Cutting out at high rpm

[Tickers]

We're having an issue with our engine dying at ~11k, and I'm wondering if anyone else has come across similar problems.

Sometimes we seem to be able to avoid the issue and reach the rev limiter, other times the car will cut. The car is coming to a slow halt, then there's a bang when the car stops.

I'm wondering if our ECU (Bosch MS4) or our CDI (M&W Pro 14) is the root cause, or if it's a wiring issue. We have a problem with the same symptoms on the dyno, but as our data logger hasn't arrived yet we can't be completely sure they're the same problem.

We'll be doing some rolling road testing tomorrow to try to find if any particular situations that cause it.

[Boffin]

What engine are you guys using?

More importantly, what coils?

[Simon Dingle]

It's a 2007 Yamaha R6 and I'm pretty sure they're the coils that came with the engine.

[Boffin]

Looking at those coils on the net (ebay/etc) how do you know that they are CDI coils?

They look near identical to the cbr 600 coils (f4 - black top denso, RR - blue top denso's), which to the best of my research and knowledge are a hei coil.

Now, a cdi on a hei coil will work, but it will hit a frequency where it all breaks down.
Perhaps this is what you are finding.

Got a spare induction ignition unit (ignitor) or someone with a LS engine (coils have in built ignitors) spare to test the theory?

[Tickers]

The service manual for the bike lists a CDI ignition system

[Simon Dingle]

Hey Boffin,

Cheers for the replies.

I've been looking a bit on the net and can't find many references to the CBR on HEI but there are quite a few CDI modules for the CBR for sale on ebay??

Having never worked with a CBR or an HEI ignition system I'm not in a brilliant position, but I thought that HEI coils had a significantly bigger top (the bit near the connector).

Tickers, is that the manual for the R6 or the CBR? Have you guys been able to replicate the problem while the car's stationary, or does it only happen at higher loads as on the dyno?

[wweissin]

Tickers,
It might be worth while to but a scope on your crank sensor to see what it is doing on the dyno. It could be possible that your crank voltage thresholds are being exceeded and not crossing a "Zero" for the ECU to recognize the tooth. Therefore loosing sync and causing the ECU to reset which would cut fuel and spark.

[Boffin]

If the service manual lists them as CDI unit's, I'd believe that over what some guy on the internet here is saying.

There is very little information that I could find on them. After lots of research, I could find one post on an internet forum that mentioned what they were. Additionally, considering the size and shape of a stock cbr ecu, I doubt there is enough size in there for the cdi module, let alone the heat.

Alex's idea sounds plausible too. Also check sync position (motec thing sorry - relative position of sync signal to the ref). Make sure it's not moving around to much, which it can is things a little loose in the top end.

[Tickers]

Running the engine up at high rpm with the computer connected we've seen that the engine is losing sync.

We've had to increase the number of teeth on our trigger wheel for the Bosch ECU, so it seems that we must be going outside the range of the sensor now it's seeing 50% more teeth.

Has anyone ever had this problem before, and if so, do you have any suggestions for higher resolution sensors that will fit in the small space available?

I'm going to ring round teams that I know are using the Bosch ECU with a Yamaha engine to see what they've done, and also give a few Superbike/Supersport teams a ring.

[Drew Price]

It's losing sync, or the crank signal is dropping out / going erratic?

Can you gap the sensor closer to the trigger wheel?

Any noise or backfeeding in the signal lines will cause weird things too, is the signal leads ground shielded all the way to the ECU?

[Kirk Feldkamp]

There a number of instances that I've run across with Motec ECU's that have warranted modifying the sensor output voltage so as to not overwhelm the ECU input. It was so common an issue that Motec added an oscilloscope to the MX00 series ECU's and created a few webinars to explain how to make things work for their products. Check out the webinars on ref/sync that they have on the website to better understand the issue. It is very very common to hear about the older Busa engines acting like they had a *hard* rev limiter around about 7000 rpm with the OE trigger wheel/sensor once you added a Motec. The solution was the modify the voltage inputs to an acceptable level and to filter things. It wouldn't surprise me if this same sort of thing was happening on an ECU that perhaps wasn't designed to go to relatively high RPM's like on a bike engine. Probably worth having a talk with Bosch about it.

-Kirk

[VFR750R]

Try resistor spark plugs if you don't have them installed already.

If you are using a bosch ECU, seems like they are happiest with a 60-2 wheel and one of the bosch inductive sensors (which most are relatively cheap). Our grand-am engines use a bosch sensor that is found at the bosch-motorsport website.

[wweissin]

If you increased the number of the teeth I would be very inclined to say that you are saturating the input on the ECU with the voltage. A common way to fix this is to attenuate the signal with a resistor. It can be a bit of trial and error or you can do the math. I would recommend starting around the 1k ohm mark and go from there.

I ran in to this specific issue with the new BMW SBK in AMA on a Pectel ECU. It would get close to the rev limit then just lose sync because of the crank sensor voltage. I ended up using 14.7k ohm of resistance to get the bike to run all they way to redline. And this is with a 60-2 wheel at well over 11k RPM.

[Tickers]

The ECU monitor is showing a lost sync, with the crank sensor dropping out (I haven't seen this with my own eyes, it's what the engine manager has reported)

The gap is currently 0.4mm, with the ECU manual recommending 0.8 ± 0.3. The signal wires are twisted, which is as standard on the bike and previous cars (which have no noise issues).

Alex, do you think it's likely to be voltage saturation due to increased teeth? It's what we'd initially guessed, but we're also worried about the obvious possibility that the size of the trigger wheel (and therefore teeth) is too small for the sensor to react to the passing tooth. Unfortunately the space in an R6 engine gives no room for increasing the size of the trigger wheel.

To attenuate the signal, where is the resistor placed into the circuit with the two-wire sensor?

[Kirk Feldkamp]

My understanding is that it's not the gap that necessarily has the largest effect on voltage... but more the speed with which the teeth pass the sensor (for a mag sensor... aka variable reluctance... aka "VR"). Basically, I think the voltage corresponds to how fast the magnetic field changes. I'm sure someone will correct me if I'm wrong. Larger wheel OD's and higher RPM's would then tend to produce higher voltages with mag (VR) sensors. I don't believe increasing the tooth count will have an effect on the voltage... just how many times it crosses the 0V level per revolution.

You can always chuck a trigger wheel up in a lathe and play with the sensor gap, tooth count, speed, etc. to prove this to yourself too. All you would need in an oscilliscope!

[Jimmy_seven]

How thick is the trigger wheel you're using? With lots of teeth and a small OD the sensor can have trouble distinguishing between teeth. Using a thicker disc can help. I think the only way to really find out what is going on is to get a scope on it and have a look at the signal.

[Kirk Feldkamp]

For reference:

h t t p : / / www . megamanual . com / ms2 / pickups . h t m

[VFR750R]

I forgot to mention before, make sure your speed sensor is not vibrating or moving around. Make sure your bracket is stiff. A vibrating sensor can double count teeth, and confuse the ECU.

Also, I'm not believing that an inductive sensor can't handle the speed. Most can register frequencies of 12-15kHz, which is waaay higher than what you're talking about. Make sure your tooth depth is deep enough, depending on the sensor, at least .150" but more won't hurt anything.

I agree on the thicker disk. Bosch recommends 5mm or thicker for their sensors. Don't forget that crank thrust and/or deflection may mean the wheel changes gap or axial location. I've seen dampers off the nose of a V8 engine deflect .025" radially using high speed video!

[Steeeeeeeeeeve]

This may be unrelated as we run a WR450 with a custom trigger wheel, but we had a problem where our sensor just wasn't working at a high enough frequency with our old 60-2 wheel. We went down to 9 or 12 teeth and that solved our problem with the high RPM cutout.

[Jimmy01]

We have had a similar issue in the past. R6 with CDI and Autronic ECU. Our issue was related to the timing of tooth pulses - Or sync (cam) signal was occuring at almost precisely the same time as the crank signal. The engine wouldn't rev past 7/8000 rpm but was immediately fixed when we moved the position of the crank sensor.

The other option relates to your trigger wheel dimensions. Thickness as others have mentioned may be the problem but the ratio of tooth width to air gap is also very important. This Honeywell document has info about trigger dimensions that they recommend for their GT1 sensors.
Honeywell GT1 (http://www.pe-ltd.com/Downloads/hall_effect_GT1.pdf)

[VFR750R]

I don't doubt that some issues were fixed by some of you guys by changing wheels, but I had a cup engine running to 10,000 with a 60-2 last year with no issues and Bosch ECU.

I suspect a single may be more susceptible to issues due to the single power pulse, and high variability in angular speed. But...that's exactly why Bosch systems use such a high resolution wheel, so it can capture the true crank position for most accurate ignition and injection timing. A cup engine getting into the rev-limiter at 10,000RPM could certainly create havoc on the instantaneous RPM signal, but that's precisely why a high resolution wheel is desired.

bosch-motorsport . com/en-US/literature/en-US/0261210136_DataSheet_enUS_T6821393291.pdf

Max speed is 15kHz. With 60 tooth wheel, that corresponds to 15,000RPM.
Now, the wheel diameter is important. When going to a smaller tooth count it creates wider gaps, and if the wheel is too small, the gaps will be too small.

[Gareth Gwilliam]

Restarting an old thread I'm afraid.

I've taken over from Tickers investigations, we have been looking into the problem for a while know and speaking with Bosch and a few other experts in the industry about what the source of the problem may be.

We are awaiting delivery of a new logging scope so soon I will have some traces to show.

VFR750R, the suitable rpm values you suggest tie up with what we calculated, and the 60 tooth 10k rpm being successful suggests that the ECU is more than capable of handling the input signal frequency with our 36 tooth disk. This all but rules out the problem being that the ECU is incapable of handling signals at that frequency.

Does anyone know what conditions the Bosch ECU (or another ECU) is looking for in order to "see" the tooth as a signal? I have been told that it is the falling voltage passing the 0v threshold which is "seen" by the ECU as the trigger.

Is there likely to be some form of hysteresis boundary to counteract small imperfections in the disk which could cause small additional undesired peaks?

We've some modified disks made up to try and create a signal from the stock sensor more like what we think the ECU will expect to see from a Bosch sensor and disk of the correct thickness, and are in the process of testing these to see if we see any improvements.

We have heard from some teams that the stock sensors can be a bit hit and miss, we are concerned this may be a placebo; most of the experts we have spoken to find it unlikely that the stock sensor would struggle with the new higher frequency requirement.

Does anyone have any experience or views they could add to that?

Oh and lastly, as many Yammy users may be aware, finding who makes some of the electronics for the bike is something of a mission, finding info on the components is near impossible. We suspect the stock sensor to be made by Denso (as the matching disk is stamped Denso), does anyone have any data sheets for the sensor? It would be nice to see what the sensor is rated for by the manufacturer.

Thanks,

Gareth

[VFR750R]

Without going back and re-reading everything, have you looked at the coil dwell times?

[Gareth Gwilliam]

Originally posted by VFR750R:
Without going back and re-reading everything, have you looked at the coil dwell times?

In short, No.

Could you explain why you feel this may be part of the problem, it's something we haven't even looked at as far as I am aware.

[VFR750R]

usually when I've seen misfires created by the coils, it's due to high cylinder pressure and too low dwell time. And, the Bosch ECU's have a table for both RPM and load (throttle position), so the coil dwell might be too short at your high rpm. It could also potentially be too high, causing some weird results. Make sure you are in the operating box you think you are, and that the dwell seems reasonable.

If you don't have the tools to measure coil primary current on the engine to develop a coil map, you can do a primitive map by starting low and moving the dwell up slowly until you get no misfires. If you do this around peak torque, you can use the same dwell pretty much everywhere. The bosch ECU's have a table where you can lower it (optimize it) at many operating points so that the coils can run cooler (less average dwell), and coils pull a lot of current, so long dwells will result in higher amp load on your battery/alternator. Generally speaking the dwells are longer at low rpm and start to shorten after peak torque. Low load (low TP%) also requires less dwell.

Or, you could steal some pickup truck LS coils from one of your team-mates (or ebay), and wire them direct to the bosch (no amp required, they have internal transistors), and mega-squirt forums have lots of information about appropriate dwell for the different part number LS coils and wiring. Hint: Dwell for LS coils are generally in the 3-5ms range.

Something else I just thought of that may cause misfiring is a poor end of injection timing, or a step in injection timing. The Bosch function definition manual is spectacular in my opinion. It should exist as a PDF in the directory you've installed modassport. If you can record data where you are having misfiring, you should be able to look for something toggling, an error code, etc.

[Gareth Gwilliam]

Originally posted by VFR750R:
usually when I've seen misfires created by the coils, it's due to high cylinder pressure and too low dwell time. And, the Bosch ECU's have a table for both RPM and load (throttle position), so the coil dwell might be too short at your high rpm. It could also potentially be too high, causing some weird results. Make sure you are in the operating box you think you are, and that the dwell seems reasonable.

If you don't have the tools to measure coil primary current on the engine to develop a coil map, you can do a primitive map by starting low and moving the dwell up slowly until you get no misfires. If you do this around peak torque, you can use the same dwell pretty much everywhere. The bosch ECU's have a table where you can lower it (optimize it) at many operating points so that the coils can run cooler (less average dwell), and coils pull a lot of current, so long dwells will result in higher amp load on your battery/alternator. Generally speaking the dwells are longer at low rpm and start to shorten after peak torque. Low load (low TP%) also requires less dwell.

Or, you could steal some pickup truck LS coils from one of your team-mates (or ebay), and wire them direct to the bosch (no amp required, they have internal transistors), and mega-squirt forums have lots of information about appropriate dwell for the different part number LS coils and wiring. Hint: Dwell for LS coils are generally in the 3-5ms range.

Something else I just thought of that may cause misfiring is a poor end of injection timing, or a step in injection timing. The Bosch function definition manual is spectacular in my opinion. It should exist as a PDF in the directory you've installed modassport. If you can record data where you are having misfiring, you should be able to look for something toggling, an error code, etc.

Thanks for the response, it is most appreciated. I'm not fortunate enough to have witnessed the problem first hand on the dyno (when the laptop is connected). But I am told we loose camshaft-crankshaft sync when the engine dies.

I'm very much not an expert in this area, I'm more of the teams general problem solver cause I'll give anything a go, so forgive me if the following sounds a little amateur.

Would I be correct in thinking that you are suggesting that the problem may not relate to the crankshaft signal, but is that the ignition settings in the ECU are causing a misfire; leading to the engine cutting out (which would also cause a loss of sync?).

[VFR750R]

Well, loosing sync could be a number of things. Sounds like the crank sensor has been investigated, but make sure the cam tooth is appropriately sized and timed for what it is expecting to see. Also, once the engine is running, you should be able to unplug the cam sensor which should force the ECU to lock the crank position wrt to TDC cyl 1 (it won't resync based on cam input). If the mis-fire goes away, then you will have narrowed the problem to the cam sensor or related items or interference.

Make sure it's also not something simple like low voltage to the coils or injectors. As speed goes up, average current goes up, and if all the coils are pulling from a common power source, the overlapping demand could cause issues. Make sure you have a good grounds too, and I imagine it's been checked, but it can't be overstated. Good grounding techniques should be used too. Make sure all grounds go to same place and beware of ground loops where voltage drops in ground wires (which are usually as high as voltage drops in positive wires), cause voltage differentials to sensors and other milli-volt sensitive items. When in doubt get out a multimeter and physically check voltage in many areas. oscilloscopes are great for this too while the engine is physically running.

An oscilloscope is exactly what you need to verify crank and cam signals as well. I'm sure even if the team doesn't have one, some science or engineering department will have one to borrow.

[Andrew Richards]

Google search "cdi dwell" time and you will rule out ignition calibration values as an issue. First it sounds like you need to verify that you are indeed losing sync, put the parameters phsok_b,phsokset_b, and synclost_c into your experiment window. Verify that the phsok parameters are reading TRUE, also put in rev_max(this should be your peak engine speed). Then take the engine to the speed that you are having the issues and verify that the phsok parameters go to FALSE and that synclost_c begins counting the number of lost phases. This will basically tell you that something is going wrong and you don't have something silly like a rev limiter set up. It is likely that there is something wrong with your crank signal. I say that because the cam signal is hall effect (more robust) and it only has one tooth to count. Also keep in mind what is happening when you are losing sync, the ecu is not seeing the camshaft signal after the falling edge of the second tooth after the gap on the crank signal. It is my assumption that the trigger wheel is in need for a redesign, however there are some things you can try before that has to be done:

1.Put an O-Scope on the crank and cam signal to verify they are lining up (again second tooth after the gap on the crank should coincide with the cam tooth.)
2.Try an inline resistor on the + side of the VR crank signal, try a 1K at first, if it doesn't work then try a 5K then a 10K.
3. Play a little bit with teethdebounce2, and gaptolerance. Bosch says not to so that just means you shouldn't change them to much from what they are. http://fsae.com/groupee_common/emoticons/icon_biggrin.gif

Good Luck.

[Gareth Gwilliam]

VFR750R and Andrew, thanks both for your suggestions.

It will probably be a little while until we are able to get to test a few things on the dyno. Knowing we have this issue we have tuned our engine more towards the mid rev range this year than normal, getting this mapped ready for our competitions is currently taking priority.

We have put in place a rev limiter in order to stop the engine cutting out during events, the problem seems to only happen at high rpm or rapidly accelerating rpm. Obviously this (hard and soft limiter) would be turned off/raised before running any tests of this nature.

Let me explain where we are now. We have produced a redesigned trigger wheel already through a bit of trial and error which we don't know if it will help or not. The plan is to run that in an engine as soon as possible, logging its trace to see if we can identify the problem if it is still there, we want to produce a rig to spin the disk up to around 15k if we are struggling to find anything. If nothing else this would allow us to discount the trigger signal as being the problem.

As for the resistor thing, been there and got the T-shirt I'm afraid; this was one of the first things we tried a few years back. We have actually found (through some of my initial tests with a disk and sensor set up on a lathe) that the stock sensor output voltage seems to be lower than the recommended Bosch sensor, so the reason for doing this seems void (unless there is another reason for doing this other than bringing down the signal voltage).

Thanks for the info on the Labels to try and log during dyno runs, I'll hold off on the TEETHDEBOUNCE2 setting until I have to I think.

Its probably going to be a week or two until I get a chance to look into much of these on the dyno, so if anyone has any further suggestions carry on posting!

Thanks,

Gareth

[VFR750R]

If the crank signal is weak, your wheel may not be iron-y enough http://fsae.com/groupee_common/emoticons/icon_smile.gif, or your gap may be too big.

[Gareth Gwilliam]

Hi,

We think we may have fixed it.

The problem appeared to be that the sensing element of the sensor was too large pick up only a single tooth.

Therefore the rising edge of one tooth would be cancelled out by the falling edge of another producing a rather odd waveform, and also reducing the amplitude of the majority of the signal (all apart from the missing teeth area).

What we had been looking for in the waveform was more the symptom rather than the cause. Another team at FSG told us that they had similar issues and fixed the problem by making the end of the stock sensor smaller by machining it down to about 1mm wide. Thanks to the guys at CAT Racing!

We're not certain its worked yet; a track test is needed yet to be sure, but it revved freely to 14000 rpm on the stands even with quick throttle application. We haven't been able to do that for a while....Fingers Crossed.

[VFR750R]

Great to hear!! You may also consider a 36 tooth wheel, which bosch also supports, or some of their other sensors. That way its not band-aided by a machined sensor which may impact reliability, or be sensitive to sensor rotation, or not be the same if you have to replace the sensor.






Hi,

We think we may have fixed it.

The problem appeared to be that the sensing element of the sensor was too large pick up only a single tooth.

Therefore the rising edge of one tooth would be cancelled out by the falling edge of another producing a rather odd waveform, and also reducing the amplitude of the majority of the signal (all apart from the missing teeth area).

What we had been looking for in the waveform was more the symptom rather than the cause. Another team at FSG told us that they had similar issues and fixed the problem by making the end of the stock sensor smaller by machining it down to about 1mm wide. Thanks to the guys at CAT Racing!

We're not certain its worked yet; a track test is needed yet to be sure, but it revved freely to 14000 rpm on the stands even with quick throttle application. We haven't been able to do that for a while....Fingers Crossed.