Feasibility of pushrod/cam-in-block engines in Formula One [OHV/overhead-valve]

[e36jon]

Greetings all

I hope I didn't miss this earlier in the thread: For OHC engines with bucket style tappets there is the option of having a curved rib on the top of the tappet. Brian G has a great examination of an F1 pneumatic valve assembly from the V10 era that has this feature. (viewtopic.php?t=15385&start=195) The BMW V10 from the M5 (Street car) also has this feature. Does this change the discussion?

On the pushrod motor front there was a system some years ago that used multiple springs distributed through the valve train to reduce the net spring force and overall component weight. Imagine a lifter with its own spring plus a pushrod with its own spring plus a rocker with it's own spring and then the valve with its own spring. Turns out that all of the mechanical bits get lighter because they aren't trying to pass the loads from the valve all the way back to the cam. Does this alter the discussion?

I really appreciate the informed discussion on this site. I almost always learn something.

Cheers,

Jon

[Hoffman900]

Greetings all

I hope I didn't miss this earlier in the thread: For OHC engines with bucket style tappets there is the option of having a curved rib on the top of the tappet. Brian G has a great examination of an F1 pneumatic valve assembly from the V10 era that has this feature. The BMW V10 from the M5 (Street car) also has this feature. Does this change the discussion?

On the pushrod motor front there was a system some years ago that used multiple springs distributed through the valve train to reduce the net spring force and overall component weight. Imagine a lifter with its own spring plus a pushrod with its own spring plus a rocker with it's own spring and then the valve with its own spring. Turns out that all of the mechanical bits get lighter because they aren't trying to pass the loads from the valve all the way back to the cam. Does this alter the discussion?

I really appreciate the informed discussion on this site. I almost always learn something.

Cheers,

Jon

It does and I am aware of both, but yes, that is an effort to get around this issue. That trick has been around since the Miller / Offenhauser and Ford Flathead days. The bucket needs to be keyed though to keep it from rotating. NHRA Pro Stock bike racers (Suzuki based engines) have been doing this since the 1980s.

Those bits were quickly found to be problematic and were from an era when people were just guessing what was going on as opposed to measuring it.

[63l8qrrfy6]

If you worked backwards checking velocity, a roller cam may have an effective tappet diameter of 50in+. A single bucket / tappet alone would be the size of the entire engine.

No it wouldn't, maybe check your maths ?

If you look at the lobe profile from Honda's Third Generation Formula 1 engine, you saw maximum intake valve velocity of 15mm/rad (.0103"in/deg) and acceleration of around 55mm/rad^2 (.03779in/deg^2). This doesn't account for multiplication through the finger follower. An older NASCAR application when multiplied through a 2.37:1 ratio rocker arm, might only be .01778in/deg (velocity) and .00209in/deg^2 (acceleration).

Ok, let's take the Honda F1 example then. The multiplication ratio for a finger follower is tiny and non-constant. It peaks around max lift when the velocity is very low. Let's assume a worst case scenario where the multiplication is 1.2 everywhere.

The Honda F1 engine you reference would require a 37.25mm dia flat bucket. Let's say this won't fit - doming it with a ~20 mm rad would drop the diameter to about 32mm. If this is not enough it can be domed further - the only limit is the contact stress which in turn is a function of acceleration.

That's exactly what e36jon is showing in the previous post. Those are not failed experiments, they have ran successfully in F1 engines - other than Peugeot, Cosworth have also used them in the older V10s too. Those engineers knew exactly what they were doing rather than "just guessing" as you suggest.

[Hoffman900]

If you worked backwards checking velocity, a roller cam may have an effective tappet diameter of 50in+. A single bucket / tappet alone would be the size of the entire engine.

No it wouldn't, maybe check your maths ?

If you look at the lobe profile from Honda's Third Generation Formula 1 engine, you saw maximum intake valve velocity of 15mm/rad (.0103"in/deg) and acceleration of around 55mm/rad^2 (.03779in/deg^2). This doesn't account for multiplication through the finger follower. An older NASCAR application when multiplied through a 2.37:1 ratio rocker arm, might only be .01778in/deg (velocity) and .00209in/deg^2 (acceleration).

Ok, let's take the Honda F1 example then. The multiplication ratio for a finger follower is tiny and non-constant. It peaks around max lift when the velocity is very low. Let's assume a worst case scenario where the multiplication is 1.2 everywhere.

The Honda F1 engine you reference would require a 37.25mm dia flat bucket. Let's say this won't fit - doming it with a ~20 mm rad would drop the diameter to about 32mm. If this is not enough it can be domed further - the only limit is the contact stress which in turn is a function of acceleration.

That's exactly what e36jon is showing in the previous post. Those are not failed experiments, they have ran successfully in F1 engines - other than Peugeot, Cosworth have also used them in the older V10s too. Those engineers knew exactly what they were doing rather than "just guessing" as you suggest.

My math is correct, wording /statement was off.. doing too many things at once. A flat tappet cam would be around a 50-60" radius roller follower. I stand corrected.

I know what he is showing. As I mentioned, this a recycled and refined concept since the 1920s. These were even used in Offenhauser Indy engines going back that far. Drag race bike builders have been doing this since 1980s, and it's a pretty common mod.

Yes, acceleration is the constraint because by doming the bucket, you are essentially designing it like a roller cam, which is a point I have made several times.

My comment about guessing was the "rev kit" for pushrods (the multiple springs throughout the system).

[e36jon]

Here are some visuals to go with the conversation.

[J.A.W.]

I'll agree with your statement PZ, the Bugatti sohc and Rollsroyce Merlin sohc setup lends itself well to a narrow valve angle and seems like a design worth revisiting.
The only drawback is removing the cam to change a centrally located spark plug, the 2 examples had side plugs, obsolete from poor combustion.

Not quite the only drawback, the R/R V12 was a bit notorious for cam/follower wear,
whether in high-power big-boost aircraft use, or as an NA tank engine.

The SOHC 4V Allison V12 equivalent featured a roller cam follower system:

[coaster]

That Allison reminds me of 1980's Mitsubishi technology, 2 shafts above cam centreline.
Buggati, Merlin have 2 shafts below centreline like an f1 finger follower setup. Both have merits, still the 4 valve sohc is lumped with the spark plug to the side, poor combustion, some power loss.
I like the Merlin Buggati style as the angularity is minimal and allows more radical valve motion whilst keeping the rocker tip close to the centre of the valve, the Allison setup sweeps the rocker tip a larger distance.

[Hoffman900]

That Allison reminds me of 1980's Mitsubishi technology, 2 shafts above cam centreline.
Buggati, Merlin have 2 shafts below centreline like an f1 finger follower setup. Both have merits, still the 4 valve sohc is lumped with the spark plug to the side, poor combustion, some power loss.
I like the Merlin Buggati style as the angularity is minimal and allows more radical valve motion whilst keeping the rocker tip close to the centre of the valve, the Allison setup sweeps the rocker tip a larger distance.

Most of the Japanese bike engines in the 1970s were this design, Japanese cars up into the 1980s too.

They make it challenging to modify for sure and they tend to wear pretty quickly at high lifts. These valvetrains all need carefully measured geometries to calculate the lobe profile. Unfortunately, most cam manufacturers don't do that and just recycle lobe designs. You really get a sense of this when you measure the lobe profile and the valvetrain lift curves of an OEM cam, then do the same with the aftermarket cams. Sometimes the derivatives do the opposite of what you want as the rocker geometry sometimes gains as it approaches the seat.

[Hoffman900]

I ran the Penske Ilmor Mercedes PC-23 past Billy Godbold and asked him about the pivot followers on the cam in this engine vs a translational follower typically used. Here is what he had to say:

On the translational vs pivoting followers, that seems to be a bit of a fad that comes and goes. It makes the cam design side at least 5 times as time consuming as you have to go through the equivalent of another coordinated transformation between the valve motion and the cam surface profile. Working around a purely transitional tappet that can be duplicated on an ADCOLE gauge is much less sensitive to minor perturbations. V&H made their first Pro Stock Bike engine with pivoting followers. The latest version is translating. Overhead based guys overlook the beauty of one translating system connected to a rotating rocker compared to two rotating systems connected by a pushrod. The pressure angles are worse on one side of the pivoting follower, but there is a very slight inertial advantage. I'll take the slight mass penalty for the reduction in sensitivity to small geometric variations.

[63l8qrrfy6]

In addition to the lower inertia there may also be a small friction benefit to using roller followers.

The 500I had needle rollers on both ends of the follower. Of these two, the load on the pivot bearing is relatively small.

A roller lifter with needle bearings would also lose power in the guide bore as well as the tie bushes.

Having said this, Ilmor only spent 6 months or so designing and developing the engine from scratch so it may well be that many of the decisions were made based on assumptions and rough calculation rather than definitive experimental results. They really struggled to get those followers to work and ended up having to make them out of S157.

[Hoffman900]

In addition to the lower inertia there may also be a small friction benefit to using roller followers.

The 500I had needle rollers on both ends of the follower. Of these two, the load on the pivot bearing is relatively small.

A roller lifter with needle bearings would also lose power in the guide bore as well as the tie bushes.

Having said this, Ilmor only spent 6 months or so designing and developing the engine from scratch so it may well be that many of the decisions were made based on assumptions and rough calculation rather than definitive experimental results. They really struggled to get those followers to work and ended up having to make them out of S157.

Absolutely on the needle bearings. I know they have had problems with the point loading / spalling and rpm issues in some of the V8 racing stuff. They are sensitive to valvetrain control. The aftermarket guys save the bushing lifters for the high loading stuff (ie: pro drag racing classes, circle track engines that take a lot of abuse). Service is higher though. NASCAR (Toyota and Ford) were playing around with these in the mid 2000s (Del West was making them):


They went back to a more traditional steel rocker with a roller tip.


NASCAR commonly uses PM M4 for the cam cores, in both flat tappet and now roller applications.

I'm really shocked that Vance and Hines tried that set up on the first gen Harley Pro Stock bike. I'm curious to who was advising them at the time. In drag racing, they can easily deal with a bit more inertia from a translating setup. Figure they're running near 500lbs on the seat and 1400lbs open with those type of valvetrains.

I still need to read the book on the PC-23. Patrick Morgan has CAD drawings of the valvetrain on his Flickr.
https://www.flickr.com/photos/10131211@N03/page2 (scroll down).
Billy said Kenny Dutweiler has a few of them and he's been able to crawl through them.

[coaster]

At 8.15 minutes he describes a high rpm pushrod setup, nascar, 9500rpm.