Trip Meter Resetting

[Bill R.]

the trip odometer is stored in volatile memory and the main odometer is stored in 2 places in an eeprom. The volatile memory is sensitive to voltage changes and that the voltage drop while cranking is enough to wipe it out. The caps stored enough power to ensure that these momentary drops didn't affect the volatile memory, that the caps took the place of the nicad batteries that were originally used for the si board on the earlier cars.
I don't claim to be an EE and this is what i was told in the past, made sense to me.

The caps don't act as batteries. The trip odo is stored in solid state memory and requires battery power to retain. Disconnect the battery and the memory is dumped. The caps in question are used in a switching power supply that powers various parts of the cluster. When they go bad the power gets very 'dirty' and the circuitry that uses it may or may not work correctly.

Just a couple of months ago I replaced every electrolytic cap in the cluster in my '89. Not only did all the cluster gremlins go away, but .... seriously, I'm not making this up and my wife can verify it ... the radio reception got a WHOLE lot better. The only explanation I have for this is that the cluster must have been emitting a lot of hash due to the bad caps and noise on the long traces (think antenna) on the board, and that was interfering with AM reception.

When you have the cluster out you should replace the high-usage lamps, like the odo, check-control, and auto-trans annunciator if you have one of those. In my cluster the odo lamp was so degraded with tungsten deposits on the glass that it was barely readable. The difference with a new lamp was amazing. BTW, not all those lamps are created equal. I bought a bunch of them and screened them by cold resistance. I put the lowest-resistance one in the odo. It's nice and bright now.

[SRR2]

Your explanation is close but not quite right. The problem is that when these caps go bad the switching regulator can't work correctly. The regulator itself becomes more sensitive to incoming battery voltage. Thus, when the regulator is working correctly it can supply the required 5V to the RAM with incoming voltage as low as ~8V or so. When it's not working right due to the bad caps its ability to deal with reduced voltage is essentially nonexistent. While it's true that caps in a sense can act as 'batteries', in this case all they're being asked to do is to be a battery for a few tens of microseconds. The design of the cluster never anticipated that they'd have to hold up the RAM for more than that.

NVRAM took the place of the NiCds in the old SI board.

And I do have an EE degree, btw.