RomRaider

It worked exactly as expected, on 1 of 2 fuel pump controllers.
Somehow one controller is stuck on 100% as soon as it gets power. Weird, because the one that functions right does the same thing when swapped in. Possibly in the harness? It did this when I first started tinkering, but it worked right up until now controlling them with an Arduino.

The Multiplier address works perfect. Leave Low/Med/High = 33/66/100 and set the multiplier = 32768. Prime is at 50% and idle is at 16.7% at the controller and 33.3% at the pump.

A pulldown resistor is needed on the signal line from the ECU.

I'll keep at it I guess. Thanks again.

EDIT: The pump was grounded through the surge tank to the chassis. Also, FYI, Fiberglass Reinforced Polymer bolts can barely be snugged down without breaking, but they work..

Good to hear is working, sort of.....

Best of luck!

I've got a Walbro 535 strapped inside a radium fuel pump hanger ready to install but I'm afraid to do so, my fuel pump control module is still the above mentioned (7W83-9D372-AA Jag) unit. I'm running AZ1G900C

I have had a good crack but can't unpick the logic that governs the priming, and then once the car has started, the PWM signal that is sent to the controller. After a small period of time from startup (maybe 15-30 seconds) the logic is clear to me, the ECU uses the MRP to govern which of the three PWM signals to send the controller, but what isn't clear is what happens up to that point from startup, I've read through the hardware and software ECU manuals and have a better understanding of how the ECU PWM control system works but am lacking some reverse engineering skills, not for want of trying!

I'm happy for the priming to be at 100 duty cycle but I don't necessarily think it needs that much fuel to prime. Directly after priming once the car has started I'm concerned about the typical behaviour which is the pump continuing to receive full voltage for x seconds and having the stock return line causing the fuel lines to over-pressure and the car running crazy rich and rinse the oil from the cylinder walls and gas my neighbours with raw fuel!

Maybe a dirty way to accomplish the desired behaviour could be using previously mentioned multiplier to say halve the PWM output to the controller (so priming is 50% and then it runs at 50% for a while, which I'm assuming would be enough for start-up) and then compensate the overall PWM multiplier with the regular MRP based (Low/Medium/High) commanded fuel pump duty table - i.e. 100% would be 200% in ECUFlash (but then the conversion of the duty cycle values would mean the pump is actually receiving the intended 100% duty signal in operation)

I hope there's a cleaner way to change the behaviour, so was wondering if anyone has taken the time to understand how the priming -> transition to regular running logic is strung together? I'd like to prime at 100% but the ECU to jump straight to the MRP based logic once the engine has fired up, if that's possible.


p.s. see image below... Thinking it might be the priming PWM reference, I've tried adjusting the offset at the below location to change the referenced address to the MRP based 'low' setting which is adjust the HEX code at 43698 from C785 to C789 but I can't see any change in behaviour - it's not the prime value it seems, priming is still at 100% duty cycle.



Thanks a lot,
Nick

I don't have your ROM disassembled, but typically they're laid out similarly.....

If you follow the reference at line 438C4 to the subroutine, this should give the locations of the priming and start up behaviour for the fuel pump.

The pump primes with ignition on; and then with cranking and post-start, runs at 100% duty according to various switches and counters laid out in that subroutine. The most important for the situation your describing is the first word value = EA6 (compared to one of the post start counters) which amounts to ~ 30s of pump running at full duty, after which it switches to the MAP related duty switching (branch in this subroutine) until the ignition is switched off.

Your post piqued my interested, so I went off and found this. I changed this to 150 (96 in hex) and now it runs for two seconds, before switching back down to 33% reported duty.

I do wonder what the role of this behaviour is, clear/fill lines? Fill pot in fuel pump assembly, for cooling and avoiding surge? Other? There must have been a reason in the first place for the engineers to do it.

I suspect that in stock systems, running 100% duty was less of an issue with over fueling, as the reg could probably take the extra flow and still control pressure. When dealing with larger pumps and flow, a stock system will definitely be over pressure, unless you've drilled out the venturi. I'm inclined to think that the engineers didn't consider it a major problem (bore wash), otherwise we wouldn't see the behaviour in the first place. Of course the volume of fuel heading into the cylinder at full boost is far greater than when sitting at idle, even if greater evapouration is occurring. No question your idle AFR will be improved though, and there will be less stink from a rich condition.

Anyhow, let me know how it goes and your thoughts.

See the attached image cheat sheet.

Firstly, thankyou very much for digging this up Daniel, I really appreciate your effort!

I tested the D12BA location; modifying the hex from 0AE6 to 00FA (I thought the binary multiplier must have been 0.008 - looking at the 32Bitbase and trying to draw a parallel between the existing hex value and one of the time multipliers in the 32Bitbase file, which would make the time 2 seconds? Could be tripping but it seemed to check out)

When this hex modification worked out as you had suggested I installed the Walbro 535 and Radium hanger. After basic testing went for a 20 minute drive, filled it up with fuel, then hot started it and to my surprise the ECU commanded a longer period (felt like about 30 seconds) of 100% duty cycle again!

I looked at the next location D12BC which had the same 0AE6 value there and changed that to 00FA as well, but the hot start was the same - it maintained a 100% duty cycle for a longer period of time.

The silver lining is that it seems the Radium hanger enables the return of a lot more fuel than the stock hangar which is what the marketing says too but the fuel pressure (@ FPR) did raise at idle conditions when the duty cycle was either 68% (what I use at medium, note that it is a true 68% unlike if you use the stock FPR, medium/66 in the ECU code actually commands a 75% 12v signal at the fuel pump) or 100% duty cycle by ~2 PSI for medium and ~4 PSI for high/100% duty.

I'm planning on upping the base pressure to 4 bar next as the ID 1050's are more than maxxed out on e85, and I'm curious whether the lower flow rate of fuel at the higher pressure will minimise those medium/high duty fuel pressure discrepancies when they're unwantingly commanded?

I'm going to keep poking around for other routines that might only be activate on hot start, but no luck yet!

Thanks

No worries!

I think you are correct about the counter. Elsewhere NSFW has looked at this and reported the most timers to increment 125 x per sec (or every 8ms) I was simply looking at the logs that I had and doing a quick eyeball calc, which was also dependent upon the number of parameters I was logging at the time.......

I've had the opportunity to have a look at the comparative code between the WRX and the STI, for both the subroutine mentioned above and the subroutine which is invoked after the counters/timers expire at 43ABA (this controls the normal behaviour after start manifold pressure/pump duty).

The only thing that really stood out was the difference between the two ROMs with respect to the tables for injector switch over points for medium and high duty at varying manifold pressure.

As the all the vehicles crack open the throttle and retard timing on start up, I suspect that the manifold pressure and subsequent pump duty created by these tables is masking the effects of the timers. Thus, your pump is still running at higher duty until later in the warm up process (even on hot start).

If you haven't changed over these tables, are running bigger injectors, and still have the stock tables for idle RPM after start etc, these may be the culprits.

It's possible that there are other switches that are being invoked in the WRX that I haven't been able to look into/log, compared to the STI, but think it's less likely. I haven't found anything to date that affects operation with hot start, but that is not to say that it doesn't exist.

I did have to change over my MAP/injector switch points to enable more 'normal' 33-100% duty operation after changing over to bigger injectors.

Anyhow, food for thought/comment.

Roger that!!! I've checked the logs on start-up and it seems you are on the money with the tables, I'm not sure if it's something to do with the engine being forged and tolerances are a little loose when things are cooler but the engine doesn't draw as much of a vacuum for about some seconds when starting up and just nudges into that -6 to -7psi vacuum range before heading back to -9.5 to -10 when it's got a little bit of heat, that might be enough to trip the duty logic into the medium setting it seems.

I've taken some time to visualise and understand the pump duty cycle changeover points (or area on the graph.) Hit a bit of a wall! It's a fairly complicated equation to balance in theory; how much pump duty cycle do you need at the pump to run the engine at various MRP and injector duty % - pump flow, duty cycle, injectors, hydraulic resistance, pump efficiency, etc - at one point this week I had to give up trying to model the required fuel theoretically.

One issue I can see with being inaccurate with the duty cycle tables here and allowing too much fuel to pass through the system, aside from heat transfer and other things mentioned in the forum, is that the fuel pressure regulator has a line in the spec (Turbosmart FPR1200) which shows what the regulation slope is at a given flow rate - so with this Walbro 535 pump, which can flow quite a bit of volume at high pressure seems that there's a pressure increase of 0.003psi per 1 LPH, sounds like nothing, but at 400 litres an hour is ~2% of fuelling inaccuracy @ 58psi just through the fuel flow rate between potential duty cycle jumps.

I guess that makes the switchover point for the pump more important, the more fuel you can potentially flow.

I can't find a pressure vs flow graph for any stock Subaru/Denso pump, but assume there's quite a bit of pressure and volume reduction with the stock pump at higher boost pressures, otherwise they wouldn't have weighted the MRP in the pump logic so heavily with the injector duty?

Anyway, I'm going to try flattening out the changevoer points with the bigger pump so it relies a little more heavily on the duty cycle percentage vs MRP than the stock config and see how the pressure looks in the logs.