What's new
What's new

In process grinding measurement

motion guru

Diamond
Joined
Dec 8, 2003
Location
Yacolt, WA
We have a rotary surface grinder that we would like to improve accuracy of finish thickness to +/- 0.0002 inches.

The parts are Rc60 - Rc65 and we are finding that the wheel loading changes significantly between wheel dressings and downward force on the wheel / chuck interface varies from 800 lbs on a freshly dressed wheel to over 3000 lbs when the cutting performance of the wheel requires dressing again.

This variation in downfeed force creates enough deflection in the chuck, grinding wheel mounting flange, and present measurement system to make hitting our desired tolerances a bit of a crap shoot resulting in under ground parts by as much as 0.0008 inches . . .

Has anyone used the Marposs inline measurement system?

MARPOSS universal measuring gauge for grinders

As I see it we could mount a measurement head on the chuck and another would differentially measure the parts being ground. Marposs says an interrupted surface is no problem as the parts go by on the rotary chuck. We can have as many as 16 parts on a chuck spinning at 25 rpm (one part hitting the sensor every 150 msec). Marposs indicates 0.1 um accuracy in this kind of application which is roughly 50x better than our desired tolerance so resolution and accuracy should be more than adequate.

I would be interested in any experience forum members might have with this or a similar product.
 
Hi Motion,
Wouldn't the change in force from dulling of the grit introduce a "springing" between the chuck and grind head, such that you may have a very shallow cone rather than a planar surface to the chuck face? Are you confident that even if you get the correct measurement at the point of contact that the rest of the part would be good?
 
Hi Milland,

Two things . . . yes, there is “pinch error” when grinding these parts, but it is a fraction of the tolerance and well within acceptable limits.

Also the geometry of the wheel and part fixturing is such that the wheel is in contact with multiple parts at any given time with the goal that the total area of parts in contact with the wheel is relatively constant as the chuck rotates.

According to the Marposs literature, the measurement is done using an algorithm that filters and evaluates a moving average of 25 parts and when the difference between the measured chuck height and part height hits target it sets a digital output. When we receive the digital output, we will back off until the spindle load drops to 20% and run for a set number of chuck revs (spark out) so that surface finish is optimum and then lift the head off the parts and transfer the chuck out into the open for the operator.

The probe actually drags across the longest dimension of the part. Hard to believe how accurate it is when you see it bouncing off the chuck surface, then bouncing along the part surface and then back onto the chuck in an interrupted fashion. They must have some really good digital filtering on the signal.
 
Fair enough, and it sounds like Marposs has a good process that would work for your needs. While I haven't used their products myself, everything I've read about the company has been positive, so I'd not hesitate to take their word on suitability.
 
While expensive Marposs is as good as it gets for such measuring. I have worked with their in-process systems since the 70s.
A bit not "normal" that you need a difference setup. Even under the mentioned huge load differences usually the table is stable and the deflection is up top.
The bouncing probe will wear but this is easily comped with a decent UI.
Two tenths means you want 20 millionths in size control and the Marposs units should handle this.
I'm assuming the feed arrangement is capable of a stable 5-10 millionths.
Trying to back off until the spindle load is at 20% may not be a good idea. A predictive algorithm based on load and variable sparkout might work better.
Attempting to "unspring" the loaded machine assembly can be difficult and spindle power is not a good indicator.
I now never attempt to "back off" in such applications. It did seem like a good idea to me a while back and I fought with it. Not so good results.

Not sure the machine design or part size but sometimes they "twist" when loaded so part size inner to outer moves and the probe only sees what it contacts.
If this happens, with all the fancy measuring, although they all check zero in the center parts off a new dress and parts off just before a dress will be different.
Bob
 
We/I ran to .0005 (sometime better)with spotters on the Blanchard chuck. I would paint my + .005 blue and my +.001 red... The +.005 was the watch-out and the + .001 start measuring.

-Perhaps wheel selection may be needed with your wheel loading up so much might go a couple tads softer .

with a fixturing a marposs could intersect the spotter and give size to make finish..if you could find an area out of the grit throw.

Seems we had a maposs on a grinder..trying to remember what grinder it was...


Perhaps the Shipman SG.. or one/few of the production grinders

Qt (one part hitting the sensor every 150 msec).
Perhaps the spotter hitting the sensor..but yes now see your telling about taking a running average that the better method because one might see the size coming up...
Agree Rc60 is a tough grind.
 
Agree Rc60 is a tough grind.
This'd cost a fortune but ... we had an Arter set up just for grinding shaper cutters with a cbn wheel. CBN cuts where a vitreous wheel rubs it off. You can touch a running vitreous wheel. Try that with cbn and there will be blood.

Or possibly try those Cubitron II wheels first ? Somewhere I have test results showing about a 4:1 difference in productivity and accuracy when form grinding carburized (61 - 63 Rc) gear teeth using the Cubitron wheels. And the tests were done by the user, not the vendor.

The Marposs is still cool but a better wheel wouldn't hurt, either.
 
We/I ran to .0005 (sometime better)with spotters on the Blanchard chuck. I would paint my + .005 blue and my +.001 red... The +.005 was the watch-out and the + .001 start measuring.

One can hold 5 tenths on a 11 or 16 Blanchard with a 10 dollar per hour operator all day long on carbide insert sized parts.
No in-process gauging, just a digital readout on the head.
Do this or loose your job.
2 tenths is different. Dust, dirt, burrs on the chuck load face....
Do you need to clean, stone, acetone swipe, lint free wipe and blow down the bottom load surface every run? You have a micron to play with.
The jump from 5 tenths to 2 is not to be taken lightly and I would put it at close to 10 times harder to do.

As a machine builder the OP needs a six-sigma inside his customer finish requirements.
Many can do 5 on an rotary grinder and you need a better wheel or this and that. Duh, my dog can do that and give such advice.
Think 20-50 millionths as your target and what you "feel" you can do or you won't meet the spec.

My Blanchards have 50 millionth scale readouts.
I can not consistency nail under 2 tenths on them on any sort of repeatable high production bias within the curve.
Now we have only been using them with such for 38 years here so many years to learn still.
When we are inside or at this is very, very hard work and the risk for scrap is high.

We don't know much about the OP needs, he simply asks about the gauge system for a good or bad.
Bob
 
This'd cost a fortune but ... we had an Arter set up just for grinding shaper cutters with a cbn wheel. CBN cuts where a vitreous wheel rubs it off. You can touch a running vitreous wheel. Try that with cbn and there will be blood.

Or possibly try those Cubitron II wheels first ? Somewhere I have test results showing about a 4:1 difference in productivity and accuracy when form grinding carburized (61 - 63 Rc) gear teeth using the Cubitron wheels. And the tests were done by the user, not the vendor.

The Marposs is still cool but a better wheel wouldn't hurt, either.

Our customer has asked us to consider a CBN wheel and we would be happy to do it, but setting up high pressure coolant and mist control isn't something we want to experiment with - we have to get this machine done and off our floor.

The wheel is a 350mm diameter type 52 Radiac 80 Grit epoxy bonded wheel . . . any idea if a CBN wheel of this configuration is available?

Are the grinding forces much lower with a CBN wheel?

350mmDia.jpg
 
....

Are the grinding forces much lower with a CBN wheel?

Generally higher than a clean dress on other steel wheels but more consistent.
You can buy a CBN resin bond in any shape or size you want. Same making as diamond and I buy really strange stuff all the time.
The maker just has have cores for the press dies in in your size.
Big ass wheel and small part if the picture is correct. 350 does begin to limit the people pressing this size as the press tonnage is high but it not real big so still many sources.

One would guess resin bond or you could go vit but how to dress/true it on the machine?
A Blanchard takes care of this being flat in a few runs, here the part is small so it's more like a big SG doing face grinding.

From the one pic if your mechanics are good holding a 2 should not be a problem if your stock is decent and removal is not huge.
Then again, maybe you are taking a quarter inch off this part, in which case I'm full of shit way past my eyebrows.
Bob
 
The wheel is a 350mm diameter type 52 Radiac 80 Grit epoxy bonded wheel . . . any idea if a CBN wheel of this configuration is available?

Are the grinding forces much lower with a CBN wheel?
Now that Carbide Bob slapped me I should shut up :D but from your description I still think a better wheel would help.

So back into the breach ... all the cbn wheels we had were made special. They make an aluminum base to your specs then coat them on the areas where you needed abrasive.

Yes, they took less force and they lasted a looong time. You could easily feel the difference. The finish is different, too. Not smeared. But that could have been a grit difference, so I should not say "Yes" so positively.

Unlike vitreous wheels tho, they never loaded up, which seems like your application is doing.

However, for a quick and dirty, the cubitron ii would be a lot easier to come by and a lot cheaper to try. I can't get to the charts right this second but yes, we're talking form grinding tooth profiles to two tenths or less so seems like it should be applicable. 2 meter diameter, carburized 9310 better than agma 12, an eight-hour grind was reduced to two. Pretty impressive.

I'd call the GE guys and see if they can give you some info. Hofler is no slouch in grinders, either. These tests didn't start from a base of South Bumston Beginner Machining School recommendations, they know their stuff.
 
Generally higher than a clean dress on other steel wheels but more consistent.
You can buy a CBN resin bond in any shape or size you want. Same making as diamond and I buy really strange stuff all the time.
The maker just has have cores for the press dies in in your size.
Big ass wheel and small part if the picture is correct. 350 does begin to limit the people pressing this size as the press tonnage is high but it not real big so still many sources.

One would guess resin bond or you could go vit but how to dress/true it on the machine?
A Blanchard takes care of this being flat in a few runs, here the part is small so it's more like a big SG doing face grinding.

From the one pic if your mechanics are good holding a 2 should not be a problem if your stock is decent and removal is not huge.
Then again, maybe you are taking a quarter inch off this part, in which case I'm full of shit way past my eyebrows.
Bob

The machine is programmed to auto dress the wheel, then acoustical touch off, then get back to grinding. We have a hydraulically loaded dressing point that stops against a precision stop and traverses the wheel using parameters from the HMI.

Never remove more than 0.007-0.012 in rough grind, then 0.001 or so finish grind.
 
Control Gauging is another option.

If you have an interrupted grind this is going to possibly be problematic with the probes coming on and off the part. Every Mattison Rotary or Blanchard I have seen that started its life off with Marposs gauging has had it removed or rendered not functional. I bought several Blanchards from Stanley Tool in the UK that all had Marposs, not a single one was functioning when we removed them from production.

As we have spoke in the past, once you start chasing tenths on a "blanchard" style machine, even a small one, you will be going crazy.

I second CarbideBobs's #8 post.
 
Our customer has asked us to consider a CBN wheel and we would be happy to do it, but setting up high pressure coolant and mist control isn't something we want to experiment with - we have to get this machine done and off our floor.

The wheel is a 350mm diameter type 52 Radiac 80 Grit epoxy bonded wheel . . . any idea if a CBN wheel of this configuration is available?

Are the grinding forces much lower with a CBN wheel?

View attachment 243504

Hello:

The Norton seeded gel and the already mentioned 3M cubitron abrasives are designed for your application. Either abrasive will be less expensive than a conventional aluminum oxide abrasive because of the increased wheel life and higher machine throughput. These abrasives require higher grinding wheel pressures to be self sharpening.

The grinding pressure can be selected based on the abrasive mix. The Norton seeded gel is available in 10%, 30%, and 50% mixes with conventional aluminum oxide abrasives. The required grinding pressure increases with increasing concentration of the seeded gel abrasive.

Another alternative is to resort to continuous dressing of the wheel. This is done for creep feed grinding. It would also work in your application. You described a handheld continuous dressing technique in a earlier post about this project. Continuous wheel dressing will not provide the wheel life cost savings or higher throughput of the modern abrasives.

You may find that these abrasives are a stock item in your wheel size.

The Marposs gage installation is an expensive and time consuming approach to fix what appears to be a less than optimum abrasive choice.

Radiac does sell ceramic abrasive grinding wheels. The web page is Radiac Abrasives | Abrasives - Radiac Abrasives I suspect that they are using the 3M abrasive. The type 52 wheel described in the original post specifies the wheel geometry ( inserted nut- cylindrical shape) and not the abrasive.
 
once you start chasing tenths on a "blanchard" style machine, even a small one, you will be going crazy.
Do you remember that little benchtop (probably only weighed twenty-five hundred pounds, Blanchard version of benchtop :D) machine they made for doing silicon wafers ? That thing should hold tenths ... it was adorable, too.
 
Do you remember that little benchtop (probably only weighed twenty-five hundred pounds, Blanchard version of benchtop :D) machine they made for doing silicon wafers ? That thing should hold tenths ... it was adorable, too.

That little 12" grinder may have been able to hold this on very small parts. B&K had one of these at IMTS.

As the machine size increases along with part size, then toss in harder material and hardness and you have many more factors which influence tolerance.
 
Do you remember that little benchtop (probably only weighed twenty-five hundred pounds, Blanchard version of benchtop :D) machine they made for doing silicon wafers ? That thing should hold tenths ... it was adorable, too.

Bliley electric had something very similar to grind crystals.
Although IIRC they would rough them on that machine,
finish to size by lapping.
 
The wheel is a 350mm diameter type 52 Radiac 80 Grit epoxy bonded wheel

View attachment 243504

Motion Guru has gone off and fixed the problem I guess.

There is a major problem with this story. Resin bonded wheels with aluminum oxide abrasives only have three applications:

They are used in cutoff wheels which operate at very high wheel pressures and temperatures. The high temperatures degrade the resin and allow the worn out abrasive to drop out.

They are used in machines which have vibration or shock loads such as the regulating wheel on a centerless grinder. The wheel is not designed to wear.

They are used when a mirror surface finish is needed. The goal in this instance is to retain the dull abrasive.

Blanchard grinders use very porous wheels. This is required to allow the large grinding chips someplace to go and provide a path for the coolant to reach the work.

A resin bonded aluminum oxide wheel should not work at all on a Blanchard grinder.

The rules are different for a CBN wheel. The sharp CBN abrasive produces a much thinner chip . The abrasive can also be self dressing. A porous wheel is not required in this instance and a resin bond is the best choice. In a roughing application a 120 grit CBN abrasive would be substituted for a 60 grit aluminum oxide abrasive.
 








 
Back
Top