Making a Fuel Pressure Regulator Tester

[revilla]

After all the modifications I’ve done to my engine it’s been left running a bit marginal on air fuel ratio on the rollers, and since I’m not running a fully programmable ECU I wanted to tweak the fuel pressure a little to enrich it. I calculated I needed about 5-6% increase in fuelling which meant an increase in fuel pressure from the standard 44psi to about 49psi (√(49/44)=1.055).

I planned to do this by squashing the fuel pressure regulator as suggested by Dave Andrews and others, but not having done it before I wasn’t sure how far to go and taking the regulator on and off the car for trial and error would be awkward due to poor access to the clip when installed. I also wanted to be able to test the calibration of the spare pressure regulators I had to make sure they were good.

It occurred to me that a pressure regulator is just a controlled leak that leaks at a constant pressure, so I thought an engine leak down tester should read a pressure regulator correctly. It shouldn’t make much difference whether it is regulating air or fuel, pressure is pressure. So I set about making an adapter as shown below.

I simply sawed the end off an old injector rail, screwed a pneumatic fitting into the cut end to match the fitting on my leak down tester and screwed a sump plug into the one injector port left to blank it off, using a little PTFE tape to seal all the threads. I bent the ends of the retailing clip up so that it no longer locks in place, allowing me to fit and remove regulators easily.

The adapter is shown below with a regulator fitted and plugged into my home-made leak down tester.

Things looked promising; with 50psi or more on the input gauge, the output gauge read a steady 44psi which is exactly what the regulator was showing on the inline fuel pressure gauge I have on my car.

The next job was to compress the internal spring inside the regulator a little to raise the operating pressure. This can be done by sandwiching the regulator between two suitable sized sockets (23mm works well to support the base, in my case an 11mm socket was just about the right outside diameter to press the top in). You can them squeeze the regulator between the sockets in a vice. There’s a little plastic ring around the regulator which supports the sealing O-ring, this need to be removed whilst compressing it and refitted later to avoid damaging it.

The aim is to roll in the top of the regulator as shown below.

After a few cycles of trial and error I had it reading 49psi on the tester. Do it a bit at a time because there's no way back if you go too far!

At which point I fitted it back in the car and checked the inline fuel pressure gauge. For this the little vacuum line needs to be left off the regulator as the vacuum signal lowers the fuel pressure under the high manifold vacuum level at idle. As you can see below, it showed a nice 49psi in agreement with the reading on the tester.

The engine fired up and ran smoothly. Idle fuel trim as reported over OBDII had reduced, suggesting that the injectors were now flowing a bit more fuel but well within what the ECU could trim out when operating closed loop. This should leave me with safer lambdas under high load conditions. We will see when I get it back on the rollers later this summer.

Job done!

[aerobod - near CYYC]

Excellent job, Edd China would be envious! Nothing more satisfying than an undetectable modification that makes things better than original.

[oilyhands]

Andrew, a 5% increase in pressure does not yield a 5% increase in flow, pretty sure flow increases at the square root of the pressure increase, so you’ll need a 12% increase in pressure to yield 5-6% increase in fuel delivery.

The rest of the post looks good though, I have a tester using an old fuel rail and a spout that I connect to my compressor that does the job quite well.

Oily

 

[revilla]

Hi Dave,

I absolutely agree with you, fuel flow varies as the square root of the pressure drop across the injector. In my original post I said:

I calculated I needed about 5-6% increase in fuelling which meant an increase in fuel pressure from the standard 44psi to about 49psi (√(49/44)=1.055).

44psi to 49psi is an 11.3% increase in pressure, I included the square root in my calculation to get a 5.5% increase in fuelling.

Is that not correct?

Cheers,

Andrew

[oilyhands]

Yup, didnt spot the details of the equation. An explanation of the increase percentage would make it clearer for the causal reader (like me) :)

Oily

[revilla]

Yes, fair point.

So for example if I want to reduce the lambda value from 0.95 to 0.90. That means reduxing the air-fuel ratio by a factor of 0.90/0.95 or equivalently increasing the fuel-air ratio by a factor of 0.95/0.90=1.055, so I need to increase the fuel flow by a factor of 1.055 (105.5% or a 5.5% increase) for the same air flow.

The fuel flow through the injector is proportional to the square root of the pressure difference across the injector, so to work out the pressure increase I need in the fuel rail I have to square that factor. 1.055*1.055 is about 1.114 so I need to increase the fuel pressure by a factor of 1.114 (111.4% or an 11.4% increase).

The standard regulator is calibrated to run at 3 Bar or about 44psi, so I need to multiply this by 1.114 to give 3.34 Bar or 49psi.

All of this only applies to areas of the load map where the engine is running open loop, i.e. not making trim corrections to the fuelling based in the feedback from the lambda sensor, such as at wide open throttle. At steady lower loads the ECU will correct for the increased fuel flow to restore the original lambda value, provided that the increase in injector flow does not go beyond the limits that the ECU is allowed to trim for (I may be wrong but it think that's usually around +-15%). In my case the long term fuel trim was showing about +8% so the ECU was finding that it was needing to increase the fuel flow relative to the base map to achieve lambda 1 at lower loads as expected due to the increased airflow I have achieved with mild headwork at the same MAP and IAT. The increase in fuel pressure should therefore bring it closer to being correctly fuelled on the base map figures and that long term trim number should come down by around 5.5%. So at lower loads the AFR should be largely unaffected and at higher loads it will be reduced to safer levels.

[oilyhands]

Perfection..

Oily

[StevehS3]

Interesting topic. It would be good to understand more about how the Rover ECU operates. I have often wondered about the degree of AFR trim the ECU could make based on the lambda reading. I also didn’t know that the ECU ran open loop at wide open throttle.

[revilla]

The standard "narrow band" lambda sensors used on our cars tend to produce a fairly on-off signal, switching from fully low voltage to fully high voltage over a very narrow range of mixtures. They are basically answering the question "is there or isn't there excess oxygen in the exhaust?". This means they tell the ECU whether it is running rich or lean, but not by how much.

The ECU gets around this by progressively enriching the mixture while the lambda read lean, then once the lambda signal switches to rich it starts to lean the mixture until the lambda read lean again. That's why you see a cyclic voltage from the lambda sensor, it's not that the lambda itself switches, it's the ECU driving the mixture slightly lean then slightly rich, using the lambda sensor to sniff out the correct mixture.

If you hold the engine at say 2500rpm and hold the throttle very steady you can sometimes even see this cycling as a small oscillation in RPM on the tacho. Incidentally the alternating lean and rich mixture is exactly what a catalytic converter needs to drive all of the various reactions.

The problem is that at full throttle and generally at higher loads, you actually want to run the engine on a rich mixture to control the combustion temperatures and prevent detonation, and around anything other than the ideal stoichiometric mixture the lambda sensor isn't giving any usable information; it will just read rich all the time, with no indication of how rich.

So under these conditions the ECU just has to fall back into its base map and doesn't use the lambda signal. So it's important to make sure that the base map results in an appropriate slightly rich mixture.

[oilyhands]

The clamp values for normal running are quite generous, Mark at Kmaps was tweaking them while he was here a while back, that is why the ECU is fairly tolerant of tuning provided you dont mess too much with the MAP signal.

I normally raise fuel pressure by around 0.5 bar when pushing the boundaries, this ensure that the base map (open loop) is rich enough but the trim to stoichiometric is within the bounds of the ECU clamp.

oily

[StevehS3]

Many thanks for the info. I can certainly understand the benefits of running rich at high load sites but I am surprised that the emissions are within limits and that the cat works correctly in those conditions. I had assumed that once the engine is hot it would always run in closed-loop mode. Good to know.

[aerobod - near CYYC]

Typically closed loop isn’t used at high loads with a wideband Lambda sensor either, as a failed or out of calibration sensor could cause a dangerous lean condition and transient conditions could also be too rapid for closed loop operation to adapt safely unless a very tight trim allowance is used, which may make closed loop pointless in those conditions.

For my MBE 9A4 ECU with wideband Lambda, I set 5000RPM and/or Throttle Site 8 as the limits below which closed loop is used together with more than 60 seconds warmup and over 60C coolant temperature. I don’t have or need a catalytic converter on my car though, due to the car classification here in Alberta.

[oilyhands]

It’s important to make the distinction between map adaptation and closed loop running, closed loop generally operates at a constant load and constant speed where the ECU will target Stoi and cycle in order to activate the cat, map adaptations are the result of lambda feedback in all situations including open loop where the ECU determines that the base line fuelling is too high/low and will trim to suit. Closed loop s a dynamic situation, adaptations are a semi permanent adjustment.

Oily

[aerobod - near CYYC]

Also that with a wideband sensor, closed loop operation doesn't have to target 1,0 / stoi, any Lambda value can be targeted at any map site, unlike with a narrow band sensor. This strategy may be detrimental to a catalytic converter, though.