Haltech fuel injection

Competition engines and ancillaries - general discussion
Post Reply
Guy Croft
Site Admin
Posts: 5039
Joined: June 18th, 2006, 9:31 am
Location: Bedford, UK
Contact:

Haltech fuel injection

Post by Guy Croft »

This thread is solely for a detailed description of components, mode of operation, calibration of Haltech fuel injection by a member who is intimately familiar with it.

Read 'referring to vendors' at http://guy-croft.com/viewtopic.php?t=429
before posting and pay particular attention to copyright.



GC
Last edited by Guy Croft on December 14th, 2006, 9:36 am, edited 1 time in total.
petert
Posts: 61
Joined: July 13th, 2006, 12:38 pm
Location: Sydney, Australia
Contact:

Haltech E6X

Post by petert »

I started working with Haltech ECU's in approx. 1990 when we developed a fuel injected single cylinder 34cc engine for the Australian Mileage Marathon competition. We set several records along the way, achieving 875mpg and 2400 mpg in our two and single seater cars respectively. Spraying 3 to 8ms of fuel @ 3 Bar into a 34cc cylinder twice per cycle, was, and still is, a difficult task to achieve.

Understanding the delivery of the fuel requirements into a cylinder is probably the most difficult thing to comprehend. Once you've got this under your belt, and understand the differences between multipoint, batch, sequential and staged injection, you're well on your way.

It will take quite some time to put this together, so please bear with me. I have started by taking some screen shots of maps and setup screens from a Haltech E6X, set up on a warm 1.9L 16V Peugeot Mi16 engine. The only mods. are increased CR, inlet cam and individual coils (replacing the distributor).

Inputs are being taken from the following sensors, typical of most modern ECU's:
- coolant temp
- air temp
- manifold pressure (used in this case for engine load + atmoshpere)
- throttle position
- crank position (standard Motronic 60-2)
- lambda (but turned off in this case)

Please let me know if you want me to elaborate on any point.
Last edited by petert on October 28th, 2006, 3:08 am, edited 1 time in total.
petert
Posts: 61
Joined: July 13th, 2006, 12:38 pm
Location: Sydney, Australia
Contact:

Trigger Setup

Post by petert »

There are many different ways to trigger an ECU, but which ever method you use it has to be ultra reliable. Depending on what features you want, you may need two trigger signals. In the case of Haltech, these are called "trigger" and "home". Home signals are used to tell the ECU when #1 cyl. is going to fire, for use with sequential injection and direct fire ignition.

The simplest method is to trigger off the negative side of the coil. Other methods include Hall effect triggers with alternate north/south magnets, vanes, reluctor, optical etc.

What ever the triggering method, it has to occur at least 60 deg. BTDC so the ECU has time to calculate fuel and ignition events. This particular ECU allows that event to occur between 60 and 110 deg. BTDC.

In this instance, the standard Motronic 60-2 crank trigger has been used. You'll notice that it has to count 120 teeth per engine cycle. No home trigger is used, but this arrangement still allows batch/multipoint injection and wasted spark ignition. Cyls. 1-4 and 2-3 will fire on alternate TDC's.
Attachments
trigger setup.jpg
trigger setup.jpg (31.81 KiB) Viewed 44539 times
petert
Posts: 61
Joined: July 13th, 2006, 12:38 pm
Location: Sydney, Australia
Contact:

Ignition Setup

Post by petert »

The ignition setup page will automatically pickup the trigger setup values from the "Main Setup" page. There has to be some way of verifying the ignition timing before going too much further. Haltech allows you to lock the software at 10 deg. BTDC (default value), regardless of map values, so you can verify this on an external timing mark. Adjustments are then made to the Trigger Angle value to ensure the timing is occuring at 10 deg. BTDC.

Haltech allows you to choose between Constant Duty and Constant Charge ignition types. This requires carefull consideration or ignitor failure is only flick of the starter away.

Constant Duty is for use with "intelligent" ignitors that can optimise the dwell for the coil being used. eg. Bosch 008 ignitor.

Constant Charge is for use with "dumb" ignitors that need the dwell time to be specified. This particular setup uses four Chev LS1 coils, wired in pairs. This can be done as the coils have built in ignitors. A charge time of 5ms works best with these coils. If your coil(s) is over heating you need to reduce the charge time.

Two ignition outputs from the ECU are used to fire each of the coil pairs. This makes for a very neat and uncomplicated installation.

Alternatively, you could fire a CDI unit (Crane, MSD, M&W, etc.) directly, without any ignitors.
Attachments
ignition setup.jpg
ignition setup.jpg (25.45 KiB) Viewed 44535 times
Last edited by petert on October 29th, 2006, 4:54 am, edited 1 time in total.
petert
Posts: 61
Joined: July 13th, 2006, 12:38 pm
Location: Sydney, Australia
Contact:

Main Setup

Post by petert »

The next screen that needs attention is the Main Setup. Most values are self explanitory. Load Sensing however, needs explanation. In this case manifold pressure has been chosen, rather than throttle position. Further still, there are various options of MAPSENSOR. The maximum load in this case is determined by a GM 1 Bar external MAP sensor. Other options are 2 and 3 Bar external sensors , for blown applications, or you can use the internal 2.5 Bar sensor.

For a N/A engine with good vacuum, the external GM 1 Bar sensor gives the best result as all 32 bars of variable load are spread over 100 kPa. For an engine with a good idle that has approx. 65-70 kPa of vacuum, there will be 20-22 bars to adjust. If the internal 2.5 Bar sensor is used, 1.5 Bar is allocated to boost, leaving just 0.5 Bar for vacuum. So you only get 9-10 bars to adjust. This will become more apparent when we look at the fuel and ignition maps.

If you have big cams with poor vacuum at low revs, or a single cylinder engine, then load sensing by Throttle Position is a better option. Note that the fuel maps of "Throttle Sensning" are very different in shape to "Manifold Sensing". I'll illustrate this later.

Also be aware that the ECU needs a MAP sensor regardless of which method is chosen. It uses the MAP sensor to determine the "absolute" pressure in the inlet manifold. ie atmospheric pressure +/- inlet manifold vacuum/pressure. This is necessary for altitude compensation.

Another interesting feature on this screen is "Dual Maps". This feature allows you store two sets of fuel and ignition maps. This is very handy for optimizing race fuels, LPG dual fuel installations or valette parking.
Attachments
main setup.jpg
main setup.jpg (29.33 KiB) Viewed 44512 times
petert
Posts: 61
Joined: July 13th, 2006, 12:38 pm
Location: Sydney, Australia
Contact:

Fuel Maps

Post by petert »

So now all the basics are entered, it's time to start looking at the fuel maps. If you've chosen your injectors correctly, you should have a range of approx. 3ms to 7.5ms, idle to maximum torque, for batch/multipoint, or 6ms to 15ms for sequential, for a N/A engine. Note that the bar heights will automatically double or half, depending on which injection method is prescribed.

I have illustrated the 4000 RPM map. There is a map every 500 rpm, creating a 3D matrix of load/injection time/RPM. I prefer to look at things in 2D mode, working at one RPM point at a time.

Two things to note are the yellow bar and the triangle on top of 0kPa. The yellow indicates the bar you're currently adjusting. The laptop arrow keys are used to move left/right and up/down to adjust the height of the bar, which in this case refers to injection time in ms. Page up/Page Down will make large adjustments.

The triangle represents the current bar the engine is running on. In this case, the engine is either stopped or at WOT. Hitting the "Home" key as you're tuning will instantly bring you to the triangle, or the current load/RPM.

Another usefull map is the Zero Throttle map. I use this when the engine has a reasonable cam and MAP sensing is selected. This map only works when 0% is registered by the throttle sensor. There are 16 bars from 0 to 2000 RPM, allowing the injection time to be altered. This gives a nice stable idle. Without this feature, the ECU would hunt accross a range of bars (injection times) as the vacuum varied due to reversion pulses. In this case the engine is forced to accept 3ms up to 2000 RPM, regardless of manifold vacuum.
Attachments
fuel maps.jpg
fuel maps.jpg (40.76 KiB) Viewed 44507 times
Zero Throttle.jpg
Zero Throttle.jpg (34.39 KiB) Viewed 44508 times
petert
Posts: 61
Joined: July 13th, 2006, 12:38 pm
Location: Sydney, Australia
Contact:

Post by petert »

Next task is ignition mapping. I have again chosen the 4000 map as it shows the change between light and heavy loads, as would normally be handled by vacuum advance on a distributor. Once again it's possible to vary the height of each bar which corresponds to degrees BTDC. On this map the advances varies from 28 deg. (WOT) to 38 deg. light loads. In constrast, the 1000 map is dead flat, all bars at 12 deg. BTDC.

One criticism I have of this system is that it's not possible to enter half (or less) degrees.

The car will drive quite nicely with maps like this, but the only way to really optimise the mapping is run the car on a dyno. Each bar should be adjusted whilst watching the torque readout. Fine tuning with an adjustable trimpot makes the task a lot smoother and quicker. The bars can be altered to the desired points later. Maximum torque is the point of optimum advance, not necessarily where it stops pinging.
Attachments
1000 RPM Ignition Map
1000 RPM Ignition Map
ignition map idle.jpg (41.44 KiB) Viewed 44474 times
4000 RPM Ignition Map
4000 RPM Ignition Map
ignition map.jpg (49.96 KiB) Viewed 44473 times
petert
Posts: 61
Joined: July 13th, 2006, 12:38 pm
Location: Sydney, Australia
Contact:

Tuning

Post by petert »

Using a cheap narrow band O2 sensor is one of the easiest ways to tune the engine if it's road registered. This will significantly reduce the time needed on the dyno. Wide band sensors tend to be the "in thing", but a narrow band sensor is perfectly adequate for a N/A engine.

The screen belows shows the default values for an NTK sensor. Note the stoichmetric voltage is 638mV. Correpsonding voltages are:

14:1 - 710mV
13:1 - 820mV
12.7:1 - 850mV

So you can see from the numbers above that ideally the engine should be returning a voltage between 638mV and 850mV, except on over run. I've verified these numbers with an M&W Uego.

I leave the O2 sensor disabled and go for a drive, or a friend does. I then watch the voltages on the data page trying to "rough in" the mixtures. Then I enable the O2 sensor, setting the RICH/LEAN % to 10%. If the ECU can lock in on the mixture you know you're fairly close. You'll know when the ECU goes in to closed loop mode as the injection times will fluctuate automatically on the data page.

Most modern engines keep the mixtures between 14.4 and 14.7:1 most of the time, except for full load. Performance engines are a lot more responsive if you keep the mid load mixtures around 13.5:1, richening up to 12.7:1 at full noise. You can really feel it in the foot as you pull the mixtures under 14:1 whilst you're cruising around. I'm not suggesting you can do the full load mixtures this way, but you'll be well on your way to shorter (and cheaper) dyno session.

Altering the ignition maps on the dyno will richen/lean the corresponding fuel maps, so it will need touching up again anyway.
Attachments
O2 options.jpg
O2 options.jpg (31.16 KiB) Viewed 44467 times
petert
Posts: 61
Joined: July 13th, 2006, 12:38 pm
Location: Sydney, Australia
Contact:

Load Sensing by TPS

Post by petert »

Here's a map from a 1.9L 8V Peugeot race engine that has 45mm TB's and is mapped by TPS. Notice the load is now expressed as a percentage, 0-100%. This engine has bigger injectors, almost twice as big as std., so the maximum injection time is significantly less than above. The idle injection time is only 1.68ms, which is boarder line as an injector typically takes this much time to open/close. Idle quality is less important in a race car, so in this case it doesn't matter. For a street car however, these injectors are a bit too big.

More importantly, note how the shape of the map is completely different to the MAP sensor fuel map above, even though they're both at 4000, and there are less usefull bars to tune.
Attachments
TPS Fuel Map.jpg
TPS Fuel Map.jpg (23.65 KiB) Viewed 44456 times
Post Reply

Who is online

Users browsing this forum: No registered users and 38 guests