cluster, but he reported:

Don't seem to have coolant, fuel level or mileage reading and ive checked both fuel level sender and coolant temp sensor

The tacho is now OK, but mechanic also later reported no oil pressure display either. The odo display was intermittent.

Another (OEM) cluster was installed in the customer's car and everything works fine. The Knightsport cluster was put into another car and the report was, "the same issues are there".

I requested checks were made on the earth connections to the cluster as, with so many faults present, there had to be a common cause. I'm unsure if this was done, but the problems persisted.

In trying to understand what is happening, I agreed to look at the cluster again, with the thought that maybe something had occurred in transit back to the customer. It arrived yesterday and I put it on bench test, only to find it passed all my tests 100%, as it did when it left me previously . Here is my latest test photo (coolant light deliberately activated to test it):

I'm completely at a loss to explain why I see a 100% working cluster and the customer reports multiple faults when installed in his 1992 FD and another car. Is there something very odd about a Knightsport cluster in a 1992 car? I've not altered anything on my bench tester to accommodate it, just plugged it in, like I have with ~20 clusters now and it just works.

I'm still thinking a earthing fault, as fuel, collant and pressure gauges all share the same earth, which is separately supplied is the harness. I asked for this to be checked, but unsure if this was done by the mechanic - who I have offered to speak to direct, but he won't call me.....

The only problem with my earthing theory is the tacho works, and it shares the earth with the three gauges - pic of my reference flexi-PCB, not the customer's, marked with the earthing path:

I'd be very grateful for any ideas to explain what is happening here?

 

[JoeLanger]

Looks like the JDM gods have lost interest in your clusters now
You have strayed from the EFR sect

Perhaps try a sacrifice

 

[Sam.rog]

Above my pay grade but is the car running a standard ecu.

Could an aftermarket ecu cause this peculiarity.
Without seeing a full schematic of the cars systems and the knightsport its going to be very difficult to diagnose.

If standard cluster works correctly in the car in question and the KS cluster does not. It does strongly point to an issue with the ks cluster.

Id be looking at the connectors. Are all the pins seated correctly in the cars loom. Any corrosion on the cars loom. Any damage to the connector on the KS cluster
Nothing daft like a burnt out pin not making good contact.
Perhaps try an electrical contact grease to improve conductivity.

 

[casey]

Thanks Sam, your input is very much appreciated.

If standard cluster works correctly in the car in question and the KS cluster does not. It does strongly point to an issue with the ks cluster.

I agree, the information available does tend to suggest that. The mysterious part is the KS cluster works perfectly on my bench test rig, as have about 20 other clusters (after necessary repairs) - both OEM standard and Mazdaspeed variants - so I'm 100% confident my test environment is sound. The KS cluster appears to be identical** to the OEM/MS clusters, and I have not had to change or adapt anything on my test rig to accommodate it, just plug in the 4 connectors, as I do with all others.

**The only obvious difference is the speedo ranges to 300km/h and has been calibrated accordingly.

I'm thinking a poor earth somewhere. Either a pin onto the cluster or, in the car - maybe a poor earth to the chassis/engine, which the sensors will rely on. All the sensors operate with a very low resistance range, so even a minor issue with earthing will be a problem. But why would a OEM cluster work and not the KS one

There was some corrosion around the bolt heads making electrical contact to gauges and PCB's on the KS cluster. I have replaced or cleaned the bolts and used a fibreglass pen to remove corrosion deposits from the copper track contact areas.

I've zero experience with after market ECU's, but this car does have one fitted so, conceivably, could be an issue. But, as the fundamental problem is zero readings on the basic gauges - coolant, fuel and oil gauges, I think much more likely a power or earth issue to the cluster is the problem. But, as far as i can see, the KS cluster has identical gauge power wiring to OEM, as evidenced by it's successful operation on bench testing.

Unrelated to the fundamental "dead" oil/fuel/coolant gauges, I have tested this KS cluster extensively over the past week and I did spot a weird "interaction" appear yesterda, between the tacho rpm and speedo reading. Increasing the rpm led to increasing instability in the speedo needle. It would wildly fluctuate, unless rpm was less than ~2000rpm. At 9,000rpm the speedo would tend to drop to 0kph, regardless of the simulated VSS speed input.

A 'scope showed the normally smooth, near sine wave, VSS signal was being "broken into" by the tacho pulse stream. Yellow trace is the VSS signal and blue is the tacho:

A removed the speedo and noted early signs of capacitor leakage:

The customer agreed that I replace all 11 caps. I have replaced 10 so far (one out of stock, on its way to me) and then retested.

Problem solved! The VSS waveform is now smooth with no tacho/speedo interaction. Ignore the non-perfect sine wave in the yellow trace below, it doesn't need to be perfect for test purposes as the Mazda circuitry is only interested in zero voltage crossing and sufficient amplitude to input to a Schmitt trigger IC , which converts it to a 5 volt logic level square wave signal.

Another PCB with failing caps, caught very early in this case.