If your CPU board hasn't been re-capped, it could be another cause of your intermittent tacho.

 

You mean replace the electrolytic capacitors(I changed 2 on the tacho pcb- 1 x electrolytic-C1 & 1 x polyster-C5)? The CPU board has very little circuitry for the tacho, mainly TR8 & IC4(as well as some resistors & caps), with pins 1,2,3 & 4 associated with the tach signal.

Are you just recommending to replace all electrolytic capacitors in general, not necessarily for the tacho?

 

Every single CPU board I've seen has had at least one badly leaking electrolytic (the ceramic, polyester, etc caps are not a problem). The leakage contents have caused light to severe corrosion damage to other components, so well work changing them all out if you have the board out to work on. C3 is usually the worst, the 1000uf power smoothing cap. I've had 2 zener diodes fail too, I suspect due to corrosion issues.

You are correct, very few of the electrolytic caps are directly asociated with the tacho signal processing circuitry, but many are power smoothing caps, so can potentially cause issues with power fluctuations, which won't end well and could create intermittent/unpredictable behaviour of some circuitry.

A new versus a bad cap taken from a recent repair job (C11, so not related to tacho in any way). But you can see the damage the leak has caused!!

 

Good progress on a small daughterboard to replace the failed (Panasonc? Obsolete, IC6) AN8363UBK air-core gauge driver chips.

IC6 came off cleanly using a professional vacuum desoldering tool, enabling 9 wired connections from the main board to the daughterboard (GND, 12v, signal in, X1, X2, Y1, Y2 and VR[2 off]) Wired using a 5 way plus a 4 way mini connector for prototyping purposes.

The board contains just 10 (realatively cheap) components and uses the existing onboard "VR" for speedo calibration. The other variable resistor on the daughterboard is to set the needle zero mph position.

I have a revised board layout already planned, which should enable it to be secured to the main board using existing fixing points.

 

Another possible cause for your tacho issue is a dry/cracked solder joint on the tacho PCB. Worth a close visual inspection and re-flowing any suspect joints.

 

Already did that months ago(along with changing C1 electrolytic cap) & thought I had it resolved as it would work 99% of the time(with occasional drop to 0rpm for a sec & back to normal), then removing cluster to trouble-shoot another issue(low coolant alarm) it reverted back to original intermittent(working when warm only).

Anyway, spent a few hours today setting up the signal generator & could not get a good signal on the collector of my transistor on my set-up(amplitude was too low...believe you need at least 750mV p-p based on this thread(https://www.rx7club.com/3rd-generati...tacho-1168633/ ..thanks RGF for the info) & was getting nowhere for a long time(tacho needle wouldn't budge on any 1-300HZ range) & just when I was about to give up somehow I got the needle to move & found that movement around the motor generated a nice clean PWM signal at the collector. This I believe is where my intermittent issue has been all along(although not sure how it even worked at all when TR8(from speedo board) has been totally open cct, as mentioned in my previous posts).

Initially I thought it was a dodgy VR1(tacho) pot as applying pressure would cause a loss of my PWM signal on the collector of the switching transistor, however measured pot whilst tapping & it resistance was fine. Managed to stop it from failing by putting pressure on the edge of the pcb(right where VR1 pot resides), so i could have just wedged something & hope its fixed, but I have desoldered the motor, applied some solder to the windings down below then resoldered back to the PCB. So far so good & any movement around VR1 no longer drops the needle to 0rpm(as well as the signal on the collector.

One question though....not sure if my motor is sticky, as when I drop power the needle either slowly drops to zero or if at above the middle(6000 rpm or higher, either stays there or flops to 9000rpm). Applying power & a signal will drive t back according to the frequency o/p to the transistor. Having compared it to the speedo needle, it is very springy & will always swing back to zero(with no power).....it that a concern?

Bench set-up

rear of Tacho pressure point

 

Well done, excellent work there! The reason for a poor signal level at tge collector of TR8 is that, if measured on a disconnected speedo PCB, it is open circuit. It relies on being connected to the tacho, to provide a pull-up resistance to the collector and a proper output level.

Your strange needle movement might be due the back, black, part of the needle "catching" on the black gauge face. I had that issue with a tacho I looked at 2 weeks ago. A little, VERY careful tweak of the needle - pulling it out slightly, cleared the problem. The needle is VERY fragile!!

 

My odometer reads way to fast for the speed I'm driving. If I drive 55 speedo reads 85 or so. Need to figure this out.

 

There is a variable resistor (VR) on the speedo PCB to set the speedo calibration. If you have access to a sine wave signal generator you can calibrate on the bench. If not, you'll need to drill a small hole in the side of the cluster casing, to access the VR with a screwdriver, with cluster cable tied in position in the car, to calibrate on the move.

 

My success was short lived, started playing up again on the bench this morning. Have literally continuity checked the whole tacho board circuit & all seems to be ok. It is very reproducable(push/pull on edge where Vr1 is comes good) but got me beat where/why its breaking down. As a last resort I resoldered all the solder pads where it goes through top to bottom of the pcb traces, success for now to the point I can't reproduce the fault so jury is still out on this.

Good work btw Casey on the AN8363UBK air-core gauge driver chip replacement daughter board....keep the FD clusters operational for many years to come hopefully.

 

I have a dead tachometer that I'm trying to fix. I didn't see too many issues with the solder joints, but I touched up a few. Components also look ok. Soldermask on the speedo board did look like it was peeling, but didn't think it would affect performance. All other gauges seem to work fine. Any thoughts?

 

Firstly, you need to feed a pulse signal into the tacho input to see if it works, range ~10 - 300Hz. That will tell you if the tacho is working in isolation and whether it's the tacho at fault or the output from the CPU board, TR8 open collector IIRC. At this point, and based on my experience, it could be either. If you are OK with electronics, this very simple circuit below will provide the pulses needed to check the tacho. From memory, I changed the 5k variable resistor to 50k and the 47uF to about 1.8 - 2.2uf to get the right frequency output and about a 2mS pulse width. The resistor and LED on the output is not required. Power this circuit and the tacho with 9-12 volts. Component cost less just a few £UK. the tacho should sweep with movement of the VR. Can be build on breadboard, no soldering needed.

Youtube ink to this circuit in use (no sound on video):

 

@casey_uk I was able to get the tacho by itself working, but it was twitchy. It wouldn't hold the RPM steady occasionally and sometimes dropped to 0, but I think that's because my function generator was only able to max 7.7vp-p. I couldn't see your division markers, but I'm guessing your pulse was at 12vp-p? Have you tried going lower to see if that could work? Or do you think there is something going on with the board?

 

You are on the easy path if you have a signal generator but more Voltage won't help. You need to be able to sink current. Check out https://www.rx7club.com/3rd-generati...t#post12619951 and you can drive it with a +1V pulse from a signal generator.
Intermittent operation - the variable resistor should be a prime suspect: https://www.rx7club.com/3rd-generati...t#post12619950.

 

I'm pretty confident that if you power the gauge with 9-12 volts, then 7.7v peak pulse input will be sufficient. The signal output from the CPU board is from a 5 volt powered chip, but into an open collector transistor output before delivery to tge tacho - the tacho circuitry provides a pull-up path for the transistor. Note, the pulses must be ground to 7.7v, not cross through zero volts. My sig gen will output a square wave crossing through zero by default, but allows an outset dc voltage to be applied to shift the zero volt level.

Let me know if this helps. If not, give me as much detail as you can and I will replicate on my test bench.

 

Last night I measured continuity between the pads on the flex circuit and the solder joints on the ribbon cable connector. I was not getting very consistent results when checking 12v, gnd, and tacho. I will try to clean up the ribbon cable connection some, but should be straight copper so should produce very low resistance. Was fluctuating between 5 adn 50ohms. I could wiggle the ribbon cable to get better and worse results.

Current setup to test just the tacho is to run 12v (more like 13.5v) from a battery to power the circuit/motor and a function generator with gnd tied to dc gnd so should be 7.7v square wave. It's all alligator or spring clips. I'll double check connections and run the test again tonight to see if it fluctuates like last time.

 

We don't have specs on the air-core driver ICs and therefore their max operating voltage. 13.5 volts ought to be OK, but the circuitry on the CPU board that provides power to the tacho regulates the voltage to slightly less than 12 volts (I measured it).

Poor continuity at the ribbon connector would explain your symptoms. If the copper fingers look clean and the connector is tight with the retainer pushed in, then inspect the flexible board very carefully for any damage.

 

@casey_uk will do! Thanks for all the support in this endeavor!

 

Just reporting that I had another FD owner send me their cluster to look at last week. The tacho was unreliable, only working about 20% of the time. On testing, the tacho needle simply wanted to sweep hard anti-clockwise against the zero stop, reading nothing on the gauge when a 10-300Hz signal applied.

Tacho removed from cluster. Same result when signal applied direct to to the tacho gauge, confirming it was not the processing of the signal through the speedo board at fault.

Gauge needle carefully removed. Always a hairy moment as there is a danger of the shaft pulling out of the gauge air-core motor, which cannot be repaired! This one was tight, but with the correct technique, came off fine.

Gauge face removed (lift and 90 degree twist) and two screws removed, so the PCB could be accessed.

A hot air gun set to 100 centigrade was applied for about 10 seconds around the components on the PCB and the tacho sprang into normal operation. Conclusion: either a dry joint or an internal IC1 bond wire failure.

Waiting for the PCB to cool (and fail again), I used an oscilloscope to check that the input signal was getting to a pin on IC1, which it was. There is no known datasheet for the obsolete IC1, but knowing how these class of IC's operate, I was expecting to see a processed "ramp" signal, the same as the input frequency, on another pin on IC1. There was none BUT, with heat applied, the ramp signal appeared on another pin! Conclusion, internal IC1 fault, probably a bond wire detachment.

Nothing to be lost now in reflowing all 22 pins on IC1. There is just a chance the heat transmitted through the pins to the bond wires, can "repair" the fault.

Wow, it worked! Picture of cluster on test rig where I soaked tested the tacho at 6k rpm for a few hours.


Cluster on test


Customer put the cluster back into his car yesterday and reported the tacho is working fine

No knowing if this will be a permanent repair, but at least it's working for now. I know others on here have gone through the same solder reflow process to restore working tacho's, with a proportion failing again after a few months. Others have had no recurrence. fingers crossed for this one!

 

Greetings from Sweden! This is my first post, bare with me (I have been lurking...).

So I've been trying to fix my brothers FD speedo and tach. I managed to get the tachometer working by replacing the 2 caps on it and reflowing all the solder. There was clear cracks in most of the joints. This seem to have fixed it. There was no signs of any leaking or blown caps.


Visible cracks

But for the speedo, I found no obvious problems with the joints. I re-flowed all the solder anyway (except the IC's), but it did not fix anything. The ODO/display is working (don't know if it adds milage, probably not?) but the speedometer shows 0. When he bought the car the speedo worked intermittently, but later stopped working completely. I have inspected the board, and all the caps and everything looks fine to me. At least not as bad as many of the boards I see here on the forums.

Some pics of the speedo board.





PCB backside







Could it be the speed sensor in the gearbox? How can I verify the signal?
Or should I just replace some caps and hope for the best? Is there a spec list somewhere of all the caps? They are really hard to read...

I should add that this is a JDM -95 FD 😊

EDIT:
Based on my findings, replacing the C3 might be a good start? But I tried to map them to a local distributor in a spreadsheet.
I'm a bit confused about C9. Gemini and ChatGPT suggests I can replace them with a polyester/film cap, and that the higher voltage is OK (for any of the caps). Thoughts?

Screenshot, should the link not work.

 

That is a very clean looking board, with no obvious signs of bulging caps or leaking electrolyte. Personally, I think it would be a waste of time to replace the caps at this point. I very much doubt it is the cause of the speedo failure. About 50% of the speedo failures I have seen is due to failure of IC6, the chip that drives the 2 coils in the "motor" that drives the speedo needle. In all cases, application of 100 C narrow stream of air from a heat gun is enough to bring the speedo back to life after about 10 seconds. After heat source removal, the speedo will work for maybe a minute, then will return to the failed state. Sadly, IC6 is obsolete. Similar chips are available, but would require removal of IC6 and a small PCB daughter-board installed in its place to provide appropriate drive voltages to the "motor" (an air-core mechanism with 2 colis at 90 degrees to each other).

See this topic for my journey into understanding these clusters and my efforts to repair them. Page 6 shows a repair to a speedo using a prototype daughter-board https://mazdarotaryclub.com/forum/in...topic=85243.75

 

Thanks, but a bit disappointing What about the VSS, are they prone to failing? As I said earlier, the speedo was working sporadically for a short period of time (on and off) at the start. But I guess that might speak more for the IC6 than the VSS...
I guess this might be a long shot from Alibaba.

Are you using another type of IC on your daughter-board? I assume this could be done using something like a ATtiny1616 and software? I'm no electrical engineer though and for sure don't know how the IC6 (AN8363UBK) work...

 

Sporadic operation of either the speedo and/or the tacho is a classic symptom of failing driver IC's, although dry joints are just another possibility.

I'd put a bet on those Alibaba chips being a waste of money. The picture has an incorrect number of pins - in fact, the pin pitch on IC6 is also narrower than standard (very unusual). I've also not been able to find a datasheet for IC6, suggesting it was a custom part made for Mazda.

I have never come across a failure of the VSS signal but, from an engineering perspective, it has to be an outside possibility.

 

We have sourced a Mazda 626 -95 cluster for about $20, lets see if it has a salvageable AN8363UBK / IC6 or not. Regarding the tacho I guess it might not be 100% confirmed working yet, as we have only tested it in the car for about 1-2 minutes. But at least it went from not working to working, so we must have done something right.

 

It did not

​​​​​​
I have reflowed the old solder on almost every joint on the PCB. I might have missed some. But I did not reflow any IC joints. I did however check the continuity of the IC6 pins to the PCB. Not sure what to do next, but reflowing the IC6 joints and replacing some caps seems like my only option right now, and hope for the best.
What are your plans for the prototype IC6 daughter board?