Dashboard instrument cluster repairs

[Paul-Knightsport]

Another cluster repaired and packaged up ready to ship back to it's owner.

Another unique fault. The primary issue was the odo display would be backlit, but blank for several minutes (or more) before any mileage data appeared. Strangely, if the cluster was switched off, then immediately back on, the data would appear after a much shorter delay. This pointed to maybe a failed capacitor not charging/discharging as it should. As we know, leaking/failed electrolytic capacitors are often an issue with FD clusters, so seemed a reasonable hypothesis. So, 'scope and multimeter probes in hand, together with a sketched circuit diagram I have painfully derived from the PCB tracks, the challenge was on to find the cause of the issue.

The first clue was that the reset pin on the NEC microprocessor chip (which processes speed data to calculate mileage) was being held low (zero volts, so switching it off) and when the pin went high (5 volts), the processor would spring to life and the data would then appear on the odo. By why/how was the reset pin being held low for so long? The circuit diagram revealed 7 components involved in producing the reset signal. One was an electrolytic capacitor. A 'scope trace showed this to be charging very, very slowly on power up. When charged enough, it operated a Schmitt trigger to send a clean high (5v) signal to the processor to start processing data for the display chip. So, cap changed, but no improvement! Next, what components were involved in charging the cap? A few resistors and a diode array (essentially 2 diodes in a TO-92 "transistor plastic package). Resistors rarely fail, and when they do, there are usually visual signs, so the array, DA2, was a potential problem. In circuit testing showed one of the diodes was suspect. I had a used, known good, spare in stock, so changed it out and that fixed the fault.

There is now the usual slight delay as the processor starts to operate, as seen in this video of the CPU board on my test box. The processor is held off deliberately for a short period at "ignition on" to allow the voltage to stabilise.

View attachment 43770


Here is the section of the PCB after changing C6 and DA2. Note the main solder (and critical) contacts are on the rear of the PCB. The front pads are simply plated through holes.

View attachment 43766

Note the early signs of black corrosion, caused by the large cap (C3) in the photo starting to leak it's electrolyte. The customer agreed I change this cap out and clean up the corrosion. In doing this, I noted ZD3 (a zener diode) looked pretty badly corroded!

View attachment 43767

So that was changed as a precaution. I have seen this component fail on two previous repair jobs!

That's much better, new cap and diode in place, with electrolyte corrosion cleaned.

View attachment 43768

Back side soldering of DA2 and C6....


View attachment 43769

There were a few other, minor, items to fix (blown warning lamp, broken trip button...) but all good on final test:

View attachment 43771

Loving the "Test Jig" 10/10

 

[Jester*]

I also love seeing nice clean soldering without burning the PCB surround.

Spent a huge amount of time during my apprenticeship trying to master that.

 

[casey]

I also love seeing nice clean soldering without burning the PCB surround.

Spent a huge amount of time during my apprenticeship trying to master that.

Thank you! I have used a soldering iron for well over 50 years. I can't remember how many soldering irons I've had in that time. They are much better now too.

 

[casey]

Also now taking a look at how to repair F100 CPU#2 modules click here

 

[casey]

Another repaired cluster is winging its way back to its owner.

Another weird fault. One that I have heard of, but not seen before. Tacho needle would rise to 3-4k rpm and move around randomly, with the ignition on but engine not started. When engine started, the owner reported it floated around, often 3-4k rpm higher than expected, for about 30 minutes, then worked normally.

On my test rig, on powering up with no rpm input to the tacho, the needle did rise to 3-4k rpm and fluctuate, but for only about 20 seconds, then worked normally.

On removing the tacho and testing in isolation, it was difficult to reproduce the anomaly, but hitting the tacho PCB with a heat gun and freezer spray do make the needle jump around. The behaviour was not the same as a failed IC reacting to changes in temperature, more like a poor solder joint.

Next step was to dismantle the tacho so I could visually examine the PCB. I was told the cluster had been sent to North America a couple of years ago for a tacho repair and this was evident as I could see many solder joints on the PCB, including all the IC pins, had previously been reflowed. However, under extreme magnification, several of the solder joints, particularly this one, looked very dubious - the solder "blob" seems to be sitting on top of the PCB pad, rather than flowing onto it. A few others looked similarly dubious:

I decided to apply plenty of flux and reflow all the IC pins, applying some fresh solder, then clean up and re-examine the joints. Happy with the results, I reassembled the tacho and retested out of the cluster. Multiple, extreme blasts with a freezer spray and heat gun was met with a rock solid needle.

I then put the tacho back in the cluster and it behaved normally. Left on for several hours with no glitches. Left to cool overnight and tested the next day and all looked 100%.

I also noted the trip to North America had resulted in a complete replacement of all the Speedo PCB electrolytic capacitors. Although I didn't remove the Speedo board, so I couldn't examine the soldering, it did look a very neat job from the topside, plus high quality, German manufactured "Wurth" capacitors had been used (red cylinders in pic).

 

[casey]

Awesome progress on my PCB design to replace failed Speedo needle driver IC's - a common failure, resulting in a "wandering" needle.

1st PCB assembled:

TBH, I could shrink the size of this PCB quite a lot, but it fits in the space and makes assembly easier.

Before committing to PCB manufacture, I'd taken quite a lot of time checking that the board would fit the space and could be attached securely and was relieved that it fitted and attached perfectly.

The blue trimmer potentiometer you can see adjusts the zero needle position and the existing trimmer on the Mazda PCB is used (as it is in the original design) to calibrate the speedo. On test, it works perfectly, with excellent accuracy up to 120mph, then tends to read slightly lower than actual at 120mph+. I don't think that's a practical issue, even at insane track speeds, who'd be looking at the speedo! I know I'd be focussed on the track and the tacho at those speeds

 

[casey]

Another tacho with the "wandering needle" syndrome. This one requires the PCB to be replaced, due to failing driver IC:

 

[MightyCondor]

I don't think that's a practical issue, even at insane track speeds, who'd be looking at the speedo! I know I'd be focussed on the track and the tacho at those speeds

Never been on the autobahn then?

 

[casey]

Another successful repair, but a bizarre one!

Cluster arrived with dead tacho and dead speedo, no needle movement on either.

Dismantled the cluster and removed the tacho. Perfect on test out of the cluster.

Removed the speedo to find this power transistor lying loose inside the casing behind the speedo

The transistor tested OK, and it's function is to provide regulated power to both the speedo and tacho, which explained why neither were functioning.

This is where the transistor should be, bottom side of PCB:

Top side f PCB - TR4 missing!!

No broken pins in there, so it looks as though the transistor had been desoldered at some point

Cleaned out the plated through holes - always a delicate task to avoid damaging the PCB solder pads, but successful.

Transistor soldered back into place - a bit of a challenge as the speedo gauge face and other components were restricting access:

Tacho and speedo back in cluster and tested fine - good calibration.

Finally, the customer asked for the speedo to be calibrated to show MPH, rather the KMH, so this was done.

 

[MKZ7212]

Amazing work FD community would be lost without people like you

 

[casey]

Another successful cluster repair. This one had the classic multiple electrolytic capacitor failures, with leaked electrolyte all over the Speedo PCB. The customer had already replaced a few caps, but cluster function was still intermittent.

Here is an example of the electrolyte that has run down the PCB and was in the process of corroding a zener diode, ZD3. This diode was replaced as a precaution, allowing cleanup around that area.

Cleaning the board took a while, but these brushes and IP Alcohol did a good job:

In the photo above, TR4 had been removed as it had failed. A replacement was soldered into place.

That fixed the intermittent operation. Final test passed OK.

 

[casey]

Busy week! Another cluster repaired AND converted from kph to mph too.

This one had an blank ODO display;

The polarising filter on the display was missing (that's why the display is so bright), so I was hopeful that replacing the filter was the issue. But no, the ODO was still blank.

This took quite a while to locate 3 faulty components:

A leaking 1000uF (C3) electrolytic capacitor. Easily spotted as it was leaking electrolyte and bulging too.

A failed 4.19MHz ceramic resonator, that provide a a master clock signal to the 4-bit microprocessor that calculates milage and converts the data to a serial data stream to an OKI display chip to drive the display.

A failed 3.9volt zener diode (ZD5). This provides a reference voltage involved in converting the high voltage VSS ( Vehicle Speed Sensor) sinusoidal signal from the engine into a logic level, 5 volt digital pulse stream.

The board was then thoroughly cleaned to remove the corrosive electrolyte. Hare is what the the electrolyte can do:

All that yellow/brown discolouration is caused by leaking electrolyte from failed capacitors:

That brought the ODO display back to life:

Last step wa conversion form kph to mph.

Before photo above. Here is after the conversion, the ODO displaying the converted total distance covered:

A

 

[casey]

BTW, this is my small, but well-equipped workshop for carrying out these repairs:

 

[casey]

And.....the last cluster repair for this week. Another blank ODO display.

See how clean is this PCB with no leaking electrolyte:

It took a while to find the fault as I have to work back quite a way from the ODO display, through the PCB circuitry to narrow down the likely fault to a capacitor, C6.

I started with the display, which was receiving data and signals correctly from its driver chip, but the data was basically empty cells, hence the blank display.

That led to the previous component in the line, a 64 pin 4-bit microprocessor (IC5) that takes a processed VSS (vehicle Speed Sensor) signal and calculates mileage travelled, converts this to a serial stream of data which is sent to the display chip. this "stock" image of PCB, shows
IC5, circled in red and the location of the faulty capacitor (C6), identified by ciruit analysis, circled in orange:

C6 was replaced and the display sprang into life:

 

[casey]

I've just shipped back a Mazdaspeed cluster after some checks and minor repairs. The customer reported a dead speedo, but it proved to be fine on test, pointing to a fault with the VSS signal getting to the cluster.

Photo of cluster on test with instruments set to ~50% readings:

Within a few hours, another Mazdaspeed cluster arrived for repair, again with report of an unreliable speedo. Initial testing showed the speedo to be 100% after 40 mins of testing. More testing to do next week though.

As an aside, here is a small selection of spare electronic components and modules that I have invested in, to ensure fast turnaround of cluster repairs. All electrolytic capacitors are manufactured in Japan, USA or Germany - no Chinese caps! Most components come from DigiKey in the USA, a great source for reliable components.

 

[casey]

Duplicate of post above, deleted.