BMW Fuel Pump Controller (Helbako EKPM2(3))

BMW Fuel Pump Controller (Helbako EKPM2(3))

Author
Discussion

Max_Torque

Original Poster:

17,166 posts

189 months

Sunday 3rd August 2014
quotequote all
Anyone know anything about these modules? Thanks to BMW selling what is technically known as a "Sh*tload" of cars, they are now cheap s/h (~£20 on ebay etc).

A quick bit of research says they are "dumb" just needing a PWM input for pump speed and they can possibly control 2 pumps (1 PWM'd 1 Switched)? Early ones are EKPM2 and later ones EKPM3.

They look like this:







They might be a cheap easy way to a speed controlled fuel pump system?

TheEnd

15,370 posts

160 months

Sunday 3rd August 2014
quotequote all
Controlled fuel pump
Functional description, see BMW Technology:

TIS -> Document -> SI Technology -> 16 02 03036

Note on function
The following control modules are involved in the controlled fuel pump function:

- Engine management system
- Electric Fuel Pump (EKP) control module

The engine management system sends the request for the fuel required at the moment to the EKP control module. The EKP control module converts this request into a corresponding engine speed for the fuel pump in the fuel tank. The EKP control module then activates the fuel pump according to the engine speed.



ie. there's a TIS document that looks like it might be handy, and it's not in the standalone TIS disc from 2007 that can be downloaded. I'll see if it can be found in ISTA

Edited by TheEnd on Sunday 3rd August 11:50

TheEnd

15,370 posts

160 months

Sunday 3rd August 2014
quotequote all
Something else I have thought of was mentions of a problem involving reprogramming the EKP to get rid of an issue, so it looks like there is some brains inside.

stevieturbo

15,755 posts

219 months

Sunday 3rd August 2014
quotequote all
Solid state relay any use ?

Max_Torque

Original Poster:

17,166 posts

189 months

Sunday 3rd August 2014
quotequote all
The "2" version looks to be dumb, and the E60 M5 info i have refers to it driving two pumps, the 1st one with a PWM output (speed controlled) and the second one just switched on at some target percentage. The control signal is a hardwired PWM signal from the engine ecu. No idea if the units have CAN buses, but i wouldn't be surprised if the later "3" type got a CAN bus interface for diagnostic reasons etc!

Think i might snaffle one off ebay for an teardown and a look see. For £20 its a cheap source of a heatsink and some power silicon if nothing else......

Sardonicus

18,144 posts

193 months

Monday 4th August 2014
quotequote all
Max_Torque said:
The "2" version looks to be dumb, and the E60 M5 info i have refers to it driving two pumps, the 1st one with a PWM output (speed controlled) and the second one just switched on at some target percentage. The control signal is a hardwired PWM signal from the engine ecu. No idea if the units have CAN buses, but i wouldn't be surprised if the later "3" type got a CAN bus interface for diagnostic reasons etc!

Think i might snaffle one off ebay for an teardown and a look see. For £20 its a cheap source of a heatsink and some power silicon if nothing else......
This ^ quietly confident they are on CAN scratchchin all the best if you decide to have a play though whistle

Max_Torque

Original Poster:

17,166 posts

189 months

Monday 4th August 2014
quotequote all
Sardonicus said:
Max_Torque said:
The "2" version looks to be dumb, and the E60 M5 info i have refers to it driving two pumps, the 1st one with a PWM output (speed controlled) and the second one just switched on at some target percentage. The control signal is a hardwired PWM signal from the engine ecu. No idea if the units have CAN buses, but i wouldn't be surprised if the later "3" type got a CAN bus interface for diagnostic reasons etc!

Think i might snaffle one off ebay for an teardown and a look see. For £20 its a cheap source of a heatsink and some power silicon if nothing else......
This ^ quietly confident they are on CAN scratchchin all the best if you decide to have a play though whistle
If it's on CAN that's fine, i'll just chuck some UDS / ISO 15765 commands at it and see what sticks!
At these volumes you can bet the processor is an Freescale OTP one, which probably rules out any easy reprogramming / repurposing stunts, but at least with CAN i can sniff the Bus on my 335d which has one of these onboard

So, I have got one off of that ol' ebay, and i'll tear it down and stick up some pics/thoughts when it arrives ;-)

shoehorn

686 posts

115 months

Monday 4th August 2014
quotequote all
Sardonicus said:
his ^ quietly confident they are on CAN scratchchin all the best if you decide to have a play though whistle
Yes,

FWIW,I was sharing beers and tales with a friend at his workshop while he was replacing this type of pump controller on a 2007/8 ish E91 M3,(single pump)the owner who wanted to later drive it to Turkey had previously had it re configured but the fault re-occurred.
I opened the old unit up out of pure interest and to be honest it reeked of being cooked and was browning here and there.


Max_Torque

Original Poster:

17,166 posts

189 months

Monday 4th August 2014
quotequote all
One of the reasons i want to "look inside" is to see what sort of input filtering / decoupling they are using. My pump controller, which does have quite a bit of EMC filtering on the input does still inject a fair bit of PWM harmonic noise out onto the power bus in the car, so it will be interesting to see how the OEM device attempts to combat that noise source......


From circuit diagrams i have found, it looks like the version 2 modules are not connected to CAN (the module may well have it, but for example, it is not shown connected on E60 M5). The later cars with the version 3 modules look like they are CAN enabled, but i am not sure if the engine ecu is sending commmands / getting diag feedback from the module over that BUS, or if it is just for diagnostics/coding etc

Sardonicus

18,144 posts

193 months

Tuesday 5th August 2014
quotequote all
Max_Torque said:
If it's on CAN that's fine, i'll just chuck some UDS / ISO 15765 commands at it and see what sticks!
At these volumes you can bet the processor is an Freescale OTP one, which probably rules out any easy reprogramming / repurposing stunts, but at least with CAN i can sniff the Bus on my 335d which has one of these onboard

So, I have got one off of that ol' ebay, and i'll tear it down and stick up some pics/thoughts when it arrives ;-)
Should be interesting wink def keep us updated thumbup

Max_Torque

Original Poster:

17,166 posts

189 months

Wednesday 6th August 2014
quotequote all
Right, here comes the science!

"My" EKPM3 was build in week 26, 2010, and has the BMW part number 7276383:

It has a large(ish) FCi connector, with 2 pin sizes (depending on the current requirement for the each pin, Split into two "cavities". Cavity 1 contains the 4 large pump output pins, labelled 1-2-3-4. Cavity 2 has 16 pins: 2 large pins for 1( VbatT) 2(GND) and 3-16 small pins for logic connections

The PCB is multilayer (looks like 4) and is 82mm x 58mm in standard 1.6mm thick FR4 material, with both a typical green solder mask and a partial thick hard conformal coating for moisture resistance (only over the "logic" part of the upper side, which is unusual)

All components are SMC, as you would expect these days, and it is a 2 sided "load" with parts on both the top and bottom side of the board.

Topside view:


Main components on top side:

Input filter capacitors: 3 x 330uF 35V in a std "panasonic G case" size

Input Inductor: 12mmx12mm SMD 3.3uH fully shielded

Downstream Filter capacitors: 1x330uf 35V "G" and 3x 220uf 35V in "E" case

Half Bridge Driver IC: Infineon BT7960B in TO263-7 package

Freescale 16bit Microprocessor MC9S12XDG128MAA 80pin TQFP package, 128K ROM,12k RAM 2K Eeprom.

4Mhz Xtal Oscilator in std can

Epcos CAN common mode choke



Bottom Side View:



Main components on top side:

On Semiconductor Voltage regulator NCV4269 5v output at 150mA, Automotive load dump rated

NXP CAN tranceiver PCA82C250 Std automotive rated fault tollerant CAN Txcvr, up to 1Mbit data rate.

Unpopulated footprint for "load dump" Transient Voltage Suppressor (TVS) diode in Dpak package

A minimal number of descrete components such as resistor networks, a SMA diode, and a few SMC caps in 0805 size



Edited by Max_Torque on Wednesday 6th August 14:51

Max_Torque

Original Poster:

17,166 posts

189 months

Wednesday 6th August 2014
quotequote all
Some detailed views:

Half bridge driver IC is "Heat sunk" to the cast ally lower case half via a field of vias and some SIL pad to ensure good thermal conductivity:

Via field: (~200 vias)


SIL pad and lower ally casting as heat sink




As soon as you see one of these common mode data chokes, especially an EPCOS one, you know you've got CAN onboard:



Voltage regulator for logic and CAN TXCVR on bottom side:


Driver IC:


Uses N & P channel Mosfet (7 & 9 milliohms RDS(on) typical, max 25Khz, with current sense output and internal overvoltage / overtemp shut down.


Here's that unpoppulated footprint on the lowerside:


A DPAK footprint, directly between VBATT and GND can really only be one thing, and that is Transient Voltage Suppression. As it is not fitted here, BMW have done a bit of a cost down, and obviously are relying on similar diodes in other modules to protect for load dump etc (also, with the power driver being 45v capable, and the Vreg being 60V capable, this is already most of the way towards withstanding load dump without any shunting diode to help


Max_Torque

Original Poster:

17,166 posts

189 months

Wednesday 6th August 2014
quotequote all
PS: No sign of a second pump driver on this version 3 module. Second set of pump output connectors are not, er connected to anything! All drive capability is for 1 pump only, via the Infineon half bridge driver, which is used to minimise "free wheeling" loses (instead of using a "dumb" freewheel diode, which has a large voltage drop and hence a large loss.

Max_Torque

Original Poster:

17,166 posts

189 months

Wednesday 6th August 2014
quotequote all
A first order estimation of the current capability of the device suggests that whilst the "peak" overcurrent shut down in the half bridge driver is up at around 40Amps, a continuous load of around 18Amps will see the device go into thermal limitation:



That also looks about right in terms of the pcb track and connector pin sizes, both of which are going to limit max continous amperage of the driven pump to something like 15A.

shoehorn

686 posts

115 months

Wednesday 6th August 2014
quotequote all
Max,
Forgive my ignorance of the finer points of this witch craft,I`m one of the other 99.9999% of the populationsmile
but would it not be easier,with your obvious skills to build something your self from scratch,
is it just some parts or the program/software bit you are after,is it just not practical to replicate due to cost,time etc. or are you looking to modify its functions/capability as a complete unit?
There is a surprising amount of eye of newt and leg of toad in there for what is a relatively simple task on the grand scale of things.
BTW that one looks better than the one I opened up,your 2nd picture shows where the pins are soldered to the board,ours was burnt crispy and cracked around some of the larger pins.

Max_Torque

Original Poster:

17,166 posts

189 months

Wednesday 6th August 2014
quotequote all
shoehorn said:
Max,
Forgive my ignorance of the finer points of this witch craft,I`m one of the other 99.9999% of the populationsmile
but would it not be easier,with your obvious skills to build something your self from scratch,
is it just some parts or the program/software bit you are after,is it just not practical to replicate due to cost,time etc. or are you looking to modify its functions/capability as a complete unit?
There is a surprising amount of eye of newt and leg of toad in there for what is a relatively simple task on the grand scale of things.
BTW that one looks better than the one I opened up,your 2nd picture shows where the pins are soldered to the board,ours was burnt crispy and cracked around some of the larger pins.
I have my "Own" fuel pump control module, that is much more capable (3pumps, 25A per pump continuous), but it is expensive to produce and complex. I wanted to investigate using a std OEM module as a base, either with modified internal hardware (ie replace the microcontroller) or with a recoded controller, or just simply using it as it stands. Seeing how the OEM module cuts costs, and deals with noise/EMC issues is also enlightening.

As you can now buy these HELBAKO modules for less than £20 second hand, if i could re-appropriate it in some fashion that would be useful. (For a comparison, just the enclosure on my controller costs £20 in the low volumes i have produced them in!)

One thing that stands out is that this module isn't that beefy in terms of connector current capability or the copper tracking on the pcb, which perhaps explains why you see the occasional overheated one!


Jonesy23

4,650 posts

108 months

Thursday 7th August 2014
quotequote all
Looking at the part number the controller is the version with Flash rather than ROM (MC9 not MC3) so is it reprogrammable in-situ? Though I guess that like the background debug it depends on what if any security is enabled.

Though not too challenging to just pull the original controller off and replace it with something else.

Max_Torque

Original Poster:

17,166 posts

189 months

Thursday 7th August 2014
quotequote all
Yeah, slightly unusually it's got FLASH. They were obviously hedging their bets in it requiring in-service reprogramming or to be used for different OEMs etc. Unfortunately the DBM pins aren't broken out to any std header, but need a "bed of nails" on the test points on the underside of the board. I could solder wires to these, but the chance of the FLASH lock not being set are about 10,000:1 ;-) Of course, i could reverse engineer the CAN bootloader functionality that exists with tools such as INPA etc, but again it ain't worth the effort.

Looking at it, the pump PWM drive is just done in that single half bridge driver, so it would be much much easier to just do a new pcb with the same driver on it (or better, with two of them for dual pumps) and at that point i'd just stick my own uC on enabling compete control of functionality.

The only question remains is if buying them s/h for the case / power switches is worth it?

The half bridge driver is about £2.50, and those FCi connectors are around £3 in low volumes, the inductor is another couple of quid, and bits like the xtal and CAN choke / can txv add up to another ~£2. So it looks like it's not quite worth it unfortunately. Finally, the plastic case and die cast enclosure are not water proof, so would limit the module to being cabin mounted.

Max_Torque

Original Poster:

17,166 posts

189 months

Thursday 7th August 2014
quotequote all
It would be interesting to see inside the EKPM2 version, and see how different that one is. Without the "hardwired" input this version 3 is of limited use to the average kitcar builder etc


I will probably carry out a thermal / current loading test on this one, just to get a decent handle on what that BTN7960 driver can do, because it looks like a useful little chip to use, and is obviously fully "automotive rated". Also, i will measure the effectiveness of the input filter used, as my module has a much more comprehensive input filter (for which, read "expensive" ;-) but that does allow operation at very low PWM fundamental frequencies, which makes the module more useful for slower/higher current devices (like cooling fans etc)

shoehorn

686 posts

115 months

Thursday 7th August 2014
quotequote all
Max_Torque said:
It would be interesting to see inside the EKPM2 version
I have a couple of working EKPM2`s would you like one opened with photos?
Or are you getting more intimate,Id be happy to pop one in the post.