What's new
What's new

General questions about rotary surface grinder

motion guru

Diamond
Joined
Dec 8, 2003
Location
Yacolt, WA
We are nearing completion of a rebuild of a Surface Grinder with automated controls and the more we get into this thing, the more questions that I am having about the final configuration of the controls and wheel dressing.

1.) Dressing the Magnetic Chucks - our customer provided a grinding "wheel" to dress the magnetic chucks. After rebuilding the gearboxes, all new bearings, shimming the mounts, etc. the chucks ran true to +/- 0.002 inches or so. The grinding wheel seemed only to be effective at polishing the chucks with roughly 1 hour of grinding required for every 0.0005" of material removal. We noted that if we dressed the wheel it seemed effective for about 75% of material removal in the first few revolutions and then went back to polishing. I bet we worked on one chuck for at least 4 hours before we changed approaches. One of my employees grabbed a dressing stick and just held it against the portion of the grinding wheel that was not over the chuck while it was grinding and this made a huge difference. More material was removed in 10 minutes than in the prior 4 hours. We ended up using this approach on both rotary chucks and now they are great

3.jpg

We have a diamond stylus for dressing the wheel mounted to a small hydraulic cylinder with guiding rod bearing for "automatic" dressing operations, but again, this seemed only effective for a minute or so before grinding effectiveness was reduced to almost nothing. We tried chuck rotational speeds from 5 RPM to 45 RPM . . . no real change. Is this normal? Was the grinding wheel loading up with the brass? If it is normal - should we design a dressing stick holder that would press the stick into the wheel at a programmable interval / force to improve grinding efficiency or is this kind of scenario only going to happen when grinding a continuous surface like the magnetic chuck / or when encountering non-ferrous materials like the brass rings on the magnetic chuck?


2.) The original OEM controls had the Spindle Downfeed as (+) to go down and (-) to go up . . . this seems backwards to me. Any reason why a grinder axis direction would be configured like this?

3.) As we consider how to control the downfeed - I am of the opinion that we have the operator set the maximum motor load they want to see (something like 75% or thereabouts) and the downfeed should automatically adjust the rate downward to keep the spindle loaded to this level. Once the acoustical sensor detects that we are within some distance of finish dimension, it would transfer over to finish grind feedrate until the dimension was reached (we are using an acoustical sensor on a linear axis to determine grind wheel position and thus thickness of parts being ground). . . does this seem reasonable? The spindle motor is rated at 40HP, the wheels are 450mm diameter, and typical grinding speed ratings of the wheels is between 28 - 35 m/s
 
1.) Dressing the Magnetic Chucks - our customer provided a grinding "wheel" to dress the magnetic chucks. After rebuilding the gearboxes, all new bearings, shimming the mounts, etc. the chucks ran true to +/- 0.002 inches or so. The grinding wheel seemed only to be effective at polishing the chucks with roughly 1 hour of grinding required for every 0.0005" of material removal. We noted that if we dressed the wheel it seemed effective for about 75% of material removal in the first few revolutions and then went back to polishing. I bet we worked on one chuck for at least 4 hours before we changed approaches. One of my employees grabbed a dressing stick and just held it against the portion of the grinding wheel that was not over the chuck while it was grinding and this made a huge difference. More material was removed in 10 minutes than in the prior 4 hours. We ended up using this approach on both rotary chucks and now they are great

********************************************************************************************


Hello M.G.

There was no mention in your post about the use of coolant. There may have been a strong temptation to avoid mixing up a batch of coolant for a quick chuck grind.

Robert
 
Hi Robert,

Yes indeed - we mixed coolant per instructions 30:1 . . . the tank is roughly 5 feet square and about a foot deep.

Ken
 
During the initial grinding attempts, was the wheel stopped after it "went dull" and checked for loading? If brass particles were found embedded in the wheel then it may have been the soft materials plugging the grinding face. Also, coolant nozzles sometimes must be optimized to ensure coolant is getting to the cutting action, not just splashing against the OD of the wheel.

It can be an issue of too low a jet velocity allowing entrained air from the wheel to "push" the coolant away, so it appears you're getting effective coolant flow, but not into the work zone.
 
No, we didn't stop the wheel and take a good look at it. We lowered the wheel down to the chuck watching the spindle current to know precisely when we actually made contact (incrementing down 0.002 inches at a time).

The spindle head is counterbalanced and raised/lowered with a 60mm acme thread and it has some backlash so from the time we first make contact (as noted by the spindle current going from zero to about 10 percent) to the time when we actually see the spindle current rise beyond about 10% is about 0.015" of further incrementing downward. Once it starts to rise predictably with downward incremental moves (we know the backlash is taken up), we load the spindle to roughly 50% of rated current and see a bit of progress in the first 30 seconds or so and then progress was painfully slow after that.

I imagine that we are putting roughly 10-15 GPM of coolant at the wheel / chuck interface when running. The flow is more of a pour than a jet.
 
A coolant "pour" may be part of the issue, it'll flood the area but may not get into the actual grinding interface. Check out the Grinding Doc on Youtube and CTE, he's got lots to say about setting up coolant:

Grinding Doc | Episode 81 | Diagnosing shoulder pain - YouTube [search the list of videos for coolant related ones] And here's one: Episode 1: Coolant velocity vs. flowrate - YouTube

Ask the Grinding Doc: Loading the wheel | Cutting Tool Engineering

On the acme screw, I'm a little surprised you didn't fit a ballscrew replacement. Was there a reason? Does this have to remain manual control?
 
The acme Nut is integrated into a right angle gearbox - the screw is clamped above and below the gearbox on the Spindle Raise / Lower assembly and does not rotate. This is a low buck rebuild with a focus on automation for a proof of concept so we spent money where needed.

We did add a linear encoder to the spindle downfeed axis but the encoder OEM shipped the wrong part. When the right part shows up, the backlash will no longer be an issue.

I went out and cycled the pump on and off a few times, very weak flow. It occurred to me that it might not be phased correctly so I rolled two of the motor leads and this made a huge difference . . . we have far more flow AND pressure now. Thanks for the links to the videos - I'll have a look!
 
Gotcha. If it ever comes to doing a more comprehensive build you might still want a ballscrew, but it may be annoying to integrate. I have a couple older Spitfire lapping machines that I intend on converting to ballscrew downfeed with some sort of control, but it's on the "to do later" list. Like a lot of things...
 
You don't need a special wheel to dress the chuck. Throw that POS away and use what you will grind with. The "dressing wheel" is either too fine, too hard, or both.
 
If the coolant is fed into the center of the wheel and the inside of the wheel is not off the outside of the chuck you will tend to "float" the wheel so make sure there is somewhere for this water to go. A 10 inch circle loaded with 30 PSI of water with no escape is a fair amount of upwards pressure.

The wheel dress for grinding the chuck should be a very coarse dress.
Best done with a single point diamond at very fast feed and the wheel lifted on the return pass.
You want a "record player" type finish on the wheel surface.
If the ways are straight often you will grind down only half the chuck and then step in finishing the center until you are dusting the outside. Bigger chucks may need three cuts, we grind , color with a sharpie, step in.
This reduces the arc of contact.
My wheels are designed to grind parts much smaller than the chuck, not full size plates (Think toolholders) so they exhibit the same problems you are having when grinding the chuck surface.
I also routinely grind (dress) chucks with a (gasp, WTF) diamond wheel. They don't like it but faster than changing wheels. You throw away a few bucks.

There is the continuous hand stick method, not recommended as a pain and just plain not very safe. Once this stick bites ya you will never want to do it again not to mention you get very wet in there.
Knowing full well your companies safety orient and protocols on controls, does this seem like a good idea? Done wrong this will take your fingers off. Sure we all do such it but once every 10 years......why do I need flash protection....
Do no harm and our end users may be a bit below the average end of the scale.

The wheel is obviously loaded, wrong grade or not dressed course enough, take your pick. Fast dresses make the wheel act softer as less grains are in contact. (know in the grinding world as unit grain loading)
You don't want a nice flat wheel or a shiny polished finish as you run the risk of burning the chuck no matter how much water you pour in. Once this happens ... :bawling:

If it stops cutting you go back and redress. Otherwise it will sit forever and do nothing.
There is not much need for a ball screw in these machines as the weight of the head is normally so high. Not sure I like the idea of a counterbalance on this type machine. Perhaps designed "light" to the actual head weight load by a fair amount?

With a scale attached on a normal machine if you downfeed and the head does not come down.. stop.
Some float is normal on the acme nut but if it does not drop withing 2 minutes to where it should be you are no longer grinding. We run a 2 minute sparkout on parts and you can watch the spindle current drop off to idle.
On a chuck grind more like a 4-5 minute spark and you should be at idle current. I would never give anything more time than this.

I highly doubt you will see anything in the face of the wheel like loaded brass or such, under a microscope you will see "rounded" grains and too large grain tails if you are used to looking for such.

Mostly the wheel is just too damn hard and won't break down. More than one way to skin this cat.
Bob
(certainly I talk too much here, hope something is of some help, it's kind of artsy fartsy, a niffy build project and much respect for your shop and capabilities.)
 
Bob - agreed that hand sticking the wheel is not acceptable. The engineer who made this call and did this work is an military veteran and experienced machinist (and fellow PM'er) who has a very clear understanding of risk and as you said - you might get away with it if done infrequently, but I will have a chat with him.

Cash - that last photo is what we wound up with in the end.

Regarding the coolant - what I did figure out yesterday afternoon is that the pump was rotating the wrong direction and coolant flow was far less before rolling the motor leads.

The owner of the grinder is stopping by today and we will discuss wheels / chuck dressing with him as they have a half dozen of these grinders and are well familiar with the process.
 
Bob - agreed that hand sticking the wheel is not acceptable. ....

I did not say not acceptable when you "have" to do it and I certainly have done it many times.
Just not recommenced as part of a production process for your average Tom, Dick, or Harry.
As you know what you or a skilled person can do all day is different that designing machines for shall we say "the general public".
The problem with error-proofing and safety on machine/process design is that you are faced with well above average idiots out there.

No way any bad intended on your guy or your shop.
Posts and other info have shown me you are workplace safe oriented more than most I know in this businesses and that's a feather you should wear proudly in your cap.

Aside, does anyone to a profilometer check on a chuck finish, and does it matter since this gage is a very short stroke.
Would not poopy or high numbers be okay if it made a plane and not a bit of a cone or sometimes cup? (which these machines love to make)
Is shiny, looks nice and low numbers better? Is crosshatch always good?
I'm not so much about looking pretty anymore, seen this go bad on these style grinders.
I had one of my engineering co-op/students do his senior year thesis on this problem as he was "rocket scientist, straight A" smart and it seemed like a good challenge to throw at him.
Bob
 
Other thing you can do if you have a VFD on the spindle motor is drop the wheel speed, 5-10% drop is plenty to make a very noticable diffrence, it will make it softer and more inclined to break down. But you have to run wheels hard on grinders like this, you need the feed pressure to keep the wheel breaking down. Takes a lot of pressure to do that on a engamnet arc that large. Dressing is important, it should be at least as fast or faster than a rough facing cut on a lathe of a part going that speed, you must shear the grains, break some lose and have it rough, a slow dress and you get the exact results your describing. Think more like the spiral on a pipe flange not the fine luster of a records finish.

As to your feed question, thats not how you generally grind, normally with grinding you just grind and spark out, theres nothing to be gained by slowing feed as you approach finished size, that will just glaze the wheel stop it breaking down and not help you. its why most the old grinders had simple stops. Best thing you can probaly do is make it so you feed down at 100% motor load or have that adjustable (there’s no point having a 40hp spindle and only running it at 30 hp) a adjustable dwell time then a auto retract.

IMHO you badly need to get some one that knows how to grind in there to show you how the machine wants to be ran, from your comments your clearly not use to grinding and grinding is not the same as a typical machining cycle at all. Its kinda counter intuitive to the average machinist. You generally want to do the exact opposite to what you would do with a typical cutting tool when you encounter a issue and its a very touchy - feely - art-form like process.
 
On adama's and Bob's comments, one reason I would prefer a ballscrew in the downfeed is to maintain "absolute" control of the grinding pressure and not allow backlash to give the opportunity for hydroplaning on the coolant. I know it's worked just fine in the past with all the Blanchard-style grinders, but why not use technological advances where appropriate?

Great thread BTW, lots of insight from the experienced hands...
 
Because the simple reason is even the elastic loading in a ball screw is a bloody mile of distance when grinding. Grinding is not millling the forces and how things behave is different.

Bob does not mean hydroplaning on the coolant for were the wheel is like a car tyre on the road, but he means the complete wheel rim acting like a skirt on a hover craft and lifting its self out of contact. So long as theres not a pressure build up in that wheel well your fine, second its sealed well enough to build pressure you will lift the head and stop the rim grinding the work.

One of the things i don't see mentioned is dresser alignment, you alude to it being on its own hydrulic slides, what is this aligned too and are you dressing the wheels rim flat or concave - convex? If your dressing even slightly convex your makeing the skidding issue a nightmare, dress convex and you may do a lot better. Not ran a vertical spindle that big, but on a smaller grinder i have had good luck reducing the contact width of the wheel rim by simply only dressing a few thou part way across the face leaveing a rim to do more of the work load.

As to spindle head counterbalance, that could also be part of your issue, on a grinder pretty much every grinder that i have encountered with a head that travels down to meet the work there all head heavy so gravity removes the slop. Every grinder i have encountered has been this way right back to designs from the turn of the last century, i know your a cleaver guy, but 100+ years of proven approach probably needs thinking over. Gravity im assuming is still much the same this year as it was any other point in the last hundred years?
 
Very likely the parts grinding wheel was/is way too fine grit for chuck grinding.. Chuck grinding wheel should self dress with break down. For .005 or so a one dress should do if using the right wheel. Likely I would stick dress such a (to fine or hard) wheel.. Tram is important, we used to do that with setting a parallel on the chuck with a same jo block same away from center....For big mill cutter grinding we would set the table dead straight to grind the chuck..then drop table to be able to grind a dish(hollow)at the cutter face..I made a lever so I could bring a 200 pound cutter to 1" off chuck..then hand drop the last inch for a soft landing...yes it had to be dead center so the dish would be .0002 /.0003 or so close run-out and .001/.002 to size. Yes this for Arter and Heald grinders.. The Blanchard we mostly kept the chuck dead flat..not much other one can do with the chuck (except canting the wheel)..
 
First off I never have ran a Rotary Surface grinder but I use acoustic sensors on my Kellenberger everyday. I love them and would not buy another grinder without one. So are you just using the Acoustic Sensor to rapid to the part or parts and then switch to using the VFD and spindle load? Are you grinding more than one part at a time? My thought is lets say if you load the table up with 20 parts and use 75% load on the spindle, what would happen if the operator loaded only 5 parts onto the table? The machine will look for 75% load and it would plow down onto the parts at a rapid speed trying to get to 75% load. I have heard great stories for my Dad about throwing parts of the table. Maybe that would not happen I done know. Cash would know more about that than me. This is how I would like it. I would use the acoustics for just the gap and then set a feedrate down in a rough to finish and the a dwell.
 
First off I never have ran a Rotary Surface grinder but I use acoustic sensors on my Kellenberger everyday. I love them and would not buy another grinder without one. So are you just using the Acoustic Sensor to rapid to the part or parts and then switch to using the VFD and spindle load? Are you grinding more than one part at a time? My thought is lets say if you load the table up with 20 parts and use 75% load on the spindle, what would happen if the operator loaded only 5 parts onto the table? The machine will look for 75% load and it would plow down onto the parts at a rapid speed trying to get to 75% load. I have heard great stories for my Dad about throwing parts of the table. Maybe that would not happen I done know. Cash would know more about that than me. This is how I would like it. I would use the acoustics for just the gap and then set a feedrate down in a rough to finish and the a dwell.

1.) Wheel wear is known (positional accuracy +/- 0.0002") using the acoustical sensor (mounted to a ballscrew actuator with carbide rod making contact with the grind wheel)

2.) Chuck Position is known using a set of grind coupons and calibration routine we have written (again +/- 0.0002")

3.) Rough Part thickness is known

Knowing these three things we servo / rapid down to a programmable clearance distance above the parts (multiple parts being ground - Chuck Template ALWAYS fully populated with parts) . . . and then rough grind feedrate down while monitoring spindle torque.

When the spindle torque is registered (depending on mode of operation) . . . if in spindle load mode - the downfeed axis loads the spindle until you get to the defined torque target . . . otherwise if in feedrate mode, the spindle feeds down at a programmed rate UNLESS the spindle gets above the max allowed torque at which time the downfeed pauses. If the pause time exceeds a specific threshold, we go through an auto dressing cycle, remeasure the wheel with the acoustical sensor and get right back to grinding.

While grinding, the acoustical sensor is parked at some point (programmable . . . i.e. ~ 0.003 - 0.005") above the finish dimension. When the sensor detects the wheel at this location, the motion controller moves the wheel down some distance related to the (0.003" - 0.005") just mentioned and go to spark out and then retract the spindle and move the chuck to the unload position.

Meanwhile the other chuck has been reloaded with parts and grinding on the other chuck now commences while the parts just ground are removed and replaced with a new set of raw parts.

I anticipate that we will start grinding within the next day or two.


As far as the dressing point is concerned - it is mounted on a sealed guided linear bearing and moved into position with a hydraulic cylinder with a hard stop. It repeats within +/- 0.0001" and we are pretty happy with how it works. We are using a air/oil intensifier with pressure only on the cap end of the cylinder and spring return with the rod end of the cylinder piped outside the grinder to a breather.

We have VFDs on the Spindle and on the chucks and servos on the downfeed axis, touch-off acoustical sensor, and the Chuck carriage to shuttle the Chucks under the Grinding Wheel as needed.

This is the first grinder of this type we have done - so it has been a learning experience. We have done cutlery / bevel grinders and have found that after we are done with them that they are far more accurate / productive than what the OEMs provide with respect to performance, repeatability, and ease of use . . . we are hoping to get the same kind of improvements by plying our talents on this grinder.
 
If you are going to try and keep up with segment wear on a grinder like this you will drive yourself nuts.

Blanchard type grinders with segments work best when you have the segments graded right and you allow them to self dress and break down by themselves.

Dressers on a machine like this are to "clean" the stones when they get glazed over. Or to "rough" the stones before you put a finish on parts.

Getting the proper grade of segments for the task at hand is #1 priority.

Is this a Goeckel or Reform machine?
 








 
Back
Top