I've put up a simple web page detailing the starting problem.
http://www.geocities.com/nsr800/999
Tuesday, September 25, 2007
Monday, September 24, 2007
Thursday, September 20, 2007
Some people are very very very important. Very.
I shot this pic a few years ago and just found it again on my photobucket account. Here we find a brilliant king of the road demonstrating his godlike parking skills at the local Lowe's. Luckily I had a camera that day. :)
I think next time I won't blank out the license plate.
I think next time I won't blank out the license plate.
Friday, September 14, 2007
Spoke with a fellow enthusiast
The bike still has trouble starting. It's not an air problem. I've ruled out fuel. Last part of the equation is spark.
On Wednesday I ruled out the battery and solenoid once and for all. Bypassed the solenoid, did a direct connect to the starter. Same result, so the solenoid isn't the problem. Connected the battery to a shop car charger/starter. Same result. Tried jumping directly off my car, same result. Battery is fine.
Other than an acute timing problem, the only two things that I concluded could be causing the problem are the ECU or the coils.
My friend Mark turned me on to Rick, another Ducatista with an 03 999. Just so happens he's also a Bimmerphile like me, he did tech work on BMWs, so he's adept at diagnostics and knows his way around a garage.
I spoke at length with Rick today. He also had a chronic starting problem with his 03. We went through the usual step by step process of elimination and arrived at ignition. He said that since I was able to connect and run the diagnostics with the VDSTS, it's not likely to be the ECU (good news). His starting problem ultimately was due to the coils, and I mentioned that my coil test indicated 800 milli-ohms, much higher than the factory specked 650 milli-ohms. Rick said he talked to a few Ducati mechanics on the mainland, including the respected BCM Ducati. Bottom line is that 999 coils do go bad.
During my research, I discovered that EVERY year from 2003-2006, Ducati updated the coils for the 999. That kind of shit doesn't happen unless there's something inherently wrong with the original design. I verified this coil succession by going to the dealer and standing over the tech's shoulder as he looked up the part numbers in the database. In the end, Rick suspected it may be a coil problem, and I concurred.
At this time, there is a set of 999 coils on ebay for about $10. I looked at it again, fortunately the seller put a large pic where the part number is visible. I looked it up and it turns out to be the outdated coils. ARRGGHH! So it's off to the dealer, I'm just going to fork out the $400 for some new coils. I really hope this solves the problem. The next step would be to replace the ECU, which I priced at $1800 at the dealer, or $1544 online.
On Wednesday I ruled out the battery and solenoid once and for all. Bypassed the solenoid, did a direct connect to the starter. Same result, so the solenoid isn't the problem. Connected the battery to a shop car charger/starter. Same result. Tried jumping directly off my car, same result. Battery is fine.
Other than an acute timing problem, the only two things that I concluded could be causing the problem are the ECU or the coils.
My friend Mark turned me on to Rick, another Ducatista with an 03 999. Just so happens he's also a Bimmerphile like me, he did tech work on BMWs, so he's adept at diagnostics and knows his way around a garage.
I spoke at length with Rick today. He also had a chronic starting problem with his 03. We went through the usual step by step process of elimination and arrived at ignition. He said that since I was able to connect and run the diagnostics with the VDSTS, it's not likely to be the ECU (good news). His starting problem ultimately was due to the coils, and I mentioned that my coil test indicated 800 milli-ohms, much higher than the factory specked 650 milli-ohms. Rick said he talked to a few Ducati mechanics on the mainland, including the respected BCM Ducati. Bottom line is that 999 coils do go bad.
During my research, I discovered that EVERY year from 2003-2006, Ducati updated the coils for the 999. That kind of shit doesn't happen unless there's something inherently wrong with the original design. I verified this coil succession by going to the dealer and standing over the tech's shoulder as he looked up the part numbers in the database. In the end, Rick suspected it may be a coil problem, and I concurred.
At this time, there is a set of 999 coils on ebay for about $10. I looked at it again, fortunately the seller put a large pic where the part number is visible. I looked it up and it turns out to be the outdated coils. ARRGGHH! So it's off to the dealer, I'm just going to fork out the $400 for some new coils. I really hope this solves the problem. The next step would be to replace the ECU, which I priced at $1800 at the dealer, or $1544 online.
Monday, September 10, 2007
Not the connectors
It wasn't the connectors. Took them all apart, cleaned them, dielectric greased them. No change.
I'm fairly certain the battery is in good condition. Nevertheless I will attempt to rule it out by using a shop battery charger/jumper during my next garage session.
And getting back to the coil testing... The factory spec is 650 mili-ohms. NO over the counter multimeter has a resolution that dips into the mili-ohm range. After some research I've discovered that is the realm of dedicated mili-ohm meters. The problem is that the cheapest mili-ohm meter out there is $350 by Extech. Might as well buy coils outright. Still have to make a trip to the dealer to find out if the updated coils for my bike are the same as the ones specked in the 749 factory shop manual.
I'm fairly certain the battery is in good condition. Nevertheless I will attempt to rule it out by using a shop battery charger/jumper during my next garage session.
And getting back to the coil testing... The factory spec is 650 mili-ohms. NO over the counter multimeter has a resolution that dips into the mili-ohm range. After some research I've discovered that is the realm of dedicated mili-ohm meters. The problem is that the cheapest mili-ohm meter out there is $350 by Extech. Might as well buy coils outright. Still have to make a trip to the dealer to find out if the updated coils for my bike are the same as the ones specked in the 749 factory shop manual.
Sunday, September 9, 2007
Thinking out loud
Well the loose exhaust closer on the vertical head wasn't the cause of my starting problem. Bike is still difficult to start. Though I've gotten a better picture of the situation with yesterday's open airbox tuning. Definitely not an air and fuel problem. Must be spark/electrical.
Previously I tested spark by connecting one coil at a time to a spare plug, and grounding the plug on the engine case. Tiny spark visible.
My bike is a 2003 999, but the factory shop manual I have is for a 2006 749. Why? Basically these manuals are virtually impossible to come by, and the ONE and only source I found doesn't carry base model 999 manuals. Normally this isn't a problem for most things, as all of the procedural info is identical. However this descrepancy could make a difference in troubleshooting my starting problem.
Today I tested my coils. According to the 749 manual, the resistance across pins 1 & 15 should be 650 +/- 50 mili-ohms. This is one case where the discrepancy between the 749 and 999 manuals could make a difference. According to their respective parts catalogs, the 06 749 uses a different coil from the 03 999. Therefore, the resistance specified in the manual won't necessarily apply to my bike.
HOWEVER, further investigation reveals that every year from 2003 through 2006, Ducati changed the 999 coils, ending up with the 2006 999, which uses identical coils to the 2006 749. Ducati has been known to supercede existing parts, even within the same model year (a fact which I experienced first hand trying to get frickin' pulley tensioner nuts, which were superceded nearly half a dozen times). I will have to visit the dealer with my findings, and see if the updated coil for the 2003 999 is indeed the same as the 2006 749.
Back to my coil test. So if the updated 03 999 = 06 749 coil, and the manual spec for the 749 coil resistance is the same for the 999, then my coils might be shot based on the readings I got. The problem is I can't say for sure because my multimeter doesn't read into the mili-ohms range; it rounds to the nearest tenth ohm. So a reading of 650 mili-ohms looks like 000.7 ohms. Not very helpful. Anyway the reading I got for my coils were 000.8 ohms. I will probably have to invest in a multimeter that reads low enough to get a meaningful reading. Much better than just blindly dishing out for new coils.
In the end though, all this may be moot. My friend Mark called and told me about a starting problem he had with his 998, which had similar symptoms. The culprit was a corroded connector. So before going any further, my next step will be to trace all my connectors.
Previously I tested spark by connecting one coil at a time to a spare plug, and grounding the plug on the engine case. Tiny spark visible.
My bike is a 2003 999, but the factory shop manual I have is for a 2006 749. Why? Basically these manuals are virtually impossible to come by, and the ONE and only source I found doesn't carry base model 999 manuals. Normally this isn't a problem for most things, as all of the procedural info is identical. However this descrepancy could make a difference in troubleshooting my starting problem.
Today I tested my coils. According to the 749 manual, the resistance across pins 1 & 15 should be 650 +/- 50 mili-ohms. This is one case where the discrepancy between the 749 and 999 manuals could make a difference. According to their respective parts catalogs, the 06 749 uses a different coil from the 03 999. Therefore, the resistance specified in the manual won't necessarily apply to my bike.
HOWEVER, further investigation reveals that every year from 2003 through 2006, Ducati changed the 999 coils, ending up with the 2006 999, which uses identical coils to the 2006 749. Ducati has been known to supercede existing parts, even within the same model year (a fact which I experienced first hand trying to get frickin' pulley tensioner nuts, which were superceded nearly half a dozen times). I will have to visit the dealer with my findings, and see if the updated coil for the 2003 999 is indeed the same as the 2006 749.
Back to my coil test. So if the updated 03 999 = 06 749 coil, and the manual spec for the 749 coil resistance is the same for the 999, then my coils might be shot based on the readings I got. The problem is I can't say for sure because my multimeter doesn't read into the mili-ohms range; it rounds to the nearest tenth ohm. So a reading of 650 mili-ohms looks like 000.7 ohms. Not very helpful. Anyway the reading I got for my coils were 000.8 ohms. I will probably have to invest in a multimeter that reads low enough to get a meaningful reading. Much better than just blindly dishing out for new coils.
In the end though, all this may be moot. My friend Mark called and told me about a starting problem he had with his 998, which had similar symptoms. The culprit was a corroded connector. So before going any further, my next step will be to trace all my connectors.
Free auxiliary tank!
The 999 injectors are the shower type, so the injectors, TBs, TB linkage, and idle screw are all completely encased in the airbox. So even simple things like setting the idle require running the bike with the airbox open, a major pain in the ass because you can't just hang an auxiliary tank to do tuning.
The stock tank can be used like an auxiliary tank for tuning, just strap it backwards to the subframe. Support it with a 2x4 so it doesn't rest on the fuel lines & fuel pump harness. In this orientation, the send & return orifices on the pump will be reversed with respect to the fuel lines. However, for tuning, it can and will run fine in this configuration. Both lines meet at the junction just before entering the airbox. This junction is non-directional, so it doesn't know which side is pressurized and which is the return. Essentially the two lines form a loop between the send and return; at one point in that loop it branches off at the junction to go to the injectors.
I used this configuration to run some Seafoam through the TBs (sending billowing clouds of white smoke into the neighborhood), and subsequently reset the TPS & idle.
The stock tank can be used like an auxiliary tank for tuning, just strap it backwards to the subframe. Support it with a 2x4 so it doesn't rest on the fuel lines & fuel pump harness. In this orientation, the send & return orifices on the pump will be reversed with respect to the fuel lines. However, for tuning, it can and will run fine in this configuration. Both lines meet at the junction just before entering the airbox. This junction is non-directional, so it doesn't know which side is pressurized and which is the return. Essentially the two lines form a loop between the send and return; at one point in that loop it branches off at the junction to go to the injectors.
I used this configuration to run some Seafoam through the TBs (sending billowing clouds of white smoke into the neighborhood), and subsequently reset the TPS & idle.
Saturday, September 8, 2007
Testastretta belt tension without a Mathesis
Some notes on setting belt tension.
Beginning with the Testastretta motor, Ducati now requires that belt tension be set by measuring the belt's vibrating frequency when plucked. Normally this requires multi-billion dollar equipment, but after much research (thanks to ducati.ms), I was able to set tension with nothing more than a microphone and a notebook computer.
A simple PC microphone should be sufficient if you can eliminate all extraneous sounds from your garage. I can't do that, but luckily I have a cheap Shure dynamic mic. While condenser mics are more sensitive, they cost a mint, and more importantly, they will pick up every aural minutia--NOT GOOD if you are trying to isolate the sound of the belt. A uni-directional dynamic mic is better if you have ambient noise like me.
I used a freeware program called Audacity. For the vertical head, the belt should be plucked on the part between the static (non-adjustable) tension pulley and the drive pulley. For the horizontal head, it should be plucked on the part between the mobile (adjustable) tension pulley and the drive pulley. Set the head in question to TDC on the compression stroke (EXTREMELY IMPORTANT), mic the belt, start recording in Audacity, and pluck a few times. Use the spectrum analyzer, keep repeating the process until the correct frequency is achieved. The factory spec is 110 hz, which corresponds to A on a bass guitar.
Beginning with the Testastretta motor, Ducati now requires that belt tension be set by measuring the belt's vibrating frequency when plucked. Normally this requires multi-billion dollar equipment, but after much research (thanks to ducati.ms), I was able to set tension with nothing more than a microphone and a notebook computer.
A simple PC microphone should be sufficient if you can eliminate all extraneous sounds from your garage. I can't do that, but luckily I have a cheap Shure dynamic mic. While condenser mics are more sensitive, they cost a mint, and more importantly, they will pick up every aural minutia--NOT GOOD if you are trying to isolate the sound of the belt. A uni-directional dynamic mic is better if you have ambient noise like me.
I used a freeware program called Audacity. For the vertical head, the belt should be plucked on the part between the static (non-adjustable) tension pulley and the drive pulley. For the horizontal head, it should be plucked on the part between the mobile (adjustable) tension pulley and the drive pulley. Set the head in question to TDC on the compression stroke (EXTREMELY IMPORTANT), mic the belt, start recording in Audacity, and pluck a few times. Use the spectrum analyzer, keep repeating the process until the correct frequency is achieved. The factory spec is 110 hz, which corresponds to A on a bass guitar.
Tuesday, September 4, 2007
Ducati factory messed up my head
Here is a story about an alarming discovery I made while doing the valves. I've posted a truncated version of this story info on the ducati.ms forums.
* * * * * *
I've always had a chronic starting problem. I load tested the battery, it checked out. The starter was fine. Wiring was upgraded, but still no change. Mind you, the bike ran fine once it ever did get started, aside from the occasional stall. It just took forever to start. Something was preventing it from "catching" during startup. You need 3 things for combustion to occur: air, spark, fuel. I tested each one with the airbox open (by default this satisfies the air component). Spark checked out. Fuel ok, I could even visually verify the fuel pulses during starting. There would be an occasional backfire and/or knock from the vertical cylinder during the failed start attempts. An acquaintance opined that it seemed as if it may be lacking in some backpressure during startup. It was approaching time for a valve adjustment anyway so I put off any further troubleshooting.
* * * * * *
Fast forward a couple of weeks to when I did my valve adjustment.
With the cams exposed, I found a design flaw in my vertical cylinder head... Let's backtrack a bit. For those not familiar with desmodromic engines, desmo valves do not have a conventional closer spring. There is an opening rocker as usual, but the closing of the valve is done by a closing rocker. This closing rocker is held snug against the closing shim by a helper spring. This detail plays a key role in the design flaw I discovered.
Normally, one end of the spring hooks onto the closing rocker, the other end hooks onto a spring nook in the head.
This is my horizontal head. This is what it should look like. I'm pointing to the indented spring nook, and you can just barely see the end of the spring tang. With that indentation, there is no way that spring is going to slip off the nook:
On the other hand...
This is the vertical head. Note that there are NO indentations for the spring tangs. So basically in this case, the springs have nothing but their own tension to prevent them from slipping off their nooks. And that is exactly what happened with the right exhaust valve. The net result of this is that the closing rocker was lying limp against the cam, and at rest the opening rocker/shim was experiencing a gap equal to the sum of BOTH the opener and closer gaps (aka a loaded gap). At the bottom of the pic you can also see that the left intake and exhaust springs are also on the verge of slipping off.
People that responded to my thread in the forum said that the vertical heads they've seen DO have the spring indentations. Therefore I concluded that someone was sleeping at the factory when my head was rolling down the assembly line.
My solution was to machine out the indentations to mimic the horizontal cylinder head (sorry I forgot to take a pic before reinstalling the cams; I was very irritated by then). I verified that this is indeed how it is done at the factory--upon close inspection of the horizontal head, there are unmistakeable machining marks where the spring indentations are. So at the factory, the heads are cast so as the nooks lack the indentations, THEN the indentations are machined in after the fact.
I managed to get 3 out of the 4 vertical springs to reside in their newly machined nooks. The left intake spring seems to have been in that posture for so long that it has "remembered" this unnatural shape, so even with the indentation, it appears to want to leave its post. Time will tell.
It will be at least a couple of days before I get everything reassembled (I have a day job). Like I said in the previous post, the cooling system is half done; you have to remove the radiator to access the horizontal head. So it's a good opportunity for a coolant flush. So anyway, I won't know for at least that long if I have solved my starting problem.
* * * * * *
I've always had a chronic starting problem. I load tested the battery, it checked out. The starter was fine. Wiring was upgraded, but still no change. Mind you, the bike ran fine once it ever did get started, aside from the occasional stall. It just took forever to start. Something was preventing it from "catching" during startup. You need 3 things for combustion to occur: air, spark, fuel. I tested each one with the airbox open (by default this satisfies the air component). Spark checked out. Fuel ok, I could even visually verify the fuel pulses during starting. There would be an occasional backfire and/or knock from the vertical cylinder during the failed start attempts. An acquaintance opined that it seemed as if it may be lacking in some backpressure during startup. It was approaching time for a valve adjustment anyway so I put off any further troubleshooting.
* * * * * *
Fast forward a couple of weeks to when I did my valve adjustment.
With the cams exposed, I found a design flaw in my vertical cylinder head... Let's backtrack a bit. For those not familiar with desmodromic engines, desmo valves do not have a conventional closer spring. There is an opening rocker as usual, but the closing of the valve is done by a closing rocker. This closing rocker is held snug against the closing shim by a helper spring. This detail plays a key role in the design flaw I discovered.
Normally, one end of the spring hooks onto the closing rocker, the other end hooks onto a spring nook in the head.
This is my horizontal head. This is what it should look like. I'm pointing to the indented spring nook, and you can just barely see the end of the spring tang. With that indentation, there is no way that spring is going to slip off the nook:
On the other hand...
This is the vertical head. Note that there are NO indentations for the spring tangs. So basically in this case, the springs have nothing but their own tension to prevent them from slipping off their nooks. And that is exactly what happened with the right exhaust valve. The net result of this is that the closing rocker was lying limp against the cam, and at rest the opening rocker/shim was experiencing a gap equal to the sum of BOTH the opener and closer gaps (aka a loaded gap). At the bottom of the pic you can also see that the left intake and exhaust springs are also on the verge of slipping off.
People that responded to my thread in the forum said that the vertical heads they've seen DO have the spring indentations. Therefore I concluded that someone was sleeping at the factory when my head was rolling down the assembly line.
My solution was to machine out the indentations to mimic the horizontal cylinder head (sorry I forgot to take a pic before reinstalling the cams; I was very irritated by then). I verified that this is indeed how it is done at the factory--upon close inspection of the horizontal head, there are unmistakeable machining marks where the spring indentations are. So at the factory, the heads are cast so as the nooks lack the indentations, THEN the indentations are machined in after the fact.
I managed to get 3 out of the 4 vertical springs to reside in their newly machined nooks. The left intake spring seems to have been in that posture for so long that it has "remembered" this unnatural shape, so even with the indentation, it appears to want to leave its post. Time will tell.
It will be at least a couple of days before I get everything reassembled (I have a day job). Like I said in the previous post, the cooling system is half done; you have to remove the radiator to access the horizontal head. So it's a good opportunity for a coolant flush. So anyway, I won't know for at least that long if I have solved my starting problem.
MBP collets
Went back to school this past month! Well sorta. Bought a book off the internet.
This summer's project: doing a desmo valve adjust. Not nearly as difficult as I psyched myself up into believing. The main problem is I insisted on upgrading to aftermarket valve retainers (MBP collets). They are a bitch to work with. Though because of this nightmare, I have become quite adept at working in tiny spaces with tweezers, and am confident that a valve adjust with OE collets would be child's play.
Aren't these the cutest little cams you've ever seen?
Ended up spending more than anticipated on tools & hardware... a midlife crisis of sorts. Fuel filter has been done, cooling system half done. Still need to do brakes, oil, reset TPS. I would prefer to have done oil prior to the valves, but certain circumstances prevented that from happening.
This summer's project: doing a desmo valve adjust. Not nearly as difficult as I psyched myself up into believing. The main problem is I insisted on upgrading to aftermarket valve retainers (MBP collets). They are a bitch to work with. Though because of this nightmare, I have become quite adept at working in tiny spaces with tweezers, and am confident that a valve adjust with OE collets would be child's play.
Aren't these the cutest little cams you've ever seen?
Ended up spending more than anticipated on tools & hardware... a midlife crisis of sorts. Fuel filter has been done, cooling system half done. Still need to do brakes, oil, reset TPS. I would prefer to have done oil prior to the valves, but certain circumstances prevented that from happening.
Subscribe to:
Posts (Atom)
