Honda Power Unit Hardware & Software

[PlatinumZealot]

'prevent hysteresis' - what nonsense ?!

hysteresis is an inherent characteristic of the material and not something that can be changed by design
metals have little hysteresis and higher strength versions of metals ie spring materials have less than little hysteresis
so the best qualified spring materials via high strength are also the best via low hysteresis

hysteresis can down in the region of ten parts per million and should be close to that in a performance valve spring
ok a race valve spring will be generating thousands of times per second that hysteresis (loss of strain energy) as self-heat
which will certainly need removing by properly designed coolant flow

force measurement in eg wind tunnel tests of race cars or when buying a piece of steak depends on the low hysteresis of metal
the load cells or balances work by measuring the dimensional change in metal with change in load
hysteresis would show as a residual error ie the output not returning to zero when the load is removed

true hysteresis is the residue after unloading and allowing the self-heat to disperse
most load cells are fortunately small enough for natural heat dispersion within the time taken by the calibration procedure
but with large load cells the calibration procedure should allow for this (but probably hasn't)
various behaviours of the strain-sensing side tend to add to an overall measurement 'hysteresis'


regarding eg triple valve springs (or similar doubles) ..... afaik
interference friction develops at incipient surge resonance of main spring(s) and by design this won't be at continuous rpm

For my simple view Hysteresis is when the rebound properties of the spring are different from the bound properites.. Making the spring slighlty unseat at high cycle speed..

I am not a "googly" type of poster.. But i did a sanity check on google amd this is what i got. I don't think i am too far off the mark.

Compression Spring Hysteresis:
The mechanical energy loss that occurs as a result of the spring ends' tendency to rotate when compressed during cyclic loading and unloading.

[63l8qrrfy6]

You are both right I think.

TC is talking about stress-strain curve hysteresis which is material specific and is negligible provided that loading is kept below the proportionality limit (and it is for a spring).

PZ is talking about energy dissipation through friction which can come from several sources such as spring/spring, spring/retainer, or contact between the coils of a single spring. This is normally manageable for a single spring and becomes atrocious for interfered springs.

[Tommy Cookers]

iirc about a million Honda CB450 and CB500Ts etc had torsion bar valve springs
so external friction hysteresis was minimal and internal hysteresis was minimal and strength high due to pure torsional loading
these engines were somewhat related to the 1964-5 F1 engine

[63l8qrrfy6]

I reckon there is still a fair amount of friction in the splines.
In terms of loading the torsion bar setup trades hertz coil contact for hertz spline contact - which may well be a bit of an improvement.

The real advantage is that dynamics are easier to deal with as there is only the torsion mode of the bar to worry about. By comparison one needs to assume about 4 degrees of freedom per coil when modelling springs resulting in 4x number of active coils of modes that have a significant participation.

As a disadvantage I would guess that the inertia is quite a bit higher compared to that of a conventional spring.

[roon]

iirc about a million Honda CB450 and CB500Ts etc had torsion bar valve springs
so external friction hysteresis was minimal and internal hysteresis was minimal and strength high due to pure torsional loading
these engines were somewhat related to the 1964-5 F1 engine

Interesting. Thanks, Tommy.

[PlatinumZealot]

Nice information there. I didn't know such a mechanism exists. The torsion bar is in two peices. One half enveloping the other. Smart. Would be interesting to see how it works for paired valves.

[wonk123]

I mean Honda made single coil springs that could handle 9,100 rpm and 400,000km of service back in 1999. I would not be surprised if they had valve springs capable of handling 13k for 7,000km 19 years later.

One of the biggest reasons for the massive power increases in the last 20 years in NASCAR and drag racing is the development in valve spring technology.
As an example pro stock engines can have valve lift of 1.2 inches and max rpm of 12,000. They don't exactly last for 7,000km though

[63l8qrrfy6]

Nice information there. I didn't know such a mechanism exists. The torsion bar is in two peices. One half enveloping the other. Smart. Would be interesting to see how it works for paired valves.

You'd just need to have one more arm coming out, rigidly connected to the first.

Or keep a single lobe with a single arm and have a valve bridge to connect the 2 valves.

Of course you'd then need to double the torsional stiffness which is quite neat as all you have to do is reduce the length of the shaft.

[gruntguru]

That would then run into a strain energy issue. Assuming an optimised design, you need twice as much metal in the spring if you want it to control twice the reciprocating mass.

The nicest thing about the torsion spring is undoubtedly its lack of resonances. Recip' mass shouldn't be a problem if the finger also serves as a cam follower. Packaging would be the main downside.

[63l8qrrfy6]

Not quite sure what you mean by the strain energy issue.
You'd just design it like a normal quill shaft - calculate the diameter to take the required torque then determine the length to achieve the required stiffness.

The fact that you have the outer shaft also contributing stiffness gives you even more opportunities to optimize it.

[gruntguru]

Always an issue with spring design. If you halve the length of the torsion bar to double the stiffness, you double the stress at a given deflection (since torque is double).

The optimum spring will be longer and thicker.

[63l8qrrfy6]

Surely lift will more or less halve with 2 valves so displacement will also halve.

[PlatinumZealot]

Three layers might have to be used then to effectivley lengthen the bar. Takes up more space though.

[63l8qrrfy6]

Three layers might have to be used then to effectivley lengthen the bar. Takes up more space though.

You don't have to lengthen the bar..

Let's assume that mass doubles and lift halves when moving to a double valve.

In order to maintain the same natural frequency you need to preserve the stiffness to mass ratio. Effectively you need to double the stiffness of the 2 valve bar.

Torque is stiffness (torsional) x angular displacement. Since you've doubled the stiffness but halved the lift (angular displacement) the torque has not changed. The radius of the shaft can remain the same as the shear stress in the bar is not a function of length. You still have to achieve the increase in stiffness which you do by decreasing the shaft length.

The other thing to consider is the change in angular acceleration. For a simple harmonic cam profile the max acceleration is given by lift x w^2. Halving the lift would halve the angular acceleration. Since the mass has doubled and T=I*a , the torque requirement can again be shown to remain constant. Of course cams use significantly more complex profiles and peak accelerations occur on the flanks rather than at max lift but I think this is a reasonable assumption.

[PlatinumZealot]

Three layers might have to be used then to effectivley lengthen the bar. Takes up more space though.

You don't have to lengthen the bar..

Let's assume that mass doubles and lift halves when moving to a double valve.

In order to maintain the same natural frequency you need to preserve the stiffness to mass ratio. Effectively you need to double the stiffness of the 2 valve bar.

Torque is stiffness (torsional) x angular displacement. Since you've doubled the stiffness but halved the lift (angular displacement) the torque has not changed. The radius of the shaft can remain the same as the shear stress in the bar is not a function of length. You still have to achieve the increase in stiffness which you do by decreasing the shaft length.

The other thing to consider is the change in angular acceleration. For a simple harmonic cam profile the max acceleration is given by lift x w^2. Halving the lift would halve the angular acceleration. Since the mass has doubled and T=I*a , the torque requirement can again be shown to remain constant. Of course cams use significantly more complex profiles and peak accelerations occur on the flanks rather than at max lift but I think this is a reasonable assumption.

I wasn't halving the lift because i was assuming full advatange is taken for more flow in a race car application. So for arguments sake i kept the lift the same. For example when BMW moved from two valves to four valves per cylinder the valve lift did not reduce. Remember the valve diameter is smaller to fit two valves in the same space so the lift would be just about the same more or less.

So. I assumed the same acceleration. The moving mass maybe be not double, but more than 1.5 times the mass of one valve maybe? So the spring has to be stiffer by that at least. To get the required stiffness i was saying it is better to have a longer bar instead of only just bigger diameter or thicker walls. The spring will be under less shear stress.