A technical question about the DFV

Probably contrary to the FIA technical regulations (without actually checking). As tightly controlled as the engines are I can't imagine them allowing anything remotely different or innovative to compete. Renault I think boasted having a similar system under development some years back (I assume they in the end lacked the technical expertise or will to make it practical) and others have claimed to have electrohydraulic systems under development. I don't think big car companies today have the engineering culture to do innovation with engine architecture, it all seems to be about safe but tortuously slow incremental development. This sort of fundamental redesign of the traditional engine must look like a potentially disruptive nightmare to the business side of the industry. If it provides a significant advantage, obviously everyone must adopt a version of it, huge costs are incurred all around and the net effect is decreased profits (at least in the time scales businessmen are capable of thinking in) for everyone as they can't just increase prices to cover the changeover costs. From a bean counter's perspective there is no likely upside at all. Probably similar reasoning for the near-spec engines in F1--what is to be gained by allowing or investing in innovation when the money can safely be spent on endless grinding aero development chasing vanishingly small gains with little potential disruptive effect or risk to the status quo. Industries with large entry barriers and relatively few players don't behave like the truly competitive free market model you are taught we operate under, there is a continuum from small businesses that actually do operate near that ideal to huge corporate operations that in practice are closer to monopolies.

perhaps this is why a company like Koenigsegg is on the path - its still firmly under the control of the visionaries, dreamers and engineers with the bean counter advising instead of steering. Oh to work for an adventurous company and not a margin-motivated profit machine.

I don't think big car companies today have the engineering culture to do innovation with engine architecture, it all seems to be about safe but tortuously slow incremental development. This sort of fundamental redesign of the traditional engine must look like a potentially disruptive nightmare to the business side of the industry. If it provides a significant advantage, obviously everyone must adopt a version of it, huge costs are incurred all around and the net effect is decreased profits (at least in the time scales businessmen are capable of thinking in) for everyone as they can't just increase prices to cover the changeover costs. From a bean counter's perspective there is no likely upside at all.

I think this is one of those technologies that is so "game changing", that any of the major car companies would grab it and run with the significant competitive advantage it offers. Re-tooling costs would be no more than say a "Vanos" type system but with far more potential. Customers would pay the initial slight premium for the fuel savings alone.

Agree with your first sentence.

Using cam gear drives with large amounts of torsional compliance can be problematic. The dampening system used on the DFV idler gear was necessary due to the very energetic shaft dynamics that existed at the nose of the DFV crank where the cams were driven from. If you add up the total amount of angular backlash in the gear train meshes, and add in the torsional deflections, it becomes apparent that the total amount of valve timing variation at the far end of the camshaft can be quite significant. Plus there are the increased dynamic load effects in the valve train components due to the torsional energy transmitted from the crankshaft.

Modern F1 camshafts are still fairly compliant in torsion, because they must be lightweight. But the individual cam lobe profiles along the length of the camshaft are each ground with slight timing variation to offset the torsion deflection of the camshaft during operation.

Some of you may also recall the spring-loaded "scissor gear" that was used on the cam drive of some Honda VFR V4 bike engines.

just read the Koenigsegg thread [not normally interested in supercars]
Doh !

not related to question below.
http://www.speedhunt...giken-tc24-b1z/

Please respond regarding belts. Im getting confused here..

just read the Koenigsegg thread [not normally interested in supercars]

Me too, but you have to be impressed by Mr Koenigsegg, the way the cars are built and his enthusiasm. If his valve actuating system works commercially, good for him. I have always thought that there is more chance of innovation happening this way than can ever come from F1 in its current form. All we need now is seamless gearchanges...

Edited by Tony Matthews, 01 March 2013 - 11:17.

Tony - I thought you had progressed to seamless lycra

who knows - maybe there is a hot line from Norfolk to Koenigsegg

The dampening system used on the DFV idler gear was necessary due to the very energetic shaft dynamics that existed at the nose of the DFV crank where the cams were driven from.

Minor point but my understanding is the dynamics/excitation was coming from the camshafts. At one resonant RPM (around 10,000 from memory) the instantaneous torque in a camshaft was greater than the output torque of the engine. I think this was prior to the addition of the TVD coupling.

Minor point but my understanding is the dynamics/excitation was coming from the camshafts. At one resonant RPM (around 10,000 from memory) the instantaneous torque in a camshaft was greater than the output torque of the engine. I think this was prior to the addition of the TVD coupling.

The only torques input to the camshaft from the valvetrain are those produced by the closing valve springs acting on the backside flank of the cam lobes. And these torques are mostly offset by the opposing torque moments created by the valve springs forces of the valves being opened. Thus, there is very little chance that a situation would exist where there was sufficient net torque produced by the valvetrain to backdrive through the cam gear drive. The main issue with racing camshafts is due to their very low torsional stiffness resulting in structural vibration modes that have natural frequencies low enough to couple with the forcing modes produced at the crank drive. With the DFV V8 in particular, with its even firing order in each cylinder bank and 1:2 ratio in the cam drive, there were lots of possibilities for coupling of structural modes in the cams and crankshaft.

While a gear drive gives the best combination of weight, volume, reliability and efficiency, a chain or belt drive does provide some amount of dampening from friction or hysteresis.

a chain or belt drive does provide some amount of dampening from friction or hysteresis.

Thank you! How hard was that guys!

Gears is the heaviest solution.. if you want a compact, reliable and efficient cam system weight gotta give. Belts should be pretty good if you are willing to sacrifice long maintenance intervals and compactness.

Edited by MatsNorway, 03 March 2013 - 11:00.

While a gear drive gives the best combination of weight, volume, reliability and efficiency, a chain or belt drive does provide some amount of dampening from friction or hysteresis.

The only torques input to the camshaft from the valvetrain are those produced by the closing valve springs acting on the backside flank of the cam lobes.

You are kidding aren't you?

And these torques are mostly offset by the opposing torque moments created by the valve springs forces of the valves being opened. Thus, there is very little chance that a situation would exist where there was sufficient net torque produced by the valvetrain to backdrive through the cam gear drive.

OK, yes you were kidding - you are now you are saying the opening events DO apply a significant moment to the cam. That much is true but the rest of your statement is incorrect. The dominant accelerations of the valve are those directed away from the camshaft ie the initial opening period and the deceleration just prior to closing. At high speed (10,000 rpm) these accelerations and their associated forces and moments applied to the cam will be orders of magnitude greater than any cam/follower contact forces associated with the valve springs - especially during the period (over the "flanks" and "nose" of the lobe) where the spring is accelerating the valve towards the cam lobe.

These "dominant" accelerations - one producing a positve torque and the other negative (because they act on opposing faces of the lobe) are typically seperated by only a little less than the camshaft duration. In the case of the DFV with a duration of about 160 camshaft degrees (320 crankshaft degrees) those torques would be about 150 camshaft degrees apart. With the lobes set 90 degrees apart (four sets on an even-firing bank and don't forget there will be only 4x2 inlets or 4x2 exhausts on each cam) there would be four dominant positive torque events each 90 deg apart and four negatives falling about 30 degrees after each positive (not the cancellations of positive and negative torques you suggest). Consequently we have a vigorous set of "driving torque" reversals through the timing gears which are equally likely to contribute to excitation of resonances in the camshaft or the drive.

The main issue with racing camshafts is due to their very low torsional stiffness resulting in structural vibration modes that have natural frequencies low enough to couple with the forcing modes produced at the crank drive. With the DFV V8 in particular, with its even firing order in each cylinder bank and 1:2 ratio in the cam drive, there were lots of possibilities for coupling of structural modes in the cams and crankshaft.

I think you will find that a hollow racing cam has a very high ratio of torsional stiffness to rotational inertia.

Edited by gruntguru, 04 March 2013 - 01:10.

http://www.not2fast....TWR-Engines.pdf

http://papers.sae.org/942503/

http://bhp-k412.p.lodz.pl/tors.htm

https://www.avl.com/...;groupId=297166

http://www.desa.ir/f..._No_292_M_3.pdf

Thanks Magoo. I lifted the following diagram from the AVL link to illustrate the two acceleration peaks per valve opening cycle. (Blue trace)

Edited by gruntguru, 04 March 2013 - 02:31.

Very interesting - I especially liked this paper - much of which appears to be written in Chinese or Greek (or a combination of the two):


http://bhp-k412.p.lodz.pl/tors.htm

Thanks Magoo. I lifted the following diagram from the AVL link to illustrate the two acceleration peaks per valve opening cycle. (Blue trace)

The valve lift trace appears to be symmetrical - why are the two accelerations different?

The valve lift trace appears to be symmetrical - why are the two accelerations different?

Since the acceleration trace is the second derivative of the displacement trace it greatly amplifies any difference, ie it is so minute you can't see it on the displacement plot. Why might they be different? The valve closing event is more sensitive to velocity (seat hammer) so perhaps the designer has allowed more scope for clearance variations in service.

Another way to look at it. The spike in acceleration is a shock to the valve and spring system and will cause oscillation for a brief period. During the early part of the opening phase this will die away with little ill effect. OTOH oscillations in the valve late in the closing phase would be detrimental to the seating process.

Edited by gruntguru, 04 March 2013 - 06:57.

gruntguru,

The calculated forces resulting from valve accelerations defined by the cam lobe kinematics does not give a complete picture with regards to torsional dynamics of the camshaft. Obviously, the net torsional moment existing at the drive nodal point of a camshaft at any given instant is the sum of the follower forces minus any frictions. And the highest instantaneous force produced by the valves would be that occurring with the exhaust valve right at the point the exhaust valve lifts off the seat, due to the fact that it must overcome both the spring seat force as well as the combustion gas pressure in the cylinder acting against the surface area of the valve face.

As for the torsional stiffness of hollow racing cams, as you noted the fact that they are hollow has little effect on torsional stiffness. However, very high rpm racing cams also tend to have very small journal/shaft diameters, and their valve trains use high relative spring rates. This results in relatively high torsional deflections in the camshaft. Since the camshafts used in race engines are made from high-strength steel alloys, the increased stress levels from the shaft torsional dynamics is not really a concern. But what is a concern is the variation in valve timing that results from one end of the camshaft to the other, due to the torsional dynamics in the camshaft and gear drive system.

And the highest instantaneous force produced by the valves would be that occurring with the exhaust valve right at the point the exhaust valve lifts off the seat, due to the fact that it must overcome both the spring seat force as well as the combustion gas pressure in the cylinder acting against the surface area of the valve face.

Don't forget the force required to accelerate the valve, which at high RPM is the greatest of the three.

Your post now seems to agree that the main source of excitation is the valve gear.

Would not belts dampen the smaller vibrations?

Basically any serious drag racing thingy ive seen got belts. Belts belts belts. all over the place.

Tension for belts load up the bearings. I would like to see the return of shaft driven cams..bulky and expensive, yes but you get gear friction efficiency but not the noise.

Edited by Powersteer, 06 March 2013 - 07:55.

Your drawing (is it a Matthews?) reminds me of another advantage of shaft drive. Reducing the diameter to make it a "quill" shaft allows some tuning of TV's. I vaguely recall the DFV having quill drives inside the hollow camshafts?

Not mine, it looks like a Bruno Betti. However, I have a V-twin engine with that type of OHC drive, from 1921.

Spiral bevel gear timing drives can have excellent mechanical efficiencies, can be designed to last the life of the engine they are installed on and package nicely in many applications.

I've seen them somewhere before too...

http://upload.wikime.../RR_Griffon.jpg

Didn't W. O. Bentley propose (or maybe actually used) a cam drive similar to the side rods on a steam loco? Possibly because I think he trained initially to be a steam loco engineer.

Not mine, it looks like a Bruno Betti. However, I have a V-twin engine with that type of OHC drive, from 1921.

Oh - you bought it new then?

Clearly it wasn't this one - it has been converted to belt drive!

http://thekneeslider...ngine-for-sale/

Clearly it wasn't this one - it has been converted to belt drive!

http://thekneeslider...ngine-for-sale/

I can't veiw that. I assume it is the Vee twin created with a pair of Merlin cylinders?

If so, if they took the cylinders from teh power take off end (ie the front) then they would have no choice but to make their own cam drive arrangements, since the Merlin's cam drives were off the rear of the engine. The Griffon's, as shown above, are at the front.

Oh - you bought it new then?

Cheeky!

Don't forget the force required to accelerate the valve, which at high RPM is the greatest of the three.

Your post now seems to agree that the main source of excitation is the valve gear.

The most energetic source, by far, of dynamic inputs is the crankshaft. If you compare the amount of strain energy from torsional vibration that is transmitted from the crank nose to the cam gear drive, versus the structural vibration energies transmitted back into the cam gear drive from the valve springs, valves, followers, etc., there is no contest. Of course, it would be correct to say that valvetrain components, such as metal springs, are more likely to suffer from dynamic excitations due to coupled structural modes, since their low structural stiffness is more likely to result in many critical frequencies within the operational frequency range.

I am not convinced.

(Peak accelerations at the crankshaft end of the timing drive) x (the low MOI of the camshaft) = (low torque values).

The dynamic torque inputs to the camshaft due to valve accelerations would be higher in my opinion. Need to find a source.

I am not convinced.

(Peak accelerations at the crankshaft end of the timing drive) x (the low MOI of the camshaft) = (low torque values).

The dynamic torque inputs to the camshaft due to valve accelerations would be higher in my opinion. Need to find a source.

It's fairly easy to establish the approx. max acceleration forces at the valves. By definition, the combined inertia forces from acceleration of the valve mass, follower, retainer, etc. should never exceed the opposing force of the spring. If you know the force characteristics of the spring and the pressure angle characteristics of the cam lobe & follower, then you should be able to calculate the instantaneous torque inputs to the camshaft from the valvetrain.

It's fairly easy to establish the approx. max acceleration forces at the valves. By definition, the combined inertia forces from acceleration of the valve mass, follower, retainer, etc. should never exceed the opposing force of the spring.

When the cam-lobe torques resulting from spring force and valve inertia act in the same direction (initial valve opening and late closing) those torques are many times those due to spring force alone.

When the cam-lobe torques resulting from spring force and valve inertia act in the same direction (initial valve opening and late closing) those torques are many times those due to spring force alone.

Should I point out that if the valve lift inertia exceeds that of the opposing valve spring force, then there would also be no contact between the valve follower and cam lobe, and thus no torque moment would be applied to the camshaft. This particular situation is referred to as "valve float", and is common with high-rpm and high-lift race cams.

The kinematics of a conventional engine valvetrain employing cams and springs is determinate in nature. This implies that the cam lobe profile and valve springs are designed such that there is a constant equilibrium between the spring forces/mass inertia balance and cam lobe profile.

Should I point out that if the valve lift inertia exceeds that of the opposing valve spring force, then there would also be no contact between the valve follower and cam lobe, and thus no torque moment would be applied to the camshaft. This particular situation is referred to as "valve float", and is common with high-rpm and high-lift race cams.

No you shouldn't because (as I explained in my previous post) during the initial opening and late closing phases, the camshaft torque generated by the spring and the camshaft torque generated by acceleration of the valve are in the SAME DIRECTION AND MUST BE ADDED. These two events represent the highest torque impulse applied to the camshaft and the torque increases with increasing rpm. Because these two events are not subject to valve float, they are deliberately designed with short durations - thus prolonging the period where spring force and accelerative force are subtractive and valve float is an issue. This latter period is therefore designed with significantly lower valve acceleration and permits lower spring force in the design.

This is probably beside the point - but if you have ever tried to rotate the cam with an electric drill on a head that is off the engine and is not properly tied down you will discover just how violent the variations in torque are as each valve is encountered. It is not surprising that variations in cam torque lead to other problems.

That is actually right on the point and I will bet that a DFV camshaft operating only four sets of inlets or four sets of exhausts would be even lumpier than the one you tried.

No you shouldn't because (as I explained in my previous post) during the initial opening and late closing phases, the camshaft torque generated by the spring and the camshaft torque generated by acceleration of the valve are in the SAME DIRECTION AND MUST BE ADDED. These two events represent the highest torque impulse applied to the camshaft and the torque increases with increasing rpm. Because these two events are not subject to valve float, they are deliberately designed with short durations - thus prolonging the period where spring force and accelerative force are subtractive and valve float is an issue. This latter period is therefore designed with significantly lower valve acceleration and permits lower spring force in the design.

By definition, to preclude valve float, the linear valvetrain mass inertia forces constrained by the cam lobe profile must always be in equilibrium with the opposing instantaneous valve spring force applied. During the valve lift event, the opposing spring force must always be equal to or greater than the excess net valve mass inertia forces produced by the cam lobe. During the valve closing event, the spring force must be sufficient to produce adequate valve mass inertia forces to keep the follower in contact with the cam lobe face. Regardless, the spring and valve mass inertia forces are mostly opposing. The only exception being, as you noted, the brief periods during initial lift and closing.

As for the max torque moments applied to the camshaft from the valvetrain, this would occur at the point of max combined spring force and cam lobe/follower pressure angle conditions. Max pressure angle usually occurs around the same cam lobe phase as max velocity.

Didn't W. O. Bentley propose (or maybe actually used) a cam drive similar to the side rods on a steam loco? Possibly because I think he trained initially to be a steam loco engineer.

Believe they did, but with eccentrics+rods rather than coupling rods and cranks. W.O. used three sets of eccentrics, at 120deg. More recently NSU used two sets of eccentrics at 90deg on their motorcycles and the Prinz car. Have a look at
http://www.nsu4.nl/e...haftsystem.html for an animation.

By definition, to preclude valve float, the linear valvetrain mass inertia forces constrained by the cam lobe profile must always be in equilibrium with less than the opposing instantaneous valve spring force applied.

Your statement (after minor modification) is correct for valvetrain accelerations TOWARD the cam lobe. When the valve is accelerating AWAY from the cam lobe, the inertial force is ADDED to the follower/cam lobe contact force ie the valve spring is not required at all during this period. This occurs twice in the diagram below - where the blue (acc) trace is greater than zero - ie from about 115 - 130 deg cam angle and from 230 - 245 deg. The period where the blue trace is less than zero (from 130 - 230 deg) is the region where contact is maintained by the valve spring. You can even see that the acceleration here is in inverse proportion to valve lift, since spring force is proportional to lift. It is possible therefore to estimate the maximum positive valve acceleration. As RPM increase, the accelerations will grow until the negative peak is some fraction (perhaps 60-80%) of the spring force - 100% would represent the onset of valve float. This represents (from the graph) a valve acceleration of 0.012 - 0.016 m/rad^2. (The scales shown should read m not mm since 0.0095 mm is clearly not a realistic value for max valve lift.) At the same RPM, the positive acceleration peak at 120 deg cam angle (of about 0.11 m/rad^2) - about 9 times the acceleration due to spring force and of course the spring force is acting on the cam lobe in the same direction, giving a total of 10 times spring force.

During the valve lift event, the opposing spring force must always be equal to or greater than the excess net valve mass inertia forces produced by the cam lobe.

Only during the second stage of the opening event. (130-180 cam deg above)

During the valve closing event, the spring force must be sufficient to produce adequate valve mass inertia forces to keep the follower in contact with the cam lobe face.

Only during the first stage of the closing event. (180-230 cam deg above)

As for the max torque moments applied to the camshaft from the valvetrain, this would occur at the point of max combined spring force and cam lobe/follower pressure angle conditions. Max pressure angle usually occurs around the same cam lobe phase as max velocity.

Good point. A more complete picture however is: (cam torque) = (axial force on follower) x (normal distance from cam axis to force vector). Cross product. You will note from the diagram that peak velocity does not coincide with peak spring force nor peak acceleration force. However, the acceleration peaks coincide with quite high velocity, whereas maximum spring force coincides with zero velocity (over the nose.)