Rocker Arm Grinding Tool Design.
 

I am going to build a rocker arm grinder. I just purchased a slightly used micrometer adjustable X Y precision stage. On this I plan to mount a diamond dressing tool for grinder wheel and an expanding arbor sized for the 18mm ID of the rocker arm bushing. Then I can hold each rocker by the same datum. I have measured the radius of the rocker arm pad as 33 mm. I will put a bearing for rotation of the rocker at this distance from the pad. I have 18mm of travel, which is just enough to true up the face of the grinding wheel with the diamond. Then I can advance the rocker towards the stone in .01mm at a time grind the face, advance the stage .01mm and grind again. I should be able to "blue print" a set of rockers by removing exactly the same amount of material from all 12. This way at least the rocker ratio will be the same between them.

http://forums.pelicanparts.com/uploa...1480979340.jpg

Watchimg. Was thinking of this exact contraption today leaving work.

I'll be a guinea pig.... ;)

I am concerned that the ball bearing slides are open, I will have to make a boot to protect them from grinding debris. This stage is designed for moving precision optics, so I will have to test to see if it is stable enough once i build on it. It does feel very rigid but the rocker arm will sit up about 2" above the current surface. I will be interested on testing it on some rockers that are complete junk before I try on the spare set of lightly used ones I have.
David

Can you pivot on a hardened shaft instead of bearings? Shaft and bushing may hold tolerance should hold better I'd imagime.

Curiously watching.

Hint

https://www.youtube.com/watch?v=HxNiWy5of7k

How do you know exactly where the center for the 33mm radius is? Maybe it doesn't matter much.

I think the arc and its centre position does matter.

We measured a '906' Rocker' with a CMM some while ago.

http://i197.photobucket.com/albums/a...ps2g61vp2y.jpg

We used these dimensions for the manufacture of our forged rockers which measure identically in terms of cam lift/duration to the 1965/1966 rockers we wanted to reproduce.

So the radius is for sure 30mm?

I didn't drive the CMM but the guys who made the measurements seem quite capable.

A rough check on the chord length and angle seems to stack up but small changes in the centre position will affect the radius and the absolute position of the pad.

We made our machining fixture from this drawing and when we checked 'our' rockers installed with an S cam they measured virtually identically to the standard 1966 rocker.

We didn't compare them to the 906 rockers as we are still making the lash caps.

I have never tried to measure one of the later cast rockers.

I was able to put my rocker on a friend's optical comparator. Made some preliminarily measurements of the stock cast rocker at 50x magnification. My radius measured 30.97mm. The pad arc length was about 43 deg. I was not really happy with how I was holding the rocker so I am going to measure again. Also this is a used but good rocker. I don't believe it had been reground but can not verify that.

KTL,
I have seen that YouTube video, that tool is not maintaining the proper arc or position. While that is a perfectly acceptable repair, each rocker is going to be unique. I want a tool that can make all 12 the same, and close to the factory dimensions.
David

So the set of used rockers I bought is a mix of B castings and 🔼 triangle castings. The one I measured today is the B. Definitely cast, rough all over, maybe sand cast, but I am no expert. The 🔼 one I pulled out of the bag to compare is MUCH nicer. Beautiful smooth surface, maybe a die cast part. http://forums.pelicanparts.com/uploa...1481163683.jpg
The shaft that came with the 🔼 rocker has negligible wear, while the B shaft I can just feel the wear with my fingernail. Is the B an aftermarket part?

My figures cam up with 31.7 for radius on the later 993 type rocker.
regards

Looks like the radius is 30 mm ("R30") over an arc of 52 degrees.

We have 3 different radii listed now.

I trust a CMM though! That should be dead on unless the rocker moved on the bed.

Rockers of this type are generally investment cast.

Sorry guys, I meant 30mm, not 30°. Rookie mistake....

I'm used to imperial.

The 906 rocker I would say is made with a 30deg rad. It is a different part from the cast rocker. I would not just jump to the conclusion that all Porsche 911 rockers share the same radius.

The rocker ratio of a 906 rocker is said to be 1.5:1 whilst standard rockers are 1.46:1 but I can believe this is within a measuring error.

The real test must be to compare with other rockers in terms of cam lift.

We have made and fitted about 40 engine sets to this drawing and have checked 3 sets in terms of cam lift.

Depending who has made the investment cast rockers I can believe there is some variation.

The man aim must be to have all the rockers the same even if the lift varies a few thou form manufacturer to manufacturer.

We are running 'our' rockers in several FIA spec 2.0 litre race cars with 906 cams and this was also part of our logic in reproducing the 906 'heel'.

I Measured my rocker again today on the comparator, could not get my numbers to repeat. I may have to find time on a cmm and check it that way.
David

ok - very good the 15 and 16mm dimensions locate the arc center. Thanks.

We are in the process of modelling the complete rocker/valve and cam so we can look at the restrictions that the cam tower and rocker cause to cam profiles and see if we can find a way of making changes.

We have the 3-d Model about 90% complete in terms of the parts but putting it all together could be tricky.

Pauter already makes some replacment rockers with improved geometry.

@$2500.......uh, no

What do you mean by 'improved' geometry? Is this the pad length or do you believe that there has been a change in the rocker ratio?

The increased length pad was developed on the 906 Rocker and used by Porsche during all subsequent developments up to the 935.

The pad length on the standard rocker is just a little too short for high lift cams.

When we made our forged rockers we used the '906' length on the heel as the primary reason was to use these rockers on Period F Appendix K Race Cars.

One of the reasons we have replicated the original design is that the inertia and 'mass elastics' of the rocker will be virtually identical to the standard part and this should eliminate any issues to do with valve spring resonances.

We have used a 300-M VAR steel which has excellent fatigue resistance and unlike tool steel will produce a definite 'endurance limit'.

I see 9M makes a CNC'd rocker too

From Pauter's site:

"Lengthening of the radiused follower “shoe” increases the usable wear surface, allowing for a cam profile aggressiveness never before seen in the 911 rocker arm arena. During development stages prototypes were cycle tested both third party and in-house, using various cams at and far beyond normal racing RPM (over 10k) with excessive spring loads for many hours. Results of these tests allowed us to provide this product with the same confidence found in all Pauter offerings. Because of the marked increase in strength, wear resistance, and weight savings, these rockers allow the engine expert to employ a range of spring pressures best suited to your engine program."

Porsche Specific - Pauter

$3600 per set.

Another UK Company modifies a 993 Hydraulic rocker to use a swivel foot and these are $6500 per set.

The RSR Rocker arm which is specifically outlawed for use in Period F have been remanufactured and are currently selling at around $7200 a set but that doesn't include lash caps.

This is what set us off making the forging tools.

If you add that the basic statement made in Appendix K then the following also applies:

3.3.5 Unless otherwise specifically authorised by these regulations, any
component of a car must have identical dimensions and material
type must be the same to the original part. Evidence of this must be
provided by the applicant.

None of the billet machined rockers meet this basic requirement and the 993 rockers were just too expensive to consider.

Standard Investment cast rockers trouble me at high engine speeds and with 906 cams as I think that they may suffer from the occasional fatigue failure.

I am not sure that you can run much more lift than around 13.5mm so making the pad longer than those used on a 906 rocker may not be useful.

This is why we are currently modelling the towers and the rest of the set up.

The Shrick Race Springs for the most aggressive cam that they offer has a force at peak lift of around 240lbs and the Eibach Springs are around 280 lbs.

This results in relatively low stress on the rocker arm and it is only the nature of investment casting and the variation in fracture toughness that is worrying.

Those prices are ludicrous....

Whatever you decide to do, to use anything other than a stock rocker will cost $$. You cannot buy quality and development for pennies.

But there should be other considerations taken in account as well. If you are going to spend $$ on new fancy Rockers, a few steps before hand should be taken.

Valves. Lighter Valves and lightweight retainers will help lower the kinematics here.

Rocker arm. Weight is the main factor that needs to be addressed here. Obviously strength, but good FEA work will allow strength to be analyzed when weight is the main design criteria.

Cam design. A proper design will allow you to lower the spring forces required to control the valve. Most of the spring force required in alot of cams is there to dampen the harmonics induced by poor design. I have seen these 2V engines with 52.00mm inlet valves run seat pressures of less than 60 Lbs and the engines run at speeds above 7500 RPM.

FYI, I have seen new aluminum rockers in development in the US where a roller is used on the cam. These will be a lot less than the costs shown above and out perform any "sliding" rocker as the friction generated in the Porsche style is a huge contributor to lost horsepower.

Roller rockers have been tried before but the consensus seems to be that they just don't fit and there are problems in getting the right ratio with the roller given the constraints of cam and housing geometry.

I think that some numbers would help us in understanding what losses are generally involved.

I used to design/manufacture 'Rotating machinery' test rigs that were used for measuring both engine friction and for valve train design and development and we also used to manufacture a range of very precise torque measuring instruments.

A Typical F1 engine from the days of the 3.5 litre V10 used to produce a friction of around 100Nm of torque at 20 000rpm.

This is effectively 280 HP which equates to about 30% of the engines total power development.

Clearly challenging engine friction used to be a very significant issue.

As speeds fall power losses improve significantly but as still worrying.

If we consider a more typical production engine the valve train will contribute around 30% to the power loss at idle but the engine is developing very little power so perhaps it is not so significant.

At 6000rpm valve train losses will be less than 10% but are much more worrying due to the amount of power now being produced.

What we need to consider is how much does camshaft bearing friction, spring loads, valve guide and seal friction, windage and the cam drive system contribute.

If you break it all down and develop a rig which measures just the sliding friction of a single lobe and follower the numbers are quite surprising.

At 75 degC a single lobe/follower requires drive torques that are typically

500rpm - 0.4Nm (0.3 ftlbs) and at 4000rpm this would fall to 0.1Nm (0.072ftlbs)

If we consider 12 lobes at 500rpm (Idle) then we are losing 0.34 HP

At 4000rpm this loss becomes 0.67 HP.

I would agree that in the limit if you could eliminate friction 0.5HP is worth having but I am not sure I would use the term huge.:)

The typical cost of DLC coating a rocker is around $30 if you use a company such as Balzers and I am not sure that it offers good value.

Couple of comments on your X-Y micrometer stage:

I highly recommend checking what max forces and moments are allowable on that stage and ensure that the reaction forces of grinding the rocker surfaces do not exceed them. My experience with precision stages such as those is that they generally do not last long working near (or above) the limits.

I spoke to my friend and asked him what level of frictional losses they saw in testing. I emailed him the numbers quoted here. He told me but asked that they stay confidential as the development and results are not for outside discussion. However, he did say they test all valve train parts on a Spintron machine and the numbers observed were nothing close to what has been posted here. He suggested either they have made a major error in testing and acquisition or the numbers posted here are estimated and not from any mechanical testing.

I asked about the new rocker arm and was told that two criteria were decided upon in the beginning. Any new rocker had to be at a lower cost than any Forged rocker currently sold in the aftermarket and it had to be better. How many would you sell was the comment he made if the cost is too high. After all, how many of these older air cooled engines need anything better and more expensive than the current cast rocker. Not many was the answer given.

Of course the numbers are for the single lobe test rig are from test results - My Company made the test system and supplied the torquemeter.

I was responsible for Sales & Marketing at that time and held all of the technical discussions with the client, wrote the initial Functional Design Specification and was present during the Acceptance tests.

The rig was driven from an Inverter controlled electric motor and calibrated to remove cam bearing friction and windage losses.

The torquemeter capacity was 2.5 Nm with mechanical overload protection in the form of 'dogs' around the signal generating teeth.

Torque shafts with different capacities were supplied as part of an overall package.

Control and Data Acquisition was written using Labview RT but it was 15 years ago.

Tricky to obtain single lobe data on a Spintron using a large motor driving a whole engine, the belt losses alone would be more then the measurements we made.

We did NOT make mistakes and supplied all leading F1 engine teams with test systems including Ilmor (Mercedes), Renault at Vitry Chatillon, Cosworth in the UK, BMW Motorsport in Munich and Toyota in Cologne.

We had a Worldwide reputation for engineering excellence and made Turbocharger dynos for Allied Signal, KKK and Holset which operated at speeds of up to 250 000rpm.

We supplied early KERS test systems and supplied Renault F1 with a gearbox test system capable of 24000rpm and accel/decel rates of around 200,000 rpm/second for gearshift development in the days of sequential gearshift.

Why would I be mistaken?

http://i197.photobucket.com/albums/a...psb3ztylcs.jpg

The rig was supplied to one of the UK F1 engine manufacturers along with many other test systems including Single Cylinder Engine Dynos and many other component test rigs.

It was used mainly in their 'Technology Division' which was concerned with carrying our Road car engine development programmes on a commercial basis.

The F1 Valve train test systems we provided were quite different in nature.

We also installed 'in-car' torque measurement inside the engine for BMW and logged 'on track' torque during testing. It was, at the time, subject to an NDA so was never publicised.

We made this system about 15 years ago and today people are boasting that they have just successfully made such a device but they have made the mistake of using strain gauges.

The main reason for changing the cast steel rocker in a 911 engine is due to the 'short' wear pad which is a source of some wear with RSR style cams and to eliminate fatigue failures on high rpm race engines.

Simply put a Cam bank in a vise with cam and rockers and spin it you will see that as you speed up the rotational force (you) the load actually tends to cancell out some what with the spring force, both sides of the cam acceleration and deceleration have spring pressure acting upon them.
Both taking away and adding to.
I agree with Chris results are not what one would think them to be.
regards

It's possible the stages won't last long, but I suspect that if I keep the feed number down to .002-.004 and pivot very slowly the grinding forces are going to be low. These would never work on a milling machine, those forces are way to much. Working on getting better measurements, but home with s sick kid today.
David

That makes sense then.

I'm sure the budgets of those F1 OEM engine companies would dwarf my friends company, many fold.

The problem with anything that is developed for these early engines, is cost. It's certainly a good idea and a part that is needed, but how many will step up and buy. As you say, only certain applications require such a part so that makes sales even more difficult.

It's great that people still develop parts for these early engines.

justn sayin lol