Camshaft design

[hoffman900]

Harold Brookshire, Jon Schmidt, Mike Shoe, and Mike Jones had a good discussion here about some of Harvey's opening and closing ramps (the "No Pulse" on the intake side and the jog on the closing side to help settle the dynamics down), which in turn resulted in Harold showing some of the acceleration curves of some of his designs:

https://www.speed-talk.com/forum/viewtopic.p ... 3&start=15

265387V Large Solid Flat
Same curve as small solid flat, with more dwell at maximum velocity


25x41635 Typical Aggressive roller Short track bracket profile


2333367V Small Solid Flat typical dwell at max velocity


Mike on the little jog on Harold's flat tappet ramps:

On the solid lifter profiles, there is a little jog before the angle changes about 80% up the first increase in acceleration, is that incidental or a feature that needs detailed control?

That would be the opening lash point. Although the opening is considered "rampless", the lash point has specific values for velocity, acceleration, and jerk that Harold chooses. The jog, and even the reveral on the closing, is simply a result the choosen ramp values.

Here is the acceleration curve for an aggressive hyd. roller I just finished.

Jon,

All my curves are done by a 5-exponent polynomial equation, Dacaman12's cam done by an Excel spreadsheet he wrote himself.
The 'little blip' is the lash point. Later on, I have learned how to get rid of it.
Even with the'little blip', solids that I have designed have gone 100,000 miles on the street, and won all sorts of races.
I want to thank you for getting the pictures out. Let me know what I did incorrect in publishing them, so I can ease your work load next time.....

UDHarold

How about a couple of 31* cams to show how different durations and lifts can all look alike????

Sure!!!

Here are two accel. curves using the exact same ramp data, but very different lift and duration #s. The shorter one is 265@.050, 182@.200, and .675 lift w/ 1.7s. The longer one is 288@.050, 209@.200, and .825 lift w/ 1.7s.

I call this family the "race winning street cam" family. These are Harold's 24hrs@Daytona Ramps.

Jon,

Some closing ramps do not have the acceleration reversal, but it is universally taught in various cam design courses, and as I understand it and use it, it is used to help absorb excess acceleration in the valve train on the closing side.
Remember, at the end of the closing ramp, the values are 0 lift, 0 velocity, 0 acceleration, and 0 jerk. So the closing ramp either has to be extremely long, or get rid of the Real Values somehow before closing.
The engine never notices that closing reversal in a bad way. I think many stock cams have it, or had it. I haven't looked at a stock cam for a decade or so.
Many cam designs used it also on the cam opening ramp, and the 'No-Pulse' ramp of Harvey's is a improved version of the reversal ramp.

UDHarold

Jon,

Revisit Harvey Crane's web-site. He recommends AGAINST using his "No-Pulse" ramp on the closing side, and he uses the acceleration-reversal closing ramp.
Constant-velocity opening or closing ramps are so "Old-School" that I have NEVER used them since I started designing in 1972. When Harvey saw my ramps in 1974, he asked me if I had been Cosworth's designer, as they used similarly-shaped opening and closing ramps, which involved constant-acceleration/constant-jerk curves.
My main curve, from the opening point to the closing point, was of the PolyDyne method, which I stopped using by 1977, when I discovered I could do everything I wanted with regular polynomial equations.
In the Spring of 1980, when I started UltraDyne, I 'invented' Multi-Segmented Polynomial' (MSP) equations, which I still use today, and they are what you call 'Spline-Fits', or 'Knots'.
All graphs shown by me are of MSP equations, or Spline-Fits........

UDHarold

It would be great if you could still see Mike Jones' curve he posted from that thread. From the conversation and looking at the acceleration curves, you can see Harold designed his ramps similarly to Harvey's "No Pulse" with very little / no ramp on the intake side and the ramp with a pulse on the closing side. Mike Jones has hinted at, and I've seen Mike Shoe confirm that this is mostly how Harold made his cams asymmetrical, and they were symmetrical from .200 on up.

Here is a profile to compare smoothing from an app Jon made vs. Blairs 4stHEAD. You can see the pulse on the opening side here:

[statsystems]

Bob,

Someday if you can, maybe you can take all these cam discussions, and group them as closely as possible and then make them stickies. I know it's a bunch of work but the information is so useful and comes up so often it may be simpler in the long run.

Hell, now that I think about it, you could do the same for induction, exhaust and similar topics.


See how good I am at making work for others????

I'm just typing out loud here.

[hoffman900]

Bob,

Someday if you can, maybe you can take all these cam discussions, and group them as closely as possible and then make them stickies. I know it's a bunch of work but the information is so useful and comes up so often it may be simpler in the long run.

Hell, now that I think about it, you could do the same for induction, exhaust and similar topics.


See how good I am at making work for others????

I'm just typing out loud here.

I could do that in time!

another one I forgot (and answered my question I had here a few minutes ago but have since deleted):
Cam with Dwell at Maximum Velocity Graph
http://www.speedtalk.com/forum/viewtopi ... =1&t=47483

[n2xlr8n]

Just when one wonders if Speed Talk is going the way of the interwebz, a thread like this one gets bumped to remind us of how privileged we are to be reading these pages.

Great thread.

[hoffman900]

On this snow day, I purchased the MJP lobe design software. I’ve been staring at all the lobe derivative curves I could for now and it’s time to plug and play.

Does anyone have any Cam Dr. data on any of Harold’s lobes? We have some acceleration charts above, but I’m curious to see the lift curve, velocity, and jerk of the opening side of his cams. Curious to see how he set up it up since they look like (and has been stated by others) they didn’t have much of a ramp on the opening side and that he had value targets for when lash zeroed out, which I would imagine would line up with the closing side lash point.

[hoffman900]

Just starting to get in the groove with this software and have learned a lot already. Not totally happy with the acceleration curve here, but this is trying to develop a profile of a Crane F5 lobe to be used in simulation. Have to say Harold left a lot of little gems in regards to cam design around the web that have been very helpful.



This is using the MJPMike's software (I cropped it out to not reveal my secrets - one afternoon playing around with this and I'm starting to sound like a real camshaft guy ). I started off with some CV ramps, but as I get more comfortable and my knowledge base improves, I'll start playing around with different types, the little jog to help settle down the valves, asymmetry, etc. This is a lot trickier than you would think though - what a PITA this had to have been before computers.

[hoffman900]

Getting the hang of this:


292 @ .020. 264 @ .050, 180 @ .200. , .398 lobe lift

Lifter Diameter: .875
Velocity: .00733
Positive Acceleration: .000374
Negative Acceleration: -.0002172
Peak Jerk: .0000519 - this seems kind of high, I'll have to work on this, either through design for smoothing. I'm really curious about the jerk values with Harold's designs. Jon used the analogy about his designs of a "point and shoot" driver, late and hard on the brakes but keeps it off the wall and puts in a fast lap. I think part of that is he was a product of his generation and he didn't have computer programs to manipulate the curve. It was just notebooks full of numbers early on.

What I'm doing is trying to go through the Crane F5 lobe family to create profiles to be used in simulations. This one is pretty close to the F264/397.

F264/397:

296 @ .020, 265 @ .050 , 176 @ .200 , .397 lobe lift.

Using some of Harold's numbers as my guide:

Since if there's anything I know, it's History, here goes.....
Prior to the widespread use of computers, very few mushroom cams were made. The earlier designers had enough trouble making cams follow their normal lifters, much less an extra-wide one.
By the mid-70's, NASCAR racers had learned enough to know that the Chrysler .904" tappets allowed more aggressive lift curves than the .842"-.874" tappets. NASCAR's compromise?---Let everyone use a Mushroom lifter. Chevrolet went to .960", and NASCAR settled on that, even though Chrysler had some 1.000" ones. Those were used in drag racing.
In 1978 I designed the Comp Cams' 310/318 mushroom cam, 268/276 at .050, .390"/.410" lobe lift. In 1980 Buddy Baker won the 1980 Daytona 500 with this cam, and it is still the fastest Daytona 500 ever run. NASCAR went to restrictor plates, and then to .874" max tappet diameter.
Cams with flat-bottom lifters, either hydraulic, solid, or mushroom, are design-limited by velocity. Here are the max safe velocities(more or less) for the various lifter diameters:
.842" .00705"/°
.874" .00733"/°
.904" .00759"/°
.960" .00808"/°
There ARE tricks around these numbers, but not all cam designers know them, or else how to use them right.....
Cams with rolller lifters, either hydraulic or solid rollers, are design-limited by acceleration and base circle diameter. A peak acceleration rate that may be un-makable for a .900" base circle may be usable for a 1.100" base circle.
There are obviously tricks around these limitations, also. The hard part is knowing the tricks, not the basic theory, and when and how to use the tricks....

UDHarold

David,

There are people in the cam design world that charge several thousand dollars for that information.
However, here it is a la UDHarold:
Almost all negative acceleration in hydraulics and solid flat tappets cause too small a radius of curvature on the nose if the negative accel rate exceeds -.000220, I do not go even that high. The open spring pressures necessary to control that rate of negative acceleration, and with that small a radius of curvature, cause almost sure wipe-out of the cam. Too much pressure concentrated in too small an area.
Roller cams are relatively OK on Negative acceleration, it just requires more open pressure, especially when the valve lift goes over 1.000". Eventually cam materials come into play, particularly when you have 1400 lbs at 1.000" valve lift, and a 1.9:1 rocker.
Flat tappet cams, both hydraulic and solid, run very high positive acceleration rates( .000400-.000500"/*2) compared to rollers. However. they are limited by max velocity before the contact point of lobe and lifter doesn't exist on the lifter anymore.
One of the signs of a good cam designer is how well you live with this limitation. I keep all my designs .017" away from the edge for over 28 years, and have had some go 100,000 miles on the street. An engine block with out-of-position lifter bores may wipe this clearance out, and therefore wipe out every cam put in it.
Roller cams, both solid and hydraulic, are controlled by positive acceleration rates, as well as base circle radius. Any acceleration rate over .000340"/*2 can cause negative raduii of curvature---The dreaded 'Inverse-Radius' cam, unless special math methods are used.
All of my 28* Major Intensity cams use such techniques.
Any positive acceleration rate under .000250"/*2 is almost a waste of time, although valve train life is tremendous.
An acceleration rate around .000300"/*2 is good for 24-hour racing, if the valve seating velocities are done right.
As is obvious, opening side rates and closing side rates may not agree in positive accelerations or opening/closing velocities, but the nose acceleration must always be the same, or there is a visible line, as well as unequal velocities, at the nose.
It took me 38 years of work to learn all this.
Good Luck!!!

UDHarold
662-562-4933
brookshire@panola.comUDHarold

I break the cam design up into 8 to 12 parts, or segments. None of them are the same duration, and I may end up with 4 on the opening side, and 5 on the closing, or any other combination.
I design each part as experience has taught me the correct values that segment needs to accomplish what I want it to do.
This information is written nowhere, except in my head.
I use a number of different types of equations, from polynomial, to what are called 'simple polynomials', to calculate the lift curve and the various derivatives.
For 27 years I have designed hydraulic and solid lifter cams with constant velocity, or as some call it, dwell at max velocity. The first 15 years I used a Texas Instruments TI85 to do so, and would add 200 or more pairs of 6-digit numbers together to make a cam design, hand-written, of course. These designs, such as the 288/296F5. were UltraDyne's most popular, and some went as much as 100,000 miles on the street, according to the customers.
Rarely, I would design cams the Old School way, with 4 equations, opening ramp, opening flank, closing flank, closing ramp. These were always cams with a lot of duration and not a lot of lift.
There are lots of ways to skin cats........

UDHarold

and Mike's

f you could make everything out of unubtainium and run 1,000# of spring pressure, that would be true, but it's not in the real world.
The acceleration limits of a roller are go beyond the most aggressive flat tappet cams.
If you tried to run the max acceleration on a flat tappet cam as high as I use on my inverse radius rollers, you'd need more spring pressure then a flat tappet cam could stand.

Here's some Max accelerations I just looked up.
Flat Tappets
Comp Cams restrictor plate cup cam: .000294"
Jones Cams restrictor plate cup cam: .000372"
My most aggressive flat tappet cam: .000422"

Rollers
My average IR cam: .000393"
A Pro stock IR cam: .000438"
A asymmetrical IR cam: .000463"

OK, I just went to my Audie Cam Pro Plus files and checked for spikes in the "jerk" curve on a couple of my cams, and a couple of other companies cams.

On the "jerk"(difference in acceleration) curve, my cams had spikes of .000013". My micro-polished cup cam had spikes of .000009". The Big Production cam company's cam had spikes of .000018". Another custom cam company's cam had spikes of .000016".
A 60 year old Winfield 64 had spikes of .000020"
A very expensive "all CNC" custom cam company's cam had spikes of .000015".

I also went and looked at one of my NASCAR designs. As designed, it has a max "jerk" of .000018". that's pretty close to the machining tolorences, and that's why I don't focus beyond "jerk".

Still have A LOT to learn, especially about ramp design (I don't think this one would work in a real engine).

[SchmidtMotorWorks]

Getting the hang of this:

If this is intended to be a limited velocity cam for a flat tappet follower, you might want to make the peak of the velocity curve more level. You can do that by moving the horizontal portion of the acceleration curve to 0.0 in that segment.

What is the concept of the software?
Do it use polynomials to design a lift curve or design an acceleration curve and integrate to velocity and lift?

Where the acceleration curve levels out on the bottom and blends out into a trough shape, you can reduce harmonics by making that more of a uniformly gradually changing shape more like an ellipse.

Don't ignore jerk, the consequences of jerk accumulate over the time that the valve train reacts.

To visualize the consequence of jerk think of towing a car with a bungee cord. If you accelerate the lead car to fast for too long, the cord stretches and builds up energy that releases in an uncontrolled way when the lead car stops accelerating by accelerating the towed car to a higher velocity than the pulling car (assuming the driver in the towed car is not dragging the brakes).

[hoffman900]

Getting the hang of this:

If this is intended to be a limited velocity cam for a flat tappet follower, you might want to make the peak of the velocity curve more level. You can do that by moving the horizontal portion of the acceleration curve to 0.0 in that segment.

What is the concept of the software?
Do it use polynomials to design a lift curve or design an acceleration curve and integrate to velocity and lift?

Jon,

It is intended to be.

It's ST user MJP_Mike's (I see him online here regulary, but doesn't post much) software: http://www.mjpsoft.dk/cam_instructions.html

I believe it uses cubic splines. How I'm using it is to add nodes, then adjust the lift value. This will have an effect on the entire curve. So to get the shape, I place a node in a spot based on its cam degrees from peak lift, then I adjust the lift value to shape the curve. To be honest, I'm really paying attention to the acceleration curve for the most part at first. It's my impression from the reading I've done that this is how a lot of lobe design is done as its a selling point for the Blair software and the Lotus Software allows the use of Bezier Curves to shape the acceleration curve. Interestingly, Harold mentioned looking at the Jerk and Snap curves in his last cam designs, but I'm not there, yet. I'm also mirroring both sides to make it symmetrical. I can turn that off however - baby steps first.

Really, I got to the point where to increase my understanding of the subject, I just need to plug and play with numbers and see what they do. I can do the math (for the most part) outside of this, so I have some concept of the underlying dynamics. Thanks to this blizzard (and subsequently work being closed), I've been able to plug away at this for the last two days. I only have about 8 hours tied up in this so far. Neels recommends it for designing lobes to be used in his software and it's an absolute bargain. It calculates values out to the 8th decimal place.

[SchmidtMotorWorks]

I tried that software a year or two ago, found it very difficult to control the shape of the acceleration (let alone jerk) curve.
This is to be expected in any software that designs at lift.

Blair's software (and one version of Lotus) works differently, they design the acceleration curve, integrate to get velocity and lift.
Then optimize to get velocity and lift to equal zero at the ends of the curves.

[hoffman900]

I tried that software a year or two ago, found it very difficult to control the shape of the acceleration (let alone jerk) curve.
This is to be expected in any software that designs at lift.

Blair's software (and one version of Lotus) works differently, they design the acceleration curve, integrate to get velocity and lift.
Then optimize to get velocity and lift to equal zero at the ends of the curves.

Did you get any further with what you were working on about ten years ago?

Being an order of magnitude and multiples cheaper than either one of of those too, this works great for my efforts. I'm figuring out how to work around that limitation, but no doubt it's more work to do. It's a pretty snazzy program though and I'm glad Mike makes it available - it's perfect for what I need (education and sim work).

[SchmidtMotorWorks]

Did you get any further with what you were working on about ten years ago?

Being an order of magnitude and multiples cheaper than either one of of those too, this works great for my efforts. I'm figuring out how to work around that limitation, but no doubt it's more work to do. It's a pretty snazzy program though and I'm glad Mike makes it available - it's perfect for what I need (education and sim work).

My software is practical for me to use but not refined enough to release as a commercial product.
I work on it a few weekends a year, mostly to make it simpler.
I mostly find myself getting off into the weeds and losing a lot of time on valve-train dynamics simulation, that is really a full time project unless I use some external physics solver.

[hoffman900]

Did you get any further with what you were working on about ten years ago?

Being an order of magnitude and multiples cheaper than either one of of those too, this works great for my efforts. I'm figuring out how to work around that limitation, but no doubt it's more work to do. It's a pretty snazzy program though and I'm glad Mike makes it available - it's perfect for what I need (education and sim work).

My software is practical for me to use but not refined enough to release as a commercial product.
I work on it a few weekends a year, mostly to make it simpler.
I mostly find myself getting off into the weeds and losing a lot of time on valve-train dynamics simulation, that is really a full time project unless I use some external physics solver.

I'm glad to hear you haven't totally abandoned it. Have you still found that it's best to design from the jerk curve and integrate back through?

[SchmidtMotorWorks]

I'm glad to hear you haven't totally abandoned it. Have you still found that it's best to design from the jerk curve and integrate back through?

I have experimented with working at velocity, acceleration, jerk and constant snap.
I always come back to acceleration.

[hoffman900]

Jon,

Looking at Harold's flat tappet acceleration graphs and Harvey Crane's Accelerated Ramps , why is there a reversal before peak acceleration? I'm assuming this is where the lash point would be on these type of ramps?

Edit: Nevermind. Answered here: https://www.speed-talk.com/forum/viewtopic.php?f=15&t=11713

Jon,

These are all tools in a cam designer's tool box.
I have been using dwell at max velocity since 1980, with 90% of the UD flat tappets being that way, and at least 50% of the hydraulics.
The UD solids and hydraulics were done on a Texas Instruments pocket calculator. All of those UD printouts were 2 sheets of notebook paper.
I have never designed a roller using dwell at max velocity---It is not needed---and I know of no designers who do.
All the VooDoo hydraulics and the new solids have dwell at max velocity.
The main questions is always: How much?
This is the only way to achieve high lobe lift with a limited-diameter flat tappet, either hydraulic or solid, and keep a reasonable nose radius.
The only other way is to use higher and higher nose accelerations to keep the max velocity on the tappet, but this requires higher and higher open load on the valve springs, and the nose radius gets smaller and smaller, and the smaller nose radius is less able to withstand the open spring pressures.
Not all flat tappets, solid or hydraulic, have dwell. In fact, looking at all cam companies, probably more do not have dwell than do have dwell. Sometimes when I do hydraulics or solids that have a decent duration and not too high of a lobe lift, I don't bother with dwell either. Some of UD's most successful hydraulics, and some damn good solids, do not have dwell. Dwell is only used when it needs to be used, as it is more difficult than designing cams without dwell.
The 'No-Pulse' ramp is best explained by Harvey Crane, as he thinks it is his baby, and he charges for it, as that is how he makes his living now.
The reversals in positive acceleration are very common, on almost all factory designs, and used by most cam designers. I use them quite commonly on closing ramps. It is just a function of the exact velocity, acceleration, and jerk values you are using, and the duration of the ramp.
The charts on Harvey's website are tremendously informative, but they do not tell you everything, and all the charts are symmetrical. You need to study the charts, and do a lot of thinking.
I hope this has been of help in answering your questions.

UDHarold

I'm not sure what the idea of the no pulse ramp is. As he mentions on the web site, it looks like it doesn't have a ramp at all. I guess there is more to it than is explained by the chart.

I don't know what the idea behind a reversal in the acceleration curve is. I can only guess that it might be to open the valve quickly for a while then avoid valve clearance issues. If it isn't that, maybe it has to do with harmonics.

The two are kind of opposites.

The "no pulse" ramp could be considered a rampless design. It has a relatively high acceleration at the seat, which is fine on the opening where the valve is "supposed" to move.

On "accelerated" ramps, the reversal usually happens BEFORE the lash point. It's a way of acheiving a relatively low acceleration at the seat with a set velocity. This works good on the closing side. It lets the system "settle" a bit before it sets the valve down.

I bet you can guess what most modern asymetrical cam designs look like.

Like Harold said, the Max velocity dwell is a must.
We started doing it a couple of years after Harold(around '83)
We called those flat tappets our E.M.V. series (Extended Maximum Velocity).
I've refined the way we do it over the years, but it's still doing the same thing.
As far as ramps go, spintron testing and my own way of thinking have moved me in a different direction them most cam designers.

Jon,

Look at the middle chart you posted from Harvey Crane's website, the 'No-Pulse' ramp.
There is a line in red towards the bottom of the chart. This is the Jerk curve, and careful study of it will reveal some secrets.
It is not easy to design, but it works extremely good on the opening side of cam lift curves.
The 'Acceleration-Reversal' ramp is used quite commonly on the closing side, and helps ease the lifter down on the base circle.
At UltraDyne, my 288F was master F1, done in April 1980.
The following UD designs were all done using the computer printout of the 288F opening side:
F9 276 243 .3450"
F18 280 247 .3500"
F8 284 251 .3533"
F1 288 255 .3600"
F4 292 259 .3677"
F13 300 267 .3823"
F12 308 275 .3933"

The other, bigger, cams were either 4-equation cams, or else Dwell-at Max-Velocity designs where I traded dwell for a lower nose acceleration and a more rounded nose, for higher open pressures and RPM.
My 288/296H8 hydraulic, .485"/.507" valve lift with 1.5s, was my first hydraulic with dwell at max velocity.
These were my more successful cams for all of UltraDyne's life.

UDHarold

The way I design asymmetricals is completely different then the way other cam design them.
Just take the clearance ramp off the opening side, and you have the way most guys design asymmetricals.

On the solid lifter profiles, there is a little jog before the angle changes about 80% up the first increase in acceleration, is that incidental or a feature that needs detailed control?

That would be the opening lash point. Although the opening is considered "rampless", the lash point has specific values for velocity, acceleration, and jerk that Harold chooses. The jog, and even the reveral on the closing, is simply a result the choosen ramp values.

Here is the acceleration curve for an aggressive hyd. roller I just finished.

Jon,

All my curves are done by a 5-exponent polynomial equation, Dacaman12's cam done by an Excel spreadsheet he wrote himself.
The 'little blip' is the lash point. Later on, I have learned how to get rid of it.
Even with the'little blip', solids that I have designed have gone 100,000 miles on the street, and won all sorts of races.
I want to thank you for getting the pictures out. Let me know what I did incorrect in publishing them, so I can ease your work load next time.....

UDHarold