Technical How did I kill my EKP?

[veer90]

There's a few possibilities.... Back EMF or over current of an open collector.

Because switching high current on an inductive load is no trivial task, often times they are switched with what's called an open collector setup. This means that there is a pull up on output and a transistor is used to "pull down" the output. This is often used because then the "pull down transistor" doesnt have to source as much current.

When you put the diode in series with the EKP, if its an open collector setup (not sure it is), then you removed the pump as a load when the voltage from the relay was applied, thus shoving much more current through the pull down transistor than would ever happen normally for the target voltage to be achieved when requested by the EKP. I'm pretty sure its not a PWM output after playing with it in the past.

The other possibility is that when ever you have a large inductive load and you switch said inductive load off the magnetic field in the coils must go somewhere. This causes the field to invert the voltage across the coil and spike the voltage up as high as it takes to flow current until the magnetic field can dissipate. Often times on relay's the protect the relay turn on circuitry with what is called a "free wheeling diode", or will also protect the relay from arcing when switching an inductive load in a similar way, as arcing the relay on shut off will reduce its life span A LOT. This free wheeling diode however is not enough. This is because you have to wait for the voltage to swap for it to take effect (diode is wired so that when the voltage is flipped it allows current flow), further diodes them selves take time to actually allow current flow, a normal diode is not fast enough. A zener diode on the other hand, is the right tool for the job. Further, a free wheeling diode can often times cause the voltage to flip back and fourth several times while the current dissipates, often causing the relay to "chatter" during shut off, bad bad news. A zener diode placed in parallel with the load, as close to the load as possible is the proper solution. They react faster, and help prevent the voltage inversion instead entering their breakdown voltage and allowing current flow in the original voltage polarity across the load until the voltage comes back down under the breakdown voltage.

Thus, when you switched both of the large inductive loads off (fuel pumps), with no protection for back EMF its likely there was a voltage spike large enough to either cause your diode to go into enter breakdown and over voltage the EKP, or a surge through the ground path of reverse voltage pulled to much current from the EKP blowing it out.


This is why an SSR or an H bridge is a better solution than the relay for inductive load switching, they have built in protection for this. What you should have done with what you're trying to do, is rather than use the diode, use two SSR's to completely swap the drive circuit for the 450 to run from the battery, and connect the EKP to a fixed resistive load (like say 7 ohms, with resistors that can handle a total of 14 watts of heat dissipation) for the duration of the hobs switch on time.

Note further, that because the diode has a 0.7v drop across it, your 450 is running 0.7 volts less than it would normally when the current is sourced from the EKP. I also hope that your relay doesnt have a NC and NO connection and have the NC connection wired to ground, as that means you just shorted the EKP to ground, another possible reason for the failure, but given that it worked to start the car up, I suspect the NC Is left floating.

So what I'm getting from all this is that I'm in way over my head and I should leave both pumps hard wired and call it a day. LOL

 

[shushikiary]

Haha, its very possible, you just have to do it right.


The good side to hard wiring them is your relay will have much longer life.

There are a few down sides however:

1. Your pumps won't prime the fuel system properly when you unlock the car, leading to shadow codes for your HPFP and possibly reducing its life as the fuel pressure isnt high enough at its first use. This is one of the firmware updates BMW made to help prevent the early death of the HPFP's.

2. If you run your tank down to 1/8 or less full you'll no longer cover the pumps in as much fuel, possibly causing them to overheat because they are no longer being cooled. Though Walbro does rate both of those pumps for external use, so you "should" be fine, not sure I like the risk, I'd rather have them run low current when not needed.

3. The draw on the battery/alternator to constantly have 45 amps worth of fuel pumps running when ever your ignition is on is not healthy either. I'm not sure that the way you've wired it would cause them to shut off during firmware update, so a 20 min flash write might completely drain your battery at that current usage.

 

[veer90]

Haha, its very possible, you just have to do it right.


The good side to hard wiring them is your relay will have much longer life.

There are a few down sides however:

1. Your pumps won't prime the fuel system properly when you unlock the car, leading to shadow codes for your HPFP and possibly reducing its life as the fuel pressure isnt high enough at its first use. This is one of the firmware updates BMW made to help prevent the early death of the HPFP's.

2. If you run your tank down to 1/8 or less full you'll no longer cover the pumps in as much fuel, possibly causing them to overheat because they are no longer being cooled. Though Walbro does rate both of those pumps for external use, so you "should" be fine, not sure I like the risk, I'd rather have them run low current when not needed.

3. The draw on the battery/alternator to constantly have 45 amps worth of fuel pumps running when ever your ignition is on is not healthy either. I'm not sure that the way you've wired it would cause them to shut off during firmware update, so a 20 min flash write might completely drain your battery at that current usage.

Good points. The FPR leaks pressure back into the tank when the car is off so I've already gotten into the habit of manually priming, hitting ignition on before starting the engine.

How low of a pressure can the HPFP handle? I've been thinking about doing boost referenced with 45 psi base pressure (33 psi boost).

I have a kill switch in the center armrest storage area for flashing, although I'm also looking into a 12v trigger that's only on when the engine is running. Fuse 39 (fuel injectors and coils) maybe?

 

[shushikiary]

So I had some drama in dealing with exactly that, and solved all of these issues.


To solve the FPR issue I put a check valve in place, but it kept failing no matter what one I used because of over pressure due to fuel boiling in the fuel rail of the PI after shut down. To solve that issue i put a second check valve in parallel but going the other direction with the primary check valve. The primary check valve has a crack pressure of 2 PSI, the secondary has a crack pressure of 90 PSI, thus it opens if the pressure at the check valve exceeds 90 PSI and saves the check valve by releasing the pressure. This has worked very well.

The HPFP, you really never want to go below 40 PSI input.

To solve the unlock priming issue I used an SSR who's given a wire from the cigarette lighter and another wire from the output of the EPK fuel pump driver. There is a diode on each so current can only flow into the SSR input, not back. The SSR then turns 12v on to a 12v to 5v power supply who runs the enable pin on a dual half H bridge that runs both pumps rated for continuous 30 amp output each half. The input of the H bridge control is a PWM signal from an arduino M0 running a PID loop I wrote that looks at fuel pressure and tries to maintain 72 PSI. The power for this arduino is sourced from the green power block in the ECU box, thus when you flash it shuts the arduino off.

So, when your ignition is on, the SSR is always on, arduino is on, and fuel pumps are driven by PWM. Arduino uses an open collector output to drive the signal wire to the H bridge control input.

Thus, when you unlock the car the EKP turns on, which turns the SSR on which then makes the fuel pumps both go full on because the arduino is also on and tries to hit target pressure. Once the ignition is on the enable is held high by the SSR always and the arduino is on controlling the pumps, but if you start a download the arduino is turned off and the fuel pumps turn off because the control line floats even though the enable pin is held up by the SSR because the ignition is on. This also makes it so that once either target pressure is hit, or the EKP turns off as normal, the fuel pumps also turn off until you turn the ignition on.

The open collector is a fail safe. If the arduino goes down for what ever reason the PWM signal will be pulled high and the pumps will turn full on. Thus if the arduino ever fails you don't shut the pumps off, they instead go full on and you don't run lean, instead you wonder why the heck so much noise is coming from the rear seat and your low pressure fuel pressure is rock solid at FPR max.

 

[marseille]

To solve the FPR issue I put a check valve in place, but it kept failing no matter what one I used because of overpressure due to fuel boiling in the fuel rail of the PI after shut down. To solve that issue i put a second check valve in parallel but going the other direction with the primary check valve. The primary check valve has a crack pressure of 2 PSI, the secondary has a crack pressure of 90 PSI, thus it opens if the pressure at the check valve exceeds 90 PSI and saves the check valve by releasing the pressure. This has worked very well.

Do you have any pictures or a diagram of this? I have a Stg 2 LPFP and recently replaced my OEM FPR in the driver side tank but that did not solve my PSI dropping to zero almost immediately after shutdown. I too have to prime with the pumps using the key on/off procedure a couple of times to get it to light. Any insight (pics/diagrams/parts) on how I can configure my 535xi to get it to hold pressure after shutdown would be greatly appreciated.

 

[shushikiary]

I'd have to pull it from the car to get a good picture, its buried between the tank and the foot well.

I can however draw a shitty paint drawing for you, LOL.

The low crack pressure check valve is this one: https://www.summitracing.com/parts/sum-220193b/overview/

The 90 PSI crack one is one I got off ebay: https://www.ebay.com/itm/1-4-MNPT-Adjustable-SS-Check-Valve-100-Psi-Crack-GENERANT-ACV-4PSS-V-100/392142510201?ssPageName=STRK:MEBIDX:IT&_trksid=p2057872.m2749.l2649

Note its stainless steel, adjustable, and uses a viton seal so it will survive ethenol.

Obviously I had to play the "converter" game to get the NPT thread on the small check valve to mate to a T that has 2 -8AN fittings on it to attach to the other valve and the fuel lines. I got all the needed fittings off summit racing. (2 x -8x-8x1/4NPT T's (all 3 sides male), 2x 1/4NPT female elbows, 1x 1/4 NPT nipple of the right length to make it all fit, 1x 1/4NPT union to mate the nipple to the check valve, 2x -8 female elbow)

I used pushlock fuel hose and fittings for the FPR/filter to the check valves, then used a stainless steel braided PTFE hose custom made by discount hydraulic hose to run from there up to the fuel rail. At the fuel rail connector I used the normal "quick disconnect" nipple that game with the CP-E PI kit but used a metal thread on type connector to connect to it rather than the plastic insert style.

I used said custom hose because its rated to over 1000 PSI, so only the check valve would ever break/leak if it didn't work. If the check valve does break (like it did once before (the low crack pressure one)) it tends to shoot its guts into the line coming from the FPR/filter clogging it. I did have a stainless PTFE hose there too, but it got destroyed because of this, so if it happens again the push lock hose is easy to replace, and it should never see more than 150 PSI on that side of things as a transient spike (and only when the pumps are on), which that hose is rated for.

It's been working now for a long time. I can easily hold a flat 90 PSI if I wanted to on 100% E85 and 600 whp/tq.

 

[marseille]

Great information. Sorry for the obvious questions, I believe I can implement the check valve system under the car prior to my short line kit for the ethanol sensor. Just trying to sort out how you are referencing the fuel rail because I see this system installed after the intank filer/fpr and before the HPFP ?

intank filter/FPR -> check valves -> ethanol sensor -> HPFP -> Fuel Rail

 

[shushikiary]

I don't have an ethanol sensor, but if I did it would go before the T that goes to the PI fuel rail.

In the paint picture I made I just left the whole "fuel rail" as a conglomeration of the HPFP, normal fuel rail, and PI fuel rail.

There is a T before the LPFP pressure sensor that goes to the CP-E PI fuel rail. The path to the HPFP is the same as normal. Its not a return fuel system so the PI fuel rail, nor the high pressure fuel rail have a FPR with return on them.

The FPR is as you wrote, its built into the "bucket" that goes into the driver side of the fuel tank, as well as the fuel filter. Its the precision raceworks setup.

The check valves are after the FPR/filer (which is a stainless steel mesh filter) and connected to the FPR output by a -8AN push lock hose that is maybe 2.5 feet long. Then the check valve output goes to the quick disconnect fitting that is screwed into the T that gives fuel to both the PI fuel rail and the the line that feeds HPFP.

 

[marseille]

@shushikiary Much appreciated!