AEM EMS-4 fuel injector impedance/deadtime questions

Hello all,

I am working on a customer 1970 Porsche 911 with a hot-rod 2.7-liter engine with higher-compression/cams (not sure which ones), twin spark plugs per cylinder (using a distributor), PMO ITBs and an AEM EMS-4 engine management system. It runs well while driving and especially at larger throttle openings, but at idle and low throttle openings, there is a lot of unburnt fuel going through the engine, our exhaust gas analyzer shows well over 2000 ppm HC at times! Engine is good mechanically and secondary ignition scope pattern of both ignition circuits is good. Fuel pressure is where it should be at about 43 PSI at idle.

After much trial and error with connection issues with the AEM EMS-4 ECU, I was able to connect with the ECU, and I was able to talk to the guy who built and tuned the car ~10 years ago. He sounds like a knowledgeable tuner/builder, said he does mostly MOTEC systems these days, but he said he did 3 different customer 911s with the AEM EMS-4 during the last decade and had numerous issues with the system and with their tech support. One of the first things I noticed on the car was a mystery electronic box on the firewall, turns out it is an injector resistor unit from a late '80s/early '90s Honda Accord! The builder claims it was needed because the fuel injectors used were on the borderline of AEM's impedance specs (which are 8.0 ohms per the EMS-4 manual) and it blew some injector drivers.

However, the fuel injectors used are Bosch part # 0 280 150 786, which are the factory-installed injectors in 3.6L Porsche 964 and the early 993 series. These are NOT low-impedance injectors, I measured 15.9 ohms on each injector, and the Honda resistor pack adds a further 12 ohms! I have not yet investigated exactly how the resistor pack is wired into the injector circuit, but the builder stated he uses 3 of EMS-4's 4 injector drivers and runs the injectors in "semi-sequential" mode using a 60-2 crank trigger (no cam sync).

I decided to check the current ramp of one of the fuel injector circuits using our Picoscope, see the first attachment of the injector waveform of the hot rod 2.7L 911 idling with the engine fully warm. I have never seen an injector current waveform with such a pronounced "hump" before the pintle fully opened. Compare this to the second waveform (third attachment) of a stock 1993 Porsche 964 that we also have in the shop, uses the exact same fuel injectors with no additional resistors, but has full sequential injection from the factory (Hall sensor in distributor), this one is at idle after cold start, and looks like every other injector current waveform that I have seen. Please also note the approximate "dead time" before the injector pintle is fully open---I forgot to set the Pico rulers in the first screenshot, but it can easily be seen.

The second screenshot is from the AEM EMS-4 "trims" menu, which shows the injector dead time/primary voltage offset curve at the lower right, and the live dead time reading is highlighted in red by the AEM software at 0.53 ms, which appears to be far less than the actual measured pintle opening time of 1.5-1.5 mS for each.

My main question is, could such high impedance in the injector circuit (presumably almost 28 ohms per injector) be the cause of the pronounced "hump" in the first waveform? I know that Ohm's law states that increased resistance will reduce current flow, but it would it not also cause it to be more difficult for the injector pintle to lift off its seat?

I am tempted to bypass the resistor pack, but I am leery of doing so based on the builder's comments about blowing injector drivers---I will check with AEM tech support, but the high-impedance resistors and the extra resistor pack do not make sense to me. Any comments from those with experience in the matter, and if running batch vs semi-batch vs full sequential has any effect on the electrical load of the injector driver transistors.

I am hoping that removing the resistor pack and correcting the injector dead time would give the injectors a better chance at achieving full atomization at idle, and while I know the entire fuel map will need to be re-done, I can start by doing the math and applying a global percentage of pulsewidth change across the full fuel map to keep the overall air-fuel ratios intact.

Chris,

What you were told about the injector drivers failing without the resistor box is correct. What you're seeing is a series of compromises due to a low quality engine management system combined with inefficient old school injectors which draw more current than modern offerings.

I respect your desire to try to optimize the vehicle as is, but don't feel you'll achieve a satisfactory result with what's in front of you. You seem detail oriented and have the equipment to determine proper engine operation, so I feel your time and the customer's interest will be better served by installing a proper engine management system. While at it, I would also switch to modern injectors (Injector Dynamics) which come with excellent injector characterization data and will offer greater precision in fuel delivery.

Hi Mike,

Thanks for the reply. I think I am already on that track with the customer, I suggested a Haltech Elite 750 and modern fuel injectors (will look into Injector Dynamics, thanks). I do not yet have experience with the Elite series, but everything I have seen and read here and elsewhere suggest it is a great entry-level ECU and Haltech's support is quite good, even in the States. And perhaps I'd be able to easily connect to the ECU and get the basic datalogging to work, unlike the current obsolete AEM crap!

This 911 has a simple setup, no idle control valve needed because it doesn't have A/C or power steering or many electrical loads on the alternator. It is currently running Alpha-N via TPS and IAT only, seems to work well with these throttle bodies. The previous tuner did install a vacuum manifold with hoses going to each intake runner, but its only current function is for the fuel pressure regulator (a blended MAP/Alpha-N mode might be overkill for this application).

The current setup is "semi-sequential" batch-fire mode using a 60-2 crank pulley trigger (via only 3 of the 4 injector outputs of the AEM EMS-4, I see the Elite 750 has 6 injector outputs), no cam sync really needed because there are no knock sensors, and while one of the main goals is to improve fuel atomization at idle and low speeds, I'm not sure fully sequential injection would provide much extra benefit because the setup is still limited by fuel injector placement.

Attached is the only pic I took of the engine bay, you can see the injector placement is far above the intake valves (possibly almost 6 inches), which is great for WOT, but not so much for atomization with a mostly-closed throttle body (injectors are also above the throttle plates!), which I think is much of the current issue. So much fuel condenses on the intake port walls at idle/low load that after engine shutoff, there is white smoke emanating from the intake ports for ~5-10 minutes from that fuel vaporizing if the engine is fully warm, which would explain the "vapor lock"/hard starting if the engine is restarted during that period. That said, the previous tuner seemed to nail the transient throttle settings, throttle response is excellent as one would hope for a hot-rod 911 engine with ITBs---I'll copy as much of that as I can! :)

So hopefully better control of modern injectors with well-known characteristics will get this engine running like it should at all RPM, as well as saving it from bore wash!

With the injector outlet above the the throttle plate, the fuel pressure regulator should be referenced to ambient, not the manifold pressure. The idea of referencing to manifold pressure is to keep the fuel pressure differential (fluid pressure in the fuel rail vs. pressure at the injector outlet) constant. This keeps the flow rate linear with injector outlet pressure.

Hi David,

That's a good point---the car is no longer at the shop to double-check, but now that I think of it, the fuel pressure bumped up while snapping the throttle as is normal with a manifold-referenced fuel pressure regulator, which prompted me to go back and look at some pictures of the PMO throttle bodies. While the injectors are mounted above the throttle shafts, the bores for injector nozzles exit below the throttle plates, and the vacuum ports are also mounted downstream of the throttle plates.

But, thanks for the reminder, this all helps as I plan the next steps for this car

https://empius.com/products/pmo-46mm-efi-throttle-body-kit-tumbled-aluminum-finish/

Looked at the pictures -- yes, that really below the throttles, so MAP referenced fuel pressure is correct.

You should be able to plenty of idle stability with that setup, you may need to just work on the linkage to the the left/right bank balance correct. That is a problem that is mostly apparent at idle. Have you used a Uni-Syn or Synchrometer to verify airflow in all throttles at idle?

Yes, that is another thing for the next visit. The throttles are balanced, but I noticed that the throttle levers weren't quite touching the ends of the adjustment screws while closed, so the idle would occasionally hang up when the engine is warm. This will be corrected, which will also require adjustment of the idle bypass screws and re-balancing. Idle is quite smooth, but the massive unburnt HCs/wall wetting thing seems to be a function of sub-optimal injectors and configuration.

We see the gamut of fuel induction setups at my shop (mostly Porsche/BMW), carburetors, mechanical fuel injection (which is quite similar to this car's setup in terms of primary inputs---throttle position versus RPM), Bosch CIS/K-Jet, all permutations of old and new EFI and GDI, and I occasionally dabble in standalone stuff as shown here. And an increasing amount of hybrid/high-voltage stuff. A lot to stay on top of :)

Thanks for the advice, hopefully this one will be optimized.