Grinding Cam profiles on a DeskCNC 4 axis mill

[NC Cams]

All the profiles have randomly located issues in them. The velocity curves are visibly rough and choppy and the jerk curves are more than choppy enough to make you seasick. We'd simply NOT ship/do something this rough.

The issues could be grinding related or they could be the result of deficiencies in the measuring proces. I'd be remisss to criticise either or since I've not measured the cams on my equipment - hence, I can't speak for the veracity of the mesturements or the dynamic predictions that would evolve from your mesaurements and/or equipment. If they are, in fact properly done and correct, I've seen and made much better.

I can say the we'd simply not grind or ship cams with the velocity deviations that are evident in the ramp and other areas of the master and copied cams. Our stuff tends to be MUCH smoother and continuous.

These issues could be the result of backlash in a gear driven system or runout or tooling deficiencies or any number of other little or substantial things - I can't tell from here but there are clearly "issues" that we'd not accept or tolerate.

We've ground cams using a supplied part as a master. Do we simply trace it as I suspect was done here? NO. We read the cam and then do selective fixing and/or smoothing. Is this effective? Our first "daytona" cams were just such copies. The client reported that our "copycat" cams made more power than the originals.

When you copy something, it is very hard to NOT copy and or to expound upon deficiencies that may previously exist in the part. If you first remove these deficiences,you can turn an adequate part into agood one. Sometimes, it is actually better to design something from scratch as there may be too many issues involved or it is simply too hard to fix what's wrong.

I don't know enough about your power requirements ("I want more" is not enough info) or the specifics of your valvetrain and/or engine. You may be looking for something from an engine in an RPM range that it was NOT specifically designed to operate.

Example. A 12000 +/-500 rpm peaky bike engine may NOT have the best power curve when you really want/need a broad, flat torquier curve that offers more grunt over the 6000 to 10000 rpm band.

Looking at a simple lift curve won't tell you how to address those needs. You may be in need of a totally different tuning strategy (shorter cams, wider or narrower lobe centers, whatever). At that point, a "custom" as opposed to a copied race bike cam may much better suit your needs.

We offer custom cam design, manufacturing and design services. We'd be happy to quote these services as needed to tailor your engine's power curve more appropriatedly to your needs.

These services could include inspecting sample cams and making specific recommendations based upon a TOTAL review of your valvetrain (springs, weights, etc) and desired operating needs. Feel free to contact me via PM for a quote for our services.

[FPV_GTp]

I am presently upgrading my mill to 4 axis.

On the spindle head (Z axis), I am going to add a diamond wheel grinder.

Right now I am not too sure how to scan an existing cam profile (from an engine intake/exhaust camshaft) into the system.

I am using a 2500 CPR encoder on the A Axis. so I would like to take a measurement for up to each of the 10000 steps in a 360 degrees rotation by a linear encoder head putting out a signal onto the encoder input of the DeskCNC controller card (same as used as feedback for the lathe spindle application). The output of this linear encoder (I still have to purchase it) has a resolution in the 0.001mm range. Then I would like to graph the resulting cam profile, eventually modify it and then produce a G code file to grind another camshaft lobe based on it. The grind is done just by rotating tha A axis and moving the grinding wheel on the Z axis. No need to move the X axis for it.

Is it actually possible to do so ? I could not find any hint on the docs. Available on the web. If yes, is there somewhere a similar application I could use for reference ?
Until now I can only find samples of scans done with the touch probe connected to the Lim input but I am afraid this method is not as accurate as if I go for the linear encoder even though I already got 2 sets of the IMservice touch probe. Also in the literature nothing hints about using the touch probe for parts rotating on the A axis.

Attachment 41427

Thanks in advance for nay hint.

Paraprop

Daniel

Hi

speak to this guy at http://www.sportdevices.com/
they had some cam-profiling device that enable scan camshafts , i couldn't find a direct link to the page .

cheers

[Paraprop]

Given that the first cam of the 3 is a Honda production cam, I should be reassured that part of the results shown should come from the lack of accuracy of my measurement method right ?

[NC Cams]

Perhaps.

IT could be technique, lack of proper preparation or even a bogus part although the Japanese stuff I've measured is darn near jewelry in quality.

It is not easy to read cams. You don't just throw them in the machine, spin it around and wah'lah, get the data.

It took us WEEKS of experimentation with known good/pre-read cams to find the proper way to read a cam. We also spent a lot of time eliminatng deficiencies in our equipment and yes, even the software.

Unless you have a known good baseline cam to use as a yardstick/reference, you can't really know if you can or should trust your measurements

[M-man]

WHat are the tolerance of the profileshape on a normal cam?

[NC Cams]

There is no such thing as generic tolerances for the profile shape of a "normal" cam.

They vary from profile to profie, engine to engine and manufacturer to manufacturer. Some engines are more sensitive to others with regard to lift deviations and others are more sensitive to abnormal changes in the velocity, acceleration and/or jerk rates.

You could have a BIG lift difference but a small acceleration and/or jerk deviation and the engine won't know the difference. Yet, with the opposite condition, you'll tear stuff up.

We supply the velocity plots to our payiing clients ONLY and ONLY when the client specifically requires it in a certified cam data chart - it is included in the quoted price of a/the cam cam analysis service and this is not necessarilly inexpensive.

Analysis price goes up if/when the cam has "issues" and we have to figure out if it is a reading problem or a manufacturing problem or simply a bad design - after all, you're paying for "right data" and a bad analysis does no one any good.

We do not offer such technical info outside that which we supply when we analyze a clients cam on a contract basis. Very few if ANY of the commercial cam grinders even supply velocity or acceleration data as standard or even optional fare to their clients - no matter what you pay for the cam.

We don't supply it to our NASCAR clients unless they are a VERY good customer and need it for problem solving purposes. in short, you ain't gonna get free cam tolerance specs on a M/B. No cam maker in their right mind gives this data out.

Why do you think that Audie, Andrews and/or Cam Doctors were created???

Answer: because cam makers have ALWAYS been notoriously close mouthed about ANTHING that has to do with the shape, dynamics and/or tolerances of their profiles. Period. Paragraph.

[vmax549]

Paraprop I have been doing what you want to do for years. It is within the realm of your Deskcnc system with the 4th axis and TP100 probe. Basically you will center the cam under the spindle and then locate a starting point on the cam . Next locate the depth of the base circle and set to Zero. Then rotate the Aaxis .025deg at a time while probing down with the spindle at each .025deg rotation. This will give you a point cloud of the cam lobe tied to a reference point on the camshaft. Next load up the points cloud in the Cad program to do the smoothing between points, THere will be a ton a points to deal with.

The grinder can be a simple wheel mounted stationary next to an x/y table that uses your Aaxis to rotate the cam for grinding while the y axis provides the interpolation to grind the profile.

Now everything HAS to be aligned with procession as far as the center of the probe to the centerline of the cam for probing and the center of the wheel to the center line of the cam for grinding. Being you are doing it straight line without a rocking arm it is fairly simple with no translation needed for the arm movement.

I would suggest you build a seperate grinder. DOn't use your mill to set up for grinding. The micro dust will eat your mill alive in no time.(;-)

Is it NASCAR perfect NOPE, will it do what you need? Absolutely Been doing it for years in small engines turning into the 10k range without problems.

We KNOW kart racing engines(;-)

(;-) TP

[Paraprop]

Vmax549,

Thanks a lot for these motivating news.

How do you fool the DeskCNC software to digitize out on the A axis ?

Are you using 3D mode ? (2D does not seem to work)

Do you switch the A to be the X axis and define Ymin and Ymax to be 0 ?

Then set Z "min" to be the base circle and Z max a bit more then the Max lift ?

Good comment about dust. Indeed it is not a problem to mount the grinder onto a dedicated ball screw setup.

Indeed my kart engines do not go much over 8000 RPM


Paraprop

[NC Cams]

The method proposed by VMAX549 will ONLY work if you are cutting a flat tappet cam with the bottom of a bottom cutting cutter. Reason: this is pretty much how a flat follower "sees" the profile.

However, if you have a roller cam or a finger follower cam, you won't even come close to duplicatiing the lift profile that you're intending to achieve. To cut the lobe you have to compensate for the fact that there is a geometry difference/offset that has to be accounted for between the cam and the radius'd follower. This is why a roller cam and a flat tappet cam of comparable durations are SO, SO much different in shape.

FWIW: we tried the 0.25 deg method and, yes it will cut/shape a cam. Yes, the engines will run BUT said cams tend to generate/impose a LOT of superfluous vibration into the valvetrain - so much that you can actually HEAR a difference in the how the system sounds while running.

The sound that the 0.25 deg "flats" (that's what you're effectively cutting) make resemble a "machine gun" to some extent. Imagine what that superfluous shaking does to a valve spring at high RPM.

NASCAR performance? To the go cart racer, their local race track win is as satisfying as any NASCAR win and, due to the $$$ consequences, a bit MORE highly valued.

This is why one of the top go cart cam grinders uses the exact same method to grind the go-cart cams he makes as he does for the high dollare V-8 cams he grinds for pro racers.

VMAX549's admonition to craft up a grinder is well said advice. The HSM pictured device should be scaleable to do what you want to do with relative ease.

Noticeable difference? If you do not consider an end of race drop off in race performance due to killed springs inconsequential, then there is no difference between a ground cam and a 1/4 deg cut cam. However, in our work with 5hp Briggs race engines, this sort of diminishment of performance and/or its affect of part durability was deemed unacceptable.

BTW if anyone saw the inspection traces of a 1/4 deg cut cam as opposed to the exact same profile generated with a ground cam, even a neophyte could readily see the difference between the two. It is very easy to choose/pick the smoother ground shape from the cut shape. When given the A vs B "blind taste test"' choice, absolutely NONE of our clients EVER chose the cut cam over the ground one when we offered them the choice 'tween the two.

The key phrase in VMAX549's post are, ".... do the smoothing between points, THere will be a ton a points to deal with...." It is easy to get/generate 1 deg lift data and have Excel graph out a smooth lift curve that SHOULD occure. It is HARD to turn that into a SMOOTHLY transitioning lift profile on a hard metal cam that does not drive the valvetrain into spasoidal fits of herky jerky motion.

ANY cam grinder learns that lesson after a while as pretty much all of them made model cams and/or masters using the interval flat cutting method over the years - most have simply gone away from it for reasons of quality, durability and speed. It has become easier and FASTER to CNC grind them when you have access to the "right" equipment.

BTW, We do it (make cams) day in and out for small, medium and large single and multi-cylinder engines that work in stock, race and prototype applications.

[vmax549]

Paraprop, I am using MACH3 cam software. It has a direct probing cycle G31.
Deskcnc uses the bed of nails approach I believe. Ask DeskCNC for advice there. Or you can do it another way, that is to switch the Y axis to the Aaxis and reset the steps to get A rotation when the Y movement is called for. Then you need to calculate the circumference of the Max lobe radius from the centerline to be able to set the A axis envelope so it represents the circumference of the lobe. Not quiet as easy but can be done.

Another thing to consider is to probe a point between the two teeth on the cam gear that has the timing dot. Probe it using the Aaxis for rotation, set the center between the teeth to zero that will give you the reference point needed so the two lobes will be timed to the gear correctly. When you grind it, do the same thing again to set the zero point of rotation.

The smoothing of the points in the point clouds is neccesary to ensure there is not the railroad tie effect in the profile(;-) once smoothed the CAM Doctor(WITHIN reason) nor the engine could tell the difference .

Just For what it is worth (;-) TP

[Paraprop]

NC Cam,

Thanks. Indeed I will only grind simple flat tappet cams. I cannot cut it with a flat cutter because I have to weld up the original cam lobes with a Rc 62 TIG welding rod else the cam lobe would not last. (my apologies if I used the wrong unit as it is a year ago that I sourced these welding rods).
I guess that using a properly aligned grinding wheel with a movement perpendicular to the cam axis should yeld the same result than a flat mill cutter.

Given that I can go yp to 10000 points per 360 degrees, I can theoretically achieve 0.036 deg. large " facets" around the lobe. That is about a 7 times finer finishing.


I do well understand your point about the quality of the end result and won't argue there. You had well highlighted the issues in the threads covering cam grinding. Nevertheless, those parts being manufactured in Thailand, they do cost very little in comparison to the U.S.

On the other hand getting a specialist in the U.S. to make me a special grind will set me back about 80 to 130 $ + shipping + 20% import duty + 7 % VAT + duty clearance agent fee. At the end, it is more than an entire new engine costs here. Quantities needed are limited too.

Paraprop

[NC Cams]

Several coments:

Simply welding HARD metal onto a lobe is not necessarily adequate for cam profiles. Reason: you have to crank the engine under marginal lube and high loads to start them. The metals have to have high scuff resistance during this guaranteed period of marginal lube. The ONLY weld material that we've ever seen work on welded cam lobes is Wahl Colmonoy #5 or the next harder alloy - I think it is #56.

Single pass welds won't hack it due to weld diliution of parent metal. You need dual pass welds at a minimum. Proper post weld stress relieve is MANDATORY. Yes, the welded cams do work - but for how long. We've found that it is not an isssue of if the weld will spall but, rather, when. Low stress applications may be more tolerant - all our stuff is high stress. Sooner or later, ALL the welded stuff we've worked with starts to come apart or, worse yet, the cams break due to the welding affects on the parent metal.

For this reason, alternative methods involving steel billet cams were developed for restorative or custom prototype applications that offer better performance, all things considered.

WHether it be steel billet or welded OEM cams, ultimately, a proper post machining/welding heat treatment is absolutely essential/critical.

Yes, you can THEORETICALLY get 10000 points MATHEMATICALLY. However, your ACTUAL machine resolution probably will not/can not be that fine. You'll be lucky to hold 0.000010" and more likely 0.0001" at best.

Again, we've been thru this already with high buck servos and steppers as well. High resolution servo's will work if you run them slowly to keep follower error low.

We've YET to see steppers NOT yield step free finishes on cam profiles. Interestingly, we've also seen them leave 'steps' on OD ground journals. Steps that are visible to cam inspection measurements as well as the naked eye.

Comments re: so called "custom grinds": We are one of the very few cam grinders/designers who does NOT offer "cookie cutter grinds". All our grinds are SPECIFICALLY designed for the specific client using THEIR parts, their operating conditions and their valve train geometries. However, just like tailor made clothing, tailor made cam profiles are not inexpensive.

So called "special grinds" offered by name grinders are essentialy "off the shelf" generic grinds that are, at times 20 years old or even more. Depending on the legacy of the design and the design philosophy used at the time, they can be OK or perhaps superb or, sadly, parts breakers.

Frankly, we've had chances to inspect/measure some of these gems and found some to be quite lacking in sophistication and/or development. Some aren't much better than stock profiles. Some are works to be proud of. There is no way to figure out which is which by looking at catalog specs.

For your application, I'd be a bit weary of anybody's $130 "custom ground cam", especially a reground or welded one.