Dual Injector Throttle Bodies for 4V Engines - Written 09/03

Updated 01/07 – slightly revised text, added graphs with air/fuel curves at the bottom.
Updated 04/08 – flow info correction for green side flow injectors.

Summary:  Hardware used on various models, how the different OEM ECU operate the injectors, some examples of actual fuel requirements and how volumetric efficiency defines fuel requirements and therefore injector requirement.

The subject of dual injector throttle bodies and whether or not they are needed is often bought up on internet forums and discussions, so I thought I’d do a (it did initially say “small” here, but I had to delete that) piece on the way they work, and why you’d want them.

Dual injector throttle bodies have been fitted as std to the following models

  • 851 Tricolour – both Strada and “Race kit”, Red or black “top feed” injectors
  • 851/888 SP1/2/3/4 – Red “top feed” injectors
  • 888 SP5, 916 SP, 996, 916/996 SPS – Green “side feed” injectors  

The red and black top feed injectors flow about  30% less than the green top feed injectors (std on single injector 851, 907, ST2, 750SSie and 900SSie models), which I used to think flowed about  30% less than the green side feed injectors.  However, a fellow named Patrick in France who emails me about his 851 quite often emailed Marelli and was told both the green end flow (IW031) and green side flow (IW724) flow the same amount.  This was after he’d tested both side by side and found they delivered much the same fuel in much the same time, confirming his experimental result.  The green side feed injectors have been used on every 4V model since 888.  If it’s an 888 or later dual injector throttle body, it has 2 of the same green side feed injectors. 

So, if you have an old 851/888 SP dual injector throttle body with red top feed injectors, you’re way behind the eight ball from the start.  The fuel maps in the Ducati eproms fitted to these models goes to the maximum map time available at 7,000 and 8,000 RPM.  At 6,500 RPM, the 888 SP4 I reported on has a torque peak of sorts of 61 Dynojet ft-lb.  My 888 Strada out torqued that bike with just one green side feed injector, running a maximum injector pulse duration of 12.43 ms at 7,000 RPM.  If you are going to bother fitting dual injectors, you need to make sure they are the later, green side feed ones.  That or replace the older red top feed injectors with some aftermarket higher flow pieces.  There are plenty available.

The race 851 got around this by running 5 bar fuel pressure regulators.  Any P7 ECU race 851 eprom will be meant for 5 bar fuel pressure pretty much – see the 851 Racing report for a full eprom application list.  Fitting bigger injectors and running the std 3 bar fuel pressure is a better overall idea than running 5 bar fuel pressure

The red and green top feed injectors have the fuel hoses connected to them by little T piece fittings, and are then held down by fixtures to the throttle body.  The green side feed injectors sit down in the throttle body itself, which has the fuel feed/transfer ports machined into it.  The injector cleaning people call them “throttle body injectors”, as they see things from the car trade terminology.

The way the ECU runs the injectors also varies.  All factory fitted dual injector bikes apart from the 996 and Euro 996S (the one with the SPS motor) have used the P7 and P8.  The 996 and 996S use the 1.6M.

P7, P8

The P7 and P8 run the dual injectors as separate entities.  One injector per cylinder was run up to a certain cut off point then the second injector was bought in.  This cut off point is set by fuel pulse width duration, and was usually between 4 and 8 milliseconds.  It became less with each later model I believe, so that the 996SPS ran them both pretty much all the time.  Below the cut off point, one injector opened for the whole map fuel number.  Above the cut off point, both injectors ran for half the map fuel number.

On a 916SP, I know the cut off worked out at around 5,000 RPM at low/mid-ish throttle, as I had to solve a wacky running problem there once.  Turned out to be the secondary injectors had been swapped between cylinders, so the numbering was set out as 1-3-2-4, looking from front to rear, not 1-2-3-4 as it should have been.  Surprisingly, it made very little difference, just didn’t feel quite right at that throttle setting.  At WOT it was just fine.

The reason they lowered the cut off point was due to some issues they were having with niceness of the changeover I believe.  With the ECU swapping from single to dual at constant throttle under certain conditions, it would have made for some problems.  Although, the particular bikes (the big cammed pre 996SPS models) always had a reputation for some crudeness in the fuel mapping area, so many just overlooked this.

The 996SPS, on the other hand, was usually just lovely.  Set up properly, and especially when fitted with the Ultimap UM171 chip, they were always the nicest engine in the range.

The cut off fuel pulse width would be adjustable in the software somewhere (I’m sure Duane knows where) so you could modify it if you liked, but it’s beyond the reach of the average player.  This notion of primary and secondary injectors has always had an air of “real hot race” stuff, and gets talked about like it’s a very special feature, but it’s not.  It’s just the way they did it, and was probably more trouble than it was worth.  Especially on a road bike where one injector would of most of the work, and the other could lose efficiency thru lack of use.

The only engines that made real use of this feature where the late ‘90s 996 Corsa race bikes.  These ran the normal dual injectors in the throttle bodies, and a third “shower” injector above the throttle bodies.  Switching between the low injectors and shower made quite pronounced differences in torque output thru the mid range in particular.  Nothing new, lots of race car engines have been using this sort of theory for years, simply because varying injector location can have RPM range specific tuning benefits.


The 1.6M ECU doesn’t differentiate between the two injectors of each cylinder, it just fires both at the same time.  There is no facility in the 1.6M software to do anything different.  Most likely, because that’s the way Weber decided to do it.  How this firing both at once relates to/causes some 996 running dual injectors to stumble at 3,500 RPM on roll on I don’t know, but it was only 1.6M controlled bikes that suffered this.  And running them on a single injector cures it.

What the 1.6M software does have is an “injector” factor.  Somewhere in the software there is a number that tells the injector driving software how many injectors are running for each cylinder.  It is possible to make this number a fraction, not just 1 or 2, which sounds a bit wacky, but this is how the software accounts for what it does with the fuel number from the map.  Realistically and practically, it is either 1 or 2 in use.  If the number is 1 (all single injector bikes) the software divides the fuel map number by one.  For dual injector bikes, the software divides by 2.

Which leads on to what the dual injector ECU software does.  In both P7/P8 and 1.6M guises, it takes the fuel number from the map, processes it as required due to environmental trims, etc, then halve it (in the P7/P8 above the cut off) before sending that pulse width to the injector drivers.  This is quite important to understand.  The maximum fuel injector pulse duration (or fuel number) the Weber maps will take is 17 milli seconds.  This means, on a single injector engine, the maximum injector pulse duration is 17ms.  On a dual injector engine, running the original dual injector software eprom, the maximum injector pulse duration is 8.5 ms per injector.  So, any bike running dual injector software isn’t ever going to flow twice as much fuel as a single injector bike.  It can only flow the same maximum amount – actually, given the opening time of an injector is considered non flow, the dual injector bike’s available peak flow will be a little less due to the opening time being a greater percentage of 8.5 ms than 17 ms.

For those fitting dual injectors to a single injector bike, this software feature needs to be taken account of somehow.  Given most bikes that used dual injectors ran P7 or P8 ECU, it isn’t really possible to get an “off the shelf” chip to just fit and suit the application.  Most run into another problem, one that some people who fit the Ultimap (FIM) 996 UM222 single injector eprom also run into.  It surprises me just how many people will fit the UM222, having been told it’s to suit single injector, and expect it to run with both injectors still connected.  It won’t run, or very badly if at all, and smells like a big fuel pump.

Simply because the software has a “1” in the injector divider, whereas the original chip had a “2”.  So the software isn’t cutting the fuel number in half any more, and both injectors are running the fuel injector pulse duration intended for only one injector.  It may sound simple, but it catches quite a few people out.

Running a dual injector throttle body with a single injector chip does increase the maximum fuel you can inject, however.  Both injectors can run up to 17 ms, so the capacity to flow twice the amount of fuel is available.  For the people who think power is all about how much fuel you can get into an engine, this is just nirvana.  Until they actually try to run it, that is.  You’d need a whole lot of engine to need both injectors running at 17 ms.

You can’t use any more fuel than the air flowing through the engine requires for correct combustion.  Some people seem to think adding the dual injector throttle bodies is just bolt on horse power.  But in reality, it’s just bolting on another injector.  If you don’t need it (if your engine wasn’t previously running lean on a single injector) then it won’t make any difference to potential power output.  The tuning hassles will most likely limit the practical power output though.

There are practical uses for dual injectors however.  Working out when they are required is not that hard, although it does require you knowing the injector pulse durations you are running or will need to run.

The reason the dual injector thing came about is that, as engine speed rises, the time for one combustion cycle (two complete revolutions) reduces.  Easy to work out.  And, at pretty much bang on 7,000 RPM, the time for one cycle is 17 ms.  So, if you are using all the available injector pulse duration at and over 7,000 RPM, you’ll be suffering from lack of fuel.  At 11,000 RPM, the cycle time is down to 10.9 ms.  For a clearer view of this, the chart below shows this all.

The red line is the cycle time available.  It doesn’t even make it onto this graphs scale until over 4,500 RPM, such is the cycle time at lower RPM.  The pink line is the maximum injector pulse duration for a single injector (maximum 17 ms) given the limitation of cycle time at any given RPM.  The orange line is the individual maximum injector pulse duration for dual injectors (per injector) run by original dual injector software.  As you can see, it is unaffected by cycle time limitations until well over 12,000 RPM.

The green line is the maximum “equivalent” injector pulse duration if you increased the fuel pressure to a single injector from the std of 3 bar (44 psi) to 4.5 bar (67ish psi).  With this extra 1.5 bar of fuel pressure, you wouldn’t be restricted until just over 8,500 RPM.  Most of the Corsa bikes came with 4.5 bar fuel pressure regulators, and the std pumps can handle this sort of load.  So raising the fuel pressure and keeping a single injector is certainly an option.  Doug Lofgren talks about this a little more in his 955 SPS report - www.visi.com/~moperfserv/955_sps.htm

The blue and purple lines come from the 996S (Euro, SPS engine) fuel map.  The blue line is the map number, the purple line what the ECU will require per injector.  From this, there are a couple of things that should be clear.

  1. This engine could quite happily run on a single injector.  It does get close to using the maximum injector pulse duration at 11,000 RPM (also known as 100% duty cycle), but it’s probably rev limited to about that anyway.
  2. The maximum fuel requirement does not occur at the max power RPM.  It occurs at the maximum torque RPM – in this case, around 8,500 RPM.  This is something many don’t understand.  The amount of fuel required is based on how much air gets trapped in the cylinder.  More air trapped, more fuel is required, and more torque is produced.  Once the engine goes past its torque peak, however, the amount of air being trapped has to be decreasing.  If it wasn’t, the torque wouldn’t decrease.  Frictional and pumping losses have an effect here, but the volumetric efficiency has the largest effect.  Remember that power = torque X RPM.  So, even though the power output is still rising, the torque output is dropping.  And, as the air trapped per cycle is now decreasing, the fuel required per cycle is decreasing accordingly.  Not proportionally, as the less efficient the engine becomes, the more fuel it needs, but certainly decreasing at a somewhat similar rate.

This is a simple error in understanding that many people make.  They assume if the power peaks at 11,000 RPM, it needs the most fuel at 11,000 RPM.  Not true, although how much less than that required at the torque peak depends on the individual engine’s state of tune, and how well it holds its torque.  So you sort of need to know what shape torque curve to expect before you get it so you can prepare for it.  At 11,000 RPM, it will need more fuel over a given length of time, but less fuel per cycle, and that’s what the fuel map deals with – injector pulse duration per cycle.

As a side note, the comment “At 11,000 RPM, it will need more fuel over a given length of time” relates to how specific fuel consumption is expressed, and how fuel is used by an engine.  Brake specific fuel consumption (BSFC) is a commonly used engineering term.  It is quantified as the amount (in mass) of fuel required to make a measured unit quantity of power in a measured time.  For the metric system, this is kg per kW per hour.  For the imperialists, lbs per Hp per hour.  The lower the number, the higher the efficiency.  A BSFC chart looks something like a weather barometric pressure chart, and is quite interesting.  Most petrol engines are most efficient at approximately 85% of full load at max torque RPM.  At low percentage of full load (low throttle) they are generally very in-efficient.  Diesel BSFC charts are quite different, as they are far more efficient everywhere.  I would show one here, but then I’d have to explain it all, and that in itself would take a whole report.  I suggest those interested get themselves down to the local university library and check out the engine section, which from memory is at ref. 629.43 (629 something anyway).

For my 888, for instance, at 4,500 RPM (torque climbing steeply, 11.28 ms fuel number), it is using 25.380 seconds of injector time per minute (if I knew how that related to flow, I’d say so, but I don’t) to make 47 Dynojet Hp – 540.00 milli seconds per Hp per minute.

At 5,500 RPM (first minor torque peak, 11.90 ms fuel number), it is using 32.725 seconds of injector time per minute to make 64 Dynojet Hp – 511.33 milli seconds per Hp per minute.

At 7,000 RPM (the torque peak, 12.43 ms fuel number) it is using 43.505 seconds of injector time per minute to make 85 Dynojet Hp – 511.82 milli seconds per Hp per minute.

At 9,000 RPM (the power peak, 12.18 ms fuel number) it is using 54.810 seconds of injector time per minute to make 104 Dynojet Hp – 527.02 milli seconds per Hp per minute.

At 10,000 RPM (fuel number 11.66 ms), when both torque and power curves are on the downward slide, it is using 58.300 seconds of injector time per minute to make 102 Dynojet Hp – 571.56 milli seconds per Hp per minute. 

Which means that, once past the torque peak (but not necessarily the power peak):

  • the fuel required per cycle is dropping (fuel number decreasing)
  • the efficiency is dropping (milli seconds per Hp per minute increasing)
  • the overall fuel use (seconds of injector time per minute) is still increasing.

Back to the story.  The ST4S and S4R come from the factory with single injector throttle bodies, the same injector as the ST4 and S4 in fact.  Maybe with the shorter duration inlet cam they felt the high RPM operation would be a little less demanding, but I’ve seen ST4S make as much power as any equivalent 996 Strada, so I don’t see any logic in that theory.  Rev limiting them a bit lower can also avoid this issue somewhat, but that’s also a pretty good idea for any 4V engine’s general health anyway - valve collets in particular.

Update – 01/07

I thought I’d add a couple of graphs showing air/fuel ratios with RPM from a couple of bikes I’ve done, just to show what happens when they get to the point where they are injector restricted.  These are all bikes I’ve run on the dyno with some sort of altered cam timing.  One benefit of the advanced inlet cam timing we run on most engines is that it brings down the volumetric efficiency after the power peak.  This reduces the need for fuel right about where a single injector may be a problem, so it’s rather convenient.

All I’m showing here is air/fuel curves.  I do a set of runs when possible to see what the engine wants in terms of air/fuel ratio and to get a picture of the power curve.  Some of these graphs show very rich or lean mixtures – don’t worry about that.  If I showed you the power curves you’d see corresponding dips where those mixtures occur.

First up is a 916 running 112/108 cam timing and an Ultimap UM071 eprom.  Green is no added fuel, red +5%, blue +10%, yellow +15% and pink +20%.  As you can see, it’s going rich above 10,000 RPM anyway.  Adding 5% makes a difference, but then from about 10,200 RPM upward the richer curves begin to converge as we run out of injector time.  The power peaked on this bike at 9,100 RPM, with the torque curve peaking at 6,800 RPM, although it was somewhat flat to 8,500 RPM when it started falling away.

Next up is the 916 I fitted 996SPS cams to, running some sort of Ultimap UM or Megazone eprom to suit the P8 ECU.  This one had 104/107 cam timing, which is inlets advanced 15 degrees from spec.  You can see the +5% and upward curves converging around 9,000 RPM.  This bike was mapped on the race track, and started going lean on the data logs around 10,200 RPM, which is where we set the rev limiter.  On this one the power peaked from 9,100 to 9,700 RPM, with the torque peaking at 7,000 RPM and holding pretty much to 8,700 RPM

Finally a 996 running the Ultimap UM222 eprom.  This bike had the cam timing set to 108/108 as we usually do and also had one of the Gio.ca.moto 52mm (I think) full exhaust systems.  As you can see once over 9,500 RPM the curves start to converge, and there’s no available enrichment left over 10,000 RPM.  But with this setup, the power peaks at 8,800 RPM and the torque at 6,600 RPM, so it’s somewhat irrelevant to rev them over 10,000 RPM regardless of the rev limiter.

[Top Of Page]

Home | Blog | Facebook | Service Enquiry | Products | Reports | The Dyno | Disclaimer | Contact Us