Another main EFI relay failure data point

[Old Guy]

Another data point for the list of main EFI relay failures - 2000 model year with approximately 76,000 miles.

I had almost convinced myself that the main relay failure was an early model 'thing' that had been fixed during later production years and that it would not be an issue for me. Clearly, Honda did not have the relay vendor address the design problems. Fortunately, I had followed the advice of other owners and had an aftermarket spare in the trunk. I was doubly fortunate in that the relay failed during a start attempt right in front of our house with ready access to tools. The engine actually started up and ran for perhaps 1-2 seconds before quitting and failing to restart. Because the MIL light was lighting up during the start sequence I knew that the ECU was powering up so the failure was likely somewhere in the fuel pump circuit. That meant the pump control relay in the main EFI relay, the fuel pump resistor, the fuel pump and the pump control output on the ECU. Since the main EFI relay is a known issue it was a logical place to start checking and replacement by substitution provided a ready fix.

For some reason I had it in my head that you could replace the relay by just popping off the panels behind the seats to expose the relay. So, I just needed to have a spare relay on hand in case of a requirement for an easy field repair. Not so on the later cars. The relay got moved and is behind the right rear trim panel. You can pull up the center rear panels by hand; but, you need a phillips screwdriver to get the right rear panel off and you need a 10 mm wrench to remove the relay bracket. On the 2000 there is a big wire bundle tie wrapped to the relay bracket so you need needle nose pliers to release the tie wrap from the bracket or an end cutter to cut the tie wrap. If like me you were thinking having a spare to do a in field repair is good enough, make sure that you have the tools on hand to do the repair. Otherwise, replace preemptively.

As has been noted in other threads on this subject, the relay failure was not caused by a component failure; but, by a soldering failure. The failure was in the fuel pump control relay circuit. Testing indicated that relay was operational and it was the output circuit of the relay that had failed - the one that carries the fuel pump current. Some up close investigation indicated that one of the solder connections for the relay contacts definitely looked dodgy and a check with an ohmmeter indicated an open circuit between the actual relay terminal (in the center) and the surrounding solder pad. You could restore the connection by applying pressure to the side of the center terminal forcing it into contact with the rest of the pad. For reference, the solder joint on the relay coil connection beside it is an example of a good looking solder connection.


The photograph below shows the back side of the relay board.




The blue circle shows the location of the faulty solder pad (after the solder has been removed). The red circle is the location of the fuel pump control relay and the purple circle is the location of the ECU control relay. I initially wanted to remove and resolder all the solder pads for both relays even though none of the other solder pads looked questionable. Removing the solder on the relay coil connections (two small round solder pads on each relay) was easy with my vacuum desoldering station. The relay contact connections were a problem because they are large rectangular terminals and the tip of my solder vacuum would not fit over the terminal. On the failed solder terminal I had to resort to a soldering iron and solder wick to remove the solder. The other side of the relay contact pair are the two rectangular terminals at the bottom of the circled areas. The relay is constructed such that the copper frame of the relay forms the connection for the relay contact terminal. Copper is an excellent heat conductor and my 60W temperature controlled soldering iron could not get the terminals heated up fast enough for the solder wick to do its job without risking melting the plastic housing of the relay. In order to get those terminals de soldered you would need something like a 140 soldering gun to dump a lot of heat into the terminal fast. I have a solder gun; but, it is a clumsy tool so I elected to just try and reflow the solder on those two rectangular pads rather than do a complete replacement. Reflowing was difficult and I might have actually made those connections worse so I would leave them alone. Re soldering the matching contact (the obviously failed one) and the relay coil terminals was easy. Close examination of the other solder pads showed no apparent damage. Because of the potential for vibration induced flexing I was expecting that the solder failures would occur where the terminal plugs are soldered to the board along the left edge of the board; but, visual inspection did not indicate any dodgy looking connections so I elected to leave them alone.

I installed the repaired relay in the car and gave it a test fire and it works fine. The new relay is back in the car and the repaired relay is now in the trunk as a spare. If it goes another 21 years before the next failure its going to be somebody else's problem.

 

[Miner]

Good info.
I repaired my relay several years ago, soldering what I saw as bad joints. A couple of times over the years it has acted up; I probably missed a bad solder joint, based on your observations. One of these days I’ll pull it out and resolder the whole relay. And maybe even get a spare, unless I can find the one from my 89 Legend that suffered from the same problem until I got a new one from the dealer. Since my wife reminds me nearly every day I don’t throw anything away, it should be in my garage!

 

[RYU]

I'm not an EE and certainly don't know my way around a circuit board but I have resoldered a bunch of things in my life and I have a question.

Since it appears the solder break is caused from long term vibration, is it appropriate to submerge the entire circuit board including capacitors etc.. into some kind of spray foam or gel? If not, why? Is it a heat issue?

 

[Old Guy]

Since it appears the solder break is caused from long term vibration, is it appropriate to submerge the entire circuit board including capacitors etc.. into some kind of spray foam or gel? If not, why? Is it a heat issue?

Potting a circuit board is a practice that was not uncommon on boards subject to vibration in the days of through-hole components. In these days surface mount devices are the normal practice and much more resistant to vibration so potting is less common unless somebody is trying to prevent some reverse engineering by looking under the hood. Potting compounds tend to be the heavy viscous gooey stuff like on the NSX crank position sensor or compounds that almost look like polyester or epoxy resin. One of the down sides to potting is that once potted, repair becomes exceedingly difficult / impractical. In the case of the NSX main EFI relay a serious impediment to potting would be the fact that the individual relays are effectively open frame relays. Any potting compound that you put into the box before assembly would get inside the guts of the relay and once set up would prevent movement of the contacts rendering it useless.

I have seen pictures of stress fractures in solder joints and they look like hairline cracks in an overall smooth blob of solder. The solder joint on the connection to the relay does not look like a stress fracture. The relay is very solidly mounted to the board with four large soldered pegs / terminals. None of the other terminals looked like the failed terminal so I don't think component vibration is the cause of my particular failure. To me, it almost looks like the original solder joint was poor or incomplete and that the joint might have failed due to localized high currents - this is the only path in the relay that carries significant current (fuel pump current). There is a phenomenon that can affect solder joints called electro chemical migration of solder which can occur when high currents cause potential differences. In the original high resolution photo you can see little whiskers have formed around the solder joint (barely visible in the compressed photo that I attached). I am thinking that the initial joint was poor / incomplete leading to higher than normal current density in the joint causing the eventual failure.

When I first popped the relay cover off, I immediately started looking at the solder joints where the terminal plugs connect to the circuit board. I was expecting to find solder fractures where the pins are soldered to the board caused by vibration / flexing of the external wire loom. All of those connections were just fine; but, that is not to say that others have not experienced failures at those points.

 

[Old Guy]

Good info.
I repaired my relay several years ago, soldering what I saw as bad joints. A couple of times over the years it has acted up; I probably missed a bad solder joint, based on your observations. One of these days I’ll pull it out and resolder the whole relay. And maybe even get a spare, unless I can find the one from my 89 Legend that suffered from the same problem until I got a new one from the dealer. Since my wife reminds me nearly every day I don’t throw anything away, it should be in my garage!

If you throw it away, a requirement for the disposed of piece will appear withing the next 4 weeks. I am confident of this. I have several stubs of different sizes of ABS sewer pipe, copper pipe, copper house wire and chunks of clear Fir, Oak and one piece of Cherry that I keep 'just in case', not to mention a box of old NSX parts and 1971 Volvo parts.

 

[RYU]

Makes sense! Thanks!

Speaking of fuel pump relays... I inspected my Bosch 40 amp fuel pump relay which I ran on it's own circuit. Even using 14 gauge wire the wire ends of the rubber insulation of the wires were brittle and discolored from the heat. The relay itself looks a bit tired as well. It's rare that I see electrical components subjected to such high load on 14 gauge. I swapped the wire to 12 gauge and changed to a new 40 amp relay. I'm so glad I checked that. I'm running the big boy Toyota Supra MK4 Denso fuel pump.

Potting a circuit board is a practice that was not uncommon on boards subject to vibration in the days of through-hole components. In these days surface mount devices are the normal practice and much more resistant to vibration so potting is less common unless somebody is trying to prevent some reverse engineering by looking under the hood. Potting compounds tend to be the heavy viscous gooey stuff like on the NSX crank position sensor or compounds that almost look like polyester or epoxy resin. One of the down sides to potting is that once potted, repair becomes exceedingly difficult / impractical. In the case of the NSX main EFI relay a serious impediment to potting would be the fact that the individual relays are effectively open frame relays. Any potting compound that you put into the box before assembly would get inside the guts of the relay and once set up would prevent movement of the contacts rendering it useless.

I have seen pictures of stress fractures in solder joints and they look like hairline cracks in an overall smooth blob of solder. The solder joint on the connection to the relay does not look like a stress fracture. The relay is very solidly mounted to the board with four large soldered pegs / terminals. None of the other terminals looked like the failed terminal so I don't think component vibration is the cause of my particular failure. To me, it almost looks like the original solder joint was poor or incomplete and that the joint might have failed due to localized high currents - this is the only path in the relay that carries significant current (fuel pump current). There is a phenomenon that can affect solder joints called electro chemical migration of solder which can occur when high currents cause potential differences. In the original high resolution photo you can see little whiskers have formed around the solder joint (barely visible in the compressed photo that I attached). I am thinking that the initial joint was poor / incomplete leading to higher than normal current density in the joint causing the eventual failure.

When I first popped the relay cover off, I immediately started looking at the solder joints where the terminal plugs connect to the circuit board. I was expecting to find solder fractures where the pins are soldered to the board caused by vibration / flexing of the external wire loom. All of those connections were just fine; but, that is not to say that others have not experienced failures at those points.

 

[Old Guy]

#14 AWG copper has current ratings up to 45 amps; but, the current rating is determined by the wire insulation; not the wire. If you want to run the wire at 45 amps you need Kapton or PTFE or a silicon jacket which is rated for operation at 200 C. A very safe low temperature rating for #14 is 15 amps which is the common limit applied for #14 house wiring (safe as houses!). The equivalent rating for #12 would be 20 amps. If you adhere to those current limits you should not be running temperatures that damage adjacent things that are not rated for extreme temperatures (like your plastic relay housing). Two other factors to consider when evaluating wire gauge. Is the wire bundled in with other wires or is it operating in a high temperature environment? If so then it may be necessary to reduce the rating of the wire because of the loss of cooling. The other factor is the length of the wire which determines voltage drop. If you have a long run of wire you may need to up the size to reduce voltage drop. However, #14 copper has a resistance of 0.0025 ohm/ft at 25 C. If you respect the 15 amp current limit that is a voltage drop of 0.037 volts per foot and even with 10 feet of wire (unlikely on an NSX) the wire related drop would be only 0.37 volts at 15 amps which is likely not material if the alternator is running in the 14.5 volt range.

If you were suffering from apparent heat damage at just the ends of the wire (the center sections of the wire remain non brittle) the problem may have been the terminations rather than wire gauge, particularly if the relay uses a spade style crimp connector. On another vintage car I had an electric radiator fan that started operating intermittently. Investigation showed that the plastic on the base of the fan relay had heat deformed on the relay output connection (high current to the fan motor). In this case I believe the actual problem was the crimp spade terminal. I think the terminal I used was probably low quality and the female terminal was not grabbing the male terminal for adequate current transfer resulting in heating right at the connection. If the female terminal slides on and off the male terminal easily you probably have a poor connection - the terminal on my fan relay definitely pulled right off with ease. I have now come to hate the common crimp terminals with the colored plastic ferrules. On all new work I use correctly sized uninsulated crimp connectors with closed (brazed) barrels. The uninsulated crimp allows you to see that the wire is properly positioned in the crimp and that the actual barrel has correctly deformed to hold the wire. I then use a heat shrink sleeve if the terminal needs to be insulated. A crimp tool with the proper dies sized for the crimp is a must. Open barrel terminals are also a good choice; but, are more work and fiddly because I do the two step crimp process. Good quality crimp terminals are not so easily available. I find the best source is DigiKey, Mouser or Allied Electronics.

 

[Albert86]

I procrastinate as always; I have had this in my stash for a couple of years.
I expect to be crying on the side of the road when my original 94 relay dies, but here is what
a newer replacement looks like:

Seems like the soldering issues have been resolved.

 

[Old Guy]

Your board appears to be identical to mine with the exception that the plug body appears to be fastened to the board by large solder blobs (or a security fastener of some sort) whereas in mine the plug body is fastened by two phillips screws. I thought the conformal coating on my circuit board was pretty poor quality. You can see ridges in the coating in the first photograph. Most of the coating is removed in my second photograph to facilitate de soldering. By comparison to yours the conformal coating on mine looked to be an example of precision assembly. Your board looks like the conformal coating was applied by a 10 year old working on their Jackson Pollock drip / splatter paint technique. Given the quality of the conformal coating I am not convinced that any issues have been resolved.