Cam timing inconsistant with each revolution
07-23-2010, 02:01 AM
#76
Inventor
Rennlist Member
Rennlist Member
The primary fact is that the Audi system works, and is proven on 928 engines.
You haven't answered my question, BTW:
Where do you see it being too loose? It keeps the belt on the crank gear when cranking, cold, hot, to 6700 rpm, acceleration, deceleration, etc. All without the band-aid anti-flap (water pump), and keep-the-belt-on-the crank-gear-during-cold-start (crank) pulley(s).
Too loose is when you see the belt flapping at high rpm, or the belt skips teeth on startup, as happens with the stock (de)tensioner system.
Too loose is when you see the belt flapping at high rpm, or the belt skips teeth on startup, as happens with the stock (de)tensioner system.
Yes, thanks.
07-23-2010, 03:54 AM
#77
Former Vendor
I'm done with you/this, for now.
If anyone ever needs an expert witness when they are suing you, when this thing fails and ruins their engine...we'll talk about it then.
07-23-2010, 12:41 PM
#78
Rennlist Member
The following description and example is purely a thought expiriment and not meant to represent reality. All opinions are invited and welcome.
There is no question that tension will effect cam timing due tot he elasticity of the belt, however this is the tension on the pulling side of the belt. The tension on the slack (tensioner) side will not affect timing. Now the tension on the slack side may effect the tension on the pulling side, but when evaluating cam timing you only need to consider the tension on the pulling side since it is what affects the stretch between the crank and relative position of the cams. In order to figure out this tension we will need to know what torque is needed to drive each relative component at the given RPM. Obviously this information is a little difficult to determine. At RPM it will be significantly different than static so I do not believe we can get where we need to be with static measurements. We really need the torque required to drive the cams say ~1500 RPM (mid rpm range of engine) or ~3000 (near red line) this force will not be the same as friction increases with RPM as well as the inertial loading of the moving valve components… One way to determine this empirically may be to have a fully assembled head off an engine and drive it with an electric motor and see how much power it takes for any given RPM. This really isn’t that simple since it would still require the oiling system to be pressurized to keep the lifters pumped up and… I would imagine all this data is out there, the trick is finding it. Gate or other belt supplies must have access to this type of information to design belts for specific applications, or engine designers must know in order to spec the correct belt size initially.
OK so I decided to play with some RANDOM numbers, and use a few generalizations:
For a typical synchronous belt system the pretension is typically designed to be between 10% and 30% of the operating load (there is of course no evidence that Porsche used this convention). If the system uses a dynamic tensioner (PKns’r) the pretension can be assumed to be additive to each run of the belt. This is not the case with a fixed position tensioner. In the case of a fixed position tensioner, the tensioner load is dictated by the fixed operating length of the belt, and the tension at the tensioner varies according to the belt elongation generated in each belt run.
Assumptions for the following examples:
The schematic frontal view I used for the belt run lengths is taken with the belt in plane with the photo.
The estimated length of each run as I measured them is ~ proportional to the actual proportions of the engine.
The power generated by the timing belt is distributed to the following components in these proportions (this is a total guess):
Cams 25% (per head)
Water pump 10%
Oil Pump 40%
Belt Units = BU
Length Units = LU
The Coefficient of Stiffness (EA) is constant over the load tensile range we are dealing with.
Stretch units = SU = BU*LU/EA
EDIT: X = tension in BU (unknown tension load generated by the tensioner in the fixed position tensioner system, this is what we are solving for to determine belt tension during operation of the engine with the fixed position tensioner) (+, positive result indicates tension load; -, negative result indicates compressive load or slack condition).
As mentioned in TB adjusting threads it is common to set the factory tension at the top of the tension range so we will assume this equates to the 30% so the working tension would be 5.3/30%=17.7 BU.
And for the PKns’r the % pretension is 3.7/17.7=20.9%
Belt units required to drive each component = BU * % of load: (% distribution required to drive each component is a completely random guess. Any suggestions here are welcome.) I ran the numbers with several different guesses this is just what I chose to use for the example.
EDIT: My initial guess had the cams carrying the majority of the load which seems much more logical to me. After playing with the munbers alittle it became evident that the further along the pulling side of the belt run the load working load was carried the worse the scenario was for the fixed position tensioner, so I randomly chose to care a large percent of the load at the closest possible position (the oil pump). This was done to illustrate that even with the short run carrying such a large portion of the load a fixed position tensioning system has dificult controlling the belt.
1-4 cam = 17.7*25% = 4.42 BU
5-8 cam = 17.7*25% = 4.42 BU
WP = 17.7*10* = 1.77 BU
OP = 17.7*40* = 7.07 BU
Dimensions of the belt runs.
Measurements where taken from Figure 1 as best I could to determine the length of each run from tangent point if the pulley to the tangency point of the pulley at the opposite end of each run.
For calculation purposes I am establishing the shortest run (crank to tensioner pulley) to be one length unit. The subsequent belt runs can be seen below in Figure 2 with each run labeled with its unit length.
The cumulative length of the runs is 1+2.58 +4.69+4.81+5.82+4.94=23.84 length units.
Figure 1: frontal view of 928 engine used to estimate belt runs
Figure 2: Unitized belt length LU. the title at the top of the pic shoud say LU not BU
Now for the dynamic tensioner system: The dynamic system holds a continuous 3.7 belt unit regardless of pulling side tension. This tension can be distributed to each belt run so that the resulting tension in each run will be the sum of the pulling tension plus tensioner applied tension. This can be seen in figures 3 (non-running condition) and figure 4 for the running condition.
Figure 3: Tension in the dynamic tensioner system with the engine not running. units are BU.
Figure 4: Tension in the dynamic tensioner system with the engine running. units are BU.
And for the static tensioner the tension is driven by belt length.
The static condition is illustrated in figure 5. From this figure the total SU can be calculated to be
SU = BU*LU since BU is constant,
SU total = BU*LU total/EA
SU total = 5.3*13.7/EA = 72.6/EA
For the fixed position tensioner SU total is a constant.
The condition during running can be seen in Figure 6.
EDIT:X represents the unknown tension generated by the tensioner position in BU.
Again SU = BU*LU however each run sees a different load this time, so
SU total = 126.4 /EA = [1*x+4.94*x+2.58 *(17.7+x)+4.69*(10.6+x)+4.81*(6.18+x)+5.82*(4.42+x)]/EA and since EA is a constant you can eliminate it from both sides.
126.4 = 1*x+4.94*x+2.58 *(17.7+x)+4.69*(10.6+x)+4.81*(6.18+x)+5.82*(4.42+x)
solving for X you get X= -0.61 or the belt run of the tensioner would actually need to be in compression to maintain the length defined by the fixed position tensioner. I played with the numbers quite a bit and with a preload based on 30% of drive load it is very difficult to not und up with the belt going loose on the tensioner in service.
Figure 5: Tension in the fixed position tensioner system with the engine not running. units are BU.
Figure 6:Tension in the fixed position tensioner system with the engine running. units are BU.
Conclusion:
Based on the above calculations teh assuming the Porsche system acts ans a fixed position tensioner system and assuming 30% pretension the OEM system will go slack at load on a running engine.
To avoid going slack, either Porsche is running the OE tensioner with significantly higher pretension than the standard 30%, and/or the Bevel washers in the factory tensioner serve a dual purpose as both a temperature compensation device and as a spring. With out this spring action it would be extremely difficult to keep the 928 belt from going slack during operation with such a short tensing run relative to the length of the pulling section of the belt.
In theory the dynamic tensioning system seem to have a few distinct advantages. First even if the pretension is below what is believed to be ideal it is highly unlikely the tensioner run will go slack. And the second is that the tension can be maintained even with a relatively short tensioner belt run section.
There is no question that tension will effect cam timing due tot he elasticity of the belt, however this is the tension on the pulling side of the belt. The tension on the slack (tensioner) side will not affect timing. Now the tension on the slack side may effect the tension on the pulling side, but when evaluating cam timing you only need to consider the tension on the pulling side since it is what affects the stretch between the crank and relative position of the cams. In order to figure out this tension we will need to know what torque is needed to drive each relative component at the given RPM. Obviously this information is a little difficult to determine. At RPM it will be significantly different than static so I do not believe we can get where we need to be with static measurements. We really need the torque required to drive the cams say ~1500 RPM (mid rpm range of engine) or ~3000 (near red line) this force will not be the same as friction increases with RPM as well as the inertial loading of the moving valve components… One way to determine this empirically may be to have a fully assembled head off an engine and drive it with an electric motor and see how much power it takes for any given RPM. This really isn’t that simple since it would still require the oiling system to be pressurized to keep the lifters pumped up and… I would imagine all this data is out there, the trick is finding it. Gate or other belt supplies must have access to this type of information to design belts for specific applications, or engine designers must know in order to spec the correct belt size initially.
OK so I decided to play with some RANDOM numbers, and use a few generalizations:
For a typical synchronous belt system the pretension is typically designed to be between 10% and 30% of the operating load (there is of course no evidence that Porsche used this convention). If the system uses a dynamic tensioner (PKns’r) the pretension can be assumed to be additive to each run of the belt. This is not the case with a fixed position tensioner. In the case of a fixed position tensioner, the tensioner load is dictated by the fixed operating length of the belt, and the tension at the tensioner varies according to the belt elongation generated in each belt run.
Assumptions for the following examples:
The schematic frontal view I used for the belt run lengths is taken with the belt in plane with the photo.
The estimated length of each run as I measured them is ~ proportional to the actual proportions of the engine.
The power generated by the timing belt is distributed to the following components in these proportions (this is a total guess):
Cams 25% (per head)
Water pump 10%
Oil Pump 40%
Belt Units = BU
Length Units = LU
The Coefficient of Stiffness (EA) is constant over the load tensile range we are dealing with.
Stretch units = SU = BU*LU/EA
EDIT: X = tension in BU (unknown tension load generated by the tensioner in the fixed position tensioner system, this is what we are solving for to determine belt tension during operation of the engine with the fixed position tensioner) (+, positive result indicates tension load; -, negative result indicates compressive load or slack condition).
As mentioned in TB adjusting threads it is common to set the factory tension at the top of the tension range so we will assume this equates to the 30% so the working tension would be 5.3/30%=17.7 BU.
And for the PKns’r the % pretension is 3.7/17.7=20.9%
Belt units required to drive each component = BU * % of load: (% distribution required to drive each component is a completely random guess. Any suggestions here are welcome.) I ran the numbers with several different guesses this is just what I chose to use for the example.
EDIT: My initial guess had the cams carrying the majority of the load which seems much more logical to me. After playing with the munbers alittle it became evident that the further along the pulling side of the belt run the load working load was carried the worse the scenario was for the fixed position tensioner, so I randomly chose to care a large percent of the load at the closest possible position (the oil pump). This was done to illustrate that even with the short run carrying such a large portion of the load a fixed position tensioning system has dificult controlling the belt.
1-4 cam = 17.7*25% = 4.42 BU
5-8 cam = 17.7*25% = 4.42 BU
WP = 17.7*10* = 1.77 BU
OP = 17.7*40* = 7.07 BU
Dimensions of the belt runs.
Measurements where taken from Figure 1 as best I could to determine the length of each run from tangent point if the pulley to the tangency point of the pulley at the opposite end of each run.
For calculation purposes I am establishing the shortest run (crank to tensioner pulley) to be one length unit. The subsequent belt runs can be seen below in Figure 2 with each run labeled with its unit length.
The cumulative length of the runs is 1+2.58 +4.69+4.81+5.82+4.94=23.84 length units.
Figure 1: frontal view of 928 engine used to estimate belt runs
Figure 2: Unitized belt length LU. the title at the top of the pic shoud say LU not BU
Now for the dynamic tensioner system: The dynamic system holds a continuous 3.7 belt unit regardless of pulling side tension. This tension can be distributed to each belt run so that the resulting tension in each run will be the sum of the pulling tension plus tensioner applied tension. This can be seen in figures 3 (non-running condition) and figure 4 for the running condition.
Figure 3: Tension in the dynamic tensioner system with the engine not running. units are BU.
Figure 4: Tension in the dynamic tensioner system with the engine running. units are BU.
And for the static tensioner the tension is driven by belt length.
The static condition is illustrated in figure 5. From this figure the total SU can be calculated to be
SU = BU*LU since BU is constant,
SU total = BU*LU total/EA
SU total = 5.3*13.7/EA = 72.6/EA
For the fixed position tensioner SU total is a constant.
The condition during running can be seen in Figure 6.
EDIT:X represents the unknown tension generated by the tensioner position in BU.
Again SU = BU*LU however each run sees a different load this time, so
SU total = 126.4 /EA = [1*x+4.94*x+2.58 *(17.7+x)+4.69*(10.6+x)+4.81*(6.18+x)+5.82*(4.42+x)]/EA and since EA is a constant you can eliminate it from both sides.
126.4 = 1*x+4.94*x+2.58 *(17.7+x)+4.69*(10.6+x)+4.81*(6.18+x)+5.82*(4.42+x)
solving for X you get X= -0.61 or the belt run of the tensioner would actually need to be in compression to maintain the length defined by the fixed position tensioner. I played with the numbers quite a bit and with a preload based on 30% of drive load it is very difficult to not und up with the belt going loose on the tensioner in service.
Figure 5: Tension in the fixed position tensioner system with the engine not running. units are BU.
Figure 6:Tension in the fixed position tensioner system with the engine running. units are BU.
Conclusion:
Based on the above calculations teh assuming the Porsche system acts ans a fixed position tensioner system and assuming 30% pretension the OEM system will go slack at load on a running engine.
To avoid going slack, either Porsche is running the OE tensioner with significantly higher pretension than the standard 30%, and/or the Bevel washers in the factory tensioner serve a dual purpose as both a temperature compensation device and as a spring. With out this spring action it would be extremely difficult to keep the 928 belt from going slack during operation with such a short tensing run relative to the length of the pulling section of the belt.
In theory the dynamic tensioning system seem to have a few distinct advantages. First even if the pretension is below what is believed to be ideal it is highly unlikely the tensioner run will go slack. And the second is that the tension can be maintained even with a relatively short tensioner belt run section.
Last edited by 76FJ55; 07-24-2010 at 01:58 PM.
07-23-2010, 01:32 PM
#79
Inventor
Rennlist Member
Rennlist Member
You need a tool to see that the belt stays on the crank gear, automatically? Glasses?
Shh! Even before you start. That was a preemptive Shh. Just know I have a whole bag of Shh with your name on it.
Sorry to break it to you, but you were done before you started. You pop up with figures based on inproper procedures, and follow with questions based on the factory tensioner. Even a non-expert can see you are not following along or even trying to learn how the new system works.
And you still haven't answered my question. Where is the belt too loose on the engine with a PKsn'r installed, if the belt stays on the crank gear in all situations?
BTW: here's a hint of what will be on the next test: 'belt management'
An expert on the original fixed tensioner system, on a good day. Please don't purport to being an expert on a dynamic tensioner system. There's ample evidence, provided by you, to show otherwise.
07-23-2010, 02:53 PM
#80
Inventor
Rennlist Member
Rennlist Member
Conclusion:
Based on the above calculations teh assuming the Porsche system acts ans a fixed position tensioner system and assuming 30% pretension the OEM system will go slack at load on a running engine.
To avoid going slack, either Porsche is running the OE tensioner with significantly higher pretension than the standard 30%, and/or the Bevel washers in the factory tensioner serve a dual purpose as both a temperature compensation device and as a spring. With out this spring action it would be extremely difficult to keep the 928 belt from going slack during operation with such a short tensing run relative to the length of the pulling section of the belt.
In theory the dynamic tensioning system seem to have a few distinct advantages. First even if the pretension is below what is believed to be ideal it is highly unlikely the tensioner run will go slack. And the second is that the tension can be maintained even with a relatively short tensioner belt run section.
Based on the above calculations teh assuming the Porsche system acts ans a fixed position tensioner system and assuming 30% pretension the OEM system will go slack at load on a running engine.
To avoid going slack, either Porsche is running the OE tensioner with significantly higher pretension than the standard 30%, and/or the Bevel washers in the factory tensioner serve a dual purpose as both a temperature compensation device and as a spring. With out this spring action it would be extremely difficult to keep the 928 belt from going slack during operation with such a short tensing run relative to the length of the pulling section of the belt.
In theory the dynamic tensioning system seem to have a few distinct advantages. First even if the pretension is below what is believed to be ideal it is highly unlikely the tensioner run will go slack. And the second is that the tension can be maintained even with a relatively short tensioner belt run section.
For all the grief I give GB (a self professed 'heavy hitter' in the 928 world
), he has made me define more and more succinctly -why- what was already obvious, that the Audi T/D based system works well.At high rpms, the belt is too long to be controlled by the factory tensioner, unless the pretension is very high. Still, the belt flaps as the pulling tension surpasses the static setting.
07-23-2010, 03:08 PM
#81
Race Car
I haven't done any calculations, but here are my thoughts on the stock tensioner system:
- First of all, the washers in the stock tensioner do not act like springs. They only expand slowly to compensate for engine expansion.
- Fixed tensioner systems no longer exist on new engines. Why is that?
- A belt spinning around gears/pulleys that have varying load is a dynamic situation and a fixed tensioner will never be able to keep the belt tension consistent under such conditions.
- The fact that Porsche had to add an idler pulley to control the TOOTHED sides of the belt proves that the belt can and does flap around even when the proper belt is used and installed/adjusted. I'm sure I don't have to explain why a flapping belt is a bad thing.
My car now has 15k miles on Ken's tensioner system. It also puts down ~420rwhp and has very high torque output even from 1,500RPM.
Dan
'91 928GT S/C 475hp/460lb.ft
- First of all, the washers in the stock tensioner do not act like springs. They only expand slowly to compensate for engine expansion.
- Fixed tensioner systems no longer exist on new engines. Why is that?
- A belt spinning around gears/pulleys that have varying load is a dynamic situation and a fixed tensioner will never be able to keep the belt tension consistent under such conditions.
- The fact that Porsche had to add an idler pulley to control the TOOTHED sides of the belt proves that the belt can and does flap around even when the proper belt is used and installed/adjusted. I'm sure I don't have to explain why a flapping belt is a bad thing.
My car now has 15k miles on Ken's tensioner system. It also puts down ~420rwhp and has very high torque output even from 1,500RPM.
Dan
'91 928GT S/C
475hp/460lb.ft
07-23-2010, 03:21 PM
#82
Former Vendor
Why so defensive?
You need a tool to see that the belt stays on the crank gear, automatically? Glasses?
Shh! Even before you start. That was a preemptive Shh. Just know I have a whole bag of Shh with your name on it.
Sorry to break it to you, but you were done before you started. You pop up with figures based on inproper procedures, and follow with questions based on the factory tensioner. Even a non-expert can see you are not following along or even trying to learn how the new system works.
And you still haven't answered my question. Where is the belt too loose on the engine with a PKsn'r installed, if the belt stays on the crank gear in all situations?
BTW: here's a hint of what will be on the next test: 'belt management'
You need a tool to see that the belt stays on the crank gear, automatically? Glasses?
Shh! Even before you start. That was a preemptive Shh. Just know I have a whole bag of Shh with your name on it.
Sorry to break it to you, but you were done before you started. You pop up with figures based on inproper procedures, and follow with questions based on the factory tensioner. Even a non-expert can see you are not following along or even trying to learn how the new system works.
And you still haven't answered my question. Where is the belt too loose on the engine with a PKsn'r installed, if the belt stays on the crank gear in all situations?
BTW: here's a hint of what will be on the next test: 'belt management'
So predicable. No data. Guesswork, followed by condenscending bull****.
Since you have never given me anything but the above, and have never had anything worthwhile to say...alll you do is market your stuff...never help anyone...you are now on "ignore".
Note that ignore is the root for ignorant, which seems very appropriate, for you.
07-23-2010, 03:24 PM
#83
Inventor
Rennlist Member
Rennlist Member
I thought you were done? 
Never help anyone?
I like this one: if electricity comes from electrons, does morality come from morons?
Never help anyone?
I like this one: if electricity comes from electrons, does morality come from morons?
07-23-2010, 05:51 PM
#84
Nordschleife Master
Simon, thanks for that! 
Very enlightening, and concisely and clearly done.
Very enlightening, and concisely and clearly done.
The following description and example is purely a thought expiriment and not meant to represent reality. All opinions are invited and welcome.
Conclusion:
Based on the above calculations teh assuming the Porsche system acts ans a fixed position tensioner system and assuming 30% pretension the OEM system will go slack at load on a running engine.
To avoid going slack, either Porsche is running the OE tensioner with significantly higher pretension than the standard 30%, and/or the Bevel washers in the factory tensioner serve a dual purpose as both a temperature compensation device and as a spring. With out this spring action it would be extremely difficult to keep the 928 belt from going slack during operation with such a short tensing run relative to the length of the pulling section of the belt.
In theory the dynamic tensioning system seem to have a few distinct advantages. First even if the pretension is below what is believed to be ideal it is highly unlikely the tensioner run will go slack. And the second is that the tension can be maintained even with a relatively short tensioner belt run section.
Conclusion:
Based on the above calculations teh assuming the Porsche system acts ans a fixed position tensioner system and assuming 30% pretension the OEM system will go slack at load on a running engine.
To avoid going slack, either Porsche is running the OE tensioner with significantly higher pretension than the standard 30%, and/or the Bevel washers in the factory tensioner serve a dual purpose as both a temperature compensation device and as a spring. With out this spring action it would be extremely difficult to keep the 928 belt from going slack during operation with such a short tensing run relative to the length of the pulling section of the belt.
In theory the dynamic tensioning system seem to have a few distinct advantages. First even if the pretension is below what is believed to be ideal it is highly unlikely the tensioner run will go slack. And the second is that the tension can be maintained even with a relatively short tensioner belt run section.
07-23-2010, 06:11 PM
#85
Rennlist Member
Join Date: Nov 2005
Location: Arkansas Ozarks
Posts: 106
Likes: 0
Received 0 Likes
on
0 Posts
I haven't even yet installed my Porkensioner, and Simon just makes my head hurt on a Friday afternoon when I've come home early from work and rattled ice cubes, but some things seem self-evident:
1. The timing belt is elastic and extends under load.
2. Load, and therefore extension of the belt, increases as a direct function of rpm.
3. Timing belt extension retards cam timing, right bank more than left.
4. Timing belt extension (I'm guessing very early) reaches an asymptotic limit, beyond which the next step is belt failure or tooth shredding.
5. The sole purpose of the factory thingie is to reduce tension as the aluminum block and heads expand, solely to avoid the asymptotic limit: because
5a. the only way to use a rubber belt to compress that thick stack of Belleville washers would be to use it as a spacer under a hydraulic piston.
6. The engineers at Audi may not have been fools to pass up 1970's technology.
7. Ken has actually done something to make our toys more reliable and therefore our lives better.
8. Smart S4 owners should just get together and buy him one.
9. I really need to get the 86.5 running.
1. The timing belt is elastic and extends under load.
2. Load, and therefore extension of the belt, increases as a direct function of rpm.
3. Timing belt extension retards cam timing, right bank more than left.
4. Timing belt extension (I'm guessing very early) reaches an asymptotic limit, beyond which the next step is belt failure or tooth shredding.
5. The sole purpose of the factory thingie is to reduce tension as the aluminum block and heads expand, solely to avoid the asymptotic limit: because
5a. the only way to use a rubber belt to compress that thick stack of Belleville washers would be to use it as a spacer under a hydraulic piston.
6. The engineers at Audi may not have been fools to pass up 1970's technology.
7. Ken has actually done something to make our toys more reliable and therefore our lives better.
8. Smart S4 owners should just get together and buy him one.
9. I really need to get the 86.5 running.
07-24-2010, 03:26 AM
#86
Rennlist Member
I got lost somewhere in the above theory. Whats the basis for assuming oil pump load is 40% ? It has a very low number of belt teeth in contact transmitting power to it.
Also I didnt notice a definition of what X was - just seemed to appear , and when it solved out as negative, it indicated a slack condition?
FWIW, at my local wrench recently he had an S4 undergoing a belt and WP change. Owner was very lucky - the tensioner pushrod (with tension warning blade on the end) was OUT of the arm socket, resting on the body of the arm at an angle....That would require a very slack belt event wouldnt it?
jp 83 Euro S AT 52k
Also I didnt notice a definition of what X was - just seemed to appear , and when it solved out as negative, it indicated a slack condition?
FWIW, at my local wrench recently he had an S4 undergoing a belt and WP change. Owner was very lucky - the tensioner pushrod (with tension warning blade on the end) was OUT of the arm socket, resting on the body of the arm at an angle....That would require a very slack belt event wouldnt it?
jp 83 Euro S AT 52k
07-24-2010, 03:54 AM
#87
I've been watching this thread, as all threads that come up on this subject.
First, when it comes to engine building, there is no one that does it better than Greg Brown. He is one man that, when he speaks about making a Porsche engine more than it is, I take what he says as Gospel. I would rather spend the extra $10k on a stroker build from him, than I would trying to do it myself.
On the other hand, I have been dealing with Kens products and think they are the ****. Pardon the expression, but the man knows what he is doing when it comes from the opposite spectrum.
Now combine the lack of connection between with the two minds of Greg and Ken. It kinda screws up those of us that install, and maintain these engines with the associated bits. I have played with both creations.
Greg prefers, and rightly so, the factory system. The factory system is not bad at all. Does it have issues? Yes. Does Kens? Yes.
Kens does not give us a timing belt warning. But so what, our car is the only system that gives us that. Why did Porsche give us a warning system? They were not confident that the system they made in the 1970's was able to keep a constant tension that was adequate for the engine. Granted, it works and Kens system bypasses the warning.
In the past 3 years I have installed 19 Porken tensioner set ups. I have used the 32Vr cam timing set up on most of those, and on follow ups with my customers, I have found no deviation after 1500 miles on either cam timing, or belt tension issues. Or cam timing issues. I will say that I have had more than one issue getting that damn timing belt grounded right so there are no problems with the light going off.
First, when it comes to engine building, there is no one that does it better than Greg Brown. He is one man that, when he speaks about making a Porsche engine more than it is, I take what he says as Gospel. I would rather spend the extra $10k on a stroker build from him, than I would trying to do it myself.
On the other hand, I have been dealing with Kens products and think they are the ****. Pardon the expression, but the man knows what he is doing when it comes from the opposite spectrum.
Now combine the lack of connection between with the two minds of Greg and Ken. It kinda screws up those of us that install, and maintain these engines with the associated bits. I have played with both creations.
Greg prefers, and rightly so, the factory system. The factory system is not bad at all. Does it have issues? Yes. Does Kens? Yes.
Kens does not give us a timing belt warning. But so what, our car is the only system that gives us that. Why did Porsche give us a warning system? They were not confident that the system they made in the 1970's was able to keep a constant tension that was adequate for the engine. Granted, it works and Kens system bypasses the warning.
In the past 3 years I have installed 19 Porken tensioner set ups. I have used the 32Vr cam timing set up on most of those, and on follow ups with my customers, I have found no deviation after 1500 miles on either cam timing, or belt tension issues. Or cam timing issues. I will say that I have had more than one issue getting that damn timing belt grounded right so there are no problems with the light going off.
07-24-2010, 02:36 PM
#88
Former Vendor
The following description and example is purely a thought expiriment and not meant to represent reality. All opinions are invited and welcome.
There is no question that tension will effect cam timing due tot he elasticity of the belt, however this is the tension on the pulling side of the belt. The tension on the slack (tensioner) side will not affect timing. Now the tension on the slack side may effect the tension on the pulling side, but when evaluating cam timing you only need to consider the tension on the pulling side since it is what affects the stretch between the crank and relative position of the cams. In order to figure out this tension we will need to know what torque is needed to drive each relative component at the given RPM. Obviously this information is a little difficult to determine. At RPM it will be significantly different than static so I do not believe we can get where we need to be with static measurements. We really need the torque required to drive the cams say ~1500 RPM (mid rpm range of engine) or ~3000 (near red line) this force will not be the same as friction increases with RPM as well as the inertial loading of the moving valve components… One way to determine this empirically may be to have a fully assembled head off an engine and drive it with an electric motor and see how much power it takes for any given RPM. This really isn’t that simple since it would still require the oiling system to be pressurized to keep the lifters pumped up and… I would imagine all this data is out there, the trick is finding it. Gate or other belt supplies must have access to this type of information to design belts for specific applications, or engine designers must know in order to spec the correct belt size initially.
OK so I decided to play with some RANDOM numbers, and use a few generalizations:
For a typical synchronous belt system the pretension is typically designed to be between 10% and 30% of the operating load (there is of course no evidence that Porsche used this convention). If the system uses a dynamic tensioner (PKns’r) the pretension can be assumed to be additive to each run of the belt. This is not the case with a fixed position tensioner. In the case of a fixed position tensioner, the tensioner load is dictated by the fixed operating length of the belt, and the tension at the tensioner varies according to the belt elongation generated in each belt run.
Assumptions for the following examples:
The schematic frontal view I used for the belt run lengths is taken with the belt in plane with the photo.
The estimated length of each run as I measured them is ~ proportional to the actual proportions of the engine.
The power generated by the timing belt is distributed to the following components in these proportions (this is a total guess):
Cams 25% (per head)
Water pump 10%
Oil Pump 40%
Belt Units = BU
Length Units = LU
The Coefficient of Stiffness (EA) is constant over the load tensile range we are dealing with.
Stretch units = SU = BU*LU/EA
EDIT: X = tension in BU (unknown tension load generated by the tensioner in the fixed position tensioner system, this is what we are solving for to determine belt tension during operation of the engine with the fixed position tensioner) (+, positive result indicates tension load; -, negative result indicates compressive load or slack condition).
As mentioned in TB adjusting threads it is common to set the factory tension at the top of the tension range so we will assume this equates to the 30% so the working tension would be 5.3/30%=17.7 BU.
And for the PKns’r the % pretension is 3.7/17.7=20.9%
Belt units required to drive each component = BU * % of load: (% distribution required to drive each component is a completely random guess. Any suggestions here are welcome.) I ran the numbers with several different guesses this is just what I chose to use for the example.
EDIT: My initial guess had the cams carrying the majority of the load which seems much more logical to me. After playing with the munbers alittle it became evident that the further along the pulling side of the belt run the load working load was carried the worse the scenario was for the fixed position tensioner, so I randomly chose to care a large percent of the load at the closest possible position (the oil pump). This was done to illustrate that even with the short run carrying such a large portion of the load a fixed position tensioning system has dificult controlling the belt.
1-4 cam = 17.7*25% = 4.42 BU
5-8 cam = 17.7*25% = 4.42 BU
WP = 17.7*10* = 1.77 BU
OP = 17.7*40* = 7.07 BU
Dimensions of the belt runs.
Measurements where taken from Figure 1 as best I could to determine the length of each run from tangent point if the pulley to the tangency point of the pulley at the opposite end of each run.
For calculation purposes I am establishing the shortest run (crank to tensioner pulley) to be one length unit. The subsequent belt runs can be seen below in Figure 2 with each run labeled with its unit length.
The cumulative length of the runs is 1+2.58 +4.69+4.81+5.82+4.94=23.84 length units.
Figure 1: frontal view of 928 engine used to estimate belt runs
Figure 2: Unitized belt length LU. the title at the top of the pic shoud say LU not BU
Now for the dynamic tensioner system: The dynamic system holds a continuous 3.7 belt unit regardless of pulling side tension. This tension can be distributed to each belt run so that the resulting tension in each run will be the sum of the pulling tension plus tensioner applied tension. This can be seen in figures 3 (non-running condition) and figure 4 for the running condition.
Figure 3: Tension in the dynamic tensioner system with the engine not running. units are BU.
Figure 4: Tension in the dynamic tensioner system with the engine running. units are BU.
And for the static tensioner the tension is driven by belt length.
The static condition is illustrated in figure 5. From this figure the total SU can be calculated to be
SU = BU*LU since BU is constant,
SU total = BU*LU total/EA
SU total = 5.3*13.7/EA = 72.6/EA
For the fixed position tensioner SU total is a constant.
The condition during running can be seen in Figure 6.
EDIT:X represents the unknown tension generated by the tensioner position in SU.
Again SU = BU*LU however each run sees a different load this time, so
SU total = 126.4 /EA = [1*x+4.94*x+2.58 *(17.7+x)+4.69*(10.6+x)+4.81*(6.18+x)+5.82*(4.42+x)]/EA and since EA is a constant you can eliminate it from both sides.
126.4 = 1*x+4.94*x+2.58 *(17.7+x)+4.69*(10.6+x)+4.81*(6.18+x)+5.82*(4.42+x)
solving for X you get X= -0.61 or the belt run of the tensioner would actually need to be in compression to maintain the length defined by the fixed position tensioner. I played with the numbers quite a bit and with a preload based on 30% of drive load it is very difficult to not und up with the belt going loose on the tensioner in service.
Figure 5: Tension in the fixed position tensioner system with the engine not running. units are BU.
Figure 6:Tension in the fixed position tensioner system with the engine running. units are BU.
Conclusion:
Based on the above calculations teh assuming the Porsche system acts ans a fixed position tensioner system and assuming 30% pretension the OEM system will go slack at load on a running engine.
To avoid going slack, either Porsche is running the OE tensioner with significantly higher pretension than the standard 30%, and/or the Bevel washers in the factory tensioner serve a dual purpose as both a temperature compensation device and as a spring. With out this spring action it would be extremely difficult to keep the 928 belt from going slack during operation with such a short tensing run relative to the length of the pulling section of the belt.
In theory the dynamic tensioning system seem to have a few distinct advantages. First even if the pretension is below what is believed to be ideal it is highly unlikely the tensioner run will go slack. And the second is that the tension can be maintained even with a relatively short tensioner belt run section.
There is no question that tension will effect cam timing due tot he elasticity of the belt, however this is the tension on the pulling side of the belt. The tension on the slack (tensioner) side will not affect timing. Now the tension on the slack side may effect the tension on the pulling side, but when evaluating cam timing you only need to consider the tension on the pulling side since it is what affects the stretch between the crank and relative position of the cams. In order to figure out this tension we will need to know what torque is needed to drive each relative component at the given RPM. Obviously this information is a little difficult to determine. At RPM it will be significantly different than static so I do not believe we can get where we need to be with static measurements. We really need the torque required to drive the cams say ~1500 RPM (mid rpm range of engine) or ~3000 (near red line) this force will not be the same as friction increases with RPM as well as the inertial loading of the moving valve components… One way to determine this empirically may be to have a fully assembled head off an engine and drive it with an electric motor and see how much power it takes for any given RPM. This really isn’t that simple since it would still require the oiling system to be pressurized to keep the lifters pumped up and… I would imagine all this data is out there, the trick is finding it. Gate or other belt supplies must have access to this type of information to design belts for specific applications, or engine designers must know in order to spec the correct belt size initially.
OK so I decided to play with some RANDOM numbers, and use a few generalizations:
For a typical synchronous belt system the pretension is typically designed to be between 10% and 30% of the operating load (there is of course no evidence that Porsche used this convention). If the system uses a dynamic tensioner (PKns’r) the pretension can be assumed to be additive to each run of the belt. This is not the case with a fixed position tensioner. In the case of a fixed position tensioner, the tensioner load is dictated by the fixed operating length of the belt, and the tension at the tensioner varies according to the belt elongation generated in each belt run.
Assumptions for the following examples:
The schematic frontal view I used for the belt run lengths is taken with the belt in plane with the photo.
The estimated length of each run as I measured them is ~ proportional to the actual proportions of the engine.
The power generated by the timing belt is distributed to the following components in these proportions (this is a total guess):
Cams 25% (per head)
Water pump 10%
Oil Pump 40%
Belt Units = BU
Length Units = LU
The Coefficient of Stiffness (EA) is constant over the load tensile range we are dealing with.
Stretch units = SU = BU*LU/EA
EDIT: X = tension in BU (unknown tension load generated by the tensioner in the fixed position tensioner system, this is what we are solving for to determine belt tension during operation of the engine with the fixed position tensioner) (+, positive result indicates tension load; -, negative result indicates compressive load or slack condition).
As mentioned in TB adjusting threads it is common to set the factory tension at the top of the tension range so we will assume this equates to the 30% so the working tension would be 5.3/30%=17.7 BU.
And for the PKns’r the % pretension is 3.7/17.7=20.9%
Belt units required to drive each component = BU * % of load: (% distribution required to drive each component is a completely random guess. Any suggestions here are welcome.) I ran the numbers with several different guesses this is just what I chose to use for the example.
EDIT: My initial guess had the cams carrying the majority of the load which seems much more logical to me. After playing with the munbers alittle it became evident that the further along the pulling side of the belt run the load working load was carried the worse the scenario was for the fixed position tensioner, so I randomly chose to care a large percent of the load at the closest possible position (the oil pump). This was done to illustrate that even with the short run carrying such a large portion of the load a fixed position tensioning system has dificult controlling the belt.
1-4 cam = 17.7*25% = 4.42 BU
5-8 cam = 17.7*25% = 4.42 BU
WP = 17.7*10* = 1.77 BU
OP = 17.7*40* = 7.07 BU
Dimensions of the belt runs.
Measurements where taken from Figure 1 as best I could to determine the length of each run from tangent point if the pulley to the tangency point of the pulley at the opposite end of each run.
For calculation purposes I am establishing the shortest run (crank to tensioner pulley) to be one length unit. The subsequent belt runs can be seen below in Figure 2 with each run labeled with its unit length.
The cumulative length of the runs is 1+2.58 +4.69+4.81+5.82+4.94=23.84 length units.
Figure 1: frontal view of 928 engine used to estimate belt runs
Figure 2: Unitized belt length LU. the title at the top of the pic shoud say LU not BU
Now for the dynamic tensioner system: The dynamic system holds a continuous 3.7 belt unit regardless of pulling side tension. This tension can be distributed to each belt run so that the resulting tension in each run will be the sum of the pulling tension plus tensioner applied tension. This can be seen in figures 3 (non-running condition) and figure 4 for the running condition.
Figure 3: Tension in the dynamic tensioner system with the engine not running. units are BU.
Figure 4: Tension in the dynamic tensioner system with the engine running. units are BU.
And for the static tensioner the tension is driven by belt length.
The static condition is illustrated in figure 5. From this figure the total SU can be calculated to be
SU = BU*LU since BU is constant,
SU total = BU*LU total/EA
SU total = 5.3*13.7/EA = 72.6/EA
For the fixed position tensioner SU total is a constant.
The condition during running can be seen in Figure 6.
EDIT:X represents the unknown tension generated by the tensioner position in SU.
Again SU = BU*LU however each run sees a different load this time, so
SU total = 126.4 /EA = [1*x+4.94*x+2.58 *(17.7+x)+4.69*(10.6+x)+4.81*(6.18+x)+5.82*(4.42+x)]/EA and since EA is a constant you can eliminate it from both sides.
126.4 = 1*x+4.94*x+2.58 *(17.7+x)+4.69*(10.6+x)+4.81*(6.18+x)+5.82*(4.42+x)
solving for X you get X= -0.61 or the belt run of the tensioner would actually need to be in compression to maintain the length defined by the fixed position tensioner. I played with the numbers quite a bit and with a preload based on 30% of drive load it is very difficult to not und up with the belt going loose on the tensioner in service.
Figure 5: Tension in the fixed position tensioner system with the engine not running. units are BU.
Figure 6:Tension in the fixed position tensioner system with the engine running. units are BU.
Conclusion:
Based on the above calculations teh assuming the Porsche system acts ans a fixed position tensioner system and assuming 30% pretension the OEM system will go slack at load on a running engine.
To avoid going slack, either Porsche is running the OE tensioner with significantly higher pretension than the standard 30%, and/or the Bevel washers in the factory tensioner serve a dual purpose as both a temperature compensation device and as a spring. With out this spring action it would be extremely difficult to keep the 928 belt from going slack during operation with such a short tensing run relative to the length of the pulling section of the belt.
In theory the dynamic tensioning system seem to have a few distinct advantages. First even if the pretension is below what is believed to be ideal it is highly unlikely the tensioner run will go slack. And the second is that the tension can be maintained even with a relatively short tensioner belt run section.
Although the numbers are guesses, on this analysis, it instantly becomes clear why Ken's system works on some applications and why I've never seen any need for his stuff. I've never seen or experienced the belt "flapping" that he talks about. I don't get cam timing variations. Nothing he has ever claimed or points out applies to anything I've seen....and I've seen a few 928 engines.
The entire difference in our two point of views comes from our point of reference. I use factory cam belts, tensioned to the correect specifications. I'm going to guess that Ken developed his system after seeing aftermarket cam belts in action.
There is absolutely no doubt (I've measured this, over and over again) that factory cam belts have very little stretch. They tighten to 5.0 belt units and will continue to gain tension with very little tensioner extension all the way past 10.0 belt units. Aftermarket cam belts (have not measured Roger's new cam belt, so this is not included in this discussion) will get to about 5.8 belt units and then not get any tighter than that. You can screw in the tensioner, forever and they stay right in that area. That's why I have always refered to them as "giant rubber bands".
Looking at this analysis, I see that with a cam belt that doesn't stretch very much (factory belt), the stock tensioner works, as designed, and controls both the tension and keeps changes in cam timing to an absolute minimum.
You can also see, that if you use a belt that stretches very badly (all aftermarket belts that I've tested....again leaving Roger's new belt out of this group) there could/will be extreme amount of slack on the "backside" of the tensioner area and the Porken tensioner system will extend and take up the slack (although I'm not sure this slack needs to be taken up, since it is on the "backside" of the load and really doesn't affect anything). Note that because of the large amount of stretch, from a "crap" timing belt, the cam timing will be severly retarded...no matter how much slack Ken's tensioner can take up...or how much the factory tensioner doesn't take up....the slack doesn't affect this.
So, I can see that if one purchases a cheap aftermarket "crap" timing belt...Porken's system will take up the slack from the belt stretching (again, not sure this affects anything, whatsoever).
Since I use high quality belts that don't (very little stretch) stretch, I will continue to use pieces (stock Porsche) that have proven themselves over millions and millions of miles and that I know I will always be able to get pieces for. The stock stuff has less hardware, has less bushings, is fairly simple and proven to be very bullet proof (I have never seen an engine failure due to the tensioner system failing....never) Sure, you have to go in and adjust it every 15,000 miles...but you should be looking at the cam belt every 15,000 miles anyway...so that's a push.
It's your money....spend it how you please. I'd just suggest buying a high quality, very low stretch cam belt....not becasue of the slack, but because of the cam timing change.
Last edited by GregBBRD; 07-24-2010 at 02:51 PM.
07-24-2010, 02:44 PM
#89
Rennlist Member
I got lost somewhere in the above theory. Whats the basis for assuming oil pump load is 40% ? It has a very low number of belt teeth in contact transmitting power to it.
Also I didnt notice a definition of what X was - just seemed to appear , and when it solved out as negative, it indicated a slack condition?
FWIW, at my local wrench recently he had an S4 undergoing a belt and WP change. Owner was very lucky - the tensioner pushrod (with tension warning blade on the end) was OUT of the arm socket, resting on the body of the arm at an angle....That would require a very slack belt event wouldnt it?
jp 83 Euro S AT 52k
Also I didnt notice a definition of what X was - just seemed to appear , and when it solved out as negative, it indicated a slack condition?
FWIW, at my local wrench recently he had an S4 undergoing a belt and WP change. Owner was very lucky - the tensioner pushrod (with tension warning blade on the end) was OUT of the arm socket, resting on the body of the arm at an angle....That would require a very slack belt event wouldnt it?
jp 83 Euro S AT 52k
JP,
thanks for for the reply. I have gone back and edited my posting with some clarifying statements that I hope will be helpful for other reading it.
to answer a few of your questions directly.
X represents the tension applied to the system generated by the position of the fixed tensioner. This is what we are trying to solve for. The result will be in BU. a positive resultant will mean that the applied force is tension, a negative result indicates that the force applied is a minus tension or compression. since we are working with a belt which will not support a compressive load this translates to a slack condition.
as for the 40% to the oil pump: I realize this is not very likely however I used this as my example to illustrate that even if this large a portion of the load was carried only by the shortest pulling run of the belt the fixed position tensioning system was inadequate to control belt tension. I started my calculations with each cam carrying 40% and the WP and OP carrying 10% each. the resultant tension in the tensioner run when solved for X was -4.38 BU.
Regardless what the percent of the force is generated by the WP and OP, based on the assumption that pretension is 30% of working load the fixed position tensioner system goes slack when more the 21% of the load is generated by each cam.
The reality is that I do not believe the stock system can be accurately modeled as a fixed position tensioner. the bevel washers used for temperature compensation do have some spring to them and therefore the system will move. Due to the high spring rate of the bevel washers the movement range of the stock tensioner is rather limited as a small change in length at a high k value equates to a large change in force. There is evidence that the bevel washer stack acts as a spring in that if it didn't the factory belt ten light would not function, so it is obvious that there is some though limited movement to the stock system.
07-24-2010, 03:39 PM
#90
Race Car
Greg, if a Porsche belt on a stock system set to the correct tension right out of the factory never flapped, why did Porsche engineers put an idler pulley on the water pump between the toothed sides of the belt that does not even touch the belt when the engine is off? To me, having an idler in such a position *proves* that the belt will flap in certain conditions, and the idler is there to stop the two portions of the belt from contacting each other causing accelerated belt tooth wear.
Dan
'91 928GT S/C 475hp/460lb.ft
Dan
'91 928GT S/C
475hp/460lb.ft