Fuel Pump EMI Noise Suppression

[Spittybug]

In the continuing search for the reasons why the Megasquirt 3.57 surface mount boards have sync issues that the 3.0 component board does not, the focus continues to be on the noisy fuel pump. The attached article helps to illustrate the point.

Calling all the electrical wizards out there (Bitsyncmaster, do you read me? :) ), can you help to translate this into an actionable item so that we can give another try at eliminating the high RPM sync loss? What would it comprise of, where to wire it in and how?

Much appreciated!

[Bitsyncmaster]

I did not download that PDF since my work internet is crawling today.

EMI is electro-magnetic-interference. Anything that creates a spark (brushed motors) will produce it. Then it gets radiated by the power wires powering the pump. Best fix for that is to install a choke in each wire up near the pump. You would need a choke that can carry 10 amps. The highest inductance is best but your limited as to the size of the choke. One of those ferite donuts or the clamp type may work with existing wire.

Another option is to filter the noise at the ECU. That would be less problem since you're dealing with low currents. But you cannot put a choke on the ignition coil signal if that's what your using for RPM measurement. If that's the case then you need to filter at the source of the noise (fuel pump).

Are you sure it's EMI and not noise on the power into the ECU?

[Spittybug]

Thanks Dave.

Not we aren't sure. The DIYautotune guys claim that there is power feed conditioning into the unit however. We've seen the o-scope output of the VR signal prior to going into MS and even at high RPMS it doesn't break down while the resulting squared tooth signal does. This got us looking the DC issue and not ignition signal issues. It's possible the noise could be from radiator fans or blower fan too I suppose. Is there a way to measure such noise to track it down?

Can you expand a bit more on adding the choke? Simple to do?

[dmc6960]

Anyone have a spare battery you can direct wire into the fuel pump for a test run to completely isolate it from the rest of the car's wiring?

[Bitsyncmaster]

Thanks Dave.

Not we aren't sure. The DIYautotune guys claim that there is power feed conditioning into the unit however. We've seen the o-scope output of the VR signal prior to going into MS and even at high RPMS it doesn't break down while the resulting squared tooth signal does. This got us looking the DC issue and not ignition signal issues. It's possible the noise could be from radiator fans or blower fan too I suppose. Is there a way to measure such noise to track it down?

Can you expand a bit more on adding the choke? Simple to do?

If you can get a large doughnut type ferite core (maybe Radio Shack) then all you do is pull the pin, and wrap as many turns of that wire in the doughnut then put the pin back.

If you can not find some cores, pm me your address and I can send you some.

[lazabby]

It's very curious as to why the 3.0 Megasquirt runs fine but the 3.57 does not.

[Spittybug]

13183 as in one of these from Radio Shack? Just snap it on the 12v supplying the fuel pump, as close to the pump as possible? Or try and get both + and - wires through it? Boy wouldn't that be great if it ends up being this easy?

http://www.radioshack.com/product/index.jsp?productId=3012599#showFullReviews (http://www.radioshack.com/product/index.jsp?productId=3012599#showFullReviews)

Shannon, it could just be minor differences in the circuits or perhaps my fuel pump isn't as noisy as yours and Ian's. Who knows? We're still digging!

[Bitsyncmaster]

It's very curious as to why the 3.0 Megasquirt runs fine but the 3.57 does not.

Maybe they changed the filter values for the RPM input. Or they have a bug in new software. If they run interrupts, those bugs are so hard to trace down problems.

[Spittybug]

Maybe they changed the filter values for the RPM input. Or they have a bug in new software. If they run interrupts, those bugs are so hard to trace down problems.

It's the same software on both boards.

[Bitsyncmaster]

13183 as in one of these from Radio Shack? Just snap it on the 12v supplying the fuel pump, as close to the pump as possible? Or try and get both + and - wires through it? Boy wouldn't that be great if it ends up being this easy?

http://www.radioshack.com/product/index.jsp?productId=3012599#showFullReviews (http://www.radioshack.com/product/index.jsp?productId=3012599#showFullReviews)

Shannon, it could just be minor differences in the circuits or perhaps my fuel pump isn't as noisy as yours and Ian's. Who knows? We're still digging!

You can try those.

The more loops passing through the core would give you higher inductance and attenuate not only the high frequencies but low ones also. Yes, I would clip one on each pump wire.

[Spittybug]

There you go Shannon, a ~$6.00 (pair) potential solution to try.

[lazabby]

There you go Shannon, a ~$6.00 (pair) potential solution to try.

There's a RS by my house so I'll go by and pick up a couple of them. I hope it works. If life could only be so easy.

[DniveK]

Seems to be an error in the PDF file, I was unable to open it. How about a High-Pass filter on the line? Usually used to stop noise from motors.

[lazabby]

I went by Radio Shack and picked up the piece that Owen found online. I clamped one on the electrical wires going to the fuel pump. I clamped the other to the wires going to the fuel sending unit. I did not loop the wires one time as suggested by Dave. I might have been able to loop it if I used one piece for each wire going to the fuel pump. I started the car and still had sync loss issues when I tried revving the car up to close to 3,000 RPM (automatic).

Just to let you know, a few weeks back I disconnected the alternator and took the belt off and started the car. Still had sync loss issues. Several days ago I removed every fuse that wasn't necessary for the car to run. Still had sync loss issues.

We're operating under the assumption that there is EMI causing the sync loss issue at high RPMs. If this is the case, then we're down to fuel pump, spark plug wires and distributor, unless I'm missing something else. The sync loss is occurring on two automobiles using the Megasquirt 3.57. One car is automatic (mine) and one is a manual.

Any more ideas of what to try?

[Bitsyncmaster]

You don't need one on the fuel sender. Put one on each wire to the fuel pump and see if that helps. I can send you some larger units you will have to wrap the wire in but I'm thinking you're pump is not the cause of the problem.

I'm thinking the filter used in the ECU for reading RPM off the coil is not optimal. I could give you one of my idle ECU boards with my filter circuit to try. I use an optical isolator and some RC filters. It would need +12 volt power and ground.

[lazabby]

You don't need one on the fuel sender. Put one on each wire to the fuel pump and see if that helps. I can send you some larger units you will have to wrap the wire in but I'm thinking you're pump is not the cause of the problem.

I'm thinking the filter used in the ECU for reading RPM off the coil is not optimal. I could give you one of my idle ECU boards with my filter circuit to try. I use an optical isolator and some RC filters. It would need +12 volt power and ground.

Tomorrow I'll re-do the setup and use one clamp core on each wire going to the fuel pump. I should be able to wrap the wires one time. I have to take off the rubber boot on top of the fuel pump in order to get at enough wire length. I'll try that and see what happens. If it still doesn't work, I may try clamping one on at the Megasquirt unit. I'll take you up on the offer on the ECU board if everything fails.

[Spittybug]

You don't need one on the fuel sender. Put one on each wire to the fuel pump and see if that helps. I can send you some larger units you will have to wrap the wire in but I'm thinking you're pump is not the cause of the problem.

I'm thinking the filter used in the ECU for reading RPM off the coil is not optimal. I could give you one of my idle ECU boards with my filter circuit to try. I use an optical isolator and some RC filters. It would need +12 volt power and ground.

Dave, I'm not sure I'm reading you properly..... we are triggering directly from the VR coil in the dizzy, not the ignition coil (tach signal). How would your board interface with this setup? What am I missing?

[DniveK]

Could be the wire from the coil to the distributor is not shielded properly, I think that would cause the most EMI.

[Spittybug]

Fully foil shielded, grounded on one end only, run independently of all other wires, away from spark wires, etc... been there tried that! :mecry:

[Bitsyncmaster]

Dave, I'm not sure I'm reading you properly..... we are triggering directly from the VR coil in the dizzy, not the ignition coil (tach signal). How would your board interface with this setup? What am I missing?

If your running the VR coil for RPM than that should be better than using the ignition coil. Is that VR coil wire shielded?

[Spittybug]

If your running the VR coil for RPM than that should be better than using the ignition coil. Is that VR coil wire shielded?

See one post up.

[lazabby]

I added one choke to each wife coming off of the fuel pump. Made no difference. I tried putting the chokes going into the Megasquirt unit. That meant trying to get five or six wires into each of the two chokes. I'm sure not sure how effective that was but it made no difference on the sync loss issue.:hmm:

Dave, coming to Houston anytime soon?

[Bitsyncmaster]

I added one choke to each wife coming off of the fuel pump. Made no difference. I tried putting the chokes going into the Megasquirt unit. That meant trying to get five or six wires into each of the two chokes. I'm sure not sure how effective that was but it made no difference on the sync loss issue.:hmm:

Dave, coming to Houston anytime soon?

I would think MS already did this but it's worth a try. Feed power to the MS through a diode and put a large cap after the diode to ground. All car power is very noisy. The cap filters the noise and the diode prevents other things from drawing power from the cap.

Do you have a schematic of the MS unit I could look at?

[Spittybug]

http://www.megamanual.com/index.html

A frustrating site. I'm not a web guy, but their URLs don't change when you navigate and that confuses me! The left navigation panel will lead you to the 3.57 board and full schematics.

[Bitsyncmaster]

http://www.megamanual.com/index.html

A frustrating site. I'm not a web guy, but their URLs don't change when you navigate and that confuses me! The left navigation panel will lead you to the 3.57 board and full schematics.

Nice schematics.

Looks like they do use the diode and capacitor but the capacitor is only 33 uf. and it shares that voltage with the stepper.
I would have driven the 5 volt regulator with it's own diode and cap (100 uf. min). ref. page 6, C16 and D10.

Also if your using the "Ignition Driver" to drive your coil, That ground on Q16 should have been a separate return line not common the the MS ground. Q16 page 5

I see they have a filter for the VR input. Have you tried adjusting the hysteresis (R52) and set point (R56)? I would need to get a scope on the VR circuit to check if it's optimal with our VR signal.

[Spittybug]

Nice schematics.

Looks like they do use the diode and capacitor but the capacitor is only 33 uf. and it shares that voltage with the stepper.
I would have driven the 5 volt regulator with it's own diode and cap (100 uf. min). ref. page 6, C16 and D10.

Also if your using the "Ignition Driver" to drive your coil, That ground on Q16 should have been a separate return line not common the the MS ground. Q16 page 5

I see they have a filter for the VR input. Have you tried adjusting the hysteresis (R52) and set point (R56)? I would need to get a scope on the VR circuit to check if it's optimal with our VR signal.

Dave, I'll have to look at the schematic to digest your first couple of points, but regarding the VR signal, we've scoped it. I chucked the dizzy in my drill press and took it up to very high speeds. The signal never appeared to breakdown. We keep the same nice shape the whole time. I looked to see if I still have the photos, but I do not. I may be able to retrieve them from when I posted them over at the Megasquirt site. Now, under very low speeds it does, but that is to be expected based on the physics of the matter. The guys at DIY had us trimming both R52 and R56 and while we can minimally change some of the behavior of just when the sync loss occurs, we are not able to eliminate it. This problem has been going on so long now that I forget all of the things we've tried and failed with. I want to say that despite the nice appearance of the incoming analog signal, the resulting square tooth wave that the MS unit converts it to for the coil driver IC ended up breaking down at higher drill press RPM. The power supply for the MS in this case was a bench top 12vdc, not a car. I'd forgotten that. That would imply that the problem isn't in the car....

To me (an observer and non-trained in such electronics), it looks like although the VR signal isn't breaking down, the 3.57 MS unit is losing its ability to process it into the clean squared wave form at the higher frequency. My 3.0 box doesn't appear to have this limitation at all.

http://www.msextra.com/forums/viewtopic.php?f=101&t=44382&start=40#p319355 This link will take you to a post in the MS forums that has the o-scope pictures and writup of what we saw. A couple of posts further down is interesting too. I hope the pictures are available to you as a guest to view. If not, I'll try to download them and post here, or you could join (its free).

[Bitsyncmaster]

If both MS circuit are the same with the same values than most likely the new PCB layout is causing the difference. I really think the common ground used for the ignition coil driver would cause a lot of noise on inputs to the MS. You could lift that pin on the driver and wire it to an extra connector pin. Then you use a ground external to the MS to that pin.

[Bitsyncmaster]

I looked at your MS forum. I see if you adjust the POTs to work at high RPMs then a problem shows up at low RPM. So the circuit cannot handle the full dynamic range. I seem to remember our VR signal runs from 1 volt/PP to 100 volts/PP depending on RPM. I looked at both the 3.00 and 3.57 schematics and it uses the same circuit with the exception of the two signal transistors. XTX554 for 3.00 and PZT2907 for the 3.57. The specs look a little better on the XTX554. So maybe that is making your 3.00 handle the full dynamic range. It could just be the new PCB layout causing problems. SMT parts work better than leaded components since the inductive leads are eliminated. So the new board should theoretically have less noise.