Search the Elsmar Cove!
**Search ALL of Elsmar.com** with DuckDuckGo Especially for content not in the forum
Such as files in the Cove "Members" Directory

Practical Data Collection Process Recommendations?

SPC_Newbie

Involved In Discussions
#1
Anyone have some ideas for efficient collection of SPC data – measuring and recording?

Let me provide some background: We are a precision machine shop. It’s basically a high mix production setup where a lot of parts (and scrap!) can be made in very short order. Staying on top of things without limiting production time of the machines is critical. We are committed to in-process inspection so we will make the necessary trade-offs but no need to give up something that a little forethought up front could minimize

We have a cell that can pull pallets of already set-up jobs and run them as time allows. Day shift typicall perform set-ups and swing and weekends run production. We often run just 1-2 parts of a larger qty job and then the next job comes in. Why I tell you this is we have run over 40 unique parts in a shift (4 machining centers in this cell). It also can provide variation as cutting tools will likely get used on other jobs before this job runs again – which could be in a week!

Each job also has a mobile cart on the production floor that contains raw material, a staging are for parts that are at different ops, a tote for good, finished parts, a place to hang a print and traveler and a place for the 1st article. There is also room to do inspection with hand tools. A large granite block and Height gage sit on one perimeter of the cell.We have 8 rows of these carts and, due to how the computer fills in time slots with partial runs, ANY of the set-up jobs could potentially run at any given time.

Needless to say our operators are kept quite busy. The operators have a limited amount of time that they can realistically spend on data collection.

We do use wireless tools at the one station. One possible way is to have tablet of some sort at each cart but that seems like a lot of overkill. I could also have a mobile station per operator but that would clog the aisles, though there may be a way we could make that idea viable. I could begin brainstorming all day/week …but I thought with your input I could narrow down a direction more quickly.

I realize I have left a lot of things open – I’ll try to provide more constraints as needed -but I didn’t want anyone to not propose something because they thought I couldn’t use it.

Thanks in advance!!
 
Last edited:

bobdoering

Stop X-bar/R Madness!!
Trusted
#2
Probably the first step is to see if the SPC you are doing is effective. Generally, you should be collecting data from each finishing tool on a job. Then, it depends on the characteristics you are measuring. For diameters and lengths, you should be doing X hi/lo-R charting. It is effective for both long and short runs. X-bar-R for such dimensions is a waste of time (if it is truly precision machining.) People like to rubber stamp X-bar-r charts because all of the big programs do them rather readily. But, for precision machining they are of little value.

Are you using a program - or paper charts?
 
S

Sturmkind

#3
Hi!

Another approach may be the use of Short Run SPC (Author: David Bothe) where the most critical Target dimensions of your skilled machinists are charted and the difference between the target and actual result are reported. This would assess trends and opportunities for the production system: operator skill, set-up targeting success, tool sharpness, and machine tool maintenance integrity.

I agree with Bob that applied SPC should be planned so that it yields real benefits and is not a mere paperwork excercise.
 

SPC_Newbie

Involved In Discussions
#4
Hi Bob - thanks for your time. Actually it was a google hit on your posts that first brought me to the forum and I initially joined just so I could download the examples/samples you provided!

I was recently hired primarily to implement an SPC program so there is a lot of work to do. What we are collecting now essentially ends up populating UMI charts (using a program).

My intentions are to implement X hi/lo-R charting as I fully buy in to your perspective. In the mean time we are still trying to put together a system that can integrate CMM data with hand tool data. What we have now has very rudimentary SPC charting. I ran X hi/lo - R charting past the developers at the software company we are using but they didn't want to commit any resources to implementing it. In fact they seemed to treat me with a bit of disdain for pushing back a couple of times!

Going off on a tangent here: part of why I mentioned that finishing tools will be used on other jobs before the original job is finished is because I am not sure how to factor this in (eventually) on the charts for X hi/lo-R as there will be wear apart from the tool and specific feature I'm concerned with

So, while I have a direction to go, and spend a lot of time trying to get software capability developed, I also need to look at practical implementation of the collection side of things. These things seem to require development in parallel :)

Cheers!
 
Last edited:

bobdoering

Stop X-bar/R Madness!!
Trusted
#5
Hi Bob - thanks for your time. Actually it was a google hit on your posts that first brought me to the forum and I initially joined just so I could download the examples/samples you provided!

I was recently hired primarily to implement an SPC program so there is a lot of work to do. What we are collecting now essentially ends up populating UMI charts (using a program).

My intentions are to implement X hi/lo-R charting as I fully buy in to your perspective. In the mean time we are still trying to put together a system that can integrate CMM data with hand tool data. What we have now has very rudimentary SPC charting.
I congratulate you, and if you need any assistance I am more than happy to help.

I ran X hi/lo - R charting past the developers at the software company we are using but they didn't want to commit any resources to implementing it. In fact they seemed to treat me with a bit of disdain for pushing back a couple of times!
I have run into the same issue. Jay Arthur of QI Macros has a template available, and Infinity QS has a technique that will duplicate this charting methodology. I have heard from some software companies that "precision machining is too much of a niche to service with a new chart." I don't see that...but suffice to say you are not alone in that battle.

I have implemented paper charts in a union environment that had X-bar R charting software - and they felt it made so much more sense that they had little complaint going to paper (rather than wasting time with wrong programs.) You may have to go through that transition for a period of time.

Going off on a tangent here: part of why I mentioned that finishing tools will be used on other jobs before the original job is finished is because I am not sure how to factor this in (eventually) on the charts for X hi/lo-R as there will be ware apart from the tool and specific feature I'm concerned with
I have read through Bothe's approach (as well as Steve Wise's books), and the assumption that the tools and machine will behave the same on various parts has to be met to have it be meaningful.

If you use the same finishing tool - but different roughing tool, that assumption is toast. Moreover, if the setup has either tool taking off more or less than the previous set-up, it will not work.

Besides, they hold some of the same assumptions of X-bar -R, I-MR and Xbar-s charts - none of which are entirely correct in precision machining.

But, with an X hi/lo-R chart as long as the high and low are between the control limits - you are good to go. You are better off setting up separate charts for each setup than trying to wrangle a shortcut to stuff a bunch of them on one.
 

SPC_Newbie

Involved In Discussions
#6
The replies and suggestions so far have reminded me I need to better nail down the kinds of data I'll be recording and how they'll be used.

Concerning the flow of parts off the machine and back to their dedicated carts, I'm still struggling with an efficient method of in-process inspection.

Having tablets at each cart seems to be too much plus co-ordinating wireless tools over so many stations probably isn't practical. Since posting earlier today I've considered a 2 place inspection table that will sit right outside the 2 load/unload stations. This way any part due to be checked will be by an inspection station anyway. For parts not being measured the operator will have to walk passed the inspection table to take the part back to the cart.

Maybe with (2) dual monitor setups we could have the inspection entry screen open on one and have a bubbled drawing showing the locations of the dims to be measured on the other. I'd like to have the entry screen come up upon the scanning of the part number. Right now that barcode is only on the traveler and the traveler resides at the dedicated cart so a little hurdle to jump there.

This approach 'may' meet my requirements but I started with the hopes of being extremely mobile/flexible, ideally being able to measure parts at thier dedicated stations and of course recording the data wirelessly. Implicit in the wireless data recording is the ability to verify you've recorded the correct dimension for the part you are measuring - probably via an available monitor??

We will be switching parts A LOT so having a way to change between parts seamlessly is needed as well.

Anyone else ever implemented or even tried something like this?
Thanks!
 

Bev D

Heretical Statistician
Staff member
Super Moderator
#7
if you are doing a lot of mixed part types automation is a great way to go. wirelessly or wired to a computer with the monitor right in front of the operators face is teh best scenario. I like paper charts too, but there are math and plotting skill issues adn teh paper chart takes longer to record. Teh downside of automated data collection is that you really need howmgrown software. there are several decent programs out there (Infinity QS is good..) but after doign this for 30 years I am firmly in the camp of homegrown software, form the data storage (collection from gage to computer is pretty standard and effective) to SPC to display to reaction plans and note taking. none of the commercially available programs come close to what I can do with homegrown software.
 
Top Bottom