Fuel Pump Relay 9434225 Examinations 1998 v70 non turbo

[MrAl]

The right way to diagnose these is with the jumper wire as noted above. The only failure mode this would not capture is the absence of the ECU enable waveform but that failure is so rare it’s not worthy of mention other than passing.

Hi Steve,

Yes that would tell us if the relay was the most likely suspect in the failure.

What i was looking for was how the ECU would tell the relay circuit to turn on or off so i could better understand the circuit of the relay itself. That may be purely academic but may help with repairing the relay instead of buying a new one. After putting out $2800 for a repair i wanted to know more.
Secondly, i wanted to know that in case the ECU is telling the relay to turn off (and thus the pump to turn off) i would look into why it would be doing that which may lead to a different sensor. The crank sensor came to mind, if this car 1998 v70 has that of course.

I also thought of another problem im not sure of but i'll start another thread for that it's pretty much different.

Thanks.

[850oldschool]

The right way to diagnose these is with the jumper wire as noted above. The only failure mode this would not capture is the absence of the ECU enable waveform but that failure is so rare it’s not worthy of mention other than passing.

This is the tried and true method for diagnosing failure after it happens. What I'm proposing is that there may be a few simple tests which can spot a deteriorating relay or pump before it fails completely and leaves you stranded.

[MrAl]

The right way to diagnose these is with the jumper wire as noted above. The only failure mode this would not capture is the absence of the ECU enable waveform but that failure is so rare it’s not worthy of mention other than passing.

This is the tried and true method for diagnosing failure after it happens. What I'm proposing is that there may be a few simple tests which can spot a deteriorating relay or pump before it fails completely and leaves you stranded.

Hi there,

I have been thinking about this too. After looking over the board and schematic, there are only a few parts that can wear.

The first of course are the capacitors. The electrolytics are always in question in devices that are over 10 years old. The Symptoms could be buzzing or something intermittent like i am seeing too now.
The test for this would be to have a good working relay and use a signal generator on the ECU line. With 12v applied (and ground) to the proper terminals on the relay (15 is +12 and 31 is ground) set the signal generator for 40Hz. The relay should switch so you see the contacts move. The top part moves down just about 1/16 of an inch so it's only a small movement.
Next, lower the frequency of the generator until the relay switches off and note the frequency.
Now to test the relay in the future, if you do the same and lower the frequency the relay should not turn off at a higher frequency than what you found when it was working good.
What this tests mostly is the 22uf capacitor and partly the 100uf capacitor.
To test the relay coil to make sure the relay inside can turn on and off, you would just apply and remove the ECU frequency and see that the relay inside turns on and off. You would do that several times and note that it always works.
In addition, it would help to load the contacts by placing a load on the relay contacts 87 to ground 31. I am not sure of what the resistance should be yet though, but i would be surprised if it was less than 1 ampere. That would require a 12 Ohm power resistor of wattage about 20 watts. If that's too much for you to obtain then try a 24 Ohm resistor rated for 10 watts. You have to be a little careful because the resistor would get hot and could melt plastic and of course keep all solvents away.

I should be able to do this test myself maybe today and i can report results back here.

So just to recap, the main components that can 'wear' are the internal relay (the one inside with the coil) and the electrolytic caps. There is a chance the 1uf ceramic can go too (top of board, looks like a paint drop with two leads) but that would be a bit rare. The electrolytics on the other hand are notorious for going bad especially when there is significant ripple current. Interestingly, during normal operation the 22uf cap would be charged and discharged to some degree about 40 times per second. Over 10 years that is a lot. I will be doing a comprehensive test on my 22uf cap too once i remove it from the board just to see if it was that one that became faulty. I will also test the 100uf cap though as that is electrolytic also. I had to wait for a solder sucker to arrive in the mail as my old one got displaced.

The only tools required would be:
Frequency generator that can put out 40Hz and be turned to a lower frequency possibly all the way down to 1Hz or lower. This can also just be a 555 timer IC set up to put out a near square wave. It would be best to have an open collector output i think but i will verify that also. 555 timers are like $1 USD and a couple caps and resistors and a potentiometer and you have a frequency generator that would be good enough. You can buy one on Amazon or other site too they are not expensive unless you want a really good one for other things too. You can get a schematic for such a thing on the web they are abundant.
You dont really need a scope, just watch that internal relay as you do the frequency sweep.

I'll provide more information as i find out more after a few more tests.

[MrAl]

IMPORTANT UPDATE

I did some testing today and found that one of the diodes is actually a zener diode. It's far right top in the schematic but i am submitting a new schematic showing the zener.

This is important because they use that zener to regulate the positive voltage supply for the chip and thus the timing. That means that the ECU signal should not go to +12v but can be much lower, +5v or even a little less. I used a TTL signal to emulate the ECU.
Ideally, the ECU should be open collector which means if using a wave generator you can use a diode in series with the generator so it can only pull down the ECU signal line and not pull it up, but with a TTL signal (roughly 0v to 3.5v or to 5v) it doesnt seem to matter.

I found an alternate to using a frequency generator if you care to test just a tiny bit less comprehensively.
That is, with the ECU lead (terminal 31b) with an alligator lead clipped to it and another lead going to ground (terminal 31) you can quickly touch and remover the connection by touching the two open ends of the jumper leads together. You have to quickly tap them together several times and you see the relay close. The frequency will be too low though so the relay will not stay closed even if you keep tapping. With a frequency generator set to 40Hz the relay stays closed.

A test point for my unit was to see what frequency the relay would turn off. For mine it was around 20Hz. So it turns on at 40Hz and as you lower the frequency it stays on but when you reach 20Hz it turns off. It also stays on up to 60Hz but i did not go any higher.
The most significant is the 20Hz threshold. Since this relay works at least sometimes that may be the correct test point for any relay. I'll know more about that soon, but it seems like a good test point. If in normal operation the ECU signal drops lower than 20Hz the relay would turn off, which seems ok although i have to data from Volvo on this test point.

Oh another thing, we dont have to power the unit (terminal 15) with +12v even +10v worked ok. That's a good sign for when the car engine is being started and the battery voltage drops a bit.

I think the above says a lot about this relay and circuit. That's probably enough of a test, but unfortunately intermittent problems are harder to test for. What i might do next is look at the capacitor waveforms and see if that tells me anything else about checking for intermittent operation. The test above though tells me it is probably working ok so the reason for any intermittent operation is not known yet i can only guess that maybe one of the caps shorts out or something as it heats up. If the dielectric is getting too bad it may even loose a lot of capacitance.

The new schematic is attached and there was only that one change, the diode was drawn as a zener of 7.2 volts. 7.2 volts was a common zener voltage used in many circuits because it was more stable then the lower voltage ones and may even be partly temperature compensated. The chip on the board is very happy with 7.2 volts i think it can take up to 15 volts on Vdd.

[MrAl]

Small update.

The frequency generator test is for the 1uf cap not the 22uf cap i made a mistake there.
To test the 22uf cap, apply a generator signal to the ECU terminal (31b) and let the internal mechanical relay close, then remove the ECU signal and see that the relay remains closed for very roughly one half second (1/2 seconds or 0.5 seconds). It may be closer to 0.4 seconds but as high as 0.6 seconds the data sheet for the IC does not specify the threshold voltage levels for when the caps charge up and that is what determines the time (as well as the cap and resistor for that section) so i can only estimate and measure my own relay circuit, which comes in close to 0.5 seconds. The RC time constant is 0.44 seconds but cap values are not that accurate usually especially the electrolytics.

So a better description of how the relay circuit works is as follows.
The ECU pulses the ECU line at about 40 Hz, that triggers the first monostable section as well as the second monostable section and the second section turns the internal relay on which closes the contacts which turns the pump on. If the 40Hz line does not stay pulsing but rather stops or goes to a low frequency, the 1uf cap times out and shuts down the second section, which turns off the internal relay which opens the contacts which turns off the pump after roughly 1/2 of a second.
So now we see why they used two sections of the dual monostable. One to simply detect the 40Hz signal is stable, and the other to act as a short term memory in case the 40Hz signal drops out momentarily while the engine is running. I suspect that the microcontroller used may have priority interrupts it has to service sometimes and so cant put out the 40Hz signal continuously but rather drops out for a short time now and then but always less than 1/2 of a second.
Oh BTW, i forgot to check if the 40Hz signal does actually go higher in frequency when the engine is revved up a bit. it may go higher depending on the speed of the engine. I might check this in the future but it isnt that important as long as a frequency around 40Hz is received by the relay circuit (or possibly higher).

Hope this helps.

[850oldschool]

It seems odd that they would have a microcontroller subject to interrupts generating the 40HZ, when that job could be done with rock solid reliability by a 555.

[MrAl]

It seems odd that they would have a microcontroller subject to interrupts generating the 40HZ, when that job could be done with rock solid reliability by a 555.

Hi,

Well usually if a microcontroller is used it is used for as much as possible without bringing in other parts especially other chips. I had created several designs using just a microcontroller alone with just a few resistors and a few small capacitors. One of them was a 'blinking' voltage meter that would blink out a code for the battery voltage on my old Hyundai. I wanted to keep track of the voltage even when i was in the house and did not want to get too too elaborate, so i made a small circuit using a low power (and low cost maybe $2) microcontroller made by the then Microchip (now Atmel). It measured the battery voltage with resolution about 4mv, then would blink a high intensity white LED which was placed in the window so i could see it from the house. The program would blink out the voltage digit by digit, first the first digit (usually '1') then the second (usually '2') then the third (could be 0 to 9) then the fourth (could also be 0 to 9) with a longer pause between each digit, and an even longer pause after the four digits. It didnt have to be super fast reading so that worked out good. I could look out my window and see it blink and just count the blinks and know the voltage, without using any RF transmitter or anything like that.
Some time later i used a web cam with the computer to count the blinks, and that worked pretty good at night. So i'd get a voltage reading like "12.38" or something like that come up in a window on the PC computer. It was pretty cool. All i needed to go with that was a cheap 5v low power voltage regulator and a capacitor to filter the DC voltage a little.

But i am just guessing that they used a ready made microcontroller it could be a microcontroller board that did the work of a more modern microcontroller chip. I built one of those also before the modern microcontroller chips became common. It used a Z80 CPU but there were many other chips that had to go with that to support that chip such as a ROM, a RAM, a clock generator, input/output decoding, stuff like that. Today's microcontrollers do all that on one chip and even the simpler $1 priced units can do a ton of stuff including read voltages.

I am also guessing that the chip or circuit board gets busy with interrupts, but it could be that it just takes time to scan all the sensors and can also pulse the ECU line to the relay at the required rate. It could also be that the 0.5 second delay is there just to make sure the pump doesnt turn off due to noise or other unforeseen events that interfere with the internal relay. If that happened and there was no delay, the pump running would become erratic.

So there are a number of guesses there when i talk about the interrupts. The delay could take care of that or just about anything else that could temporarily interfere with the ECU frequency. I wouldnt mind if it was even longer though, like 2 seconds. The circuit could be easily modified for that though which is something i will be thinking about.

So i guess you are into electronics and microcontrollers too then? That's cool.

[850oldschool]

Al, what little I know about electronics I learned in the process of keeping older CNC machines operating when I worked in a factory. I probably know just barely enough to ask someone like you a few questions and learn a little more. I'm really enjoying what you're writing here and learning a lot from it. I'd never encountered a monostable multivibrator before.

What you're revealing about how the relay circuit works aligns with what I've seen in the machine tool world, where all the logic operates at TTL voltage, which then has to be interfaced to the higher voltage used by the motors, solenoids, etc. that actually move things around in the physical world.

[abscate]

I’m a student of the 1970s and have published many circuits in Horowitz and Hill. Mine have the dark light bulb on them.

[MrAl]

Al, what little I know about electronics I learned in the process of keeping older CNC machines operating when I worked in a factory. I probably know just barely enough to ask someone like you a few questions and learn a little more. I'm really enjoying what you're writing here and learning a lot from it. I'd never encountered a monostable multivibrator before.

What you're revealing about how the relay circuit works aligns with what I've seen in the machine tool world, where all the logic operates at TTL voltage, which then has to be interfaced to the higher voltage used by the motors, solenoids, etc. that actually move things around in the physical world.

Oh ok. Well if you look up the part number of that monostable chip you can see the waveforms that will tell you how that works.
Basically all it does is when it receives a trigger signal it sets the "Q" output to high and it discharges a capacitor, then the cap starts to charge again. When the capacitor voltage gets up to a certain point if there is not another trigger signal then the Q output goes low, but if there is another trigger the Q stays high. The Q of one of those turns the relay on so you can see if the trigger stops coming in the Q goes low and the relay turns off, thus the pump turns off.