HELP! Funny Starting Quirk on 885EFI 2000 Tiger

[tigerhund]

I bought a Battery Tender Plus and charged my wife's battery.  The bike starts much more reliably now, with only a short pause, if any, before the starter engages when the button is pushed.



Bike had been ridden a lot of short distances and perhaps was not getting fully charged.



Problem is still there, but it is much worse with battery not fully charged.



Hmm....I wonder if the alternator is failing?

[wonko the sane]

Hook up a volt meter to the battery, with the engine running (in a well-ventilated area of course)  my Honda service manual specifies 13.5-15.5V/5000 rpm.  My guess is that Triumph specifications would be similar.  

Perhaps somebody with a Tiger service manual could enlighten us.

[tigerhund]

...Thanks.  I have a Triumph shop manual for my bike and will give it a check...

[tigerhund]

If anyone who might be watching this knows what the voltage output/RPM should be for a (1999) Tiger, I would appreciate your input.  I searched my Triumph shop manual in detail last night and did not find it specified anywhere.  My guess is it is similar to those specified by Honda as the systems must be very similar.



Thanks,

Mike

[acollbpse]

If your battery is weak, you make the job of the EFI system much harder.  Even with a fully charged battery, the load of a starter motor drops the battery voltage significantly during cranking.  With a weak battery the voltage drop can be too much for the EFI system to overcome.  This is a problem with all EFI sytems (not just the Sagem system in the Triumph).



Here's a description of what happens during startup - forgive me if I miss something as this is from memory and its been years since I looked into this stuff.  I'll throw some numbers out that are from a completely unrelated motorcycle (single cylinder) because they are the only ones I know off the top of my head.



Lots of cool stuff happens when you turn the key on and then thumb the starter.  When you first turn the key on, you may hear the priming pulse from the EFI system running the fuel pump (this is why the indicator lights dim).  The EFI system does this to make sure the fuel system is up to pressure for starting (probably 3.0bar).  This is critical because the injectors are rated for a given mass flow rate at a given pressure.  If the pressure is too low, the EFI system doesn't know, it uses the normal starting pulse width time and you get a lean condition (and probably a no-start condition).  If your system leaked down since the last time it ran (likely on older bikes, leaky injectors, leaky pressure regulator) this priming pulse gets even more important.



After the priming pulse (about 3 seconds) the EFI system waits for you to press the starter button.  Sometimes bike manufacturers will send a start signal to the EFI system from the button to tell the computer "hey, someone hit the button", but much of the time they just let the EFI system figure out the engine bits are are moving by looking at the crank position sensor.



If things go well the crank starts moving and the computer (ECU) looks at the incoming crank sensor waveform.  Many cars and bikes now use a VR (variable reluctance) crank position sensor.  This is a great sensor because its almost bulletproof - a coil, some iron, and a magnet, no fussy transistors to get cooked in the crankcase.  It also gives you a great (cheap) way of finding a timing mark -- you change the tooth pattern on the gear-like object the VR sensor is looking at.  You can make some teeth bigger, smaller, or missing - and the ECU now knows where something is in the engine because of the resulting sensor waveform (it doesn't have to be TDC, it can be far away from TDC, you just need to know where it is when you write the ECU code).  I say gear-like because you wouldn't be able to mesh anything with a gear that had missing or different sized teeth.  They need to create this as a dedicated feature in the engine instead of looking at something in the tranny or cam drive system.



Ok so your the ECU, you've seen the crank wheel pulses going by, you're waiting with baited breath to see the timing mark (tooth pattern) and you're not doing anything else (no fuel, no spark) because quite frankly you have no idea where the engine is until you see that mark.  All you can do at this point is run the fuel pump.  So we've got a fully charged battery so everything is working super and the motor is spinning at the typical cranking speed.  You get your timing mark so now you can start computing your start fueling and ignition timing.



Since the ECU is smart it knows that it can use a special startup ignition timing profile while cranking (instead of the normal running map which likely has lots of advance in it).  It also uses the coolant temp sensor to compute all fueling based on temperature, although there's a special cranking map that gets used for starting.  Some systems use a modifier (enrichner) table added to the normal table for starting and some use a completely different table, and some just use fixed start pulse enrichment values (no RPM or TPS or other table index data) - I don't remember what the Sagem system uses but I'd guess a single row table (I can't remember what the input variable is).



So the ECU sends some fuel into the intake tract and when the crank gets to the appropriate position then fires off the number one cylinder ignition coil.  If you're lucky you get ignition and the cylinder starts to accelerate down rapidly.  If you're trying to save some money on your EFI system you can skip the camshaft position sensor and just use this acceleration to signifiy you were lucky enough to spark on the power stroke.  This gets tricky in multi cylinder engines because as much as that piston wants to accelerate down there's other cylinders that want to use that work to pump/compress air themselves, which reduces the acceleration you'll see at the crank sensor.  This scheme probably works best on 1 or 2 cylinder motors.



But lets say you've got a camshaft position sensor (I don't know if Triumph sprung for one or not) so you always know when its time to spark cylinder number 1.  Maybe you're really lucky and that first ignition pulse was all you needed and the engine ticks over beautifully.  Its possible, but not likely.  Its more likely that you'll need several full cycles of the motor to actually get it running.



Now, there's sometimes a transitional period between cranking and running as far as the ECU is concerned.  Lets say 0-300 RPM is cranking, and above 400 is running.  Why does this matter?  Well if you're using special starting fuel maps and ignition timing you need to decided when to stop using them and switch to the normal running map.  So there will be some RPM (or a range) where the motor is hopefully accelerating through where the ECU holds its breath, makes the sign of the cross, and jumps to a potentially completely different set of fueling and ignition timing numbers.  You really don't want to spend a lot of time at this transitional range otherwise you will be a very confused motor and ECU and jump between maps.



Great - that was the perfect case, now lets see what can go wrong!



So we started with a healthy battery at 14.0V, why did the instrument lights dim?  Well fuel pumps are total current pigs - they consume anywhere between 2A-7A of current when they're running.  If you did nothing else but run your fuel pump, you could drain your bike's battery in maybe less than an hour (depending on the pump and the battery and other loads of course).  So the battery voltage may drop say 1V-2V because of the fuel pump current load (you can measure it if you want by putting a volt meter across the battery and turning the key on, expensive DVMs work better or an analog meter because the priming pulse is short enough the digital meter sample time may average it out).  The lights get bright again because the ECU turns off the pump and the load goes away.  If you've got a good battery it will likely bounce right back to 14.0V when the pump goes off.



Now you hit the starter button, and you ask for lots of power in a very short time from the battery.  Depending on engine friction, compression and lots of other stuff that initial current draw is going to vary.  That initial current draw can easily be in the 150A-350A range (we're talking about the initial rotation of the starter motor rotor from rest, not even the first rotation of the engine - we might not have even engaged the sprag clutch at this point).  Normally that initial draw is over quickly (tens of milliseconds) and once things start moving the current can drop to something  more reasonable, say 50A or so.  If the engine is warm, if it ran recently and still has lube on bearing surfaces, if its worn and compression is lower (and friction too) - the cranking current can be even lower.  However if its stone cold, hasn't run in two weeks, and pretty new - you might see higher starter currents because of dry bearing surfaces, tight parts (cold) and thick oil (cold).



If you watch the voltage at the battery terminals, its going to drop significantly during cranking - going from 14.0 to 10.0V is really easy when you're sucking hundreds of amps out of the battery.  If you've got a WEAK battery it will drop even lower, 8V or so.  If you've got a really WEAK battery it may go down to 6V.  This is while cranking, this isn't the case where the engine doesn't even move (that's a dead battery - click, rur, click, click, rur...) and you can forget it if you get that sound.



What does this reduced voltage mean to the ECU?  Well remember the ECU has to open the injectors for a specific amount of time to get a specific amount of fuel.  So too little fuel will give you a lean condition and a likely no-start condition.  No problem right?  We primed ok so we have full fuel system pressure right?  Well, not exactly.  If the voltage sags enough the injectors will take longer to open (they open quickly at 14V and slower at lower voltages).  So no problem, you're a smart ECU - you'll just measure the voltage and correct for it right?  If it droops you'll add a little more time to the injector pulse during starting to keep the mixture where its supposed to be.  You can even measure the injector output at reduced voltages and get really close with a special low voltage cranking map.  Although this map is likely only good for brand new injectors -- older injectors may flow less (partially clogged) or atomize poorly (deposits) which can translate to a lean condition during starting.



If the voltage goes low enough (6V or lower) you may run into a big problem - ECU reset.  At some point the computer just gives up and starts over, thinking the power just got turned on.  I forgot to mention that the starter current can be cyclic with engine loading (compression stroke = more current = more voltage drop).  During periods of low current the voltage can bounce back up.  So the ECU probably has some kind of undervoltage supervisor part that is used to give the CPU a clean reset signal at power-up (when you turn the key on).  If you keep going above and below the threshold while cranking you'll be resetting the CPU with each cycle and its never going to start because its too busy housekeeping and booting and never has enough time to get to firing a cylinder.  You observe a crank crank crank but not even a sputter from the motor.



As if things weren't bad enough with the starter, when the ECU detects crank motion what is it going to do?  Turn the fuel pump back on!  This means that when your voltage is already sagging, the ECU is going to add the additional load of the fuel pump and drop the voltage even lower.  This is why its a really good idea to have load reduction relays in motorcycles to turn the headlights and cooling fan and other stuff off when the starter button is pressed - the ECU needs all the voltage it can get.



This also can explain why a bike that will not start with the starter button can be bump started.  If you leave the key on, the engine starts moving when you dump the clutch but you don't have the voltage drop of the starter load so the ECU has enough voltage to run properly from the weak battery.



The problem of a bike starting with the button after you bump it a bit is a little tougher to explain.  My best guess is there may be some wear in the starter motor commutator and you've got a bad or weak spot.  You press the button but nothing happens because that commutator position is worn out.  Why does this spot wear out?  Well the starter motor could in theory randomly come to rest at any position, but permanent magnet starters will tend to cog a little bit because the magnets will pull on the iron of the rotor (you've got gaps in the iron because of the winding poles).  So lets say you've got something like 12 commutator positions, you might now have only 6 park positions of the motor (assuming 2 brushers) so your chances go from random to 1 in 6.  How does bumping the bike a bit help this?  Well again I'm only guessing but its possible when you move the engine by the bumping, it might back-drive the sprag clutch on the starter motor and put it to a new position.  The next time the motor fails to spin, try the age-old remedy of tapping on the starter motor with a hunk of wood and a hammer.  It might start right up with no bumping required.  Or a weak battery could be all that's needed to make the weak spot in the commutator act up.  If it won't start, instead of bumping it try jumper cables - if it turns over right away then you know the low battery contributed to the problem.



So what else can go wrong?  Well now that you know voltage is important to CPU booting and injector timing there are some other things to check.  If you've got a bad ground connection on the ECU it can cause additional voltage drops (the size of the voltage drop depend on what the ECU is doing - spark, fuel pump, injectors, or any other loads the ECU controls such as cooling fans).  Weak battery conditions will be aggravated by bad ground connections, and ground connections can degrade with age and vibration and corrosion.  You should check the ground lugs and make sure they are clean and tight.



Ok so the grounds are good, what else?  Well one problem can be slow cranking speeds.  Although the threshold for 'tickover' may be 400 RPM and normal cranking can be 300 or so, there may be a lower limit to the cranking RPM.  Why is this?  The VR sensor may let you down at this time.  The sensor works by detecting changes in the magnetic field in very close proximity to its tip.  Very close proximity means about 1mm-2mm or so.  If you're a manufacturer you need to worry about machining tolerances and assembly tolerances on parts, and the tolerance stack-up for a VR sensor can get pretty nasty.  Depending on the design you have to control the air gap distance from the sensor tip to the gear wheel, the position of the tip (think left/right/up/down - in/out is airgap).  You need to control the concentricity of the ring to the crank, the position of the crank relative to the case (left/right/up/down and in/out), and finally the angular 'aim' of the sensor when you make its mounting features.  



So you've got to find a balance between having enough of a gap so that the gear wheel won't rub into the sensor, but not too great of a gap that the sensor won't function.  The sensor distance from the target significantly changes the amplitude of the signal that is produced.  The VR sensor is typically not excited by the ECU, it generates an AC signal in response to the moving gear teeth.  The down side to this sensor is that the amplitude goes down with increased air gap and it also goes down with slowly moving teeth.  So if you turn the crank wheel too slowly, the VR sensor may never see the teeth.  Or the ECU may see teeth but it can't find the special timing pattern because that pattern is lost due to the low amplitude signal output.  If it never finds the timing pattern you'll likely never get any spark and never any fuel.  You'll observe a weak battery cranking a motor but no cylinder firing.  Try bump starting in second or third gear, maybe that will be fast enough to send enough signal to the ECU.



So what else can go wrong?  Well the ECU may have special safeguards in it to prevent you from flooding the engine while starting.  What does this mean?  Lets say you crank for 5 seconds and the engine dumps a ton of extra fuel in for a cold start.  If the motor doesn't start, the ECU may begin to reduce the amount of enrichment on subsequent start attempts (so you don't flood it).  So by the fifth time you try to crank the motor the amount of enrichment could be much different.  Some manufacturers even have a 'flood clear' mode where you crank the motor with a wide open throttle.  In this mode the ECU will suspend injector pulses while the throttle is in that position so you can pump the excess fuel out of the motor by cranking.



So how does this manifest itself as a no-start if it sounds like you get too much fuel?  Well lets say you ran low on fuel and the fuel system got air in it.  The ECU thinks the injectors are spraying fuel so it cuts back on enrichment at start.  In reality you were 'shooting blanks' as it were because the fuel system is still purging air and not up to pressure.  As you keep re-trying to start it keeps trying to put less and less fuel in, assuming the fuel system is functioning normally.  You get a lean condition and potentially stay lean and get a no-start condition.  What should you do?  Try turning the key off, waiting 5 seconds or so, then turning the key back on again.  You'll get another fuel system priming pulse and hopefully the ECU counter on start attempts will be reset and you'll get full enrichment again.



That's all I can remember for now - hopefully some of that will be helpful.  I should add that the ECU can do a pretty good job of detecting some faults and logging them (MIL codes or malfunction indicator lamp codes).  Some codes will be special to Triumph but you might want to plug any old OBD II scan tool in - if you just read codes you might get enough info to help you find a problem (open/shorted injector or faulty fuel pump relay or something that is in the OBD II standard).  Sagem must comply with the US spec on OBD II codes for a base subset so you'll always be able to get some information.

[wonko the sane]

Thanks, acollbpse, that's what makes this site not only the best Tiger site, but probably the best site period.  :smt041

[iansoady]

Ace account of what goes on and very helpful. Many thanks.



Just looking at the manual - the Tiger does have a camshaft position sensor.

[tigerhund]

Wow and thanks.  This is outstanding education and thanks for making time to write it.



I bought a basic multi-meter and I have 14.4V at the battery terminals with the engine idling, so I think my alternator is good.



Am going to ruminate on what I just learned and we will go from there.  I am leaning toward having a weak battery, even if it is only one year old.  When I have just used the Battery Tender to top the battery off, the engine fires off immediately with no hesitation of the starter motor.



Best,

MJ

[acollbpse]

Well I hope you can get to the bottom of your issue.  I bought my '01 Tiger used because I was sick of messing with my old bike and just wanted to ride a bike instead of work on it all the time.



Good luck!