Electrical Engineers Needed: DC Motor control?
 

I have this 12vdc waterpump that comes from a BMW and I want to make it go round & round. I apply voltage to where BMW guy says to and it doesn't do anything. I also tried bring the other pins to ground or 5vdc just for laughs. This is using a lab power supply which may not have the current capacity but it doesn't even register any current at all. Per the local BMW tech for reference:

"pin 3 is b+ and pin 4 is ground. As far as the other 2 pins the one is the signal from the electronic t-stat for on and speed demand and the other is a signal from the motor electronics for speed and various other communications. If you need the specifics on these I can get them but you should only need the power and ground to run the pump at 100%."

What kind of DC motor is this? Some sort of brushless servo thing? There's 3 leads going into the windings and there's a trick circuit that sits on top.

Actually the important questions are:
How do I get this to work with no help from BMW or the pump manufacturer (Pierburg CWA200)? Are there commercially available controllers for this type of DC motor? Ideally a programmable circuit that varies speed with a 0-5volt engine coolant temp sensor :) But I would be happy just getting it to run at 100% and using a standard thermostat too.

Thanks in advance...Pics attached!

Interesting pump. The coils (windings) look similar to the layout of the eddy current coils on the brake of my dyno. Hard to tell by the pics, but there is not an actual armature? It looks like no windings on the shaft, is that correct? The coils would then be N-S-N-S... I cant actually explain how this works, but the controller for the eddy brake is DC-PWM.

Hope that is of some help.

The best I can see, there are no windings on the shaft.

So there may be permanent magnets on the shaft, can you see, or figure a way to tell?

Quick google finds this page;
http://www.ms-motor-service.com/cont...cls=02&pcat=20

If its a brushless DC motor it must have permanent magnets and will be controlled by a tyristor controller. (Pardon my english possibly...)
Same as for whats used in RC cars\planes\boats etc...

It does say it already has the necessary electronics so there should be a schematic available somewhere. Maybe from a BMW service manual?

Sell those on ebay as electric turbos.

You know it will work :P

Dann

Are you sure it's good? I've seen a few of those dead at my friends shop.

It's supposedly brand new but there's some moisture inside the seal housing as if it's been run before.

i dare you to dyno that as an electrical supercharger.

OK I promise to do so...IF you all help me get the damn thing rotating.

what pins do the large capacitor connect to? id expect those to be power and ground.

Makes sense. Pin 4 goes to ground (aluminum case) which is also connected to one side of the large grey capacitor. The other side of the capacitor is connected to Pin 1, the "signal from thermostat" according to the BMW guy who isn't sending me the circuit diagram. Should I reassemble and send 12vdc to Pin 1 & see if I let the smoke out?

I mean I'm no EE. I just know how to solder.

Yes, it's definitely some kind of "brushless servo thing." :D

Specifically, a permanent-magnet brushless DC motor, same as the capstan motor in a tape deck or the fan in the back of your PC.

The "trick circuit" is a three-phase inverter. And all else being equal, it should free-run when energized. Motors like this don't usually require any sort of external sync signal (IOW, they generate their own timebase.)

What I suspect it does require is an external trigger to cause it to turn on when commanded. Specifically, that pin that the tech mentioned as a "signal from the electronic t-stat for on and speed demand." But what is the required polarity and voltage of this signal?

Exactly. I will attempt to find out. BMW guy said it should spin without a signal to either of the extra 2 signal terminals.

I tried sending 12volts from a battery (incase the lab powersupply's limited current was doign something funky) to pin 1, the pin that's connected to the capacitor, and no luck. I also tried sending +12v to both pin 1 and pin3 with no luck. Obviously while pin 4 was grounded. I didn't try pin 1,2, &3 at 12vdc, but yesterday I tried pin 1 & 2 at 5vdc and pin 3 at 12. I also tried grounding pin 1 & 2 while 3 was at +12

Does that circuit board have anything on it which could reasonably be construed as a microcontroller?

I ask because I was looking at the datasheet for that motor at the Pierburg website, and they indicate that it is available in version which is controlled by "digital signals like the LIN-bus." (LIN is similar to CANBUS, used for simpler, lower-datarate applications)

http://www.kspg-ag.de/pdfdoc/kspg_pr...olant_pump.pdf

Given that this part was designed for a BMW, I would not be in the least surprised if they chose to use a network interface to control a pump.

I attempted to remove the board from the aluminum case but the epoxy is pretty significant. I cannot see a computer chip thru the various little windows but I wouldn't be surprised there was one in there. I saw that on their .pdf as well.

I know BMW likes to do things the hard way if at all possible.

Is there something simple (inexpensive simple permanent-magnet brushless DC motor controller) that I can connect to the 3 motor wires that can make it spin? I remember building a Y type circuit back in school but that was 12 years ago. I'd need to purchase and go at this point.

I am thinking it's easier to attach the impeller housing to a standard 2 wire motor or easier still just run the damn OEM waterpump. It would be nice to clean up the front of the engine and give more room for the turbo.

Thanks for your help Joe.

i thought that WAS your turbo :)

fwiw: http://www.bmw-planet.com/diagrams/release/en

these are badass. some of the images arent coming through on my computer (looking at the e90, speically the wiring diagram for the cooling system)


I think once you find the proper application, you need to look for the "temperature control" diagram.

I found on BAV auto that the waterpump P/N for the oldest BMW's with electric water pumps is "11 51 7 586 925," which I'm guessing is basically the same as yours. I pulled up the attached diagram, which makes me guess that Pin 4 is ground, Pin 3 is +12v, and some type of signal differential between 2 and 3 is what runs the controller.

Good luck. This would be really awesome if it ends up working.

As pointed out, that's a 3-phase (see 3 large connections to the windings) permanent magnet (PM) brushless DC (BLDC) motor. It's "DC" because a circuit, an inverter, with 6 MOSFETs, converts DC to 3-phase AC. The inverter uses a position (hall-effect) sensor to sense the position of the rotor (which has permanent magnets) and flips the phase of the AC at the right times (right position) of the rotor to generate torque.

Fun fact, an alternator is the inverse of a BLDC motor if you replace the rotor's "electromagnets" with a bunch of permanent magnets.

Many BLDC motors (such as computer fans) just take power and ground and simply turn the load as fast as it will go which will be a function of how much voltage you apply.

As pointed out as well, this particular fan's board may take a DC or a digital comm signal to tell the inverter how fast to turn the motor.

The above schematic suggests that pin 3 accepts a pulldown signal, perhaps DC, PWM, or a comm signal. PWM is probably the most likely. You may be able to find the pump's datasheet or application notes somewhere.

Here's a tip when testing random signal input pins - do not directly connect them to 12V or gnd. Connect them via a 100 or 1k resistor. The resistor will limit the current and prevent possible damage. Having said that, automotive electronics are protected against connections to 12V and to gnd... Also if you have a bench power supply with a knob to adjust max current, set it as low as possible and only turn it up slightly as needed.

Lastly to check your pump, measure the current that it takes into the 12V. It should probably between a fraction of a mA to a few mA, if the pump doesn't run. If it's zero one of the fuses is blown (the little yellow things in the photo of the guts)

T-tim,

In the 'guts' pic:
I'm betting the li'l silver 'E's and 'forks' at 6 o'clock are (were) fuses.
I'd also gander a guess that the part with the qr code on it is a controller, or maybe a package of fets. Idk.

I'm pretty sure that they're just connecting wires to join things that shouldn't move (the circuit board) to things that might move a little bit (the various external contacts). That sort of fusion-bonding is a pretty cheap and simple process to automate as an alternative to soldering. Crack open the VAF sensor on a 1.6 Miata and you'll see something vaguely similar.



Random thought: what does the mechanical interface between the motor and the impeller look like? I'm wondering if you could easily transplant a conventional DC motor (such as are used by ShurFlo, et al) onto the pump housing.

Thanks everyone for the help and description of how this thing works. It's all coming back to me now.

Next will be to disassemble the impeller housing from the main body & see how that's all held together. I can see that the shaft is hollow and that the back side seal cup thing is pressurized. It's kinda trick in a "way too complicated than it needs to be" sorta way. Just like the motor control. I'll price out a typical 2 wire DC motor. Again, cheapest is the mech. pump already on there....there's nothing wrong with it. This pump was sitting around and I was given the go-ahead to see if it would work.

I get a blip of current on the powersupply display briefly when I connect to Pin 1. I haven't been using a resistor while trying all of this...oops!

That's the big capacitor charging.

OK got it to work. Pierburg clued me in to the actual Pin assignments, 1 is supply voltage, 2 is PWM input, 3 you don't have to worry about. So after hooking 1 & 2 to +12v and waiting about 5 seconds, it turns on. It's the whole waiting thing that I didn't try. This spins at 3500rpm when pin 2 is set to 100duty cycle.

The PWM input does slow it down (after that ~5sec lag) and there is a wide range of PWM frequencies that will work. I wasn't operating the frequency generator but the lab guy said he was trying a bunch and it all worked. I can find out a definite frequency that works.

Is there anyone who can make me a tuneable PWM circuit that uses a coolant temp sensor as an input and allows me to control/tune the coolant/duty cycle curve? Otherwise I may wire it 100% and use a thermostat, but according to the info from Pierburg, this is a 30gal/min pump at full tilt.

Or possibly a circuit that isn't closed loop...perhaps a potentiometer that controls duty cycle that I set at whatever, and I can use the AEM to trigger a relay to 100% at whatever coolant temp, perhaps a few degrees before the cooling fans turning on.

+12vdc square wave.

Thoughts?

Easy peasy.

Give me the desired temp range, e.g. first turns on at 90*C, full power at 100*C.

Then the voltages at the thermosensor at the above min and max temps.

Do you want it in an Arduino or in an analog circuit?


BTW if you control speed with temperature, you will need a separate small pump for the heater circuit. And the thermosensor should be bathed in the coolant that is about to leave the head to enter the heater.

I would put this sucker in the upper rad hose, to reverse flow the coolant (coolest coolant into head first). Reduces detonation.
And use Evans NPG+ coolant - will prevent steam pockets that reverse flow is more prone to, (IIRC).

How much is that sucker anyway?

Jason,

Thanks for your help! I am not sure the difference between Adruino and analog. I will search that out. This is still a month or two away to needing this.

I was thinking about that heater circuit just now, and whether or not to use a thermostat. I guess one of the main reasons for the circuit is to delete the thermostat...then yeah, the entire coolant has to heat up along with the engine as the pump spins slowly and the heater will not work as well. I think the pump should always spin at say 30% duty then increase as temps rise. THis may be a PITA. So keeping a thermostat in there may be better & simplier to debug. The heater would work, the engine should heat up quicker (important as the cylinders were incorrectly bored on the large side and they are 2618) and the thermostat will help control temp as well. I could also then run the pump wide open if needed. Have heater core return between radiator and pump.

I was also staring at the engine when I got home, reverse flow looks simple to do. Both coolant in/outs are right next to each other

EDIT: VP looked up the invoice for the pump, retail was ~$550 remanufactured, from BMW. I have seen them online for much less however.

DDDuuuudde this Arduino thing looks pretty cool. Only ~$25 from amazon too, plus some sort of case. I have to see if it can output PWM at 12 volts instead of 5 :dunno: I'm reading thru some of the sample codes and I haven't seen where you control frequency yet. But that looks simple to program the 'curve' as needed though!

Turning a 5V PWM signal into a 12V pulldown is cake.
Just use something like this, with a 1k gate resistor.
http://search.digikey.com/us/en/prod...0-5-ND/1039352

Negatory. If you spin at 30% duty and pump coolant through the radiator slowly, the engine will take forever to warm up.

Inserting a t-stat will defeat one of the advantages of having an electric pump - reduced load.

Why not use a small pump to pump through the heater until the main pump turns on?

If you insist on a t-stat, then route the heater as a bypass, and set up the logic so that the pump is forced to run at some minimum speed only when the heater blower is turned on and the engine is cold.

BTW do you know how much current the pump draws flat out?
If you're gonna measure it, make it pump water out of a bucket. See also how high it can shoot the water up vertically, to give an idea of how much head pressure it can deliver.

Nothing made by the 3rd party OEM / ODM manufs on rockauto?

Not sure how to use that gizmo to convert 5vdc to 12. Not so simple to an electrical idiot such as me.

Adds cost, hoses, complexity, points of failure, takes up room, etc. I'm trying to keep it cheap and simple. I am only looking at this pump because it's free and I may be able to use the extra space front&center for big turboz.

not sure I see the logic behind running it at a min speed only when heater fan is on. To me it makes more logic to run it at min speed based on cold engine coolant temp regardless if the heater blower is on or not.
I do not know current draw and me setting up the stuffs to measure that is unlikely.

EDIT: I will be using a 100amp alternator.

I haven't searched.